The National Institute on Aging Macroeconomic-Demographic Model U.S. DEPARTMENT OF HEALTH AND HUMAN SERVICES Public Health Service National Institutes of Health National Institute on Aging NIH Publication No. 84-2492 June 1984 Preface With its broad mandate to study the biomedical, behavior- al, social, and economic aspects of aging, The National Institute on Aging (NIA) is interested in research on var- ious phenomena of population aging. Moreover, the Na- tion is engaged in a great debate over long-term problems of economic security for the elderly. This debate focused recently in the President's Commission on Pension Policy (1979-1981). The National Institute on Aging joined with the President's Commission in an Interagency Agreement to create a MacroeconomicDemographic Model suitable for study of the effects of population aging on the econo- my and the income of the elderly. The work presented here was initiated under contract to the President's Com- mission. At the expiration of the Commission, responsibil- ity for the model development was transferred to the National Institute on Aging. Joseph M. Anderson of ICF Incorporated and Edward A. Hudson of Dale W. Jorgen- son Associates constructed the model under NIA Contract Number 1-AG-12107. The NIA Macroeconomic-Demographic Model (MDM) represents an exciting new approach that exploits demo- graphic information to provide a picture of the phenom- ena of population aging. It combines a population model with a macroeconomic model and a disaggregated labor supply and demand model. The information developed in this core modeling system is used in five models of com- ponents of the retirement income system, including ma- jor Federal programs-social security, public and private employee pensions, Supplemental Security Income, and Medicare. A comprehensive picture of the interrelation- ship between the public and private sector programs can be developed in the context of long-run analysis. The MDM is now being employed in analyses for the Federal government. The model was first used to study recommendations and estimate costs and benefits of var- ious proposals for the President's Commission on Pen- sion Policy. The National Commission on Employment Policy has used the model to investigate the economic effects of 1982 legislation that changes the distribution of Medicare costs and the impact of the changes on employ- ment of older members of the work force. The National Institute on Aging is supporting research on aging and health expenditures and maintains a fully operational ver- sion of the model. Additional research on labor markets and pensions is supported by the Office of the Assistant Secretary for Planning and Evaluation, Department of Health and Human Services. The results of these research efforts will be incorporated into the NIA MDM and report- ed at a later date. Federal involvement with the Macroeconomic Growth Model of the MDM began in the late 1960s when the Federal Office of Emergency Preparedness and the De- iii HEP 2d Us A787 Pud/ partment of the Treasury supported the development by Dale W. Jorgenson and Edward A. Hudson at Harvard University of a small economic growth model for policy analysis. In the 1970s, this Macroeconomic Growth Model was integrated with a 9 sector Interindustry Model in order to study the interactions between energy and the economy. The Federal Preparedness Agency (now part of the Federal Emergency Management Agency) and the De- partment of Energy supported this important disaggrega- tion of the supply side of the economy. The next generation of this model was a sophisticated 36 industrial sector model, called the Dynamic General Equilibreum Model (DGEM). While this research proceeded, Joseph M. Anderson added a population model to the basic Macroe- conomic Growth Model and disaggregated the labor mar- ket in order to study the effects of population aging on the economy. This work led directly to development of the MDM under his direction at ICF Incorporated. This devel- opment entailed considerable research on social security and private and public employee pension models. The combination of a macroeconomic model, a population model, and a labor market model provided the required framework for the implementation of the pension and Federal program models to study the retirement income system, This monograph describes the initial version of the MacroeconomicDemographic Model as of March 1982, when it was completed. In applications of the model for Federal government agencies and other organizations since that time, additional developmental work has been done and some modifications have been made. Develop- mental work is continuing under the guidance of NIA. However, the current version of the NIA Macroeconomic- Demographic Model is essentially the same as that de- scribed in this monograph. Joseph M. Anderson, the project director, authored this monograph. The author acknowledges the extensive and invaluable contributions of Dale W. Jorgenson and Ed- ward A. Hudson, who developed the original Macroecon- omic Growth Model and provided crucial guidance and assistance in the integration of the modeling system. The Macroeconomic Growth Model development, in turn, was made possible by the creation of a macroeconomic growth accounting system by Jorgenson and Laurits Chris- tensen. Construction of the Labor Market Model was made possible by the development of a comprehensive labor force data base by Jorgenson and by Frank Gollop and Peter Chenloy. William McNaught and Mark Minasi devel- oped several components of the retirement income mod- eling system, performed the vital tasks of system integration, and drafted early versions of parts of this monograph. Anne Brizendine, David Kennell and John Sheils of ICF also made valuable contributions. Signe We- trogan of the U.S. Bureau of the Census provided valuable data for the population model. William S. Cartwright, served as the project officer at the National Institute on Aging and as editor of this monograph. Bonita Bailey and Elaine Minor of ICF cheerfully and conscientiously typed the many drafts of this monograph and the voluminous reports which proceeded it. The views expressed in this monograph are those of the author, Joseph M. Anderson, and should not be ascribed iv to any branch of the Federal government or to any of those who were consulted or who commented on the manuscript. William S. Cartwright, Ph.D. Chief, Demography and Economics Office Epidemiology, Demography, and Biometry Program National Institute on Aging, NIH Contents Preface iti Chapter 1 Introduction 1 Important Features of the Model 1 Components of the Model 2 Overview of the Report 4 2 Population Model 7 3 Macroeconomic Growth Model 9 Introduction 9 Description of the Model 9 Macroeconomic Accounting System 10 New Features of the Model 11 4 Labor Market Model 13 Introduction 13 The Supply Sector 13 The Demand Sector 19 The Solution Algorithm 20 5 Social Security Model 21 Iitroduction 21 Overview 21 Contributions (Taxes) 22 Benefits 24 Future Work 28 6 Private Pension Model 31 Introduction 31 Overview 31 Coverage, Participation, and Vesting 32 Contributions 34 Estimation of Recipient Population 35 Benefits 36 Pension Plan Asset Calculation 37 Future Work 37 7 Public Employee Pension Model 41 Introduction 41 Overview 41 Estimation of the Number of Public Sector Employees 42 Coverage, Participation, and Vesting 42 Contributions 43 Benefit Recipients 45 Contents (Continued) Benefits Pension Plan Asset Calculation Future Work 8 Supplemental Security Income Model Introduction Overview of the Modeling Approach Benefits to the Aged Benefits to the Blind and Disabled 9 Medicare Model Introduction Overview Structure of the Medicare Model An Alternative Approdch 10 Base Case Simulation Introduction Population Estimates Macroeconomic Growth Estimates Labor Market Estimates Social Security Estimates Private Pension Estimates Public Employee Pension Estimates Supplemental Security Income Estimates Medicare Estimates Summary Appendices Equations of the Macroeconomic Growth Model Equations of the Labor Market Model Modeling the Substitution Between Age Groups Equations of the Social Security Model Equations of the Private Pension Model Equations of the Public Employee Pension Model Modeling Public Employment Growth Report Generating Capabilities of the Macroeconomic- Demographic Model Summary Tables for the Base Case Simulation TOTEmO OW — Bibliography vi 45 46 46 49 49 50 50 52 55 55 56 56 57 59 59 59 62 66 70 75 78 82 83 86 87 95 98 103 107 110 113 115 125 145 List of Tables Tables 4-1 4-2 4-3 4-4 4-5 5-1 5-2 5-4 5-5 5-6 5-7 6-1 6-2 6-3 8-2 8-3 8-4 8-5 8-6 8-7 8-8 9-1 Title Estimated Coefficients of the Unemployment and Participation Equations, Young and Middle-Aged Groups Age 16-54 Estimated Coefficients of the Unemployment and Participation Equations, Elderly Groups Age 55 and Older ' Coefficients of Total Hours Equations Limits on Participation Rates Used Within the Labor Market Model Estimates of the Elasticity of Input Price with Respect to Input Quantity Used in the Labor Market Model ‘Compensation of Employees: Wages and Salaries, and Fringe Benefits, 1970-1979 Estimated Factors to Convert Covered Earnings to Taxable Earnings 1982 Scheduled Combined Employer-Employee OASDI Tax Rates, 1970-2055 Social Security Retirement and Disability Incidence Rates by Age and Sex Disability Termination Rates by Age and Sex Taxable Wage Base and Average Wages Used in the OASDI Calculations, 1951-1979 Secondary Benefit Levels as Percent of Primary Benefit Type Private Pension Coverage Rates by Age and Sex Distribution of Workers Covered by a Private Pension Plan, by Plan Type, 1975 Proportion of Workers Covered by a Private Pension Plan Who are Participating in the Plan, by Age and Sex Proportion of Workers Participating in a Private Pension Plan Who are Vested by Age and Sex Vesting Rates for Public Employees Who Are Participating in a Pension Plan by Age, Sex, and Sector Pension Coverage Rates for State and Local Government Employees, by Age and Sex Proportions of Employees of State and Local Governments Covered by a Pension Plan Who Are Participating in the Pension Plan, by Age and Sex Number of Persons Receiving Federally Administered SSI Payments and Total Amount of Payments, 1974-1979 Recipients of Federally Administered SSI Aged Benefits, by Age, 1975-1979 Percent of Population Receiving SSI Benefits for the Aged, by Age Group, 1975-1979 Average Monthly SSI Benefit Amount for Persons Receiving Federally Administered Payments, by Reason for Eligibility, 1977-1979 Recipients of Federally Administered SSI Disability Benefits, by Age, 1975-1979 Recipients of Federally Administered SSI Benefits for the Blind, by Age, 1975-1979 Percent of Population Receiving SSI Disability Benefits, by Age, 1975-1979 Percent of Population Receiving SSI Disability Benefits for the Blind, by Age, 1975-1979 Personal Health Expenditures by Age Group, 1978 vii 16 17 18 19 20 23 23 24 25 25 27 28 33 33 33 34 44 44 44 49 51 51 51 52 53 53 54 55 List of Tables (Continued) Tables Title 9-2 Annual Personal Health Care Expenses For Population Over Age 65, 1978 9-3 Medicare Costs Per Capita by Age, Sex and Service, 1977 9-4 Trends in Real Per Capita Health Expenditures, 1965-1979 10-1 Projected U.S. Population by Age Group, 1980-2055 10-2 Population Distribution and Dependency Ratios, 1980-2055 10-3 Comparisons of Projected Populations in 2040—Macroeconomic-Demographic Model, Census Bureau, and Social Security Actuary 10-4 Projected Values of Principal Macroeconomic Growth Variables, 1980-2055 10-5 Comparison of Selected Annual Growth Rates in Real GNP Estimated in the Macroeconomic-Demographic Model and Assumed by the Social Security Actuary, 1981-2055 10-6 Projected Values of Macroeconomic Outputs and Inputs, 1980-2055 10-7 Comparison of Simulated and Actual Values of Principal Macroeconomic Variables, 1970-1979 10-8 Projected Values of Aggregate Labor Market Variables, 1980-2055 10-9 Projected Employment by Age and Sex, 1980-2055 10-10 Projected Hourly Compensation Rates by Age, 1980-2055 10-11 Ratio of Hourly Compensation Rates of Selected Age Groups to the Compensation Rate of Prime Age (35-44) Workers, 1980-2055 10-12 Projected Values of Average Annual Compensation Per Worker by Age Group, 1980-2055 10-13 Comparison of MDM Estimated Growth Rates in Real Compensation with Growth Rates in Real Wages Assumed by the Social Security Actuary, 1981-2000 10-14 Comparison of MDM Estimated and Actual Labor Force Participation Rates for Selected Groups, 1970-1979 10-15 Comparison of MDM Estimates and Actual Civilian Employment Levels by Age and Sex, 1970-1979 10-16 1983 Projected Numbers of Beneficiaries and the Levels of Benefit Payments by the OASI and DI Systems, 1980-2055 10-17 1982 Projected Tax Revenues and Expenditures of the OASI and Combined OASDI Systems, Selected Years, 1990-2055 10-18 1982 Projected Cost Rates and Tax Rates of the OASI and Combined OASDI Systems, Selected Years, 1980-2055 10-19 Comparisons of OASI Cost Rates Projected by the Macroeconomic-Demographic Model and the Social Security Actuary, 1981-2055 10-20 Comparisons of Simulated and Actual OASI Beneficiary Populations by Benefit Type, 1970-1977 10-21 Projected Numbers of Active Gross Participants in the Private Pension System by Age and Sex, 1980-2055 10-22 Projected Beneficiaries and Benefit Payments in Private Pension Plans by Type of Plan, 1980-2055 viii 56 57 58 60 60 61 63 64 64 65 67 68 68 68 69 69 70 71 72 73 74 74 76 76 List of Tables (Continued) Tables Title 10-23 Projected Private Pension Fund Balances by Type of Plan, 1980-2055 10-24 Comparisons of Simulated and Actual Participants in and Contributions to the Private Pension System, 1970-1979 10-25 Comparisons of Simulated and Actual Beneficiaries and Benefit Payments From the Private Pension System, 1970-1977 10-26 Projected Beneficiaries and Total Benefit Payments in Public Pension Plans by Type of Plan, 1980-2055 10-27 Projected Average Benefits Paid by Public Employee Pension Plans by Type of Plan, 1980-2055 10-28 Projected Public Pension Fund Balances by Type of Plan, 1980-2055 10-29 Projected Beneficiaries and Benefit Payments of the Supplemental Security Income System, 1980-2055 10-30 Comparisons of Simulated and Actual Beneficiaries of Supplemental Security Income 1974-1979 10-31 Projections of Future Medicare Expenditures for Selected Programs, 1980-2050 10-32 Total Benefit Payments From the Retirement Income System by Source, 1980-2050 10-33 Index of Average Benefit Payments From Various Components of the Retirement Income System, 1980-2050 10-34 Total Income Received by the Elderly As Percent of the GNP, 1980-2050 77 77 79 79 81 81 82 83 84 84 85 85 List of Figures Figures 1-1 10-1 10-2 10-3 10-4 10-5 10-6 10-7 10-8 10-9 10-10 Title Structure of the Macroeconomic-Demographic Model Trend in U.S. Population by Age, 1970-2050 Trends in the Dependency Ratio, 1970-2050 Trends in Major Macro Variables, 1970-2050 Average Hourly Compensation Rates, 1970-2050 Trends in SOC SEC Beneficiaries by Type, 1970-2050 Trends in OASI Taxes and Payments, 1970-2050 Comparison of OASI Cost Rates, 1980-2050 Trends in Private Fund Balances by Type of Plan, 1970-2050 Trends in Pension Fund Balances by Type of Plan, 1970-2050 Trends in SSI Benefit Payments by Benefit Type, 1980-2050 61 62 62 67 71 73 74 78 80 83 Chapter 1 Introduction emographic trends indicate that population aging will be a significant factor in the future evolution of the U.S. econ- omy. The changing age structure of the population will affect the income level of the elderly population as well as productivity, consumption, saving, and investment. This monograph describes a long-term Macroeconom- {pte Model of the U.S. economy which was eveloped for the National Institute on Aging (NIA) to investigate these critical areas. The model concentrates especially on the retirement income system and illumi- nates problems related to meeting private and public commitments to the elderly during a period of demogra- phic and economic change. The model particularly takes into account the productive capacity of the economy, which in the end must meet the needs of the population for current consumption, for investment to increase pro- ductivity and to maintain a high level of employment, and for the income support of the retired population. Important Features of the Model Several features characterize the analytical approach of the NIA Macroeconomic-Demographic Model. These fea- tures were established in the initial design of the model and provided the analytical structure that guided its devel- opment. They are the source of many of the model's strengths and provide its limitations. Key features of the model design included: e The use of a large amount of demographic information, e Explicit represenation of the process of long-term economic growth, e The use of a general equilibrium framework, ® Representation of the structural features of the major pension systems, e A comprehensive, integrated approach. The core of the model is its representation of the proc- | ess of demographic change and economic growth. From “the beginning, a major objective of the model was to take advantage of the tremendous amount of information about the potential future size of the U.S. population, by sex and single year of age, that can be produced by a relatively simple population model, such as the one de- scribed in Chapter Two, or that is provided by the U.S. Bureau of the Census or the Social Security Administra- tion Office of the Actuary. Because of the long-term nature of the process of demographic change, it is possible to project the size and composition of the working age and retired population in some detail with a reasonable de- gree of confidence two to three decades into the future. Beyond that, the influence of future fertility rates that cannot be predicted with confidence become important. However, for any well-specified path for future fertility, the U.S. population can be projected in detail far into the future. The size and structure of the future population is a major determinant of the future level and structure of economic activity and, especially, of the future condition and behavior of the components of the retirement income system. The NIA Macroeconomic-Demographic Model (MDM) was designed to make full use of this large body of valuable information. That objective led to the develop- ment of a Labor Market Model that disaggregates the labor force into twenty-two age-sex groups and analyzes the behavior of each one. The behavior of the household sector in the Macroeconomic Growth Model of the MDM is keyed to the size of different age groups. The pension system models follow on an annual basis each cohort as it passes through the age-sex categories of the labor force and then retires. The model thus represents the life cycle economic behavior of each birth cohort, by sex, in terms of its labor market participation, employment, hours worked, earnings, and retirement income. Information about various cohorts at a single point in time and the experience of a single cohort through time is used and reported. In this way the broad economic aspects of the aging process, for cohorts born in different periods, can be investigated and compared. The other major determinant of the future condition of the retirement income system is the future state of the economy—wealth, income, employment, and savings. The model explicitly focuses on the determinants of long- term growth—capital and labor input and technological change. As described in Chapter Three, it is based on a unique economic growth accounting system developed by Christensen and Jorgenson, and a neoclassical model of the process of long-term economic growth developed by Hudson and Jorgenson. The growth model focuses on the role of factor inputs in determining the levels of aggre- gate output and on household preferences in determin- ing both the levels of factor inputs and the composition of output. It is through these basic determinants of the level and structure of economic growth that demographic change and the pension system may affect the future econ- omy, which will, in turn, affect the pension system and the well-being of the aged population. The process of economic growth, the input levels of capital and of labor of different ages and their prices, and the quantities and prices of output are estimated in a THE NATIONAL INSTITUTE ON AGING MACROECONOMIC-DEMOGRAPHIC MODEL general equilibrium framework. The many interactions and feedbacks from one set of economic processes to another are taken into account, and a set of equilibrium values for all variables, i.e., a solution to the model, is found. This approach captures several important charac- teristics of an economy. First, feedback effects and second and third order interactions, sometimes unexpected, are captured. Second, the system tends to adjust to any exter- nal change to achieve as closely as possible the path deter- mined by the underlying preferences of consumers and objectives of producers. The model applies the large amount of information provided by the core population, macroeconomic growth, and labor market models to a detailed and com- prehensive set of models of the components of the U.S. pension system and two important transfer systems. The pension system models attempt to represent at a highly aggregate level the actual institutional structures of the social security system and private and public employee pension systems. The specific provisions of the Social Security Act determine tax payments and benefit calcula- tions in the Social Security Model. Actuarial calculations based on prototypical benefit formulae determine contri- butions and benefit levels in the Private and Public Em- ployee Pension Models. The Supplemental Security Income Model and the Medicare Model detail two Federal government programs that are important to the elderly population. This detail provides several strengths. First, it permits model simulations to be compared with historic data de- scribing the operation of each system, with which practi- tioners are familiar. Second, it permits specific policy and statutory changes to be represented. For example, alterna- tive ways of indexing social security benefits can be simu- lated. Therefore, specific aspects of the overall pension system and specific policy changes can be studied. Third, the structure of the models of individual systems can be examined and evaluated by experts familiar with those specific institutions. The modular design of the overall model permits easy modification or replacement of any of the specific components. The Macroeconomic-Demographic Model significantly extends our capabilities for research concerning econom- ic-demographic interactions and the retirement income system. Its population and labor market detail, depiction of long-term aggregate economic activity, and representa- tion of three pension and two transfer systems provides a comprehensive look at a wide range of income security issues within a single model. While there are many areas for potential improvement in the model structure, it cur- rently provides a useful tool for both pension policy re- search and research in other policy areas requiring long- term analysis. The theoretical framework of the model is provided by the neoclassical theory of economic growth. This frame- work focuses on the determinants of long-run productiv- ity and economic growth. It analyzes the determination of investment, consumption, and output; aggregate relative factor shares (labor and capital); substitution between fac- tors; and productivity change. Capital is modeled as a homogeneous, aggregate factor that depreciates and is replaced and accumulated through investment. Prices and quantities of outputs and factor inputs are determined through the interaction of supply and demand in competi- tive markets. This theory predicts that the lower the rate of interest, other things being equal, the greater the capital intensity of production and the greater the net national product per worker. Thus, policies which change savings and the interest rate have direct effects on the national product per worker. Also, policies which affect supplies of labor and capital have direct effects on economic growth. Use of a long-term model is highly appropriate for analysis of the interaction of the retirement income sys- tem and the economy. Social security and other pension systems represent long-term commitments, and the level of their benefits depends fundamentally on the produc- tive performance of the Nation's economy. Short-run, Keynesian type models are less appropriate because they focus on the determinants of aggregate demand, given a fixed capital stock, rather than the long-run determinants of the Nation's income and wealth. Components of the Model The Macroeconomic-Demographic Model is composed of a core macroeconomic and demographic modeling sys- tem, and a set of five peripheral models that depict the operation and behavior of the major components of the retirement income system. The core model has three ma- jor parts: a population projection system, a macroecono- mic growth model, and a labor market model. The five major elements of the retirement income system that are modeled are the Old-Age, Survivors and Disability Insur- ance System (OASDI), the private pension system, the public employee retirement system, the Supplement Se- curity Income (SSI) system, and the Medicare system. The Population Model replicates the U.S. Census Bu- reau population projection methodology. It projects the total U.S. population by age and sex for each year from 1970 through 2055. Fertility rates, mortality rates, and im- migration are determined exogenously. The user speci- fies an ultimate completed cohort fertility rate. An appropriate set of age-specific fertility rates is then calcu- lated and the corresponding population is projected. Mor- tality and immigration can also be varied exogenously by the user. The Macroeconomic Growth Model is an adaptation of the Hudson-Jorgenson four sector long-term econometric forecasting model. It depicts the formulation of working, spending and savings plans by households, and produc- tion, investment, and employment plans by businesses. It projects the demand for and supply of goods and services and depicts the equilibration of demand and supply by price adjustments and changes in consumption and pro- duction decisions. This long-term economic growth mod- CHAPTER 1 el is characterized by a more careful depiction of the determinants of supply than most other econometric fore- casting models, which focus on the determinants of aggre- gate demand. The demographically disaggregated Labor Market Mod- el depicts three basic aspects of the labor market: the demand for labor; the supply of labor; and the simulta- neous determination of labor and capital services input along with compensation, output, and employment. The derived demand for labor inputs is investigated by model- ing the aggregate production technology of the private U.S. economy, focusing on the substitutability among age groups in the production process. Labor supply is mea- sured in total annual manhours worked for each of twen- ty-two age-sex groups. Total manhours worked by each group is the product of the group’s population, labor force participation, employment rate, and annual hours worked per year. The labor supply-demand block is fully integrated into and solved simultaneously with the Ma- croeconomic Growth Model for the input levels and prices of capital services and of labor, for the unemploy- ment and participation rates of each age-sex group, for the level of output, consumption and investment, and for oth- er economic variables. The three pension system models and two Federal transfer models are integrated with the core Macroecono- mic-Labor Market Model. In many instances, these models use fixed actuarial assumptions rather than a system of behavioral equations. In discussing each of these models we mention plans for improvements to be made as model development continues. The Social Security Model depicts the determination of contributions into and benefit payments from the retire- ment (OASI) and disability insurance (DI) systems. Annu- al contributions are derived from the estimates of total compensation by age and sex generated by the Labor Market Model by estimating covered earnings and the taxable earnings base and applying statutory and project- ed tax rates. Total annual benefit payments are calculated by estimating the average benefit level and number of beneficiaries for each of fourteen benefit categories. A primary insurance amount is estimated for individuals classified by year of birth, sex, and initial year of eligibility by applying the statutory provisions for the calculation of average indexed monthly earnings (AIME) to the hypo- thetical earnings records of typical individuals in each age-sex cohort and using the statutory benefit formula. Average payments for the fourteen types of benefits are keyed to the estimated primary insurance amounts. The model then projects balances for each of the trust funds each year from 1970 through 2055 and can be used to estimate the level of tax collections that would be re- quired to finance projected benefits and the implications of alternative social security policies. : The Private Pension Model depicts the aggregate be- havior of three types of plans—defined benefit plans, de- fined contribution plans, and individual plans (IRAs, Keoghs, TSAs, etc.). Private pension coverage, participa- Introduction tion, and vesting rates, by age and sex, were estimated using the Pension Supplement to the May 1979 Current Population Survey (CPS). Contributions to each type of plan are estimated by applying appropriate contribution rates to the earnings of each age-sex group. The model applies age-sex-specific retirement benefit acceptance rates to estimate the population of beneficiaries and calcu- lates average pension benefits by applying prototypical pension benefit formulae to the estimated earnings re- cords (for defined benefit plans) or contributions (for defined contribution and individual plans) of the individ- uals of each age-sex cohort. Pension fund assets for each type of plan are derived from total contributions and benefit payments and the rate of return projected by the Macroeconomic Growth Model. The Public Employee Pension Model specifies that all public employees are in defined benefit plans (in fact, 98 percent of public employee participants are so covered). Public employment is divided into seven sectors: Federal Civil Service, military officers, military enlistees, state and local hazardous duty, state and local general administra- tive, state educators, and local educators. The seven sec- tors are distinguished because the characteristics of the work forces and pension plans differ significantly among these groups. Given these distinctions, coverage, partici- pation, and vesting rates and benefit acceptance probabil- ities are estimated using the same techniques as used by the Private Pension Model, drawing on data from the actu- aries of the Federal Civil Service and the Department of Defense as well as the May 1979 Current Population Sur- vey. Contributions and benefit payments are calculated separately for each sector of employment. The Supplemental Security Income (SSI) Model depicts separately the operation of the programs for the blind, for the disabled, and for the aged. The eligible aged popula- tion is projected by applying an income distribution mod- el to the population and earnings levels forecast by the Population and Labor Market Models to estimate the earn- ings, social security benefits, and other income of the elderly population. Age-specific SSI participation rates cal- culated from Social Security Administration (SSA) data are applied to the eligible population to estimate the number of beneficiaries. Average benefits are projected using data on average Federal benefit payments and average state supplements. Blind and disabled beneficiaries are esti- mated by age and sex applying historic incidence rates to the projected population. The Medicare Model forecasts revenues similarly to the Social Security Model. Expenditures are forecast by disag- gregating total beneficiaries and expenditures by age, sex and category of service. There are up to 26 age-sex groups for each of six services (inpatient hospital care, home health care, physician services, etc.). For each age-sex- service category expenditures are projected by applying estimated ratios of recipients per capita and expenditures per recipient to the projected population of the age-sex group. Figure 1-1 depicts the operational linkages between THE NATIONAL INSTITUTE ON AGING MACROECONOMIC-DEMOGRAPHIC MODEL Figure 1-1 Structure Of The Macroeconomic-Demographic Model Population ER Macroeconomic Labor Growth Market y y Y Y Y Social Private Public Supplemental Medicare Security Pension Employee Security Pension Income these models within the Macroeconomic-Demographic Model. At the start of any simulation year, the Population Model initially forecasts the new size and composition of the population. These population figures are principal inputs into the Macroeconomic Growth Model and the Labor Market Model, which operate simultaneously to project levels of aggregate economic activity and the labor market outcomes for twenty-two different age-sex groups. These projections of the state of the economy and the disaggregated labor market are inputs into the simulation of each of the three pension system Models and the two transfer income models currently included in the MDM. Successful simulation requires ensuring the consisten- cy of the results across each of the models within a single year and across years. For the most part, the version of the Macroeconomic-Demographic Model reported here op- erates sequentially. In a simulation year, each model is solved in succession until all have been completed. At this point, the next year’s simulation starts and the solution process begins again. The only major exception to this - sequential structure is the simultaneous solution of the Macroeconomic Growth Model and the Labor Market Model. Because of this structure, each of the pension system models can be solved independently from the other models, and the operation of the entire Macroeco- nomic-Demographic Model is enormously simplified. The Macroeconomic Growth Model and the Labor Mar- ket Model are the only models in the Macroeconomic- Demographic Model which actually simulate market processes. Each has one or more demand and supply relationships that jointly determine an equilibrium set of market outcomes. In the Macroeconomic Growth Model, Newton's method for solving a set of simultaneous equa- tions is used to find the market equilibrium. In the Labor Market Model, the Gauss-Seidel method is used to solve the equations of the model. The major macro models— Population, Macroeconomic Growth, and Labor Market— employ many lagged variables in their equations, ensur- ing that one year’s results play an important part in deter- mining the next year’s results. Consistency is maintained between models because the outputs of the core macroeconomic modeling system, described in. Chapters 2 through 4, serve as inputs to the pension models described in Chapters 5 through 9. In other words, the entire model operates from a consistent set of accounting relationships. All of the economic varia- bles—compensation, employment, GNP, etc.—are de- fined identically throughout the model. The model operates in essentially a “one-way” mode. Models called earlier in the solution sequence for any year affect the solutions of all later models, but no reverse linkage exists. In this version, the results of the pension models do not affect aggregate economic behavior or labor market decisions. The only exception to this “one- way” rule in this version is the simultaneous solution of the Macroeconomic Growth and Labor Market Models. A high priority for continued development of the model is the incorporation of the important linkages between pen- sion systems and the macroeconomy—the influence which social security and employer pensions may exert on labor supply and savings rates and the role pension assets play in determining interest rates and capital alloca- tion. In the base case simulation discussed in Chapter 10, we attempt to account for some of these macroeconomic effects of pension policies. Work is now underway which incorporates these feedbacks from the pension system to the macroeconomy directly. Overview of the Report Chapters 2 through 4 discuss the Population Model, the Macroeconomic Growth Model and the Labor Market CHAPTER 1 Model. Chapters 5 through 9 discuss the five retirement and transfer income models. Chapter 10 presents the base case simulation and validation of the model from 1970 to 1979. The discussion in each of these chapters is non-techni- cal. The objective is to give the reader a general under- standing of each model. Readers interested in more detail should refer to the appendices. These present the equa- tions of each of the five principal models within the Ma- Introduction croeconomic-Demographic Model—the Macroeconomic Growth Model, the Labor Market Model, the Social Securi- ty Model, the Private Pension Model, and the Public Em- ployee Pension Model. These appendices also discuss new analyses completed in the course of model develop- ment on the substitutability of different age groups of labor in the production process and on projection tech- niques for public employment levels. Chapter 2 Population Model ne of the key design objectives of the Macroeconomic- Demographic Model is to investigate the impact which future demographic change will have on both macro- economic and pension system performance. The Popula- tion Model projects the future size and demographic composition of the population. The MDM Population Model uses the basic methodology as well as projections of demographic parameters that were developed by the U.S. Bureau of the Census. The Population Model starts with a base year population—an estimate of the size of the U.S. population by single year of age, race and sex— provided by the Census Bureau.! The model then projects the size of each population subgroup in each future year by applying age-sex-race-specific survival rates and immi- gration levels. The number of births is projected by apply- ing age-race-specific fertility rates to the estimated population of females ages 14 through 49. The estimates of fertility rates, survival rates, and immigration levels used in the current version of the model are those devel- oped by the Census Bureau.’ For each sex and race, the population each year is esti- mated as follows: t+1 t t t N =NXs +M i+1 i iii (2-1) where N. is the population age i at the midpoint of year t; t . . sis the proportion of the population of age i that survives to age i+ 1 between the midpoint of year t , and the midpoint of year t+ 1; t rosortloinds} : M is the net immigration level of age i between the I midpoint of year t and the midpoint of year t+1 Next year’s population is simply the fraction of this year’s population which survives to be a year older plus any net immigration. The Population Model uses constant net im- migration levels.? Equation 2-1 estimates the population of all persons one year of age or older. Births are estimated by applying a set of age-race-specific fertility rates to the female popu- lation of ages 14 through 49. 49 N+! = > i=14 ft X F! X sd + M, (2-2) t+1 where N ° is the population less than a year old in year t+ 1, to - . . f, isthe fertility rate in year t of women of age i, pb . on F, is the number of women of age i at the midpoint of year t; to . . Sinics s, is the proportion of infants born between the midpoint of year t and the midpoint of year t+ 1 surviving to the midpoint of year t+ 1; t Mis the number of newborn immigrants between the mid- point of year t and the midpoint of year t+1. Note that survival rates are applied to the infants born in year t and that net immigration of infants occurs. The Population Model includes three different sets of projected fertility rates—the Census Bureau's Series I, Se- ries II, and Series IIL. In the version of the Population Model used for the estimates reported in this monograph, Series I rates converge to an ultimate completed cohort fertility rate of 2.7 births per women by 2015, Series II to 2.1, and Series III to 1.7. The model can produce projec- tions of population based on any set of specified fertility rates. In addition, weighted averages of two series of fertil- ity rates can be used to produce projections that approxi- mately correspond to fertility series intermediate to those specified. For example, a population approximately cor- responding to a completed: cohort fertility rate of 1.9 would be produced by averaging the rates corresponding to 1.7 and 2.1. The Population Model does not forecast the future pop- ulation. It projects the population according to a specified set of fertility rates, mortality rates, and immigration. Be- cause the current population is known, and because sur- vival rates do not change greatly in the short-run and are already high for most ages, so that even large relative changes in mortality rates produce small relative changes in survival rates, the future working-age population, at least through the year 2000, can be projected with some confidence. All the estimates of the Macroeconomic-De- mographic Model are conditional on a set of specified 'Any base year can be used. Accuracy of future projections is increased by using the most recent estimates. The model currently uses the 1980 Census population. In this monograph we compare MDM estimates of several future variables with estimates developed by the Social Security Administration (SSA). These SSA estimates used the 1979 estimated population as a base. Consequently, for the projections reported in this monograph we used the 1979 estimated population as a base. For the simulations for 1970-1978, Census Bureau estimates of the actual popu- lation are used. “For the estimates presented in this monograph, the Census Bureau projections of fertility rates, survival rates, and immigration are those developed for the population projections reported in “Projections of the Population, 1977-2050", Current Population Survey, Series P-25, Report Number 704, July 1977. Census Bureau projections developed in 1983 and based on the 1980 Census are used in the 1983 version of the model. 3In the current version of the model, the user can select the initial year immigration level and can specify a trend rate of change of immigration. demographic parameters that are not, themselves, esti- mates or forecasts. Inasmuch as those specified param- eters will differ from the actual future values, errors are introduced into all the MDM variables. Due to the nonsta- tistical nature of the Population Model's parameters, specification of the exact size of this forecast error would be quite difficult. Chapter 3 Macroeconomic Growth Model Introduction he Macroeconomic Growth Model provides the core of the Macroeconomic-Demographic Model because it pro- jects the long-term trends of the United States economy. The Macroeconomic Growth Model is fully integrated with the Labor Market Model, so that forecasts of all major macroeconomic variables-GNP, investment, savings, etc—are completely consistent with all labor market var- iables—wages, employment, hours worked, etc.! These aggregate results provide much of the information re- quired for simulating each of the retirement income sys- tems included in the Macroeconomic-Demographic Model. The Macroeconomic Growth Model depicts the behav- ior of the principal components of long-term economic growth. This design leads to several differences between this and conventional macroeconomic forecasting models which focus on the short-term determinants of aggregate demand. First, both inputs to and outputs from produc- tion are represented—capital and labor inputs and con- sumption and investment outputs are explicitly considered. The accounting system in the model encom- passes consistently the prices and quantities of all four of these components, and each of these types of goods and services is considered in terms of its effects on each of the four economic sectors: Second, the definitions of these four types of goods and services are consistent with their role in economic growth. Capital services are the effective flow of produc- tive services obtainable, in the specified year, from the beginning-of-year capital stock. Labor services are the ef- fective flow of productive services obtainable from the hours of labor time supplied by the household sector. Consumption is the flow of current goods and services consumed in the specified year. Unlike the National In- come and Product Accounts, consumption excludes pur- chases of consumer durables and includes the value of services from the existing stock of housing and other consumer durables. Investment is the output of goods devoted to increasing the capital stock of the economy. It includes the purchases of consumer durables. Third, the supply side of the economy—the growth of The combined Macroeconomic Growth and Labor Market Model forms one, fully integrated model, all components of which are solved simul- taneously. The two major submodels of the integrated model are dis- cussed in two separate chapters of this report to permit careful review of the elements of each. The Macroeconomic Growth Model was devel- oped at Dale W. Jorgenson Associates under the direction of Edward Hudson and Dale Jorgenson. The Labor Market Model was originally developed at Harvard University under the direction of Joseph Anderson. productive capacity through time—is explicitly modeled. In each year, productive capacity of the private domestic sector is defined via a price possibility frontier reflecting the state of technology and the constraint that capital and labor inputs cannot exceed available supplies. Over time, capacity expands through growth in the input of capital services, arising from the accumulation of capital stock; through change in the input of labor services, arising from demographic change and change in labor-leisure choices; and through technical change, which is included in dis- embodied, input-embodied, and output-specific forms. The Macroeconomic Growth Model is adapted from the Hudson-Jorgenson long-term macroeconomic model. The version of the Hudson-Jorgenson model that was de- veloped for the Macroeconomic-Demographic Model, which incorporates demographic influences on the pat- tern of consumption behavior, is described in Hudson, Kimbell, and Maguire (1981). Description of the Model The Macroeconomic Growth Model identifies four gener- al types of commodities in the economy—capital services, labor services, consumption goods and services, and in- vestment goods. These are defined so that they account for all transactions. There is a market, or a set of markets, corresponding to each of these commodities. Four groups of decision-makers participate in these markets— households, producers, government, and the rest of the world. Households supply labor services to producers and own capital stock and other assets. From the associated income, households purchase consumption and invest- ment goods and also save. The key information provided by the model of household behavior is the demand for consumption in the present period and the allocation of the present consumption between the consumption of goods and services and the consumption of leisure time. The most important determinant of this joint consump- tion decision is the household's full wealth. Full wealth is the accumulated value of the household sectors financial and real assets and the present discounted value of both total household current and future time resources and income transfers to households in the current and all future periods. An intertemporal consumption function “Appendix A contains a detailed description of all equations and varia- bles of the Macroeconomic Growth Model. THE NATIONAL INSTITUTE ON AGING MACROECONOMIC-DEMOGRAPHIC MODEL describes how the household sector allocates the con- sumption of this wealth between present and future per- iods. Output from the Labor Market Model, described in Chapter 4, determines the level of work effort by house- holds. With this information, leisure time consumed and consumption of goods and services can be estimated.’ Producers purchase labor and capital inputs and supply consumption goods and services and investment goods. The output of investment goods going to the private sec- tor expands the capital stock, increasing future productive capacity. Investment is separated into categories of resi- dential structures, consumer durables, and producer cap- ital goods. The equations of the production sector determine the supplies of consumption and investment goods output and the demand for inputs of capital and labor services. Technical change augmenting the input of each factor is modeled, and the overall level of labor efficiency is represented in terms of these productivity components. The values of consumption output, invest- ment output, and capital input relative to the input of labor services are determined as functions of the prices of these inputs and outputs and of output-specific or input- specific technical efficiency as related to a time index. The government sector purchases labor services as well as consumption and investment goods and also operates an extensive tax-transfer system. Tax rates and govern- ment expenditures enter into the representation of al- most all the markets of the private economy. These government variables can be drawn together into a set of accounts expressing the tax revenues and expenditures implied by these variables. This representation of the pub- lic finances leads to the calculation of the government deficit and of net claims on government by the private domestic sector. . The rest of the world sector enters into the markets for labor services, consumption goods and services, and in- 3A crucial difference between the original version of the Hudson-Jorgen- son model and this Macroeconomic Growth Model is the solution sequence for the determination of the demand for consumption by households. The Hudson-Jorgenson model includes an explicit con- sumption function to estimate demand for consumption goods and services. The demand for leisure time is then the difference between full wealth consumed in the present period and consumption of goods and services. The quantity of labor services supplied is then calculated as the difference between time available and leisure time consumed. The Macroeconomic Growth Model reverses the Hudson-Jorgenson procedure. The quantity of labor services supplied is calculated by the Labor Market Model. The quantity of leisure time consumed is then the difference between total time available and labor services supplied. Consumption demand for goods and services is the difference between full wealth consumed in the present period and leisure time consumed. “These government accounts refer to the combined Federal and state and local governments exclusive of social insurance funds. Since the social insurance system is based on taxes used exclusively to support social insurance benefits and since all revenue ultimately returns to the private sector in the way of benefits, social insurance funds are treated as part of the private sector and are not included in general government in the model described in this monograph. Changes in the current trust fund levels have a direct impact on current wealth and hence on private consumption. vestment goods. Summary accounts, similar to those con- structed for the government sector, are also constructed for the rest of the world sector. The supply of and demand for goods and services from the rest of the world sector are exogenous. All four decision groups act simultaneously in all four economic markets. The model represents these market processes, depicts the interaction of the behavioral pat- terns of each decision group, and determines the eco- nomic configuration at which the markets are in balance. - Each year, therefore, prices and quantities are determined "so that there is balance in each of the four broad econom- 10 ic markets. Over time, conditions in each market change in re- sponse to altered tastes, technology, and availability of productive inputs. Demographic change is a major deter- minant of household and producer behavior. Inputs, out- puts, and relative prices change in response to both population growth and change in the age structure of the population. Demographic processes, therefore, do not simply change the numbers of workers and consumers. They influence the interaction of supply and demand in all markets. The solution of the market system over successive years, with each year’s solution satisfying the constraints of capacity and behavior particular to that year, provides a representation of the growth over time in the economy. The information generated by the Macroeconomic Growth Model, then, is a sequence of market outcomes which provides a picture of the structure of the economy at each point in time and of the rate and nature of the change in this structure over time. Macroeconomic Accounting System The Macroeconomic Growth Model is based upon an eco- nomic growth accounting system developed by Laurits Christensen and Dale Jorgenson.’ This system was specifi- cally designed to identify and measure consistently the aggregate variables involved in economic growth. Be- cause of this, the system differs slightly from the National Income and Product Accounts (NIPA), which is the com- monly used framework for economic accounting. To achieve commonality with this conventional framework, and so to facilitate the use of the growth model in forecast- ing, the model includes a systematic method of relating the growth model variables to the NIPA aggregate accounts. The bridge between the two accounting systems is a set of stochastic equations and identities that permit all the growth model forecasts to be expressed in NIPA terms. The stochastic equations map the growth model variables, such as real consumption, real investment, price of con- This accounting system is described in Christensen and Jorgenson (1970). CHAPTER 3 sumption, and price of investment, into the correspond- ing NIPA variables. The identities then aggregate expenditure and output into gross national product, real GNP, and the remaining aggregates in the standard nation- al income accounts. The principal difference between the growth and NIPA accounting systems arises in the treatment of household sector capital. This capital comprises residential property, automobiles, and other consumer durables. These types of capital are exactly analogous to producer’s capital— they are durable assets yielding economic services over a period of years. For this reason, household capital is ex- plicitly treated as capital in the growth model accounts. Purchases of consumer durables are included in invest- ment, not in consumption, and the imputed value of ser- vices from this capital is calculated and included in consumption. This yields a consistent system of accounts in which investment includes all additions to capital stock and in which consumption comprises only current goods and services. The NIPA system does not provide this same consistency in the treatment of consumption, investment, and capital. The NIPA system includes in consumption an imputation for the services from owner-occupied residen- tial structures but not the services of other forms of con- sumer durables. Also, the NIPA system includes purchases of household capital in consumption, not in investment. ¢ 11 Macroeconomic Growth Model These features of consumption and investment are the principal difference between the growth model and the NIPA accounting systems. New Features of the Model Some aspects of the Macroeconomic Growth Model are the same as those in previous models developed by Hud- son and Jorgenson.® However, there are many new fea- tures incorporated in the version of the growth model developed for the MDM. The data base was updated. For the version of the MDM reported in this monograph, the data base covered the period up to and including 1978, with most series going back as far as 1928. Most signifi- cantly, the model's specification has been reformulated to incorporate explicitly demographic influences on house- hold behavior. The age structure of the population direct- ly affects consumption, savings, and investment composition. Through these interactions, demographics affect the entire structure and performance of the econo- my.” The Macroeconomic Growth Model retains its sup- ply side features, incorporating the determinants of labor supply and capital expansion and explicitly including the supply constraints of the economy. For example, the economic model presented in Hudson and Jorgenson (1974). "This work was completed by Hudson and Jorgenson under contract to ICF. Chapter 4 Labor Market Model Introduction he Labor Market Model projects the labor market status of twenty-two demographic groups that make up the U.S. labor force age 16 and over. It depicts three basic aspects of the labor market: the supply of labor, the demand for labor, and the simultaneous determination of labor input and compensation for labor services. The solution of the Labor Market Model is fully simultaneous with that of the Macroeconomic Growth Model, so that the projections for all major labor variables are consistent with the projec- tions for related variables such as output, capital services, consumption and investment. Labor supply is measured in total annual hours worked by each of the twenty-two age-sex groups. Younger age groups are divided into age categories of 16-17 and 18-24 years of age. Prime age groups are separated into ages 25- 34,35-44, and 45-54. Because labor market and retirement behavior change rapidly after age 55, older workers are disaggregated into smaller age groups-ages 55-58, 59-01, 62-64, 65-67, 68-71, and 72 and older. The Labor Market Model produces estimates of the number of workers, la- bor force participation rate, total hours worked, average compensation level, and unemployment rate of each of these twenty-two groups. These variables serve as data for the five retirement income models. The Labor Market Model links the Macroeconomic Growth Model to the models simulating the five retire- ment income systems. The Macroeconomic Growth Mod- el provides long-term projections of the growth of the United States economy. Its projections of assets and cap- ital services are primary inputs for the Labor Market Mod- el. Population projections from the Population Model are also input for the Labor Market Model. Labor Market Mod- el projections of employment and compensation are im- portant inputs to the three principal pension models- Social Security, Private Pensions, and Public Employee Pensions. The Supply Sector The supply sector of the Labor Market Model estimates labor force participation rates, unemployment rates, and «hours worked for twenty-two age-sex groups. These varia- bles are used to determine total labor input, which is a key variable of the Macroeconomic Growth Model, and com- pensation levels, which are key variables in the three pension models. Although many investigators have analyzed the labor supply decisions of workers, few have attempted compa- rably disaggregated analyses on a time-series basis. Most studies of labor supply are based on cross-section data. 13 The specification of labor supply used in this model is informed by the same theoretical framework as the inves- tigations of labor supply conducted in the 1970s during the consideration of welfare reform.! Changes in the basic specification were required to embed this Labor Market Model within a larger general equilibrium model and to obtain stable time series forecasts. The supply equations? produce estimates of total hours worked by various age-sex groups. Total hours worked in the private economy can be SHresied in terms of the following equation: = ((N, — My) X R, x (1-U,) — Gy) X Hg, (41) where L,, is total private labor input measured in hours worked for sex s and age group a, N,, is the noninstitutional population of sex s and age group a, M,, is the population in the Armed Forces of sex s and age group a, Ry, is the civilian labor force participation rate of workers of sex s and age group a, U,, is the unemployment rate of workers of sex s and age group a. G,, isthe number of civilian employees of sex s and age a of Federal, state and local governments, and H,, isthe annual hours worked per worker of sex s and age group a. The Labor Market Model provides estimates of the aver- age participation rate, unemployment rate, and hours worked for each age-sex group. Population values are obtained from the Population Model described in Chapter 2. The projection of civilian and military public employ- ment is described in Appendix G. The modeling of the participation decision applies stan- dard microeconomic theory to the problem of allocating time across competing uses—Ileisure and work. By work- ing, individuals earn income with which they can pur- chase goods. The model assumes that the participation decision reflects a determination by each individual whether the value of at least one hour worked, measured in terms of the goods which can be bought with the wages earned, exceeds the value of the additional leisure time which could have been consumed instead. 'Examples of these studies are provided in Cain and Watts (1973). 2A listing of all variables and equations of the model is contained in Ap- pendix B. THE NATIONAL INSTITUTE ON AGING MACROECONOMIC-DEMOGRAPHIC MODEL The principal variables in the labor force participation equation for each age-sex group are the average hourly compensation of the age-sex group, the average level of private assets per capita in the previous year, the unem- ployment rate of the age-sex group, and a time trend. Ra = f(W,, AL, Us T) (4-2) where Rg, and Uj, are as defined above, W,, isthe average hourly compensation of sex s and age group a, AL is the level of per capita assets in the economy in the previous year, and iT is a time trend. The theory of labor supply recognizes two opposing influences on individuals. As market wages increase, per- sons may tend to work more because the value of each hour worked increases. Economists refer to this as a sub- stitution effect—individuals desire to substitute work hours for leisure hours as the latter become relatively more costly in terms of foregone wages. At the same time, increases in market wages create higher incomes allow- ing individuals to consume more of all goods and ser- vices, including leisure. Economists refer to this as an income effect—individuals may tend to work less at high- er income levels. We include both a wage term and an income term in our equations to capture both effects. We used the level of private assets per capita to repre- sent life cycle or permanent income. It is far from clear what the best measure of income should be for this equa- tion. The average income level of the specific demogra- phic group is clearly important. However, in households with multiple workers, the labor force decisions of young workers, elderly workers, and prime age female workers are often made jointly with those of prime age male work- ers. In those cases, the economy-wide aggregate average income may be a better measure than an income variable specific to a single demographic group. It is also impor- tant that our measurement of income levels for a specific age-sex group be independent of that group’s labor-relat- ed income if we are to obtain unbiased statistical results. Furthermore, to capture long-term trends, a measure of permanent—or long-term expected—income may be su- perior to one of current income. Total private assets, lagged one period, and the average wage rate are primary determinants of the consumption of goods and services and leisure and of aggregate labor supply in the specification of the Macroeconomic Growth Model. To specify the labor supply equations for the disag- gregated age-sex groups of the Labor Market Model con- sistently with the Macroeconomic Growth Model, the lagged value of aggregate private assets per capita and the 3Labor income is the product of the quantity of labor supplied and the wage rate. Since the quantity of labor supplied by each group is what we are estimating, use of labor income introduces a bias into the analysis. age-sex group specific wage are included as primary de- terminants of labor supply in the Labor Market Model .* Considerable evidence has been collected that indi- cates that labor force participation of demographic groups is sensitive to labor market tightness.> The rate of unemployment of the demographic group is included in the group’s participation equations as an indicator of the tightness of the labor market for that group. It is not clear "a priori what the sign on the unemployment rate variable 14 in a participation equation should be. An increase in un- employment may have two opposing effects on the size of the labor force. Discouraged workers may drop out of the labor market when unemployment rates increase due to the larger search costs required to find a job. However, additional workers may enter the labor market to supple- ment the reduced incomes of households experiencing layoffs. The unemployment rate of each age-sex group is itself endogenous to the model. It is specified to be a function of the overall cyclical tightness of the labor market—as indicated by the unemployment rate of males in the prime ages, 25 through 54—as well as the proportion of the total labor force accounted for by the group, and the ratio of the Federal minimum wage to the average rate of com- pensation of the group. Ug = fUpm Sa Ma) (43) where Uy, isthe unemployment rate of sex s and age group a, Up is the unemployment rate of prime age workers (a proxy for the tightness of the labor market), Sea is the share of the labor force of sex s and age group a, and M,, is the ratio of the minimum wage to the wage of sex s and age group a. The following considerations led to the inclusion of a variable measuring the share of the total labor force ac- counted for by that group in its unemployment equation. Study of the production characteristics of labor force age groups indicated that different age groups are not perfect substitutes in production. Therefore, an exogenous shift in the supply of labor of different demographic groups requires an adjustment in their relative prices. It is well known, however, that wages are sticky, and in particular that changes in relative wages take place only with a lag. While relative wages are adjusting toward a new equilibri- um, excess supply will be manifested as changes in demo- graphic group rates of unemployment. Shifts in supply are “Mean income data by age-sex group is unavailable prior to 1966. We experimented with a number of income and asset variables and found none that produced clearly superior results. We chose to use the lagged asset formulation described above because it was the most readily inte- grable into the larger projections framework. >See Tella (1964, 1965), Dernberg and Strand (1964, 1966), Bowen and Finegan (1969), Cooper and Johnston (1965), Mincer (1966). CHAPTER 4 reflected in changes in the relative size of the labor force of a given demographic group. The proportion of the total labor force accounted for by the group is the relevant variable rather than the absolute size of the group or its rate of growth, since change in a demographic group that was part of a general increase in the size of the population (or stable age structure growth) would not change factor proportions. This argument predicts a positive relation- ship between the proportion of the labor force accounted for by a demographic group and the group’s rate of unemployment. Any element impeding factor price adjustment would serve to increase the level of unemployment. The Federal minimum wage is one barrier to downward adjustment in wages. A variable measuring its effects is included in the unemployment equations. Use of a unique set of time series data on U.S. labor input and compensation, developed by Frank Gollop and Dale Jorgenson (1977), combined with data on participa- tion and unemployment from the U.S. Bureau of Labor Statistics, permitted estimation of the labor supply equa- tions for the large number of disaggregated demographic groups. We supplemented these data with Current Popu- lation Survey data to obtain the greater disaggregation needed for analyzing the labor markets for elderly workers. At least four of the variables in equations 4-2 and 4-3 are jointly determined: the unemployment rate, the participa- tion rate, the labor force share, and the wage rate. The unemployment rate enters directly into the participation equation. Participation, in turn, influences labor force share, which enters the unemployment equation. The group wage influences participation but is, in turn, influ- enced by labor supply. The wage also interacts with unem- ployment through its effects on demand. Several of the explanatory variables in each equation, therefore, may be correlated with the disturbances. The disturbances may be contemporaneously correlated for other reasons. Estimation of the labor force participation and unem- ployment equations by ordinary least squares would pro- vide parameter estimates that may be biased, inconsistent, and inefficient. Both equations, therefore, were estimated jointly by a modified version of Zellner-Theil, three-stage- least-squares (3SLS). The unemployment rate of prime age males, minimum wage, lagged level of per capita assets, and time trend variables were considered to be exogenous variables. Additional instruments used in each regression to purge the endogenous variables of correla- tion with the disturbances were the ratio of the group's population to the total population age 16 and over and the quantity of capital services input in the US. private sector. The size of the group’s population is clearly exogenous to the labor market, but it is correlated with the size of the group's labor force share. The quantity of capital services is correlated with wage rates because of its effects on labor productivity. The equations for participation and unemployment were estimated with all variables, except the time trend, in 15 Labor Market Model logarithmic form, so the resulting coefficients can be in- terpreted as elasticities. The time trend is entered as a reciprocal so that its influence on the projected values diminishes over time. (In this form, a negative coefficient on the time trend variable indicates a decelerating in- creasing trend. A positive coefficient indicates a decelerat- ing decreasing trend.) Estimated coefficients of the participation and unem- ployment equations are presented in Tables 4-1 and 4-2. In Table 4-1 we present the results for the groups age 16 through 54. These equations were estimated using annual observations for 1948-1978. In Table 4-2 we present the results for the groups age 55 and older. Equations for those groups were estimated using annual observations for 1963-1978, because of limitations on data for participa- tion rates of older workers. Table 4-1 shows that the results for the unemployment equation are quite good. Most of the coefficients are sig- nificant at the five percent confidence level, and almost all exhibit the expected sign. The influence of the cyclical level of labor demand, as measured by the coefficent on prime age male unemployment rates; is highly significant and close to one in most cases. Younger age groups ap- pear to benefit less from tight labor markets than other groups. Unemployment of most groups increases if their relative size in the labor market increases. This indicates that the demographic groups are imperfect substitutes for one another. Increases in the minimum wage affect young males and all female groups but have almost no impact on prime age male groups. Average wage rates for prime age male workers are much higher than those of other groups and a smaller proportion of prime age males are affected by the minimum wage. Results for the participation equations are not as clear cut. The participation rates of male workers appear to be relatively insensitive to the wage rate. In several equations relatively high wage elasticities are observed for female groups. Since for many females non-market work in the home is an alternative to labor force participation, the larger importance of market wages in their participation decision is not surprising. The positive sign for the assets coefficient in three of five female groups is unexpected. Our time trend variable may not fully capture the large increase in participation rates in female groups over tiie past 30 years.® The asset variable, which grows steadily over the period, may be picking up some of the trend effects. The positive relation- ship between aggregate assets and female labor force par- ticipation for several groups generated unconstrained forecasts of very high levels of participation. These were constrained in the development of the base case simulations. Table 4-2 presents analogous results for the disaggre- gated older worker groups. Significance levels of all equa- °In fact, in three of the five cases the time trend coefficient is positive, indicating a declining pure time trend, contrary to expectations. THE NATIONAL INSTITUTE ON AGING MACROECONOMIC-DEMOGRAPHIC MODEL Table 4-1 Estimated Coefficients of the Unomploymem and Participation Equations, Young and Middle-Aged Groups Age 16-54! Unemployment Equation Participation Equation Prime Age Group Minimum 5 5 Group Unemployment Share Wage Ratio Constant R’ Wage Assets Unemployment Trend Constant R° Male 16-17 0.532 0.914 0.282 5732 91 62a —0.27b —0.42a 0.13b 5.032 51 (0.04) (0.17) (0.07) (0.65) (0.15) (0.11) (0.07) (0.06) (0.84) 18-24 0.804 0.372 0.23b 2402 90 0.012 —0.122 —-0.01 —0.102 4.324 83 (0.05) (0.11) (0.09) (0.26) (0.07) (0.08) (0.01) (0.02) (0.39) 25-34 1.112 -0.25 0.12 -0.29 98 0.102 —0.158 ~~ —0.012 —0.02¢ 5.494 75 (0.04) (0.16) (0.08) (0.40) (0.02) (0.02) (0.002) (0.01) (0.11) 35-44 0.954 0222 —0.01 0352 99 0.01 —0.052 —0.0032 —0.0042 4.952 91 (0.02) (0.04) (0.04) (0.12) (0.01) (0.01) (0.001) (0.003) (0.08) 45-54 0.952 0.882 —0.04 1.728 98 0.04b —0.212 —-0.012 —0.082 0.454 97 (0.04) (0.11) (0.06) (0.25) (0.01) (0.02) (0.002) (0.01) (0.03) Female 16-17 0.424 0.27¢ 0.472 3.432 83 0.992 0.16 —0.534 0.052 1.03 49 (0.05) (0.14) (0.10) (0.60) (0.18) (0.16) (0.08) (0.08) (1.30) 18-24 0.584 0.582 0.61¢ 3612 87 -—0762 1.122 0.08¢ 0.0072 —4.162 93 (0.06) (0.10) (0.11) (0.25) (0.19) (0.15) (0.04) (0.03) (1.01) 25-34 0.594 0.272 0.50 2332 87 1.65b —1.09¢ —0.432 —0.042 9.48b 42 (0.05) (0.09) (0.08) (0.26) (0.65) (0.62) (0.13) (0.05) (4.02) 35-44 0.662 -0.25 0.53 0.65 88 0.02b 051b 0.02 —0.0842 —0.21 99 (0.05) (0.34) (0.09) (0.86) (0.22) (0.20) (0.03) (0.02) (1.28) 45-54 0.702 -0.35P 0.442 0.05 91 1.812 —0.93b -0.14 0.512 8.844 85 (0.04) (0.16) (0.11) (0.36) (0.39) (0.35) (0.09) (0.12) (2.10) 'Standard errors in parentheses. “R? Statistics are not strictly correct for a three stage least square regression since its residuals do not sum identically to zero. We present R? values here to provide some indication of the goodness of fit of our model. The largest (in absolute value) sum of residuals in these 20 equations is .00045. 4Significant at 1 percent level. bsignificant at 5 percent level. CSignificant at 10 percent level. 16 CHAPTER 4 Table 4-2 Labor Market Model Estimated Coefficients of the Unemployment and Participation Equations, Elderly Groups Age 55 and Older’ Unemployment Equation Participation Equation Prime Age Group Minimum Group Unemployment Share Wage Ratio Constant R- Male 55-58 0.69 8.874 — 4.73 17.22b N/A (0.55) (255) (1.50) (7.70) 59-61 0.68 4.094 —2.492 7.20¢ N/A (0.31) (1.09) (0.84) (3.90) 62-64 0.6% 1.892 —2234 -088 NA (0.19) (0.46) (0.51) (1.76) 65-67 0.48 -1.07 0.67 -787 NA (0.27) (0.58) (0.73) (2.40) 68-71 0.18 —1.24 -0.28 -10.11 NA (0.26) (0.63) (0.73) (2.81) 72+ 0.03 -0.38 —1.40b —7.23 40 (0.23) (0.38) (0.62) (1.78) Female 55-58 0.25 1.97 —2.23b 1.75 NA (0.32) (3.37) (0.94) (11.92) 59-61 0.508 0.02 —1.21b -499 NA (0.18) (117) (0.51) (4.86) 62-64 0.56 —1.60 0.03 -11.53 NA (0.15) (0.94) (0.41) (4.21) 65-67 0.664 —250 1.01 -16.64 N/A (0.30) (2.13) (0.84) (11.03) 68-71 -0.29 —4.49 -0.73 -2870 NA (0.47) (1.83) (1.19) (9.90) 72+ —0.03 -3.02 -0.56 -2147 NA (0.36) (1.42) (0.84) (7.87) !Standard errors in parentheses. Wage Assets —0.01 —0.604 (0.14) (0.16) 0.02 —1.092 (0.18) (0.21) 0.01 2412 (037) (0.41) -051 —2.874 (0.43) (0.69) -0.36 —2.33b (0.67) (0.90) -039 1.11 (0.36) (0.98) 051 —037 (0.56) (0.26) 1.14b —0.83b (0.52) (031) 0.77 —1.864 (0.76) (0.56) 0.73b —2.57a (0.37) (0.41) 039 —1.73b (0.59) (0.64) 025 —1.82a (0.36) (0.45) Unemployment Trend -0.01 —4.08b (0.01) (1.87) —-0.01 —7.874 (0.01) (2.10) -0.02 —17.62a (0.02) (4.81) —0.052 28.434 (0.05) (7.55) —-0.12¢ —26.11b (0.07) (10.22) —-0.07 22.31¢ (0.06) (11.61) =0.05 1.06 (0.03) (9.50) —0.04¢ 5.11 (0.02) (9.53) —0.06 —11.66 (0.04) (13.61) —0.02 —15.38¢ (0.04) (7.54) —0.112 —10.85 (0.05) (13.14) —0.08¢ -10.27 (0.04) (7.86) Constant 1.702 (0.44) 3.08¢ (0.52) 6.954 (1.13) 8.854 (2.04) 6.41b (2.61) — 5.694 (3.30) —-0.49 (1.54) -0.21 (1.64) 3.40 (2.50) 5.514 (1.52) 2.42 (2.50) 1.94 (1.66) rR: 95 97 97 97 91 96 .60 65 89 77 94 2R? Statistics are not strictly correct for a three stage least square regression since its residuals do not sum identically to zero. We present R? values here to provide some indication of the goodness of fit of our model. R* values are not shown if the sum of residuals exceeds .1 in absolute value. aSignificant at 1 percent level. bsignificant at 5 percent level. CSignificant at 10 percent level. 17 THE NATIONAL INSTITUTE ON AGING MACROECONOMIC-DEMOGRAPHIC MODEL tions are lower because the smaller. sample available for these estimates reduces the number of degrees of free- dom. For the unemployment equations, prime age unem- ployment rates are the most important explanatory variable. Large significant effects of labor force share on unemployment rates are observed for male groups below age 64.7 For the participation rate equations, wages again seem to have a negligible effect for males. The asset coefficients are generally of the expected sign, are relatively large in absolute value for males, and are highly significant. Unem- ployment coefficients in the participation equations are of ‘the expected sign but are generally insignificant. Four of the male time trend coefficients are negative, which, be- cause of the reciprocal specification of the trend variable, indicates an increasing participation rate over time, if oth- er independent variables are held constant. Results for the older female participation equations are consistent with the simple theory of labor supply. The signs on all three economic variables agree with our ex- pectations. Negative time trend coefficients for the four oldest groups indicate increasing participation rates with time, holding other variables constant. The second portion of the supply section of the Labor Market Model analyzes the average annual hours worked by workers in each of 14 age-sex groups.® Conceptually, the model for this analysis might be similar to that for the participation decision—both wages and assets/income levels could be important. We experimented with several specifications and obtained the best results by specifying that annual hours are a function of the level of per capita assets and a time trend. H,, = f(AL, T) (4-4) where Hg, is the annual hours worked of sex s and age a, AL is the level of per capita assets in the economy in the previous year, and T is a time trend The equation was estimated in logarithmic form and, as in the participation equation, the time trend is specified to be a reciprocal exponential. Table 4-3 shows the results of these estimations. The assets coefficient is negative, as expected, in 11 out of 14 cases. In ten of these instances, the coefficient is signifi- cant. The assets coefficient is never significant when it is positive. The time trend coefficient is relatively small for males, except for those age 16 and 17. The time trend “Significant negative coefficients on the minimum wage variable sug- gests that increases in the minimum wage may reduce the unemploy- ment rate of several older groups, perhaps because these groups substitute for younger groups whose average marginal productivities are lower. 8Data on hours were not available for the disaggregated older age groups. 10 Table 4-3 Coefficients of Total Hours Equations Group Assets Trend Constant R? Male 16-17 06 70¢€ 635 b 27 (18)3 (28) (51) 18-24 —46b 20¢ 8.49 b 89 (05) (08) (14) 25-34 -27 -.07 828 b 75 (04) (06) (.10) 35.44 -16b -.01 802 b 70 (.03) (.04) (.08) 45-54 -a1 -.02 7.88 b 48 (03) (04) (.08) 55-64 -15 - 01 792 b 63 (.03) (.04) (.08) 65+ — 41 28¢ 840 b 82 (07) (11) (20) Female 16-17 90 99 ¢ 387b 35 (23) (.35) (.64) 18-24 -12 59 b 733 b 78 (07) (10) (19) 25-34 04 48b 701 b 57 (07) (10) (18) 35.44 -.06 23b 7.40 b 51 (05) (.08) (14) 45-54 -11¢ 11 759 b 52 (04) (07) (12) 55-64 -.10 15¢ 7.56 b 55 (04) (07) (12) 65+ -31b 4sb 791 b 70 (.10) (15) (27) Standard errors in parentheses bsignificant at 1 percent level. CSignificant at 5 percent level. Significant at 10 percent level. coefficient is positive and significant in six of seven cases for females. Since a positive coefficient represents a de- clining trend, these results suggest the increase in female labor force participation has been associated with a de- crease in the average hours worked by female workers. This would occur if many new female workers work in part-time jobs. Although most of the trend coefficients are small, the trend components can become quite large when extrapo- lated over the lengthy horizon of this model. To avoid this problem, the model permits the user to impose limits on the variation in unemployment rates and participation rates. Trend unemployment rates were constrained with- in plus and minus 50 percent of their average values over the period 1947 through 1978.° Participation rates of women were constrained not to exceed those of males of For groups above age 54, the time period is 1963 through 1978. CHAPTER 4 the same age. Participation rates of workers over age 54 were not permitted to exceed the rates of the next youn- ger group. Finally, participation rates of all age groups were constrained within the broad limits shown in Table 4-4. Table 4-4 Limits on Participation Rates Used Within the Labor Market Model Male Female Age Lower Upper Lower Upper 16-17 0 .60 0 .60 18-24 0 85 0 85 25-34 0 975 0 975 35-44 0 975 0 975 45-54 0 965 S53 965 55-58 45 95 25 95 59-61 40 95 20 95 62-64 35 95 10 95 65-67 25 95 .10 95 68-71 15 95 05 95 72+ 05 95 .02 95 When these estimates of hours, unemployment rates, and participation rates are combined with estimates of the non-institutionalized population and government employment, an estimate of total labor input of each age- sex group is derived from equation 4-1. The Demand Sector The demand for labor is derived from the demand for the output of the production process. Modeling demand for labor requires modeling the aggregate production tech- nology of the private U.S. economy. The Labor Market Model is particularly concerned with the substitutions be- tween productive factors, including age groups of labor and capital, and the interaction between changes in factor proportions and changes in relative prices, wages, and interest rates. Ideally, one might model the production process at a level of disaggregation similar to that used for the supply side equations. It is unlikely, however, that differences in substitution characteristics among similar highly disag- gregated age-sex groups are significant—e.g., that the sub- stitution characteristics of males age 25-34 are greatly different from those of males age 35-44. Furthermore, available data and estimation techniques might not pro- vide robust estimates for the large number of parameters that would correspond to detailed disaggregation. Conse- quently, for the demand analysis, the labor force was dis- aggregated into three age groups, 16-24, 25-54, and 65 and over. A four factor production function was estimated, for which separate inputs are the services of young workers, middle-aged workers, older workers, and capital. The production characteristics of the various inputs determine 10 Labor Market Model a set of demand price elasticities that show the percentage change in the price of any input which is associated with a one percent change in the quantity of input demanded. The production frontier, expressing the maximum out- put available from a given quantity of inputs, is approxi- mated by a translog function. InQ =a, + 2 alnX i +% 2% 2b;InX;InX ij (4-5) where Q is output and X; is the quantity of input i—young, middle-aged, or elder workers, or capital services. Substitution and demand elasticities can be derived from estimates of the parameters of this production func- tion.'? If input markets are competitive, the translog speci- fication implies that each factor’s share—the ratio of payments to that factor to the value of total output—can be expressed in terms of the translog parameters and the logarithms of each input’s quantity. The translog param- eters can be estimated from these factor share equations. Estimation is facilitated by applying restrictions to these equations based on the economic theory of production. It is customary to assume that the production technology (1) is symmetric, (2) exhibits constant returns to scale, and (3) is concave. The symmetry condition means that changes in input X; affect the share of X; in the same way that changes in input X; affect the share of X;. Under an assumption of constant returns to scale and competitive markets, the sum of each factor’s share of total revenue exactly equals the total revenue earned from sales of the product. The concavity assumption is equivalent to the well-known property of diminishing returns to any input. An additional restriction was placed on the parameters of the translog production function in order to assure that the four factor specification of production technology used in the Labor Market Model was consistent with the two factor specification used in the Macroeconomic Growth Model. We specified that the labor inputs could be aggregated consistently. It was assumed that the rate at which one labor input can substitute for another, leaving total output unchanged, is unaffected by the quantity of capital." At the same time, we allow differing elasticities of substitution between each pair of labor categories, so that the equilibrium wages of different age groups of workers are sensitive to their relative proportions. After imposing these restrictions, translog factor share equations were estimated. The coefficients characterize the demand curves for each of the four inputs. Because The properties of the translog function and derivation of elasticities of substitution and of demand are discussed in Appendix C. "In a revised version of the model this restriction is relaxed. THE NATIONAL INSTITUTE ON AGING MACROECONOMIC-DEMOGRAPHIC MODEL both factor quantities and prices are determined simulta- neously, estimates of these equations by ordinary least squares would be subject to simultaneous equations bias. Two-stage ‘least squares, therefore, was the appropriate estimation technique, using a set of exogenous instru- mental variables to purge the input quantities of correla- tion with the disturbances in the equations. To estimate these equations we used the labor input and compensa- tion data for the period 1947-1979 developed by Gallop and Jorgenson. For the four factor translog production function, esti- mates of the partial elasticity of input price with respect to quantity change for each factor is displayed in Table 4-5.'2 The interpretation of this elasticity is the following. For a small proportional change in the quantity of one factor, other inputs fixed, the prices of all inputs relative to the price of output must adjust for the new input configura- tion to be in equilibrium. The elasticities describe the resulting proportional changes in the prices of all factors of production. When all of the inputs are changing, the proportional change in the price of one input (say Y) will be the weighted sum of the proportional changes in the quantities of all of the inputs (Y, M, O, K), the weights being the partial elasticities of price of one input (Y) with respect to quantity of each of the respective inputs, e.g., (4-6) ® er , pn gro dO + pre dK dp” dy where _— = proportional change in the price of Y, —_ P proportional change in the quantity of Y, etc., and Ey, = elasticity of the price of Y with respect to the quantity of M, etc. A set of these equa- tions, one for each of the four inputs into the aggregate production sector, is used to depict the factor price adjust- ment process on the demand side. The Solution Algorithm The solution sequence of the Labor Market Model is as follows. First, preliminary estimates of labor inputs are derived from the supply sector. They are based on esti- mates of the population of each age-sex group from the Population Model and initial guesses about wage rates and other economic variables for each demographic group. Given the estimates of labor input levels and the level of capital services, which is determined by the size of the capital stock at the end of the previous period, the equa- tions of the demand sector then estimate how wages of each age-sex group and the price of capital would re- “Derivation and calculation of the elasticities are shown in Appendix C. 2°N Table 4-5 Estimates of the Elasticity of Input Price with Respect to Input Quantity Used in the Labor Market Model? Quantity Change Young Middle Older Price Change Capital ~~ Workers Workers Workers Capital — .496P 010° .060P 015P (.017)¢ (.0003) (.002) (.001) Young workers 0620 —175> —257 — 050 (.002) (.084) (.083) (.083) Middle workers 062 —.044> —270° —.035 (002) (044) (.050) (.050) Old workers 0626-035 —.144 -261° (.002) (.049) (.050) (.050) 3Estimates assume 1972 factor share values. Significant at the 1 percent level. Standard errors in parentheses. spond to these input levels. The solution procedure of the model, based upon the Gauss-Seidel methodology, iter- ates between estimates of wages (demand sector) and quantities of labor (supply sector) until consistent values are obtained. The model starts by using equations 4-1 to 4-4 to pick out quantity values as points on the supply curve corre- sponding to the initial guess about wages in each group. Next the solution procedure determines how wages would change to correspond to those quantities along the demand curve. The shape of the demand curve is summa- rized in the set of demand elasticities shown in Table 4-5. If supply and demand are not consistent, the model speci- fies that wages will adjust toward their equilibrium val- ues—upward if an excess demand exists; downward if an excess supply exists. Changes in the wages for young, middle-aged, and old workers and a change in the price of capital services, corresponding to the changes in quanti- ties of inputs are calculated using a set of equations like 4- 6. Using these wage estimates for three classes of workers, all twenty-two wages are adjusted proportionally to the wage changes of their aggregate groups. This new set of wages is then input into the supply sector and new sets of labor quantities are calculated as the basis for the next iteration. The model continues to iterate between supply and demand until the two are consistent. The solution procedure continues until all changes are less than a pre- scribed tolerance level. For the simulation reported here that tolerance is 0.1 percent. At that point the preliminary solution values to the Labor Market Model are passed back to the Macroeconomic Growth Model. Chapter 5 Social Security Model Introduction he Social Security Model projects the number of benefi- ciaries, the average and total levels of benefit payments, the revenues, and the Trust Fund balances of the Old Age and Survivors Insurance and the Disability Insurance (OASDI) programs. The combined Old Age and Survivors Insurance and Disability Insurance programs are the larg- est element of the United States pension system. In 1980, 115 million workers, 64 percent of the population age 16 and over, were covered by OASDI at their place of em- ployment, and 131 million, 82 percent, were fully insured as a result of current or previous coverage. Ninety-five percent of males age 60 through 84 were fully insured; 60 percent of females of these ages were fully insured. In 1980 the combined OASDI Trust Funds had total revenue of 118 billion dollars and disbursed to over 35 million beneficiaries 119 billion dollars, a figure equal to 4.8 per- cent of GNP. The OASDI system is financed on a pay-as-you-go ba- sis.! Income from current contributions (payroll tax pay- ments) is used to pay current benefit payments. Almost all payroll taxes are collected from individuals age 16 through 64. Ninety percent of OASI benefits are paid to retired workers, their spouses or their aged survivors. Consequently, the financial balance of the system is very sensitive to the age structure of the population. The retire- ment of the large “baby boom” cohorts, which will occur from 2015 through 2035, will put a heavy financial strain on the system. The Social Security Administration (SSA) Office of the Actuary estimated in 1981, in its mid-range cost projections, that OASI expenditures will exceed 15 percent of the taxable payroll after 2030. The actuary’s 1981 estimated expenditures of the combined OASDI (re- tirement and disability) systems are about 17 percent of taxable payroll after 2030. Under a set of optimistic as- sumptions, the actuary’s estimated OASDI expenditures are 12 to 13 percent of taxable payroll during the period 2030-2055. Under a set of pessimistic assumptions the actuary’s estimated expenditures are 23 to 28 percent of taxable payroll during that period. . These factors suggest two observations. First, there is considerable uncertainty concerning what the burden of social security will be in the next century—expenses may "The Social Security Model described in this chapter was completed in 1981 and depicts the structure and condition of the Old Age and Survi- vors Insurance and Disability Insurance systems at that time. Important provisions of OASDI, affecting both benefit payments and revenues, were changed by the Social Security Amendments passed in April 1983. The discussion and description in this chapter refer to the provisions of the system before the 1983 Amendments. range from 12 to 28 percent of payroll. The range of variation depends on demographic developments (fertil- ity and mortality experience) and economic develop- ments (real wage growth and unemployment levels). Second, expenditures are likely to increase significantly above the current level (about 11 percent of taxable pay- roll). Such a sharp increase may create serious political and economic strains. The impact of the social security system on the U.S. economy and society cannot be expressed in terms of the size of expenditures alone. OASDI may greatly influence patterns of work and retirement, as well as savings and capital accumulation. These patterns, in turn, affect each other, the level and rate of growth of output, and the Social Security system itself. Over the past decade, many studies have investigated the specific impacts of social security on economic behavior, e.g., savings rates, labor force participation, and retirement choice. However, the social security system heretofore has not been modeled in a general equilibrium framework encompassing the com- plete pension system and economy. Use of such a com- plete model may be necessary to understand the interactions among the economy, the social security sys- tem, and the other elements of the pension system. The present model allows detailed analysis of the ef- fects of demographic and economic change on the social security system under different scenarios. The model's ability to examine effects of specific cohorts’ movements through the system is particularly valuable to the forecast- ing of the financial health of a pay-as-you-go system such as social security. Overview The Social Security Model produces the following infor- mation each year for its simulation period, 1970-2055: ® numbers of OASDI beneficiaries, by age and sex and type of benefit; ® average OASI and DI primary and secondary benefits; total OASDI benefits paid, payroll taxes collected from each age-sex group; total OASDI taxes collected; trust fund balances. To calculate these estimates, the model uses the follow- ing information from other parts of the Macroeconomic- Demographic Model: THE NATIONAL INSTITUTE ON AGING MACROECONOMIC-DEMOGRAPHIC MODEL e otal population by sex and age, projected by the Population Model; ® compensation of each of 22 age-sex groups, estimat- ed by the Labor Market Model. The Social Security Model has two basic parts—a sys- tem of relationships that estimates contributions (payroll tax payments) and a system that estimates benefit payments. Payroll tax payments are estimated using projected compensation levels from the Labor Market Model. The methodology is similar to that used by the Social Security Administrations Office of the Actuary to provide long range cost estimates for OASDI.? The estimated total com- pensation of each age-sex group is converted into a hypo- thetical tax base, to which the combined employee- employer tax rate can be applied. This is done by adjusting compensation to estimate covered earnings, tax- able earnings, and the hypothetical tax base. The benefits sector of the Social Security Model esti- mates total annual OASDI benefit payments and average benefits of OASI and DI primary beneficiaries.” First, the model estimates the number and benefit levels of primary beneficiaries in each age and sex group who survive from the previous year. The number of newly retired or dis- abled primary beneficiaries of each age and sex is estimat- ed, and their benefits are calculated. The number of secondary beneficiaries of each age and sex is estimated, and their benefit levels are calculated. All benefit informa- tion is summed to calculate total benefits. The current version of the Social Security Model oper- ates relatively independently from the other models of the MDM—except for use of input variables mentioned above. The model is currently being revised to allow the results of the Social Security Model to influence other behavior in the economy such as labor supply and savings decisions’ and to represent the interactions between pri- vate pensions and social security. “Appendix D lists all variables and equations of the Social Security Model. 3The Actuary is required by law to provide projections of the cost of the OASDI system for a 75 vear period. The term of the Actuary’s 1981 forecast (1980-2055) coincides with that of the Macroeconomic-Demo- graphic Model (1970-2055). The basic methods used by the SSA Office of the Actuary for estimating the long range costs of the OASDI System are described in Steven F. McKay (1980). “A “primary” beneficiary is one who receives a benefit based on contri- butions he or she has made to the system. An individual who becomes disabled may become a primary beneficiary of the Disability Insurance (DI) system if he has sufficient covered earnings. All fully insured work- ers are eligible at age 62 to become primary beneficiaries of the OASI system. Some individuals are entitled to benefits because they are de- pendents of primary beneficiaries. These individuals are designated “secondary” beneficiaries. Their benefits are called secondary benefits. >A summary of the debate about the effect of social security on aggregate savings rates is given in Cartwright (1981). Principal papers are Feldstein (1974), Munnell (1974), Barro (1978), Darby (1978), Leimer and Lesnov (1980, 1981). Contributions (Taxes) The Social Security Model estimates the level of annual contributions (payroll tax payments) into the OASDI Trust Funds each year from 1970 through 2055. Total private sector compensation of each of twenty-two age-sex groups is estimated by the Labor Market Model. To esti- mate contributions, total compensation must be convert- ed to an appropriate tax base. This is done in five steps: 1. Total private compensation is converted to total pri- vate earnings. Total private earnings is converted to total private earnings in covered employment. Total private covered earnings is converted to tax- able private earnings. . Taxable private earnings is adjusted to take into ac- count military and state and local government work- ers covered by social security. . Taxable earnings is converted to an adjusted taxable earnings hypothetical tax base. 2. 3. The model could easily incorporate alternative formula- tions of any one of these steps to produce different esti- mates of OASDI contributions. 1. Calculation of total private earnings. In order to estimate total wage and salary earnings, fringe benefits are subtracted from the estimates of total compensation of 22 each age-sex group. The proportion of total compensa- tion accounted for by fringe benefits has increased stead- ily in recent years, as shown in Table 5-1. Compensation of each age-sex group is converted to earnings by multiply- ing by a conversion factor, F;, which is calculated as follows: F, = .841 Xx e000 where t = 1 in 1980, 2 in 1981, etc. The factor F, decreases by 0.2 percent per year. The rate of decline of the ratio (the negative coefficient on t) was estimated for the Social Security Administration by Robert Russell (1977). The estimated ratio of wage and salary earnings to total compensation in 2055 is .734. 2. Covered employment. Because some private sec- tor workers are not covered by social security, wages and salaries in the private sector must be transformed to total earnings in private covered employment. The ratio of private sector earnings to earnings in private covered em- ployment has been very stable, at about 98 percent, since 1970. A constant coefficient (F, = 0.980) was used for the simulations. 3. Taxable earnings. The next step converts covered private earnings to taxable private earnings. This requires an adjustment in covered earnings for the fact that earn- ings above a specified maximum are not taxable. The maximum was $17,700 in 1978, $22,900 in 1979, $25,000 CHAPTER 5 Table 5-1 Compensation of Employees: Wages and Salaries, and Fringe Benefits, 1970-1979 (Billion Dollars) Supplements to Wages and Salaries Social Security Model Wages and Employer Other Salaries as Wages and Contributions to Labor Percent of Year Compensation Salaries Social Insurance Income Compensation 1970 609.2 546.5 30.1 32.6 89.7 1975 931.1 805.9 60.1 65.1 86.6 1976 1,037.8 890.0 70.4 77.4 85.8 1977 1,156.9 984.0 81.1 91.8 85.1 1978 1,304.5 1,103.3 94.7 106.5 84.6 1979 1,459.2 1,227.6 108.9 122.7 84.1 Source: Survey of Current Business, various issues in 1980, and $29,700 in 1981. After 1981 it is adjusted annually according to the increase in the average wage. An objective of the legislation that established the level and automatic adjustment procedure for the taxable maxi- mum earnings was to establish the maximum at a level such that 91 percent of all earnings would be covered. Adjustment factors in use in this version of the model are based on estimates prepared by the Social Security Ad- ministration, Office of Research and Statistics. These fac- tors appear in Table 5-2. 4. Adjustment for non-private wage earners. Mili- tary personnel and many state and local government per- sonnel participate in the social security system. The private sector tax base that has been estimated by steps 1 through 3 must be adjusted to take into account these public sector workers. Based on 1978 data, an adjustment factor of 1.16 was estimated and is applied to the private sector tax base to get the total tax base. This factor is held constant throughout the forecast period.® 5. Adjusted taxable earnings. To simplify the calcu- lations and to illustrate the financial status of the system, the last step converts the actual taxable earnings base to a hypothetical tax base. Following the convention estab- lished by the SSA Office of the Actuary, we define the hypothetical tax base as that level of earnings which yields the actual level of net contributions when the com- bined employer-employee OASDI tax rate is applied to it. The hypothetical tax base differs from the result of the previous four steps because (1) self-employed individ- uals pay a rate only 75 percent of the combined rate,” (2) “The Social Security Model could be revised to make this factor variable and to take into account the fact that new Federal employees will be covered by'social security after January 1, 1984. "This was changed by the 1983 Amendments. 23 dual job-holders receive refunds on taxes they paid above the taxable maximum, but their employers’ taxes are not refunded, and (3) workers pay a tax on only part of tip income. This conversion factor, estimated from 1972 Social Security and Bureau of Labor Statistics data, is .886. Annual OASDI payroll tax collections are estimated by applying the combined employer-employee statutory tax rates to the adjusted taxable earnings base. Table 5-3 displays historical tax rates and those established in legis- lation as of 1982 for all future periods. These statutory tax rates were not expected to be adequate to pay current statutory benefits over the period of the early 1980's and after 2015 and were changed in 1983. The model is being revised to reflect the new payroll tax schedule and can easily incorporate alternative payroll tax rates or the use of general revenues as a source of program funding. Table 5-2 Estimated Factors to Convert Covered Earnings to Taxable Earnings 1973 8180 1974 8530 1975 8400 1976 .8410 1977 8456 1978 8445 1979 .8861 1980 8949 1981 9066 1982 9082 1983 9093 1984 9110 1985 9132 1986 9136 1987 9135 1988 9140 1989 9150 and thereafter THE NATIONAL INSTITUTE ON AGING MACROECONOMIC-DEMOGRAPHIC MODEL Benefits The Social Security Model estimates the level of OASI and DI benefits paid annually from 1970 through 2055. Total benefits paid in any year are the sum of benefit payments to new primary beneficiaries (retired and disabled), benefit payments in current-payment status to existing primary beneficiaries, and benefit payments to secondary beneficiaries. Total new benefits are calculated by estimat- ing the population of new primary beneficiaries by age and sex and calculating average awards for these new beneficiaries by age and sex. The model calculates exist- ing primary benefits in current-payment status by estimat- ing the population of currentpayment primary beneficiaries by age and sex and calculating average bene- fit payments by updating previous current-payment bene- fits. The model calculates total benefit payments to secondary beneficiaries by estimating the number of re- cipients of each type of secondary benefit and the average benefit of each type. This section describes how these tasks are performed. 1. The Population of Primary Beneficiaries The model calculates a primary beneficiary population by sex and single year of age from ages 31 through 84, and for ages 85 and over. It does this using an incidence rate method.® The disabled population of each sex and age from age 31 through 61 is calculated each year as the sum of the previously disabled who remain disabled and the newly disabled. The population of previously disabled is 8The “incidence” or “stock-and flow” approach, and the “prevalence” or “state” approach are alternative methods of estimating a subpopulation with a given characteristic from a population that could potentially have that characteristic. The incidence method uses incidence rates, which are transition probabilities, e.g., the probability of becoming retired, or the probability of getting cancer. Using the incidence method, the stock of retirees is calculated by estimating the flow of individuals from work- ing status into retirement and the flow of individuals out of retirement because of death or return to the labor force. The prevalence method uses prevalence rates, which are probabil- ities of being in a certain state or of having a certain characteristic, e.g., the probability of being retired or the probability of having cancer. Us- ing the prevalence method, the stock of retirees is estimated simply by applying the probability of being retired (or the estimated proportion retired) to the relevant population. For example, using the prevalence method: Rf = Ni Xp where: R? = number of retirees age a in year G N? = total population age a in year pd = the proportion of the popula- tion age a in year t that is ex- pected to be retired (the prevalence rate). Note that, given a population with unchanging shares of each age group, prevalence rates can be derived from incidence rates. Table 5-3 1982 Scheduled Combined Employer-Employee OASDI Tax Rates, 1970-2055 Year OAS! DI OASDI 1970 7.30 1.10 8.40 1975 8.75 1.15 9.90 1976 8.75 1.15 9.90 1977 8.75 1.15 9.90 1978 8.55 1.55 10.10 1979 8.66 1.50 10.16 1980 8.66 1.50 10.16 1985 9.50 1.90 11.40 1990 10.20 2.20 12.40 and thereafter Source: 1982 Annual Report of the Board of Trustees of the Federal Old Age and Survivors Insurance and Disability Insurance Trust Funds. estimated by subtracting from the disabled population of the previous year the estimated number who die and the estimated number who return to the labor force. The number of newly disabled is estimated by multiplying the disability insured population of each sex and age from age 31 through 64, estimated by the Population Model, by the appropriate age-sex specific disability incidence rate. The youngest age of the primary disabled population is 31 because, with the exception of the blind, one cannot, under the 1977 amendments, qualify for a disability award prior to that age. The retired population of each age and sex is estimated each year as the sum of the previously retired population who remain retired and the number of new retirees. Age- sex-specific survival rates are applied to the retired popu- lation of each age of the previous year to estimate the number of survivors in the current year. To this group is added the number of new retirees, estimated by applying age-sex-specific retirement incidence rates to the popula- tion of each age and sex: (52) RE = REP X sil + No» x rs where R32 = number of retirees age a and sex s in year t; s*ls = the probability that a retired indi- vidual of sex s age a-1 in year t-1 will survive, to age a; Ns population age a and sex s in year t; rs = probability that an individual age a and sex s in year t will retire (the retirement incidence rate). The incidence rates for retirement were estimated from historical Social Security Administration and Census data and are held constant over the simulation period. The rates were calculated by dividing the number of new retir- 24 CHAPTER 5 Table 5-4 Social Security Retirement and Disability Incidence Rates by Age and Sex Retirement Disability Age Male Female Male Female 31 0.0 0.0 0.0022 0.0018 32 0.0 0.0 0.0022 0.0018 33 0.0 0.0 0.0022 0.0018 34 0.0 0.0 0.0022 0.0018 35 0.0 0.0 0.0030 0.0026 36 0.0 0.0 0.0030 0.0026 37 0.0 0.0 0.0030 0.0026 38 0.0 0.0 0.0030 0.0026 39 0.0 0.0 0.0030 0.0026 40 0.0 0.0 0.0043 0.0039 41 0.0 0.0 0.0043 0.0039 42 0.0 0.0 0.0043 0.0039 43 0.0 0.0 0.0043 0.0039 44 0.0 0.0 0.0043 0.0039 45 0.0 0.0 0.0068 0.0056 46 0.0 0.0 0.0068 0.0056 47 0.0 0.0 0.0068 0.0056 48 0.0 0.0 0.0068 0.0056 49 0.0 0.0 0.0068 0.0056 50 0.0 0.0 0.0118 0.0095 51 0.0 0.0 0.0118 0.0095 52 0.0 0.0 0.0118 0.0095 53 0.0 0.0 0.0118 0.0095 54 0.0 0.0 0.0118 0.0095 55 0.0 0.0 0.0208 0.0154 56 0.0 0.0 0.0208 0.0154 57 0.0 0.0 0.0208 0.0154 58 0.0 0.0 0.0208 0.0154 59 0.0 0.0 0.0208 0.0154 60 0.0 0.0 0.0260 0.0164 61 0.0 0.0 0.0260 0.0164 62 0.1370 0.1200 0.0270 0.0100 63 0.1370 0.1200 0.0230 0.0080 64 0.1370 0.1200 0.0180 0.0060 65 0.2820 0.1560 0.0 0.0 66 0.0470 0.0260 0.0 0.0 67 0.0470 0.0260 0.0 0.0 68 0.0470 0.0260 0.0 0.0 69 0.0470 0.0260 0.0 0.0 70 0.0230 0.0260 0.0 . 0.0 71 0.0230 0.0260 0.0 0.0 72 0.0230 0.0260 0.0 0.0 Source: Retirement rates are estimates. Disability rates from Social Se- © curity Administration, Termination Experience of Disabled Worker Beneficiaries Under OASDI (1979). ees in each age-sex group in 1977° by the group's total population. Disability incidence rates were obtained di- rectly from the Social Security Administration Actuary.'® Retirement incidence rates were not available from the SSA Actuary, because the SSA Actuary forecasts the num- Social Security Administration, Social Security Bulletin, Statistical Sup- plements, 1970-1979. i 19Social Security Administration, Termination Experience of Disabled Worker Beneficiaries under OASDI (1979). > Social Security Model Table 5-5 Disability Termination Rates By Age and Sex Age Male Female . 31 0.0813 0.0627 32 0.0813 0.0627 33 0.0813 0.0627 34 0.0813 0.0627 35 0.0753 0.0605 36 0.0753 0.0605 37 0.0753 0.0605 38 0.0753 0.0605 39 0.0753 0.0605 40 0.0713 0.0597 41 0.0713 0.0597 42 0.0713 0.0597 43 0.0713 0.0597 44 0.0713 0.0597 45 0.0700 0.0388 46 0.0700 0.0388 47 0.0700 0.0388 48 0.0700 0.0388 49 0.0700 0.0388 50 0.0685 0.0510 51 0.0685 0.0510 52 0.0685 0.0510 53 0.0685 0.0510 54 0.0685 0.0510 55 0.0678 0.0451 56 0.0678 0.0451 57 0.0678 0.0451 58 0.0678 0.0451 59 0.0678 0.0451 60 0.0690 0.0406 61 0.0690 0.0406 Source: Social Security Administration. ber of retirees on a prevalence basis. The Actuary forecasts the disability population on an incidence basis. Table 5-4 shows both the estimated retirement incidence rates and the SSA Actuary’s disability incidence rates. If the trend in retirement behavior over the past two decades toward earlier retirement continues, these rates may not be reliable predictors of future experience. At the end of this chapter we suggest further research to im- prove the projections of these rates. Disability is terminated by re-entry into the labor force or By death. The disability termination rates shown in Table 5-5, calculated from data provided by the Social Security Actuary, combine the probability of labor force re-entry and mortality. According to the specification of the model, once a worker initially accepts a social security retirement bene- fit and is assigned a Primary Insurance Amount (PIA), he remains classified as retired and retains that PIA for the remainder of his life. The model does not take into ac- count the small number of workers who resume working Social Security Administration, Zermination Experience of Disabled Worker Beneficiaries under OASDI (1979). THE NATIONAL INSTITUTE ON AGING MACROECONOMIC-DEMOGRAPHIC MODEL after retirement and accumulate sufficient additional cov- ered earnings to change their initial PIAs. Consequently, in the model, retirement termination occurs only through death. The Social Security Model uses the same survival rates as the Population Model, which were estimated by the SSA’s Office of the Actuary. 2. Primary Benefit Levels The model estimates the average new primary benefit awarded each year to workers of each age and sex becom- ing disabled or retiring that year, and the average benefit level in each year for all retired or disabled primary bene- ficiaries of each age and sex. The average benefit award of each group of primary beneficiaries will vary by sex, date of birth, and date of retirement. Two individuals of the same age will have different benefits if they retired in different years. In the computation of the average primary benefit in current payment status of each age and sex for each year, the benefits of workers who retired in different years are averaged, weighting the benefit calculated for each retirement year by the number of individuals who retired in that year. Each year the model first estimates an average benefit award for each age and sex of new primary beneficiaries, then updates the previous average primary benefit level of each age and sex to take into account the new benefit awards. To estimate an average benefit award for each age-sex group of new primary beneficiaries, the model: (1) adjusts prior earnings histories from the Labor Market Model; (2) estimates the value of Average Indexed Month- ly Earnings (AIME); (3) computes the appropriate Primary Insurance Amount (PIA); and (4) calculates the actual an- nual benefit. The calculation of an average benefit award for new retirees of a given age and sex starts from an earnings history obtained from estimates of average earnings levels in prior years, which are generated by the Labor Market Model. Since the earnings level for a given age-sex group in a given year represents the mean of all labor income, but Average Indexed Monthly Earnings (AIME) calcula- tions only recognize earnings below each year’s taxable maximum, the average earnings value for each year must be adjusted downward. The model assumes that the earn- ings of each age and sex follow a log-normal distribution, for which the mean is the Labor Market Model's estimated mean earnings and the coefficient of variation is constant at its 1979 level. By integrating this distribution over the range of taxable earnings in each year of the simulation, the model estimates the average taxable earnings which should enter the AIME calculations. The results of this procedure are conceptually equivalent to the F3 factor used in the contributions calculations. However, a more involved procedure is necessary to estimate benefits than for taxes because the earnings profiles of each age and sex group must be adjusted individually. Next, the model estimates the Average Indexed Month- ly Earnings (AIME). This is done in the manner prescribed by the Social Security Act, including the appropriate use of “old law,” “new law,” or “transitional” formulae.!? A Primary Insurance Amount (PIA) is then computed from the AIME according to the statutory benefit formula. The PIA is the sum of three separate percentages of por- tions of the Average Indexed Monthly Earnings. In 1979 (the first year the formula was in effect), the PIA was equal to 90 percent of the first 180 dollars plus 32 percent of the amount between 180 dollars and 1,085 dollars, plus 15 percent of the amount over 1,085 dollars. The graph of the PIA as a function of the AIME generated by this formula is a series of connected line segments, the slopes of which (190, .32, and .15) diminish as AIME increases. Conse- quently, the PIA is a smaller proportion of AIME as AIME increases. The “bend points” in that formula in 1979 were 180 dollars and 1,085 dollars. Because dollar values in the Macroeconomic-Demographic Model are expressed in 1972 dollars, the “bend points” had to be expressed in 1972 dollars. The 1977 amendments indexed these bend points to the level of average wages. The average wage indices “used in this version of the model are calculated from average annual compensation levels simulated by the La- bor Market Model. Historical average annual compensa- tion levels, shown in Table 5-6, are used for the period 1951 through 1979. The model also uses historical data for the taxable wage base, the maximum amount of an indi- vidual’s wages subject to the payroll tax. These amounts are shown in Table 5-6. PIA’s are adjusted downward or upward for early or late retirement, as prescribed by law. Benefits for persons retiring prior to age 65 are reduced by %, of 1 percent for each month that retirement precedes age 65. Benefits for persons delaying retirement past age 65 are increased V4 of | percent for each month that retirement is delayed.'? Estimates of average benefits calculated according to this procedure are biased upward for several reasons. First, there are no benefit reductions due to the applica- tion of the earnings test. Second, because the current version of the model does not include information about the number of persons in each family receiving benefits, benefit reductions due to the Maximum Family Benefit (MFB) cannot be estimated explicitly. Third, the assump- tion that the average worker earned the average earnings of his age and sex each year of his career may be inaccu- rate. The average earnings variable simulated by the Labor Market Model is the average earnings of all workers with earnings. In actuality, many workers have years with zero earnings. In the recalculation of the AIME, the lowest five years of indexed earnings are dropped. These would, of course, include the years of zero earnings if any. However, a worker who had more than five years of zero earnings would have zero earnings years included in the calcula- tion of his AIME, whereas zero earnings workers are not 12McKay (1979). 3This was changed by the 1983 Amendments. CHAPTER 5 Social Security Model Table 5-6 tration data on average new PIA’s and average benefits over the period 1960 to 1978. The value of the reduction Taxable Wage Base and Average Wages Used in the factor increases with time. One reason for this is that the OASDI Calculations, 1951-1979 increased participation of women in the labor force will {1n1972 Dollars) increase the proportion of families with two primary beneficiaries, causing the maximum family benefit limita- Year Kasih Wage io tion to become more important. The aggregate nature of the model makes it difficult to 1951 6,153.85 4784.89 represent the types of behavior and economic situations 1952 5,940.59 4906.47 that account for the divergence between simulated aver- 1953 5,930.81 5172.06 age PIAs and actual average benefit payments. Each of the 1954 5,705.23 5001.01 problems mentioned above requires the addition to the 0s > ny model of information about the distribution of workers 1957 6531.88 5663.64 and secondary beneficiaries. In Section E we suggest fur- 1958 6,213.02 5434.62 ther work which would improve these benefit estimates. 1959 7,079.65 5687.02 After the calculation of a new primary benefit level for a SA a each age and sex, the current average primary benefit of 1962 © 6694.56 5785.22 each age and sex is updated by averaging the old average 1963 6,629.83 6072.71 primary benefit level and the level of the newly awarded 1964 6,495.26 6192.58 benefits, weighting by the number of recipients of each. 1965 6,290.96 6105.79 Benefits in current payment status are indexed each Ie ay Ct year by the Social Security Administration to reflect in- 1968 9443.10 6745.47 creases in prices as measured by the Consumer Price 1969 8,873.72 6705.07 Index. The current version of the Social Security Model 1970 8,685.97 6888.91 uses real prices and expenditure levels (dollars of con- 2%) ia Sa stant purchasing power in terms of consumer goods).' 1973 10,074.63 7071.04 Thus, no changes occur in the current prices of consumer 1974 11,448.40 6965.10 goods and, therefore, no indexing of benefits is required. 1975 11,576.35 7086.14 The benefit for each current disabled worker or retiree is 1976 11,751.15 7086.39 calculated in the year in which he or she became disabled 0 as oa or retired and remains constant over the remainder of his 1979 14,521.24 7172.96 or her period of disability or retirement. Source: ICF calculations based on Social Security Administration data. taken into account in the Labor Market Model in the calcu- lation of average age-sex earnings. For this reason, the model may overstate the average AIME. On the other hand, a worker who had fewer than five years of zero earnings would be able to drop a low non-zero year. In the model, the five years that are dropped are always the first five years, since with trend growth rates of average earnings the first five are always the lowest. If there is, in fact, considerable fluctuation in the (non-zero) earnings of a large number of workers, the dropped years may on average be lower than the first five years. In that case, the model may tend to understate the average AIME, but this effect is unlikely to be important. To correct the upward bias, the estimated new average benefit for each group in each year is reduced by an estimated adjustment factor, BENDEC, calculated as follows: BENDEC = .954 — .0031(YEAR) where YEAR equals 1 in 1960, 2 in 1961, etc. (53) BENDEC was estimated from Social Security Adminis- 3. Numbers of Secondary Beneficiaries The model estimates numbers of secondary beneficiaries by benefit type. Eleven secondary benefit types are modeled: ® Dependents of Retirees: — Aged spouse — Child — Spouse with child ® Dependents of Disabled Workers: — Aged spouse — Disabled spouse — Child — Spouse with child ® Survivors: — Aged widow(er) — Child — Parent — Widow(er) with child. "That is, the average price of consumer goods is the numeraire. Rela- tive prices can change in model simulations, but the average price of consumer goods remain constant. THE NATIONAL INSTITUTE ON AGING MACROECONOMIC-DEMOGRAPHIC MODEL Secondary beneficiaries are not independently mod- eled by age and sex. The model uses the forecasts of the number of secondary beneficiaries provided by the Social Security Actuary’s long-range forecasting methodology.'® The Actuary forecasts secondary beneficiaries quinquen- nially. The model linearly interpolates the intervening years. 4. Average Secondary Benefits The model estimates average secondary benefits by sec- ondary benefit type. The model uses average primary re- tirement benefit levels to estimate average secondary benefit levels for dependents of retirees. Similarly, it uses average primary disability benefit levels to estimate aver- age secondary benefit levels for dependents of disabled persons. Survivor’s benefits are increased from their 1970 level at the same rate that average retirement benefit lev- els increase. Each secondary benefit is set by law as a proportion of its corresponding primary benefit. Those proportions are shown in Table 5-7. The appropriate primary benefit (re- tirement or disability) is multiplied by the percentage specified by law to estimate the average secondary benefit for each of the eleven secondary groups. Table 5-7 Secondary Benefit Levels As Percent of Primary Benefit By Benefit Type OASI—Dependent of Retiree: 50% Aged Spouse Child 50% Spouse with Child 50% OASI—Survivors: Aged Widow(er) 100% Disabled Widow(er) 71.5% Child 75% Widow(er) with Child 75% Parent 75% DI Child 50% Spouse with Child 50% Aged Spouse 50% Source: William M. Mercer, Inc, “Old Age, Survivors and Disability Benefits Under the Social Security Act,” 1981. SMcKay (1980). Future Work J The Social Security Model represents the basic features of the OASDI system within an integrated economic-demo- graphic model. It depicts the determination of tax collec- tions and benefit payments and how economic and demographic change affects both. It is useful for forecast- ing the effects of economic and demographic change on the social security system and for analyzing the effects of ~ alternative policies. IRQ In three areas additional work could potentially im- prove the richness of the model and its simulation capabilities. 1. Improvement of the Behavioral Representa- tion of Retirement and Disability and Integra- tion with the Labor Market Model Presently, the retirement and disability incidence rates are exogenous and constant. We recognize that economic and demographic factors will tend to influence these rates. We are currently analyzing the relationship between these rates and other economic and demographic variables. Work is in progress to model the interaction between social security benefit levels, the labor force participation and employment of the elderly, the acceptance of social security benefits and wages, income, and other aspects of labor market behavior. 2. Alternative Techniques for Modeling Secondary Beneficiaries As mentioned earlier, the model's principal purpose is to depict the influence of alternative economic/demogra- phic scenarios on the social security system. Secondary beneficiaries account for about thirty percent of benefit payments and are one of the most volatile factors in the OASDI system. For instance, high female unemployment probably increases the number of secondary (wife) bene- ficiaries. This affects OASDI two ways: the unemployed wives do not contribute taxes, and the additional secon- dary beneficiaries increase benefits. Currently, we use estimates of the number of secondary beneficiaries provided by the SSA Office of the Actuary. The model's richness and accuracy might be improved if the determination of secondary beneficiaries by age and sex was represented in the same way that the determina- tion of primary beneficiaries is. We recognize that a complete modeling of the number of secondary beneficiaries would be a large scale project. While primary beneficiary status results mainly from one of two events—disability or retirement—secondary bene- ficiary status can arise from a variety of relationships to primary beneficiaries and can cease because of a variety of changes in status. CHAPTER 5 Social Security Model 3. Representation of the Effects of the Social the potential feedback effects of social security on the Security System on the Economy economy are represented in the current version.'® Poten- tial linkages have been developed. The richness and com- As noted above, the social security system may affect labor pleteness of the model will be increased significantly by force participation of various age-sex groups, capital accu- analysis, estimation, and modeling of the affects of social mulation, or other aspects of the economy. Only a few of security on economic behavior. For example, the social security trust fund is incorporated in the private sector wealth accounts in the Macroeconomic Growth Model and thus affects savings, investment and labor input. 29 a A Lr Er Chapter 6 The Private Pension Model Introduction he private pension system is a major component of the U.S. retirement income system. Almost ten million people received benefits totaling 29 billion dollars from private pensions in 1980, and 35 million workers participated in a private pension plan. Gross contributions to private pen- sions were about 70 billion dollars in 1980, and private pensions held about 500 billion dollars in reserves at the end of 1980—17 percent of the total financial assets of households.! In 1950 private pension fund reserves to- talled 12 billion dollars, about three percent of household financial assets. Consequently, the private pension system has gained importance not only for the elderly but also for the country’s financial sector and capital markets. Individuals may formulate their retirement income plans in a life cycle context. Changes in pension policy, therefore, may affect young workers as well as current retirees. Similarly, the full effects on the pension system of economic and demographic changes are experienced only in the long-run. For example, in the 1990s when the post-war “baby boom” cohorts begin entering middle age, their members will begin to accumulate considerable capital, both through pensions and other forms of savings for retirement. This will substantially affect the capital markets. When these large cohorts retire several decades later, the subsequent dissaving may also have major ef- fects on the capital markets and on the economy. In order to model the effects of changes in today’s pension policies, therefore, we must use a simulation period ex- tending into the middle of the twenty-first century. Overview The Private Pension Model produces the following infor- mation for each year of its simulation period, 1970-2055: ® The number of workers covered, participating, and vested, in each of three private pension plan types (defined benefit, defined contribution, individual plan); e Total contributions for each of the three pension plan types; Total pension fund assets in 1980, including public employer plans, equalled 771 billion dollars, 26 percent of household financial assets. (Board of Governors of the Federal Reserve System, Flow of Funds Accounts, 1981.) See Wachtel (1980) for a discussion of this phenomenon. 3Appendix E lists all equations and variables of the Private Pension Model. 31 ® The number of retirees, by age, sex, and pension status (the three pension plan types, or no pension); ® Total benefit payments for each of the three pension plan types; ® The average benefit per retiree by age, sex, and num- ber of years retired, for each of the three pension plan types; and e The level of assets held by each plan type. For each simulation year, the Private Pension Model uses information from the Labor Market Model concerning: ® Average wages, by age and sex; e Total numbers of workers, by age and sex. For simulations of the 1970-2055 period, the Private Pension Model uses the following initial inputs: e Values in 1969 of the number of retirees, total pen- sion assets for each plan type, and unfunded liabili- ties of defined benefit plans; e Parameters, such as retirement rates, life expectan- cies, mortality rates, coverage rates, participation rates, vesting rates, job change rates, and real rates of return on pension investments. Estimates of private pension coverage, participation, and vesting are reported for 20 age-sex groups. The ages are: 16-17, 18-24, 25-34, 35-44, 45-54, 55-61, 62-64, 65-67, 68-71, 72 and over. Estimates of other variables are carried out for single years of age. In a funded defined benefit pension plan, benefits are financed by two sources—the assets that have been accu-’ mulated through prior years’ contributions, and current contributions. The role of the two depends on the level of funding, the maturity of the plan, the age of the workforce, and other factors. In defined contribution and individual plans, benefits are provided exclusively by the assets accu- mulated through past contributions. The Private Pension Model depicts the contribution, asset accumulation, and benefit payment processes for all three types of plans. Because the model aggregates all private plans into only three types, this process is represented at a high level of aggregation for each type of plan. The Private Pension Model performs five tasks: determi- nation of the covered, participating, and vested working population; calculation of the contributions made for them; estimation of the population of retired beneficia- ries; calculation of the benefits paid to them; and determi- nation of the resultant pension asset levels. THE NATIONAL INSTITUTE ON AGING MACROECONOMIC-DEMOGRAPHIC MODEL Coverage, Participation, and Vesting The Private Pension Model first estimates the population by age and sex of workers who are covered by a pension plan (work for an employer who sponsors a plan), and those not covered by a pension plan. It apportions cov- ered workers among either a defined benefit, a defined contribution, or an individual (IRA/Keogh)* plan. It then determines by age and sex how many of these covered workers are plan participants. Finally, it determines how many of these plan participants are vested in each type of plan. For example, the number of covered, participating, and vested workers age a and sex s are estimated using the following equations: Cast = Wasi X PC, (6-1) Past = Cast x PP as (6-2) Vasu = | X PV, (6-3) . where Cosi = number of covered employees of sex s, age a, in year t, W, se = number of employees of sex s, age a, in year t; Poss = number of participating em- ployees of sex s, age a in year t; Vast = number of vested workers of sex s, age a in year t; PC, = proportion of employees of sex sand age a who are covered by a private pension plan; PP, = proportion of covered employ- ees of sex s and age a who are participating in a pension plan; PV, = proportion of participating em- ployees of sex s and age a who are vested in a pension plan. Coverage A covered worker is usually defined as a worker whose employer sponsors a pension plan and whose job classifi- cation allows him/her to participate after minimum age and service criteria are satisfied. Due to data limitations, however, the model defines a covered worker somewhat differently: in the model, covered workers are all those whose employer sponsors a pension plan. The model uses a “prevalence” rate approach to esti- “There are two typical types of employee-initiated private pension plans- IRAs and Keoghs. At the time of this research, an Individual Retirement Account (IRA) allowed a person who is self-emploved or emploved at an establishment which does not have a pension plan to maintain one for himself and provided tax advantages. Individuals are now no longer restricted if covered by a plan at work. A Keogh plan is intended for the self-employed and has a larger maximum contribution level. 32 mate the number of covered workers.> This approach is well-suited to the aggregate design of the model and pro- vides great flexibility in examining alternative trends in future pension coverage. Age-sex-specific private pension coverage rates are applied to estimates of the labor force by age and sex provided by the Labor Market Model. For each age-sex group, the coverage rate is equal to the ratio of covered private workers to total private part-time and full-time workers in the Special Pension Supplement to the May 1979 Current Population Survey (CPS).° This in- cludes self employed workers as well as wage and salary workers. (See Table 6-1.) Those workers not covered by one of the three types of plans go into the “not covered” category. Although aggregate private pension coverage has in- creased significantly in the past, it has increased at a de- clining rate. Consequently, there is some uncertainty about how much coverage will change in the future. Sev- eral studies both in and outside the government have concluded that private pension coverage and participa- tion may not increase substantially in the future.” Howev- er, these studies have examined aggregate, not age-sex- specific rates. Because it is not clear to what extent age-sex coverage rates will change in the future, this specification of the Private Pension Model assumes no change in these rates over time. This does not represent our estimate of what future trends in age-sex-specific coverage rates will be, but rather our recognition of the great degree of uncer- tainty concerning these trends. Given the model's flexibil- ity, alternative assumptions can easily be incorporated. Even with constant age-sex coverage rates, the aggregate coverage rate can be expected to climb somewhat as the “baby boom” cohorts enter older age groups having rela- tively high coverage rates.” The workers who are covered are apportioned among the three types of pension plans according to proportions shown in Table 6-2. In this initial version of the model, all age-sex groups were distributed in the same proportions. These proportions were estimated from ICF analysis of data reported by pension plans to the Department of Labor.” See Chapter 5, footnote 8 for a discussion of the prevalence rate and incidence rate approaches to estimation of a subpopulation with a given characteristic. “The results obtained from a special survey commissioned by the Presi- dent's Commission on Pension Policy for coverage (Table 6-1) and participation (Table 6-3) were similar to these CPS values. “Studies by Beller (1980), Rogers (1979), Greenough and King (1976), President's Commission on Pension Policy (1981) come to this conclu- sion. A study for the Employee Benefit Research Institute (Schieber and George, 1981) concludes that private pension coverage will increase substantially in the future. 8A version of the model is now being developed at ICF Incorporated that permits the user to specify future trends in coverage and participation. °ICF Incorporated, A Private Pension Forecasting Model, (1979). CHAPTER 6 Previous ICF analysis of trends in pension plan cover- age indicate that, following the passage of the Employee Retirement Income Security Act (ERISA), there was a mea- sureable shift toward the formation of defined contribu- tion plans. Although this shift may have diminished somewhat in recent years, there is some evidence that this may be a long-term trend. Although the model assumes that the initial distribution of workers by plan type will remain constant, alternative assumptions for the trends in plan formation may be employed. Currently, available data do not permit identification of a clear long-term, post-ERISA trend. Participation A worker participates after satisfying the participation re- quirements of at least one pension plan. Estimates of the number of participants by age and sex are generated using a prevalence rate method. Age-sex specific participation rates are applied to estimates of the covered population to produce estimates of participants by age and sex. The participation rates were estimated from ICF analysis of the Special Pension Supplement to the May 1979 CPS and are assumed to be constant through time.!° These rates are shown in Table 6-3. Assuming no changes in ERISA partici- pation rules and constant coverage rates, the specification of time-invariant participation rates seems reasonable. However, if ERISA participation standards were to be al- tered, participation rates would be likely to change. While these rates reflect 1979 post-ERISA experience, they were also used to simulate the pre-ERISA years, 1970-1973. The same participation rates are used for the defined benefit and defined contribution plans. Because “individ- ual” plans are almost all self-elective, we assume that 100 percent of the individuals covered by a plan also participate. { Vesting A worker is vested if he/she has a non-forfeitable right to employer financed pension benefits. This definition in- cludes both partially and fully vested workers. The model estimates the number of workers vested in a pension plan by age, sex, and pension plan type using the prevalence rate approach. The model applies an age-sex-specific vesting rate to the population of workers of each age and sex participat- ing in a pension plan. The vesting rates for the category of individual plans are, by definition, equal to 1.0 for all ages and sexes. The rates for the defined benefit and defined “These rates were estimated from Current Population Survey data by dividing the number of workers reported to be participants by the number reported to be covered. Consequently, any discrepancies due to differences in the definition of covered workers in the survey data are accounted for in the forecast numbers of participants. 33 Private Pension Model Table 6-1 Private Pension Coverage Rates By Age and Sex Age Male Female 16-17 404 406 18-24 404 406 25-34 594 524 35-44 .622 .487 45-54 .609 499 55-612 .606 437 62-64? .606 437 65-67° 262 270 68-71° 262 270 72+° 262 270 aData are for ages 55-64 bData are for ages 65+ Source: ICF analysis of May 1979 CPS. Table 6-2 Distribution of Workers Covered By A Private Pension Plan, By Plan Type, 1975 Type of Plan Proportion Defined Benefit 61 Defined Contribution 31 Other .08 Total 1.00 Source: ICF analysis of the 1975 DOL Form EBS-1 filings. Table 6-3 Proportion of Workers Covered By A Private Pension Plan Who Are Participating in the Plan, By Age and Sex Age Male Female 16-17 126 105 18-24 .620 498 25-34 .889 .761 35-44 939 817 45-54 957 .860 55-58 975 .885 59-61 954 915 62-64 915 867 65-68 .768 753 69-71 .507 .769 72+ 715 618 Source: ICF analyses of the May 1979 CPS. THE NATIONAL INSTITUTE ON AGING MACROECONOMIC-DEMOGRAPHIC MODEL contribution plans were estimated using data from the Special Pension Supplement to the May 1979 CPS. Be- cause this survey contains no information on the type of pension plan, we assumed identical vesting rates in de- fined benefit and defined contribution plans. These are shown in Table 6-4. These rates are held constant over the simulation period. As with participation rates, the specifi- cation of constant vesting rates over time assumes ERISA vesting standards remain unchanged. Contributions The Private Pension Model estimates total annual contri- butions to each of the three types of plans (defined bene- fit, defined contribution, individual) for each year of the simulation period. For each of the three types of plans, the model determines the average contribution for a partici- pant of each age-sex group, multiplies that by the number of participants in each group, and sums the products over all the ages and sexes. The main task is to determine the average age-sex-specific contributions for each of the plans. Average Contribution Calculation for Defined Benefit Plans In a defined benefit pension plan, the benefit paid upon retirement is determined by a benefit formula. In a fully funded plan, contributions must be made to the plan during an employee’s career, so that at retirement, suffi- cient assets have been accumulated to pay the benefits. The contributions made to such a plan on behalf of each employee each year are determined by the employee’s age, tenure, salary, the pension plan’s benefit formula, the actuarial costing method used by the plan, and actuarial assumptions concerning future wages, turnover, age at retirement, life expectation, interest rates, etc. The estimated cost of accumulating assets sufficient to pay the expected future benefit accrued by an employee in a given year is called the “normal cost” of the pension for that worker. The sum of the normal cost for each of the participants is the normal cost of the plan. ERISA requires that every employer with a defined benefit plan contrib- ute annually an amount equal to the plan’s normal cost and amortize unfunded liabilities over a prescribed peri- od of time."! The model determines contributions for each worker in the defined benefit plan by calculating the actuarial normal cost for that worker.!? The normal cost is deter- mined by the actuarial funding or costing method and the assumed prototype benefit formula. "Unfunded liabilities are discussed on page 6-14 below. 12For a more detailed discussion of pension funding, see Howard Wink- levoss (1976). 34 Table 6-4 Proportion of Workers Participating in A Private Pension Plan Who Are Vested, By Age and Sex Age Male Female 16-17 399 441 18-24 284 331 25-34 481 482 35-44 729 .620 45-54 837 724 55-58 889 839 59-61 .869 872 62-64 879 .821 65-68 883 916 69-71 989 862 72+ 815 918 Source: ICF analysis of the May 1979 CPS. The benefit formula applied to the model's defined benefit retirees is a career average unit benefit formula. The annual benefit is specified to be equal to one percent of the average salary over the worker’s career for every year worked. For example, a worker retiring with a job tenure of 30 years and an average annual salary of $10,000 would receive 30 X 10,000 X 1% = $3,000 annually as a pension benefit. Alternative benefit formulae can be easily incorporated into any simulation. An actuarial funding method well suited to calculating normal costs for this type of benefit formula is the Ac- crued Benefit Cost Method. The Accrued Benefit Cost Method is sufficiently complex that only the essentials of the method are described here. The method requires assumptions about normal retirement age, mortality, job change rates, and interest rates (all discussed below). There are three major steps in calculating required contributions under the Accrued Benefit Cost Method. First, one determines the expected cost at retirement (present value) of the stream of future benefit payments accrued by the worker in the current year that will begin when the worker retires and end when the worker dies. Second, one discounts that cost from the expected year of retirement to the year in which the contributions are being made. Third, one adjusts that cost to reflect the possibility that the worker will not retain the job long enough to be vested or will die before retirement. The first step—determination of the cost at retirement of the stream of benefit payments—is a simple actuarial discounting calculation. This is the present value ("pre- sent” in the first year that benefits are received) of the stream of payments, i.e., the purchase price of an annuity with that payment per year. Calculation of this present value requires information concerning annual benefit amount, interest rate, and life expectancy. The second step converts that amount into a present value in the year of the service for which the contribution is being made. The present value of the annuity in the year of retirement is discounted to the present value in the current year. CHAPTER 6 The third step adjusts the present value calculated in step two for expected job change and mortality. If every worker remained on a job long enough to be vested and lived to retirement age, the calculation of the normal cost would be completed after step two. However, some work- ers will lose accrued benefits, either because they leave before they are vested or because they die before they retire.'> Consequently, the average contribution can be reduced to take into account the fact that not every worker for which a contribution is being made will collect a bene- fit. The normal cost under the Accrued Benefit Cost Meth- od is then computed. This method requires assumptions concerning mortal- ity rates, job change rates, rates of return, and annual wages. The mortality assumptions used in the current version of the model are from the UP-1984 table.!* The turnover assumptions used are from the T-6 turnover ta- bles.!> A constant real rate of return on pension assets of 1.85 percent is used.’® A normal retirement age of 65 is assumed. The model assumes that all persons in an age- sex group receive the average annual compensation esti- mated for that group by the Labor Market Model. The model could incorporate alternative assumptions for any of these factors. The experience of a defined benefit plan sometimes differs from the probabilistic and behavioral assumptions used to calculate normal costs. For example, a plan might have contributed in 1940 for someone retiring in 1980. In 1940, average lifespans were shorter than in 1980. Hence, insufficient contributions may have been made in 1940 for 1980's retirees, due to the incorrect mortality assumption. If experience differs from assumptions so as not to favor the plan, an unfunded liability will arise.!” Unfunded liabi- lities also can arise if the plan increases benefits for prior service or increases benefits for those already retired. We do not attempt to project new unfunded pension liabilities after 1980. However, unfunded liabilities exist as of 1980 and must be funded. Under ERISA plans must amortize their post-ERISA unfunded liabilities over no more than thirty years (forty years for multi-employer plans).'® We estimated the additional annual contribution 3The current version of the model does not provide for a joint and survivor option. l4Fellers and Jackson (1975). 157-6 table from Crocker et. al. (1955). Ibbotson and Sinquefield (1979). "To define “unfunded liability” it is useful first to define another term, “actuarial liability”. Actuarial liability is the excess of the present value of future benefit payments over the present value of future normal cost contributions. Unfunded liability is the difference between actuarial liability and current assets. 'SERISA requires plans to amortize past service liabilities over no more than 30 years (40 years for multi-employer plans). Experience gains and losses must be amortized over no more than 15 years (20 years for multi-employer plans). At present the model assumes multi-employer plans fund unfunded liabilities at the same rate as single employer plans. The Multi-employer Pension Plan Amendments Act has recently modified the funding standards applicable to multi-employer plans. 35 Private Pension Model required to amortize over thirty years the estimated value of unfunded liabilities in 1980.!° The model adds this amount to the defined benefit plan’s normal cost contri- butions every year from 1981 to 2010. Average Contribution Calculation for Defined Contribution and Individual Plans A defined contribution pension plan is one in which the contribution rate or level is established by the pension plan provisions, while the benefit can vary. The contribu- tion formula used by the model to simulate defined con- tribution and individual plans is a specified percent of total compensation. For the simulations reported in this monograph, the assumed contribution was specified to equal 8.4 percent of the annual compensation of each worker who participates in a defined contribution or indi- vidual plan. This estimate is based on analysis of pension plan data reported to the Department of Labor.?° While the model could incorporate variations in this parameter by age or over time, the current formulation holds the parameter constant at 8.4 percent for all defined contribu- tion participants throughout the simulation. As in the estimation of contributions to defined benefit plans, all workers in each age-sex group are assumed to earn the average compensation of that age-sex group. Contributions are assessed at that rate for each worker beginning at age 34 and continuing until his or her retire- ment. A real rate of return on pension assets of 1.85 per- cent is assumed. After the contributions for each age-sex group for each of the three plans are determined, each plan’s contribu- tions are summed to produce total annual contributions for each plan. Estimation of Recipient Population The model estimates the population of private pension benefit recipients by age, sex, and pension plan status, using an incidence rate approach.?! First, it estimates the number of new recipients by age and sex. Second, it ages the recipient population of the prior year one year and reduces that population due to mortality. Third, new re- cipients are added to the surviving population of previous recipients to estimate the recipient population at the end of the period. “Unfunded liabilities in 1980 were estimated as S50 percent of plan assets. The 50 percent estimate is from an analysis of Fortune 500 firms by Johnson and Higgins (1979). 29ICF Incorporated, A Private Pension Forecasting Model (1979). 2The model defines a recipient as someone who has accepted a private retirement pension benefit. This is the appropriate definition for this model, as the number of recipients is used to determine total benefits paid by pension plans. THE NATIONAL INSTITUTE ON AGING MACROECONOMIC-DEMOGRAPHIC MODEL Estimation of New Recipients The number of new private pension benefit recipients each year for each age-sex group is estimated, and these recipients are distributed among the three pension plan alternatives: (1) defined benefit plan, (2) defined contri- bution plan, and (3) individual plan. Although we know that some retirees receive benefits from more than one plan, we do not currently have sufficient data to model multiple benefit receipt across different plan types. The approach of the Private Pension Model is to assign an individual to that plan which provides his largest, or pri- mary, private pension benefit. Use of a simple distribution of recipients across individual plans is consistent with that approach. Retirement benefit acceptance rates for each cohort are developed in a two step process. First, the proportion of the cohort that will ultimately receive a private pension benefit is estimated. Second, the model estimates the dis- tribution of benefit acceptance decisions over single years of age, 55 to 72. That is, it estimates the conditional prob- ability that a member of the cohort will accept a private pension benefit at each age from 55 to 72, given that he or she will ultimately accept a benefit. The product of the ultimate acceptance proportion calculated in step one and the elements of the distribution estimated in step two provides the age-sex specific benefit acceptance rates. Information about cohort retirement behavior was de- rived from the Special Pension Supplement to the May 1979 CPS. From it we estimate the number of persons 72 years or older receiving a pension benefit. We assume that all persons eligible for a private pension have claimed it prior to age 72. It is unreasonable to assume private pension benefit acceptance behavior will remain unchanged in the future. Today's rapidly increasing private pension beneficiary population reflects the rapid growth in private pension coverage since World War II. The proportion of the elder- ly population receiving a private pension benefit grew about 5 percent annually during the 1970’s.2% This growth rate is approximately equal to the growth rate of the pen- sion coverage rate during the 1950’s—S5.5 percent per year. Since trends in pension coverage are known through 1980, we use these trends to predict benefit ac- ceptance rates twenty years later. Specifically, we assume that the percentage of each co- hort reaching age 55 which will eventually receive a pri- vate pension grows at an annual rate of 5.5 percent from 1970 to 1979, 1.3 percent from 1980 to 1995 and is con- stant thereafter. The average annual growth in pension coverage was 1.5 percent annually between 1960 and 1975. 22Retired population data from Pension Facts, 1980, total population data from the CPS, and assumed mortality of 4.3 percent per year for the retired population were used to develop aggregate retirement inci- dence rates. 36 We then estimate the cohort-specific benefit acceptance rates that are consistent with the ultimate cohort benefit acceptance rates. We estimate the probability of accepting a private benefit each year from age 55 to age 72, using data based on the experience of a large corporate pension plan. Although this particular plan is a defined benefit plan, we use its experience for defined contribution and individual plans as well because comparable data for these plans were not available. Finally, the recipient popu- lation is distributed over the three types of plans in the same proportion as the population of active recipients. Current Retirees The model estimates the number of current retirees by age, sex, and plan status by aging the population of the ‘previous year’s retirees one year, and reducing each age- sex-pension status group’s population by a factor to reflect mortality. The mortality rates are from the Bureau of the Census and are the same as those used in the Population Model and in the Social Security Model. Estimated current and new retirees in each age-sex-pension status group are summed to yield total retirees in each age-sex-plan status group. Benefits The Private Pension Model calculates average benefits by age, sex, pension plan type, and date of retirement; total benefits paid by each of the three types of plans; and average benefits per retiree paid by each type of plan for each simulation year. It estimates the average benefit re- ceived by a recipient of each age, sex, year of retirement, and type of plan, then multiplies that estimate by the number of retirees of that age, sex, retirement date, and plan. It sums across all retirement dates, ages and sexes to produce estimates of total benefits paid per plan. Finally, it divides each plan’s total benefits by its total retirees to calculate average benefits in each plan. A retiree’s benefit in real terms in each of the types of plans is determined when he or she retires. The retiree retains that real benefit for the remainder of his/her life. The benefit will vary by year of birth, sex, and year of retirement. Each year the current benefits for all earlier retirees have been determined in previous simulation years.?> The major task of the benefits component is to compute benefits each year for each age-sex-plan cate- gory of new retirees. Benefits of the defined benefit plan are calculated differently from those of defined contribu- tion and individual plans. 23To determine estimates of the stock of retirees and current benefits for 1969, a simplified version of the model was run. Input data used for this version were actual workforce and compensation data over the period 1951-1969. CHAPTER 6 Calculation of Benefits for the New Retirees in Defined Benefit Plans The defined benefit formula is one percent of career aver- age compensation for every year in the plan. The Labor Market Model estimates average compensation for the members of each age-sex cohort each year of their work- ing lives. At the end of their working lives, an age-earnings profile has been created for each cohort. At retirement, total career compensation is calculated by summing the average compensation earned in each year of participa- tion in the plan. Total career compensation varies by sex, date of birth, and date of retirement. One percent of this total is the average annual pension benefit for the mem- bers of that age-sex group who retire in that year. The plan participation tenure of all new retirees aged 55 to 65 was assumed to be 21 years (the last 21 years before retire- ment).?* The computation of contributions to both de- fined benefit and defined contributions plans is based on the same tenure assumption. Calculation of Benefits for New Retirees in Defined Contribution and Individual Plans The model calculates benefits for each age-sex group of new retirees for both defined contribution and individual plans in the same way. Benefits in defined contribution and individual plans are not determined by a benefit for- mula, as in defined benefit plans. They are determined by the value of career contributions plus interest accumulat- ed at retirement. The stock of accumulated contributions plus interest for each retiree is converted into a life annu- ity, which then provides a constant stream of benefit pay- ments (similar to the benefit payments from defined benefit plans). The level of these annual benefit payments depends on the value of the annuity (in this case, the value of the accumulated contributions plus interest), an assumed rate of interest, and life expectancy assumptions. The model assumes a real interest rate of 1.85 percent annually and uses the life expectation assumptions implicit in the Cen- sus Bureau mortality rates used in the Population Model. The value of accumulated contributions is determined by assuming that throughout each retiree’s work history he or she has had an amount equal to 8.4 percent of his/her annual compensation contributed to a pension account, and that account has been accumulating interest at the rate of 1.85 percent compounded annually. Such an account is constructed by the model for the members of each age- sex group, and the sum of accumulated contributions at retirement is converted into an annuity, as described above. 24The tenure assumption is based upon an analysis of tenure data for all workers age 65 from the May 1979 CPS. 37 Private Pension Model After the average age-sex specific benefits of new retir- ees are determined, the model sums the benefits for each plan across ages, sexes, and years of retirement of all retirees to produce total benefit estimates for each plan. These are divided by the total number of retirees in each plan to determine average benefits per retiree in each plan. Pension Plan Asset Calculation The Private Pension Model calculates total assets held by each of the three types of pension plans each year of the simulation period. This is done by adding to the value of the previous year’s assets the interest on those assets and the year’s total contributions and subtracting the year’s total benefit payments. Interest on one-half the difference between total contributions and total benefit payments is also added to adjust for the fact that, while assets are accumulating continuously, the model calculates asset changes and interest only once annually. The formula for each plan is: ASSETS, = (1 + r) X ASSETS, + (TC — TB) xX (1+ 1/2) (64) value of assets in year t interest rate total contributions in the current year total benefits in the current year. where: ASSETS, r TC TB Il The real rate of interest is assumed to be 1.85 percent annually. : Data on pension plan assets in the initial simulation year are obtained from two sources. Data for total assets for all three plans in 1969 are from the Federal Reserve Board's Flow of Funds accounts. These assets are appor- tioned among defined benefit and defined contribution plans according to the estimated coverage levels shown in Table 6-3. Future Work The Private Pension Model is a simplified representation of the private pension system, integrated with the long- term model of the labor market and the economy. The Private Pension Model depicts both the actuarial structure of the pension system and how the behavior of that system is affected by demographic and economic change. It can trace the long-term development of the pension system and how the system may be affected by alternative pen- sion policies or economic policies. In several areas addi- tions or enhancements to the model would permit it to provide a richer or more accurate representation of the private pension system. THE NATIONAL INSTITUTE ON AGING MACROECONOMIC-DEMOGRAPHIC MODEL Trends in Defined Benefit, Defined Contribution and Individual Plan Coverage The model currently distributes participants between de- fined benefit, defined contribution, and individual plans in a constant proportion. The estimated distribution is based on 1975 data. It assigns each participant to one plan only. Since 1975 there has been a trend in new plan forma- tion away from defined benefit toward defined contribu- tion plans. Individual plans have also grown significantly. While the trend toward defined contribution plans has decelerated recently, it is expected to continue. In light of the uncertainty about future trends, we did not incorpo- rate a trend in the version of the model used to produce the simulations reported here. A simple time trend could be estimated. However, the data about the numbers of workers covered by defined benefit and defined contribu- tion plans is incomplete. In light of our uncertainty about recent trends in the distribution of workers over plans and about how those trends will develop in the future, we choose to use a constant estimate. It would be easy for the user to specify a trend in the distribution of participants by plan type. Many workers are now covered by more than one plan. That phenonomen is not captured in the model. For sim- plification, we allocated each worker to a single type of plan and took account of multiple benefit receipt by ad- justing average benefit levels. By providing information about the various combinations of full and part-time work and social security and pension benefit receipt, workers could be systematically allocated to more than one pen- sion plan type. As more data become available about the types of pension benefits received by workers, this part of the model can be improved. : Trends in Coverage and Participation The model uses constant age-sex specific coverage and participation rates, based on 1979 CPS data, for the simula- tion of future years. Aggregate private pension coverage and participation vary, therefore, as the age-sex composi- tion of the labor force varies. However, there may also be trends in the age-sex rates themselves. Unfortunately, data on age-sex specific coverage and participation for other years that are comparable to the 1979 CPS data are un- available. We are investigating the possibility of estimating trends in age-sex specific coverage rates that would be consistent with projected trends in aggregate coverage. Unfortunately, there is not widespread agreement con- cerning future trends in aggregate coverage. Retirement and Pension Benefit Acceptance Retirement is influenced by the availability of social secu- 38 rity and private pension benefits, and benefit acceptance, in turn, is closely related to retirement. Those relation- ships are not well articulated in the current version of the model. Research is now underway to link explicitly the determinants of labor force participation in the Labor Market Model, retirement, and benefit acceptance in the Social Security Model and in the Private Pension Model. This research may permit a more accurate and richer representation of future trends in retirement and pension benefit acceptance. The Interaction Between Social Security and Employer Pension Plans Changes in the social security system affect private and government employer pension plans. Many of these plans are explicitly integrated with social security, i.e., contribu- tions and benefit payments are keyed to social security contributions and change automatically when social secu- rity provisions change. Some employer plans are implicit- ly integrated with social security in that the existence and size of social security was taken into account when they were developed, and they are modified when social secu- rity is changed. In the development of a base case simula- tion with the current model, the simulations of social security, private pensions, and public employer pensions were developed to be mutually consistent. However, changes in social security do not directly affect the behav- ior of employer based pensions. Modeling of the links between the systems would improve the completeness of the model and the ability to analyze the effects of pension policy changes. Benefit and Contribution Levels Contributions and benefit levels are currently estimated using techniques that replicate, in a simplified manner, the actuarial calculations that must be performed for actu- al pension systems. Benefits and contributions are calcu- lated in a consistent manner. However, because of the aggregate nature of the model some links between contri- butions and benefits are not represented. For example, cash-outs of benefits before retirement in defined contri- bution plans are ignored in the current version. Receipt of dual benefits is not modeled. Experience gains and losses due to economic change, that must be amortized, are not modeled. Further work might permit a more accurate representation of the determination of contributions, benefits, and asset accumulation. Inflation and Real Benefit Levels Most private pensions are not indexed for changes in the cost-of-living. From time to time, many plans increase nominal benefits on an ad hoc basis to make up for the CHAPTER 6 effects of inflation or to increase real benefits. On average, however, during periods of rapid inflation such as the past decade, real private pension benefits have declined. The Macroeconomic-Demographic Model expresses all values in constant dollars of 1972 purchasing power. Pri- vate pension benefits are based on real earnings levels and, once awarded, do not decline in real terms. This is not an accurate reflection of actual behavior during an inflationary period, although it may be accurate over a long-term period. The Macroeconomic Growth Model does not have a monetary sector. It was not designed to forecast the rate of inflation. We do not believe it would be useful to attempt to forecast the rate of inflation over the long-term period of the model—75 years. However, the model can easily simulate the behavior of the economy and individual prices with an exogenously specified rate of general price increases. It would be useful to develop the capability of simulat- ing the effects of inflation on the private pension system. First, we could model the actual behavior of private pen- sion plans in an inflationary environment—the degree of benefit indexing that currently exists and trends in index- ing, and how ad hoc adjustments actually adjust for infla- tion. We could experiment with alternative adjustment 39 Private Pension Model rules and trace their implications over time for real bene- fit levels and fund balances. Inflation may affect the real value of the accrued liabili- ties and of the assets of pension plans. It also may affect the real rate of return to plan assets. We could model alterna- tive responses of assets and liablities to inflation under various exogenously specified rates of inflation and exam- ined the implications of these alternative response pat- terns for pension plan funding. Feedbacks to Other Model Components Like the Social Security Model, the Private Pension Model runs recursively. It uses the simulated results of the Macroeconomic Growth Model and the Labor Market Model as inputs, but does not affect the operations of other models in either current or future years. An important task for future development of the Private Pension Model is the incorporation of feedback links be- tween it and the core Macroeconomic Growth and Labor Market Models. These feedback links should depict the impacts of pension participation and funding on personal savings behavior and labor force behavior, and possible tradeoffs between pension contributions and wage levels. Chapter 7 Public Employee Pension Model Introduction n 1979 public employer pension plans paid over $33 bil- lion in benefits to 5.2 million beneficiaries. Over 12 mil- lion public sector employees participated in pension plans,' compared to 33 million participants in private pen- sion plans. The largest single public employer plan, the Federal Civil Service Retirement System, accounted for twelve billion dollars in benefits (1.6 million beneficia- ries) and collected sixteen billion in employer and em- ployee contributions. The other major Federal retirement plan—the Military Retirement System—paid ten billion dollars in benefits in 1979 (1.3 million beneficiaries). Since it is funded out of general revenues, it had no trust fund contributions. State and local governments comprise the remainder of the public sector, where an estimated seven thousand pension plans paid eleven billion dollars in benefits (2.3 million beneficiaries) and took in contri- butions of $21 billion in 1979. Demographic forces will be relatively less important to the future of the Public Employee Pension System than for the Social Security or Private Pension systems. The size and age distribution of the military workforce is more sensitive to the historical frequency of major wars than to demographic influences. Similarly, the size and age distri- bution of other public sector workforces may reflect in part the historical timing of new political programs and initiatives. Overview The Public Employee Pension Model is very similar to the Private Pension Model, described in Chapter 6. The model estimates for each year from 1970 to 2055: ® Number of employees by age and sex in each of seven sectors of public employment; ® Number of covered, participating, and vested work- ers in seven sectors of public employment and those covered by no pension plan; e Number of retirees, by age, sex, and sector of employment; 'In addition, another 2.2 million were in the armed forces. While all military personnel “participate” in the military retirement system by the conventional definition (see below), they vest only after 20 vears, so few active military personnel actually expect to receive a military pension benefit. #(stop)Appendix F lists the equations and variables of the Public Employ- ee Pension Model. 41 e Total annual benefits paid by public employee pen- sion plans; e Total annual contributions and assets for those plans operated on a funded basis; ® Average benefits for each plan. ~ For each simulation year, the Macroeconomic Growth, Population, and Labor Market Models provide the Public Employee Pension Model information concerning: Average annual wages, by age and sex; Total civilian employment by age and sex in the pri- vate and public sectors; Population age 5 to 17, which is used to estimate numbers of local educators; Population age 18 to 24, which is used to estimate numbers of state educators; Total income, which is used to estimate numbers of hazardous duty and state and local administrative workers. Estimates for twenty age-sex groups are reported— each sex in each of the following ten age groups: 16-17, 18-24, 25-34, 35-44, 45-54, 55-61, 62-64, 65-67, 68-71, 72 and over. These are the same age groups used in the Private Pension Model. Many of the estimates and calcula- tions are carried out for single-year-of-age groups. The model divides public employment into seven cate- gories: Federal Civil Service, military enlistees, military officers, state and local hazardous duty workers (police and firefighters), state and local general and administra- tive workers, state educators, and local educators. Each of the seven sectors of employment has the same type of pension plan—a defined benefit plan—with different benefit rates and funding provisions. These seven sectors are distinguished by one or more of three characteristics. The first is pension plan structure and funding. For example, the military has 20-year cliff vesting, but the Federal Civil Service has 5-year cliff vest- ing. The military system is pay-as-you-go, but the civilian systems are partially funded. The second distinguishing characteristic is the demographic structure of the work- force. For example, the military and hazardous duty work- forces are younger than the workforce of state teachers. The last characteristic is the set of determinants of total employment in each sector. State educators, for example, are usually employed at a state college, and the demand for their services is determined partly by the number of persons of college age. The Public Employee Pension Model performs six ma- THE NATIONAL INSTITUTE ON AGING MACROECONOMIC-DEMOGRAPHIC MODEL jor tasks. First, it estimates the number of public employ- ees by age, sex, and sector. Second, this population of public employees is used to estimate the populations of covered, vested, and participating workers, as well as the population of persons retired from public sector employ- ment. Third, total contributions for workers are deter- mined in all sectors except the military and state and local hazardous duty sectors, where pension funds are speci- fied to be funded on a pay-as-you-go basis. Fourth, the number of retirees from public sector employment, by age, sex, number of years retired, and pension status is estimated. Fifth, total benefits and average benefits for each sector are computed. Sixth, once total benefits and total contributions are calculated, asset levels can be com- puted for all of the sectors with funded pension plans. Estimation of the Number of Public Sector Employees The first task is to estimate the number of public employ- ees in various categories of employment which have dif- fering pension system characteristics. We assume that all public employee pension plans are of the defined benefit type.> However, the funding and benefit provisions of these pension plans vary across the various sectors of public employment. Furthermore, the demographic char- acteristics and retirement behavior of the workforce vary considerably from one sector to another. Consequently, it is important to model carefully the size of the workforce in the various sectors of public employment. The Federal Civil Service is modeled separately due to its size (28 percent of all public employees in 1979), low rate of growth, and the specific provisions of its pension plan. Almost all Federal civilian employees (94 percent) are covered by a single pension plan, the Federal Civil Service Retirement System. The military, which accounted for 15 percent of public employment in 1979, is modeled separately due to the pay-as-you-go funding of its retirement system, the unique factors affecting the size of the workforce, and the unique retirement characteristics of military personnel. The armed forces are divided into two groups, officers and enlisted personnel, because those groups differ in aver- age age of entry and retirement from the system. Both groups participate in the same system. State and local government employees are divided into hazardous duty workers (police and firemen), administra- tive workers, state educators, and local educators. The latter three groups have similar defined benefit pension plans. However, the size of each of those workforces is determined by significantly different factors. Hazardous duty workers generally have a different pension funding scheme and earlier retirement ages than other state and local employees. Local educators are generally elemen- 398 percent of public employees who are covered by pension plans are in defined benefit plans. . 42 tary and secondary school teachers, the number of which is determined in part by the population of children age 5 to 18. State educators are mostly college teachers, the number of which is responsive to the college age popula- tion, assumed in this model to be persons age 18 to 24. The model projects the number of public employees by age, sex, and sector by first projecting the total number of employees in each sector, then distributing that total among the twenty age-sex groups. The total for each sec- tor is projected with an equation estimated by ordinary least squares. The equation specifications and coefficients are discussed in Appendix G. They were estimated using data from the Bureau of Labor Statistics.* The total em- ployment estimates for each sector, except the military, are then distributed by age and sex using the age-sex distribution of private workers from the Labor Market Model. The military is assumed to maintain the age-sex distribution observed in 1979 throughout the simulation period. Coverage, Participation, and Vesting The Public Employee Pension Model estimates the num- ber of persons covered and not covered by a pension plan by age, sex, and public employment sector. It estimates the number of persons participating in a plan by age, sex, and sector, and the number of persons vested in a plan by age, sex, and sector.’ The Public Employee Pension Model uses a prevalence method to make these estimates, as does the Private Pen- sion Model. The basic procedure is identical. The only difference is that in the Public Employee Pension Model there are seven employment sectors, but only one type of plan in each. The actual prevalence rates differ by sector as well as by age and sex. Estimation of these rates is discussed in this section. To estimate the number of persons covered by a pen- sion plan in each sector of public employment by age and sex, the model first applies an estimated age-sex-sector specific coverage rate to the workers in each age, sex and sector. To estimate the number of participants, an estimat- ed age-sex-sector specific participation rate is applied to the covered workers in each age, sex, and sector. Finally, an age-sex-sector specific vesting rate is applied to the participants of each age, sex and sector to estimate vested workers. For the Federal Civil Service and the military, all work- ers are covered and participate, so coverage and participa- tion rates for those sectors are set equal to 100 percent. Vesting rates for the Federal Civil Service were devel- oped using 1979 age and tenure data. The Federal Civil Service Retirement System vests participants completely “See the tables for employment by sector in Bureau of Labor Statistics, Employment and Earnings. >The terms “covered”, “participating”, and “vested” in a pension plan, as used by this model, are defined in Chapter 6, pages 74 through 80. CHAPTER 7 after five year’s service. The Federal Civil Service Office of the Actuary provided ICF Incorporated with the age, sex, and tenure distribution for 1979.° We then assumed that all persons with five or more years tenure were vested and used the proportion of persons in each age-sex group with five or more years tenure as age-sex specific vesting rates. These rates are shown in Table 7-1. For the military, the model assumes throughout that all enlisted persons enter at age 19, and all officers enter at age 23. (These assumptions are used by the Department of Defense actuary.) Since the military retirement system has 20 year cliff vesting, the number of enlisted persons vest- ed was set equal to the number of enlisted persons age 39 or over. The number of officers vested was set equal to the number of officers age 43 or over. At the state and local government level, age-sex specific coverage, participation, and vesting rates were estimated based on ICF analysis of the May 1979 Current Population Survey Special Pension Supplement.” These rates are re- produced in Tables 7-1, 7-2 and 7-3. The current version of the model assumes these estimates remain constant for the entire simulation period, although trends in the rates could easily be incorporated. The relatively high rates of coverage for state and local employees makes this as- sumption relatively insensitive to potential future trends. These rates are used for all four state and local govern- ment sectors. The model specifies that all public employees with no pension plan are employees of state and local govern- ‘ments. All Federal employees are assumed to be covered by a pension plan. In future work we plan to examine how profiles of plan participation may change in response to current and alternative retirement policies and the impli- cations of policy changes. Contributions The model estimates total annual contributions for each of the funded plans—the Federal Civil Service Retirement System and all state and local plans except hazardous duty workers. It does this for each sector by determining aver- age contributions required for each age-sex group, multi- plying the average contribution for each age-sex group by the number of participants in each age-sex group, and summing across all ages and both sexes. All pension plans are assumed to be of the defined benefit type. Contributions to a defined benefit pension plan are determined by the benefit formula, the actuarial costing (funding) method, the age of the work force, and assumptions concerning job change and mortality.® The °Unpublished data from Office of the Actuary of the Federal Civilian Retirement System. "Bureau of the Census, Current Population Survey (May 1979), Special Pension Supplement. 8Determination of the level of contributions to a defined benefit pension plan is discussed in greater detail in Section E of Chapter 6. Public Employee Pension Model specific benefit formula of each public sector plan varies with the sector and will be discussed below. The specified actuarial costing method in each plan is a modification of the accrued benefit cost method. As in the Private Pension Model, the T-6 turnover rates, Census Bureau mortality rates, and an annual real rate of return on pension assets of 1.85 percent are assumed. Each participant is assumed to earn the annual compensation of his/her age-sex group.’ In the Public Employee Pension Model, the benefit for- mulae are based on the highest one or three years’ earn- ings rather than a percent of career average earnings, as in the Private Pension Model. This requires a modification in the accrued benefit cost method, because with a high one or high three benefit formula, a pension plan sponsor must estimate future wage levels. Estimation of future wages requires an assumption concerning the rate of real wage growth! Simulation of the Labor Market Model yielded an estimated annual rate of change of real wages of 1.8 percent. This rate of wage growth is assumed for actuarial costing calculations. The objective of this approach is not to forecast precise- ly the actual contributions that will be made to public ~ pension plans, but rather to indicate one funding pattern 43 that would provide actuarial balance. This pattern can be compared to alternative potential contribution levels to assess the future financial health of the public employee retirement system. Because of the modular structure of the model, alternative funding methods can be used. Unfunded liabilities existing in 1980 are amortized over a 70 year period.!! However, the model assumes no new unfunded liabilities arise after 1980. This version of the model does not incorporate potential trends in ad hoc benefit changes, which might give rise to additional amor- tization payments. However, in future versions of the model we will consider modeling the unfunded prior service costs associated with projected benefit formula changes.'? Unfunded liabilities are not calculated or amortized for the pay-as-you-go plans. For the Federal Civil Service Re- tirement System, the 1980 unfunded liabilities were re- ported to be $350 billion.'® Estimates of total unfunded liabilities of state and local government pension plans °In fact, Federal employees earnings are greater than average private sector earnings. In recognition of this fact, Federal Civil Service Retire- ment benefits are adjusted upward. Howard Winklevoss (1976). Seventy years was the period used in 1980 by the Federal Civil Service Retirement Service to amortize unfunded liabilities. For a further dis- cussion of unfunded pension plan liabilities, see Section E of Chapter 6. Benefit indexation for inflation, which is common in public employee plans, will also create unfunded liabilities. Such indexation does not occur in this model, because all values are stated in dollars of constant purchasing power, i.e., there is no change in the average price level. '3This is an upward adjustment from the Fifty-Seventh Annual Report of the Board of Actuaries of the Civil Service Retirement System (1980), which estimated $128 billion in 1979. THE NATIONAL INSTITUTE ON AGING MACROECONOMIC-DEMOGRAPHIC MODEL Table 7-1 Vesting Rates For Public Employees Who Are Participating in A Pension Plan By Age, Sex, and Sector Military Services Federal Civil State/Local Service Enlistees : Officers Employees Age Male Female Male Female Male Female Male Female 16-17 0.00 0.00 0.0 0.0 0.0 0.0 01 04 18-24 0.34 0.58 0.0 0.0 0.0 0.0 33 26 25-34 0.81 0.71 0.0 0.0 0.0 0.0 .58 43 35-44 0.95 0.84 0.5 0.5 0.1 0.1 73 51 45-54 1.00 0.99 1.0 1.0 1.0 1.0 76 57 55-61 1.00 0.99 1.0 1.0 1.0 1.0 67 .03 62-64 1.00 0.99 1.0 1.0 1.0 1.0 67 63 65-67 1.00 0.99 1.0 1.0 1.0 1.0 67 63 68-71 1.00 0.99 1.0 1.0 1.0 1.0 .67 63 724 0.00 0.00 0.0 0.0 0.0 0.0 0.00 0.00 Source: Federal Civil Service Retirement System, Office of the Actuary; Department of Defense; ICF analysis of May 1979 CPS. Table 7-2 Table 7-3 Pension Coverage Rates For State and Local Proportions of Employees of State and Local Government Employees, By Age and Sex Governments Covered By A Pension Plan Who Are Participating in the Pension Plan, Age Male Female By Age and Sex 1617 510 510 Age Male Female 18-24 .880 .840 25-34 940 900 16-17 130 130 35-44 , 960 900 18-24 590 560 45-54 920 .890 25-34 860 730 55-61 .860 .880 35-44 910 760 62-64 .860 .880 45-54 890 750 65-67 .860 880 55-61 750 740 68-71 860 .880 62-64 750 740 72+ WY 000 65-67 750 740 en 68-71 .750 .740 - Source: ICF analysis of May 1979 CPS. 72+ .000 .000 Source: ICF analysis of May 1979 CPS. 44 CHAPTER 7 range from $150 billion to $270 billion.'* For the simula- tions reported in this monograph, we assumed that the unfunded liabilities in 1980 were $200 billion.'> Benefit Recipients The Public Employee Pension Model estimates the num- ber of recipients of pension benefits by age, sex, sector, and year of benefit acceptance. An incidence rate method is used, similar to the method used in the Private Pension Model. The existing population of recipients from the previous period is aged one year, and its size is reduced to reflect mortality. The number of new recipients, by age, sex, and sector, is estimated and added to the existing population of previous recipients. New Recipients The model forecasts new recipients by age and sex in each sector as a proportion of the workers in the sector-specific retirement ages (39-72 for military enlistees, 42-72 for military officers, and 50-72 for all others).'® Different age- sex specific retirement rates were estimated for each sec- tor. Federal Civil Service rates are from the 1980 Federal Civil Service Actuary’s report.!” Military rates are from unpublished data provided by the Department of De- fense.'® All state and local employee retirement rates are the same as those used in the retirement calculations of the Private Pension Model. Existing Recipients The model estimates existing public employee pension recipients by age, sex, and sector by aging the population of the previous year’s recipients and reducing each age- sex-sector population by a factor to represent mortality “House Pension Task Force, Report on Public Employee Retirement Systems, 15 March 1978. 15This assumption can easily be changed to incorporate other estimates, such as those from the study of public employee plans conducted by the Urban Institute for the Department of Housing and Urban Develop- ment and the President's Commission on Pension Policy. “This differs from the Private Pension Model, which forecasts retirees based on retirement-age populations. “Retirement,” in the Public Em- ployee Pension Model, means acceptance of a benefit from one of the seven public employee plans. Many public employee “retirees,” espe- cially military, take other jobs after retirement rather than withdraw from the labor force. VFifty-Seventh Report to Congress from the Federal Civil Service Retire- ment System, op. cil. "Unpublished data from the Office of the Actuary, Defense Manpower Data Center, Department of Defense. 45 Public Employee Pension Model experience. The mortality rates are from the Bureau of the Census'® and are based on analysis done by the Social Security Office of the Actuary. They trend downward over time. The populations of existing and new recipients are summed to yield total recipients in each age, sex, and sector group. : Benefits We specify that each of the seven sectors of public em- ployment has a defined benefit pension plan. To deter- mine total benefits paid, we need to know: e Number of beneficiaries of each type; and ® Average annual benefit paid per beneficiary of each type. Annual benefits will vary with age, sex, sector of employ- ment, year of retirement, and average tenure. During each simulation year, the model estimates aver- age annual retirement benefits for all persons retiring in that year by age, sex, and sector of employment. Those benefits then remain fixed, in 1972 dollars, for the remain- der of the life of each beneficiary. The benefit estimation procedure is similar to that used in the Private Pension Model. While each of the seven sectors’ plans is of the defined benefit type,? the funding methods and benefit formulae of the plans vary considerably. The military and hazardous duty plans, as previously noted, are modeled as unfunded pay-as-you-go systems. The others are assumed to be actu- arially funded. This, of course, does not capture the very diverse nature of funding schemes used for state and local pension plans. However, the modular structure of the program used to implement the model can easily be aug- mented to incorporate the variety of state and local gov- ernment plans as more detailed models are developed. Annual benefits are calculated according to formulae that are representative of each group. These formulae are described in more detail in Appendix F. For Federal civil- ian employees, we used the 1981 benefit formula of the Federal Civil Service Retirement System. This provides for a benefit equal to 16.25 percent plus two percent for each year worked over ten years, up to a maximum of 80 per- cent, multiplied by the three highest years’ salaries. For military personnel, we used the 1981 formula of the Mili- tary Retirement System. This specifies a benefit equal to 2.5 percent for each year of service (to a maximum of 75 percent) multiplied by the last year’s basic pay. For the four state and local employment sectors—hazardous duty, “The same mortality rates are used in the Population Model and in the other pension system models. Approximately 2.2 percent of public sector employees are in defined contribution plans only , according to the final report of the House Pension Task Force on Public Employee Retirement Systems, March 15, 1978. THE NATIONAL INSTITUTE ON AGING MACROECONOMIC-DEMOGRAPHIC MODEL general administrative, state educators, and local educa- tors—we developed representative retirement benefit formulae based on data from the House Pension Task Force report. For hazardous duty plans, the annual benefit is equal to 50 percent of final year pay. For the other three sectors, the annual benefit is calculated as two percent per year of service (up to a maximum of 80 percent) times the average of the highest three annual salaries.?! To determine new retirement benefit levels, the model needs to estimate job tenure upon retirement. Presently, the model assumes a constant tenure for all non-military retirees through time.*? The assumption of uniform entry age and retirement age for the military allows the model to determine military tenure. To initiate solution of the model, we required data on variables for years prior to 1970 such as retirement bene- fits for workers who retired before 1970 by age, sex, and sector. Historic data on these variables often is not avail- able in the form required by the model, since they are specific constructs used by the model. In order to provide these data, a simulation was run over the period 1946- 1976, using the same parameters employed elsewhere in the model and historical information where available. Pension Plan Assets Calculation The Public Employee Pension Model calculates total as- sets held by each of the four funded pension plans each year of the simulation period. For each funded plan, the assets at the end of the current year are equal to the assets at the end of the previous year, plus the interest income earned by those assets, plus the year’s total contributions less the year’s total benefit payments, plus interest on one- half the difference between last year’s assets and this year’s assets.”> The real rate of return on the assets is assumed to be 1.85 percent per year. Future Work The Public Employee Pension Model represents, at a high- ly aggregate level, the essential features of the complex retirement systems of the various governments. It can usefully track expected trends in the status and behavior of those systems and can analyze the aggregate implica- tions of policy alternatives. However, there are several areas where disaggregation and model accuracy could be enhanced. ?!These formulae can be modified to take into account additional infor- mation that is available, for example, from the public employee pen- sion system study conducted by the Urban Institute for the Department of Housing and Urban Development. . 22presently the same tenure structure is assumed for non-military work- ers as in the Private Pension Model. 23See Equation 6-4. 46 Public Sector Employment Estimation Additional work could improve the estimates of the size and demographic composition of employment in each of the seven public sectors. In the current version of the model the age-sex-composition of employment in all of the non-military sectors is assumed to be the same as that of the work force as a whole. This assumption may not be valid for rapidly growing sectors, such as state and local general and administrative, or for slowly growing sectors such as the Federal Civil Service. Data on the demogra- phic composition of the Federal Civil Service has been collected. Demographic data for the other sectors may be difficult to develop. Nevertheless, it may be possible to improve the estimates of the size and composition of each sector. The endogenous estimation procedure for public sec- tor employment could then be linked to the Macroeco- nomic Growth Model to improve that model's estimates of government purchases of labor services. Public Sector Compensation The model assumes that compensation of each age-sex group is determined based on marginal productivity in the private sector. Age-sex specific wages, hours, and com- pensation estimated by the Labor Market Model are ap- plied to the Public Employee Pension Model. However, the public sector accounts for a larger proportion of total compensation (about 21 percent) than of total employ- ment (about 16 percent). An effort could be made to model the determination of public sector earnings inde- pendently of private sector earnings. Retirement Age, Tenure, Post-Retirement Employment, Multiple Pension Benefits The current version of the model does not attempt to represent the complex life cycle employment and retire- ment behavior that, in fact, characterize many public sec- tor employees. Because average retirement ages are lower in several of the sectors of public employment than in the private sector, acceptance of a private sector job or another public sector job after retirement from an initial public sector job is not uncommon, especially for military retirees. Modeling these phenomena requires more detail and a more micro-level approach than is appropriate for the Macroeconomic-Demographic Model. Such detail is not necessary to capture the essential features of the be- havior of each pension system. However, lack of such detail limits the usefulness of the model for analysis of the distributional impacts of pension policies affecting public employees, such as universal social security coverage. An effort could be made to estimate the allocation of public employee pension benefit recipients to categories other CHAPTER 7 than retirement and to simulate a variety of post-retire- ment employment and pension benefit acquisition paths. Retirement age and tenure in public sector employ- ment could also be modeled in a more comprehensive fashion. As suggested for the Private Pension Model, an effort could be made to link the estimation of retirement benefit acceptance with labor force participation. For sev- eral sectors of public employment, post-retirement em- ployment would explicitly have to be considered. One approach to this problem may be the development and integration of a work history microsimulation model with the Macroeconomic-Demographic Model. Such an integrated system might then permit relatively detailed representation of public sector and private sector employ- ment and earnings histories, pension plan tenure, retire- ment ages, post-retirement labor market activity, and pension benefit receipt. Ad Hoc Benefit Increases and Inflation The Civil Service Retirement System and the Military Re- A Public Employee Pension Model tirement System are completely indexed for changes in the cost-of-living. Consequently, modeling them in real terms provides an accurate representation. The other public retirement systems, like the private pension sys- tem, are not fully indexed. Ad hoc benefit increases in those systems are common however. The accuracy of the model might be improved by more detailed analysis of the affects of inflation on public sector pensions and the process of adjustment to inflation in non-indexed plans. Ad hoc changes in real benefit levels, even in fully in- dexed plans, as well as others, has occurred historically as the level of real earnings has increased. It may be useful to attempt to model these changes endogenously. Pension costs could also be input explicitly to the determination of government expenditures in the Macroeconomic Growth Model. In the current model, trends in pension costs are considered implicitly in the exogenous estimation of trends in total government expenditures and transfers. Chapter 8 The Supplemental Security Income Model Introduction he Supplemental Security Income (SSI) program pro- vides monthly cash payments in accordance with uniform, nationwide eligibility requirements to persons with limit- ed income who are aged 65 and over, blind, or disabled. Established by the 1972 amendments to the Social Security Act (P.L. 92-603, signed October 30, 1972), the program replaced Federal grants to states for old-age assistance, aid to the blind, and aid to the permanently and totally dis- abled. It is administered by the Social Security Administra- tion. The basic Federal cash payments are financed from general funds of the United States Treasury (a system of required and optional state supplementary payments is discussed below).! Under the SSI program, each eligible aged, blind, or disabled person living in his own household is provided a monthly cash payment from the Federal government that is sufficient, when added to his countable income, to bring his total monthly income up to a specified level ($238.00 as of July 1, 1980, $357.00 for a couple if both are eligible). If the individual or couple is living in another’s household, the guaranteed level is reduced by one-third. For institutionalized persons, the eligibility requirements and payment standard depend on the type of institution. A maximum of $25 a month is set for persons in public or private institutions who receive more than 50 percent of the cost of their care from the Medicaid (medical assis- tance) program under Title XIX of the Social Security Act. Eligible persons in private institutions whose care is not met from Medicaid funds may receive the standard pay- ment applicable to individuals living in their own households. The Federal payment is based on the individual's count- able income. The first $20 a month in OASDI or other earned or unearned income is not countable. Also disre- garded is $65 a month of earned income plus one-half of any earnings above $65. Generally, individuals are not eligible for payments if they have resources in excess of $1,524 (or $2,250 for a couple), excluding the reasonable value of a home, automobile, household goods, and per- sonal effects and life insurance with a face value of $1,500 or less. States are required to supplement the Federal SSI pay- ments where the Federal payment does not maintain the December 1973 income level of recipients who were transferred from the former state public assistance pro- grams as of January 1, 1974. States also have the option to supplement the SSI floor for all or selected categories, regardless of previous state program eligibility. In December 1979, about 1.9 million persons age 65 and over were receiving SSI benefits, almost eight percent of the elderly population. Total SSI benefits for the aged from Federal and state sources were 230 million dollars a month, in that month, an average monthly benefit of $122.67. Table 8-1 Number of Persons Receiving Federally Administered SSI Payments and Total Amount of Payments, 1974- 19792 Amount of Payments During Year Thousands of Persons (Millions of Dollars) Total Aged Blind Disabled Total Aged Blind Disabled 1974 3,996 2,286 75 1,635 $5,097 $2,414 $126 $2,557 1975 4314 2,307 74 1,932 5,716 2,517 127 3,072 1976 4,236 2,148 76 2,012 5,900 2,420 134 3,346 1977 4,238 2,051 77 2,109 6,134 2,364 142 3,628 1978 4217 1,968 77 2,172 6,372 2,342 148 3,882 1979 4,150 1,872 77 2,201 6,867 2,421 162 4,286 “Includes state supplements administered by the Federal government. Source: Social Security Bulletin, Annual Statistical Supplement, 1977-79. "The SSI Model includes only Federal benefits and state supplements administered by the Federal government. A small number of states administer their own supplements. These are not included in the model. Table 8-1 shows the total number of recipients of SSI benefits and total payments administered by the Federal government for the years 1974-1979. SSI benefits to the aged in 1979, administered by the Federal government, THE NATIONAL INSTITUTE ON AGING MACROECONOMIC-DEMOGRAPHIC MODEL were about 2.4 billion dollars,” about three percent of the total payments of all retirement income programs. In 1976, ten percent of all aged units? received supple- mental security income, eleven percent of those units over age 72. Thirteen percent of nonmarried individuals age 65 and older received SSI benefits. Supplemental Security Income, therefore, is an impor- tant component of the retirement income system. Its size alone, however, does not fully reflect its importance. SSI is the major source of income for the elderly poor and is one of the two major cash components of the U.S. welfare system. By providing a national, means-tested source of income for all the elderly poor, SSI reduces the pressure on social security to serve as an assistance program. Overview of the Modeling Approach The SSI Model, like the other four retirement income models, is based on the belief that useful information can be developed about the system by focusing on the time series behavior of aggregate variables of the system, disag- gregating them by age and sex, and investigating how those age-sex specific variables are related to the size and composition of the population and to other age-sex spe- cific economic variables. Unlike some other efforts to model SSI and other transfer programs, this model has no household detail.* Low income individuals, rather than households are the units of analysis. The other pension system models described in Chapters 5, 6 and 7 use infor- mation about the average economic characteristics or be- havior of each age-sex group (i.e. mean wage, labor force participation rate, unemployment rate of the entire age- sex group). Information about averages must be trans- formed in the SSI Model, however, because the population served by the SSI system is at the low end of the income distribution and, therefore, is not typical of the entire age-sex group. %In addition, about 100 million dollars of state supplements were admin- istered by the states. 3An aged unit is a married couple living together, or an unmarried person. These data are from U.S. Department of Health, Education and Welfare, Income of the Population 55 and Over, 1976. “Microsimulation models such as the Transfer Income Model (TRIM), the Micro Analysis of Transfer to Households model (MATH), and the Dynamic Simulation of Income Model (DYNASIM) are examples of modeling approaches that rely on household data. These models use survey data concerning the income and SSI receipts of individual households, simulate how the behavior and number of each type of household will change through time or under alternative programs, then aggregate and tabulate the SSI and other transfer payments to each household to estimate the total expenditures and beneficiaries of the programs in various demographic and socio-economic categories. These models generally use exogenous estimates of the population and macroeconomic variables to control the totals calculated by aggregating the individual household estimates. Consequently, their approach to estimation of the aggregates may not differ markedly from the approach described here. Microsimulation models can provide more distribu- tional detail than the Macroeconomic-Demographic Model. ~ Nn In other respects, the SSI Model is similar to the Social Security Model. Social Security and SSI are each com- posed of one basic program that is established by law and is national in scope unlike the private pension “system” or the “system” of public employee plans.> The SSI Model, the beneficiary population is disaggre- gated by age. Age-specific participation or incidence rates are developed for each of the three causes of eligibility (age, blindness, and disability). These incidence rates then are applied to projections of the future population by age to estimate the number of beneficiaries. The inci- dence rates for blindness and for disability are assumed to remain constant. The incidence rate for aged participation is functionally related to the average income level. To calculate total benefit payments, an average Federal benefit and average state supplement is applied to the estimated beneficiary population. It is assumed that these average benefits remain constant in real terms (1972 dol- lars). Like the other pension system models and the Macroeconomic Growth Model, the SSI Model expresses all income flows in 1972 dollars. Benefits to the Aged Benefits to the aged are estimated in three steps: (1) The total eligible population of each age and sex is estimated as a function of aggregate real income per capita and the age-sex-specific average wage. (2) The number of SSI participants is estimated by ap- plying an estimated age-sex-specific participation rate to the eligible population. (3) Total SSI benefit payments are estimated by multi- plying the number of participants by the average Federal and state benefit payments. Participants Estimation of the distribution of participants by age and sex is analytically the most difficult. The population eligi- ble for SSI benefits are the low-income, low-wealth aged. The size of that population depends on the relative in- come distribution and the average level of income as well as the distribution of wealth. We assume that the relative income distribution can be characterized by a log-normal distribution, and that it is stable throughout the forecast period, in the sense that the coefficient of variation is constant. However, the mean income level rises. Conse- quently, the tail of the income distribution below a fixed income level covers a smaller and smaller population as time passes and the entire distribution moves upward. We assume that most of the current statutory provisions of the >The SSI system has both optional and mandatory state supplements, so its benefit levels differ from state to state. The effect of state supple- ments is modeled at the aggregate level. CHAPTER 8 SSI program remain unchanged throughout the projec- tion period. Since those provisions imply a constant real income eligibility standard (i.e. an absolute as opposed to a relative concept of poverty), the relative size of the eligible population diminishes as average income grows.’ Each year the population below the constant real eligibil- ity standard can be estimated by integrating over the area of the distribution below that level in that year. Table 8-2 shows the number of recipients of SSI bene- fits for the aged in each of four age groups. Table 8-3 shows the percent of the total population of each of the four age groups acounted for by recipients of SSI benefits for the aged. Both the number of recipients and the pro- portion of the population fell steadily over the period 1975-1979. Table 8-3 shows the annual proportional rate of change of the proportion of each age group receiving SSI aged benefits. Average Benefit Levels Table 8-4 reports the average SSI benefit amount paid in 1977,1978 and 1979, in current and in 1972 dollars. There is no apparent trend in the constant dollar benefit amount. For the initial forecasts, we assumed that the average real benefit would remain constant. Because OASDI benefit payments are counted as un- earned income, after the monthly exclusion of 20 dollars, . SSI benefits are reduced dollar-for-dollar when OASDI benefits are received. As real OASDI benefit levels rise, average SSI benefit payments may decline. The rise in OASDI benefit levels has been an important factor in the decline in the number of SSI recipients, reported in Table 8-2. Combining the elements discussed in this section, total SSI benefits for the aged were estimated by the following equation: ’ TB; = AB; X (RN); X Nj (81) where: TB, = total benefits to age group i in vear AB, = estimated average benefit to age group iin year (RN), = estimated recipients-per-capita of age group i in year t N, = projected population of age group i in year t. The population estimates are provided by the Popula- ‘tion Model. “We depart from the current provisions in one area. Current law speci- fies a constant nominal asset test. Individuals are not eligible if they have resources in excess of $1,524 ($2,250 for couples). With inflation, that implies a shrinking real asset test. We assume that the asset test is indexed to permit a constant real level. This change has been recom- mended by the 1979 Advisory Council on Social Security and others. Supplemental Security Income Model Table 8-2 Recipients of Federally Administered SSI Aged Benefits, By Age, 1975-1979 (Thousands of Persons) Age Group 1975 1976 1977 1978 1979 65-69 482 399 336 293 262 70-74 623 599 585 561 515 75-79 505 485 480 474 470 80 and Over 692 664 650 638 625 Total 2,307 2,148 2,051 1,968 1,872 Source: Calculated from Social Security Bulletin, Annual Statistical Supplements, 1975, 1976, and 1977-79, Tables 162, 163. Table 8-3 Percent of Population Receiving SSI Benefits for the Aged, By Age Group, 1975-1979 Annual Rate Age Group 1975 1976 1977 1978 1979 of Change 65-79 595 482 398 342 3.02 —.1559 70-74 10.79 10.12 952 882 7.82 -.0773 75-79 12.62 11.96 11.79 11.36 11.00 —.0338 80 & over 15.29 14.13 13.40 12.88 12.23 —.0543 Source: Calculated from Social Security Bulletin, Annual Statistical Supplements, 1975, 1976, and 1977-79. Table 8-4 Average Monthly SSI Benefit Amount For Persons Receiving Federally Administered Payments, By Reason for Eligilibity, 1977-1979 1977 1978 1979 Total Current $ 124.52 129.61 155.65 1972 $ 85.97 83.11 89.71 Aged Current $ 96.62 100.43 122.67 1972 $ 66.70 64.40 70.70 Blind Current $ 159.20 164.40 212.27 1972 $ 109.91 105.42 122.34 Disabled Current $ 150.36 154.82 181.71 1972 $ 103.80 99.28 104.73 Source: Calculated from Social Security Bulletin, Annual Statistical Sup- plement 1977-79, Table 150. THE NATIONAL INSTITUTE ON AGING MACROECONOMIC-DEMOGRAPHIC MODEL Benefits to the Blind and Disabled Tables 8-5 and 8-6 show the number of SSI benefit recipi- ents by age group for disability and for blindness, respec- tively. Tables 8-7 and 8-8 show the proportion of each age group receiving benefits. Unlike the ratios for aged bene- fit recipients, the proportions for most blind and disabled beneficiary age groups have been fairly stable over the 1975-1979 period. This should not be surprising. While the number of low income aged is probably related to current and past average income levels, the number of low income blind and disabled is probably not greatly affected by average income levels. Individuals in these categories—with severe physical impairments and in pov- erty—may represent a small and rather stable proportion of the population that is largely independent of average income levels.” For these simulations, we assumed that the proportion of the population receiving SSI benefits for disability and for blindness remains constant at the 1979 level. The average real benefit payment for disability and for blindness, shown in Table 8-3, does not show a trend.® Consequently, we assumed the real benefit will remain constant. Projections of SSI disability and blindness benefits by age group for the remainder of the century were made using a methodology similar to that for aged benefits, described by equation (8-1). - Table 8-5 Recipients of Federally Administered SSI Disability Benefits, By Age, 1975-1979 (Thousands of Persons) 1975 1976 1977 1978 1979 Children, Age Less than 5 11.5 12.9 16.5 59 26.9 31.6 37.1 10-14 39.1 46.0 532 161.4 1723 15-17 26.5 314 36.4 18 and Over 19.8 26.4 27.0 30.5 33.6 Total Children 123.8 148.2 170.1 191.7 205.9 Adults, Age 18-21 70.5 72.7 79.5 81.2 77.8 22-29 204.4 216.2 228.8. 235.6 243.4 30-39 191.7 201.3 213.3 219.8 229.4 40-49 273.1 268.4 267.6 263.4 253.3 50-59 515.5 510.6 515.9 508.9 492.7 60-64 372.6 367.1 356.8 350.5 335.1 65-69 164.6 210.6 256.0 287.1 295.2 70-74 9 11.2 15.5 27.7 59.8 75-79 4 3.7 1.9 4.0 4.0 80 and Over 2 1.9 39 2.0 2.0 Total Adults 1,808.9 1,863.6 1,939.3 1,980.2 1,994.7 Total 1,932.7 2,011.9 2,109.4 2,171.9 2,200.6 Source: Calculated from Social Security Bulletin, Annual Statistical Supplement, 1975, 1976, and 1977-79, Tables 162, 163. "For example, in 1975 one-third of the disabled adults awarded SSI payments were diagnosed as mentally ill. Of these, one-third were suffering with psychoses and one-third were mentally retarded. Of the retarded, most had an IQ of 49 or less, or were too severely impaired to be tested. Kochar, Satya, “Representative Payments Under the SSI Pro- gram, August 1977,” Research and Statistics Note, No. 9, U.S. Department of Health and Human Services, September 16, 1980. ®The real benefit level fluctuates over the three vear period, reflecting the effects of the lagged adjustment for cost of living changes during a period of accelerating inflation. CHAPTER 8 Supplemental Security Income Model Table 8-6 Recipients of Federally Administered SSI Benefits for the Blind, By Age, 1975-1979 (Thousands of Persons) 1975 1976 1977 1978 1979 Children, Age Less than 5S 4 4 5-9 8 1.0 10-14 1.0 13 4.2 4.7 5.2 15-17 7 8 18 and Over 1.4 1.4 9 1.0 1.0 Total Children 4.3 4.9 5.1 5.8 6.2 Adults, Age 18-21 25 2.4 25 2.6 2.6 22-29 8.7 9.5 9.5 9.7 9.6 30-39 6.5 6.7 6.6 6.7 7.1 40-49 85 8.5 7.8 7.5 7.3 50-59 13.3 13.4 125 12.2 12.0 60-64 9.1 9.1 8.2 7.9 7.6 65-69 7.2 7.8 8.1 8.1 7.8 70-74 4.4 4.4 4.8 4.8 5.2 75-79 36 35 4.0 39 39 80 and Over 6.5 6.2 83 8.0 7.9 Total Adults 70.1 . 71.5 723 71.4 71.0 + Total 74.5 76.4 77.4 77.1 77.3 Source: Calculated from Social Security Bulletin, Annual Statistical Supplement, 1975, 1976, and 1977-79, Tables 162, 163. Table 8-7 Percent of Population Receiving SSI Disability Benefits, By Age, 1975-1979 1975 1976 1977 1978 1979 Children, Age Less than 5 072 .084 .108 5-9 155 .182 216 10-14 191 232 277 254 275 15-17 210 247 .288 18 and Over 120 157 159 .178 195 Adults, Age 18-21 427 C433 467 479 454 22-29 .728 .740 779 .787 .796 30-39 748 .769 .768 .758 .760 40-49 1.191 1.184 1.188 1.160 1.130 50-59 2.297 2.256 2.250 2.211 2.133 60-64 4.033 3.941 3.819 3.714 3.500 65-69 2.032 2.523 3.034 3.350 3.40 70-74 .015 .189 252 435 .908 75-79 .010 .091 .047 .096 .094 80 and Over .004 .040 .080 .040 .039 Source: Calculated from Social Security Bulletin Annual Statistical Supplements, 1975,1976, and 1977-79, and from U.S. population estimates published in Current Population Reports, Series P-25, Nos. 614, 870. 53 THE NATIONAL INSTITUTE ON AGING MACROECONOMIC-DEMOGRAPHIC MODEL Table 8-8 Percent of Population Receiving SSI Disability Benefits for the Blind, By Age, 1975-1979 1975 1976 1977 1978 1979 Children, Age Less than 5 .002 .003 59 .005 .006 10-14 .005 .006 .007 .008 .008 15-17 .006 .006 18 and Over .008 .008 .005 .006 .006 Adults, Age 18-21 .015 014 .015 .015 .015 22-29 031 .033 032 .032 .031 30-39 025 .026 024 023 024 40-49 037 .037 034 .033 .032 50-59 059 .059 .055 .053 .052 60-64 .098 .098 .087 .084 .079 65-69 .089 .094 .096 094 .090 70-74 .076 074 .078 .075 .079 75-79 .090 .086 .098 094 094 80 and Over 144 132 171 161 155 Source: Calculated from Social Security Bulletin Annual Statistical Supplements, 1975,1976, and 1977-79, and from U.S. population estimates published in Current Population Reports, Series P-25, Nos. 614, 870. 54 Chapter 9 Medicare Model Introduction his chapter describes the Medicare Model of the Macroe- conomic-Demographic Model. A model of the Medicare program is included in the Macroeconomic-Demographic Model because medical expenditures are an important component of the living costs of the elderly. Any complete assessment of future retirement income levels and poli- cies should include “in-kind” income. Medicare is the most important source of in-kind income for the elderly. Although the current version of the Medicare Model is quite simple, it does allow policymakers to examine the effects of demographic influences on Medicare costs, holding constant real expenditures for each age-sex ser- vice group. It provides a base which could be easily ex- panded into a more fully developed model in later versions of the Macroeconomic-Demographic Model. The retirement income needs of the elderly are heavily affected by personal expenditures for health care services. As Table 9-1 shows, health care expenditures per capita “for the elderly in 1976 were approximately 2.7 times high- er than those for individuals age 19-64. Expenditures for those over age 65 represented approximately 29 percent of total health care expenditures. As a result, total health care expenses change disproportionately as the relative size of the elderly population increases. The costs of medical care, therefore, are a major concern of those addressing retirement income issues. The Medicare program, enacted in 1965, pays for a ma- jor portion of the health care expenditures for the elderly. As shown below in Table 9-2, the Medicare program paid approximately 44 percent of all health care expenditures by the elderly and 75 percent of hospital expenditures in 1978. Medicare also pays for over one-half of all physician expenditures of the elderly, but pays for only a minor amount of nursing home care. In 1980 about 24.5 million elderly individuals received Medicare benefits costing a total of $28.3 billion, an average benefit of $1,200 per participant. : The Medicare program is a broad program of health benefits. To become eligible for Medicare benefits, an individual must either be 65 years of age or older, or disabled, if under 65 years of age. The program has two parts: hospital insurance (Part A) and medical insurance (Part B). Medicare hospital insurance pays a major portion of the costs of medically necessary inpatient hospital care, and, after a hospital stay, inpatient care in a skilled nursing facility and care in the home by a home health agency. Medicare medical insurance pays a portion of the medi- cally necessary doctors’ services, outpatient hospital ser- vices, outpatient physical therapy and speech pathology services, and a number of other medical services and 55 Table 9-1 Personal Health Expenditures By Age Group, 1978 Total $ billions % Per Capita Under 19 Years $ 199 11.9% $ 286 19-64 Years 98.7 58.8 764 65 Years and Over 49.4 29.4 2,026 Total $1679 100.0% $ 753 Source: Charles R. Fisher (1980), p. 66. supplies that are not covered by the hospital insurance part of Medicare. Medical insurance can also help pay for necessary home health services when hospital insurance cannot pay for them. Although not included in this model, the Medicaid pro- gram, also enacted in 1965, pays a significant portion of health expenses for the elderly. In particular, Medicaid paid 43 percent of the elderly’s expenditures for nursing home care in 1978. In 1980 about 5.1 million elderly individuals received Medicaid benefits costing $4.3 bil- lion, an average benefit of $800 per participant. To become eligible for the Medicaid program, an indi- vidual must be either aged, blind, disabled, or in a family with dependent children, and have insufficient income and resources to meet medical costs. Income and re- source limits for eligibility vary by state. The program is financed by Federal and state funds, with Federal pay- ments ranging from 50 percent to 83 percent of total Medicaid payments, depending on the state. The formula is based on the per capita income of each state—the high- er the per capita income, the lower the federal aid. Medicare, Medicaid and other public programs pay for approximately two-thirds of all health care expenditures for the elderly. Medicare represents the most significant public program for these services. Because Medicare is by far the largest in-kind benefit program and is almost universal in coverage, we focused initially on the Medicare program only. Although Medic- aid pays for a portion of the elderly’s health care costs, it is not currently included in the model. Private payments for health services are also excluded. Although health care expenditures in the future will be heavily influenced by Medicare benefit coverage and changes in the delivery of health care, the emphasis of the Medicare Model is primarily on the demographic factors that will produce changes in Medicare expenditures. THE NATIONAL INSTITUTE ON AGING MACROECONOMIC-DEMOGRAPHIC MODEL Table 9-2 Annual Personal Health Care Expenses For Population Over Age 65, 1978 ($ Billions) Other Services Private Medicare Public Hospital Care $ 26 $15.8 $2.7 Physician Care 3.6 5.0 3 Drugs, glasses, other 31 2 5 Nursing Home Care 6.8 4 5.4 Other Services 2.0 5 4 Total $18.2 $21.8 $9.4 Percent of Total (36.8%) (44.1%) (19.1%) Source: Fisher (1980) p. 89. Overview The Medicare Model was developed to make maximum use of the demographic information produced by the Population Model. The basic Medicare Model disaggre- gates annual Medicare expenditures into detailed age-sex- service categories and estimates expenditures—per- recipient, recipients-per-capita and expenditures-per- capita in each of those categories. The methodology estimates total recipients and total expenditures for each age and sex and type of service by multiplying the project- ed population in each age-sex group by the estimated levels of recipients-per-capita and expenditures-per-re- cipient. In the simulations, real expenditures-per-capita are held constant for each age, sex and service group throughout the projection period. In the future develop- ment, the expenditures-per-capita coefficients could be permitted to vary to reflect time trends and the effects of changes in income levels, health status of the population, benefits, and health care delivery systems. Medicare expenditures are disaggregated into the fol- lowing services: e inpatient hospital care ® home health care (Part A) ® home health care (Part B) ® outpatient care eo skilled nursing home care e physician services and other medical care For each of those services, Medicare expenditures are disaggregated into the following age categories for each sex: less than 35, 35-44, 45-54, 55-59, 60-64, 65-66, 67-68, 69-70, 71-72, 73-74, 75-79, 80-84, and greater than 85. Structure of the Medicare Model Annual Medicare expenditures and the number of Medi- care recipients from 1966 through 1977, disaggregated into each of the six health care service categories and each 56 (Percent Total of Total) $21.2 (42.9%) 8.9 (18.0) 3.8 (7.7) 12.6 (25.5) 2.8 (5.7) $49.4 (100.0%) (100.0%) of 26 age-sex groups of recipients (156 categories in all) serve as the basic data of the Medicare Model. Medicare recipient and reimbursement data were derived from published and unpublished tables prepared by the Office of Research, Demonstrations, and Statistics in the Health Care Financing Administration. For each type of service, age, and sex group estimates of total expenditures were divided by the number of recipients and by the popula- tion to estimate expenditures per recipient and per capita. The number of recipients of each service, age and sex were divided by the population of each age and sex to estimate recipients-per-capita for each type of service, age and sex group. These time series of expenditures-per- recipient and recipients-per-capita are the basic data used to estimate the coefficients of the Medicare Model. The estimates of Medicare expenditures per recipient and per capita by type of service, age and sex are deflated by the GNP deflator to obtain the real average cost of each of the six health-care services by age and sex in 1972 dollars. This adjustment ensures that Medicare cost esti- mates are measured consistently with estimates from the Macroeconomic Growth Model. These estimates are shown in Table 9-3. The coefficient for real expenditures-per-capita for each service, age and sex was set equal to the value of that ratio in 1977 for the entire simulation period. The coeffi- cients are then multiplied by the projected population in each year of the simulation period, 1970 through 2055, to obtain projections of Medicare costs by age, sex and health care service. Algebraically the projection method- ology may be written as follows: NE;, = (Eu/P;) NP, (9-1) where: NE; = Projected expenditures in 1972 dollars for age i, sex j, and service k. Ej = 1977 expenditures in 1972 dollars for age i, sex j, and service k. P; = 1977 population of age i and sex j. NP; = Projected population of age i and sex j. Source: ICF calculations from Medicare claims data. This simplified approach is intended to provide a start- ing point for a more complete representation of the deter- minants of health care costs. By assuming constant real Medicare costs per person by age and sex, this model only highlights the effects of demographic change in Medicare expenditures. These projections provide a benchmark for comparisons with models permitting expenditures-per- capita to vary and for comparisons with retirement in- come estimates. However, in its current form, it is not well suited for projecting trends in the Medicare program un- der a variety of assumptions about changes in benefits or the health delivery system which might have major effects on average Costs. An Alternative Approach All health care expenditures have increased rapidly over the last 15 years. As illustrated in Table 9-4 below, total health expenditures per capita in 1972 dollars increased by more than five percent per year over the 1965-79 peri- od. However, real Medicare expenditures per capita in- creased by approximately nine percent per vear, CHAPTER 9 Medicare Model Table 9-3 Medicare Costs Per Capita by Age, Sex and Service, 1977 (1972 Dollars Per Person) Inpatient Skilled Home Physician Outpatient Home Age Hospital Nursing Health A and Other Hospital Health B Males 35 1.17 0.00 0.01 0.33 0.54 0.01 35-44 6.21 0.03 0.04 1.87 2.33 0.03 45-54 12.66 0.07 0.11 3.81 3.29 0.06 55-59 25.51 0.17 0.27 7.68 4.28 0.11 60-64 48.81 0.44 0.61 14.04 6.10 0.24 65-66 295.93 2.34 3.23 108.10 18.38 1.01 67-68 303.47 3.17 3.60 108.06 19.79 1.09 69-70 353.49 3.78 4.54 122.78 22.27 1.21 71-72 377.94 5.11 5.86 132.65 21.72 1.77 73-74 446.85 5.89 6.98 149.58 24.00 2.04 75-79 516.92 «9.05 8.58 160.87 21.53 3.12 80-84 560.47 14.02 11.55 159.81 18.09 3.51 85+ 603.03 21.44 12.80 153.55 17.18 5.83 Females 35 0.89 0.00 0.01 0.28 0.45 0.00 35-44 3.93 0.02 0.05 1.29 1.71 0.03 45-54 9.20 0.06 0.13 3.02 2.95 0.07 55-59 19.90 0.19 0.33 6.36 4.13 0.15 60-64 31.64 0.36 0.58 9.67 4.90 0.24 65-66 232.55 2.54 3.36 92.01 19.04 1.19 67-68 237.24 2.80 3.82 92.73 17.52 1.34 69-70 272.12 3.51 5.05 104.41 19.41 1.75 71-72 294.71 4.87 5.75 111.00 19.58 2.30 73-74 360.79 7.40 8.26 128.41 20.44 2.45 75-79 434.94 12.24 10.58 140.91 19.06 3.77 80-84 504.76 20.18 13.37 147.96 18.01 5.46 85+ 519.57 29.10 13.71 138.94 17.67 7.37 somewhat more rapidly than the average. The assumption of constant per capita real Medicare expenditures, used in the simple version of the model, is not consistent with the historical experience of the program thus far. Great con- cern has been expressed about the underlying causes of this trend and the expectations for this trend in the future. Over the last 15 years, a number of factors affected the rapid growth in Medicare expenditures per capita, includ- ing greater life expectancies for the eligible population, broader benefits with possibly lower out-of-pocket ex- penses, and general patterns of health care cost increases that have affected all individuals. These general increases have stemmed from basic characteristics of the health system such as the terms of reimbursement for services, changes in health system capacity, the use of new technol- ogy and treatment techniques, and price changes. In the Medicare Model, real Medicare expenditures per capita for each age, sex, and service are held constant over the entire projection period. In an alternative analysis, we "estimated expenditures per capita over the forecast peri- +:57 od as a function of the relevant population size, per capita income level and a time trend. For each age-sex group, the following equation for expenditures per capita was estimated: THE NATIONAL INSTITUTE ON AGING MACROECONOMIC-DEMOGRAPHIC MODEL Table 9-4 Trends in Real Per Capita Health Expenditures, 1965-19792 (1972 dollars) Source of Payment 1965 1970 1975 Third Party Payments Private $ 59 $ 83 $114 Medicare — 37 57 Medicaid — 49 49 Other 59 36 68 Subtotal 118 206 288 Direct Payments 126 137 137 Total $245 $342 $426 Average Annual 1979 Rate of Growth $137 6.0% 80 8.6 59 2.1 75 1.7 351 7.8% 164 1.9% $515 5.3% aThe figures are for total expenditures divided by total population. They are not comparable to the figures in Table 9-3. Source: Health Care Financing Review 2 (1) Summer 1980. (E/N) = f; (Ni, Yo, ©) (9-2) (E/N)y; = Medicare expenditures per person of age- sex i on medical service j, in year t. Ni Population of age-sex group i, in year t. Y, = Aggregate real, personal, disposable income per capita in year t. t Year. Each equation was estimated by ordinary least squares for annual observations over the period 1966-1977, assuming a linear specification for f.! The data which we used were disaggregated by sex, age group, and service type. Changes in the Medicare program in 1969 necessitated a transformation in some of the 1966-1968 data. Prior to 1969, Medicare provided outpatient hospital services through both Hospital Insurance (Part A), -and Supplementary Medical Insurance (Part B). Thus, the data tables from those years have separate entries for outpatient services under the two parts of the program. For comparability with later years’ data, we combined the Part A and Part B outpatient services data for 1966 through 1968 into a single category for each year. However, these equations have not vet been used to project Medicare expenditures in the current version of the Medicare Model. The period 1966-1977 was the start- up period of the Medicare System. The rapid growth of Data from the Office of Research, Demonstrations, and Statistics in the Health Care Financing Administration. Data for the vear 1970 were unavailable. expenditures per capita over that period may not be char- acteristic of the behavior of a mature system. Consequent- ly, the equations estimated using 1966-1977 data are not reliable over the long periods required in the Macroeco- nomic-Demographic Model. There are certain limitations in the Medicare data avail- able through HCFA which explain some of the difficulties we encountered in developing reliable projections from SQ equation 9-2. Most significantly, no data are available from HCFA for 1970. HCFA analysts maintain that the 1970 data are not reliable enough to be made available for research or publication. Another limitation is created by lack of disaggregation by service. Inpatient hospital services are not disaggregated into components such as room and board, ancillaries, or surgery. The category “Physician and Other Medical Services” includes a broad range of ser- vices provided by physicians, specialists, and allied health professionals. Additionally, utilization data such as days of care and number of visits are limited to the following: e inpatient hospital days of care, 1977-1980 e skilled nursing facilities days of care, 1969-1980 ® home health number of visits, 1974-1980. Demographic breakdowns of utilization data are not avail- able for any service. Chapter 10 Base Case Simulation Introduction he Macroeconomic-Demographic Model is designed to | analyze the long-term implications of demographic, eco- / nomic, and policy changes for the U.S. retirement income system. Consideration of any set of policy changes re- quires a standard of comparison—often referred to as a base case simulation. A base case simulation is a projec- tion of what the future might be like if present trends or policies continue. A policy alternative or other change may be evaluated by altering the base case assumptions to reflect the policy or change of interest, performing a new simulation, and comparing the results of the new simula- tion with those of the base case. The base case projections describe one potential path of future development for the retirement income system. These projections are not predictions of the future values of the variables. Rather, they are projections of what these values may be if trends continue along the paths assumed, population growth and technological change follow the courses assumed, and there are no major institutional changes. Simulations have been developed using alterna- tive assumptions regarding future fertility rates, mortality, and technological change. The base case described in this chapter appears to be a reasonable intermediate case. The future behavior of the economy and the retirement income system can be divided into two periods. The peri- od 1980-2030 will witness the passing through of the life cycle of individuals already born. Because a large majority already live to retirement ages, future reductions in mor- tality, which are expected, will not greatly change the size and structure of the working age population. Consequent- ly, we can project, within a reasonable range, the size and structure of the labor force until about 2010. We project a continuation of the declining trend in mortality. This trend implies an increase in the size of the elderly popula- tion. Most of the change in the size and structure of the elderly population through the year 2040, however, will result from changes in fertility that have already occurred. Because of the great fluctuations in fertility that have oc- curred since the 1930’s—especially the post-World War IT baby boom—the age structure of the population will vary greatly over this period. About the year 2000, individuals will begin to enter the labor force who had not yet been born in 1983. Our knowledge of future fertility is even more uncertain than that for mortality, and changes in fertility can have a great- er impact on the size and structure of the population than changes in mortality. Consequently, after about 2010, pro- jections of the size and structure of the labor force, which influence all other macroeconomic variables, are more sensitive to assumptions about the future course of demo- graphic parameters, especially fertility. We assume that fertility rates trend smoothly to a given ultimate rate—2.1 births per female in this base case simulation—and then remain constant. Consequently, the growth and structure of the population are characterized by smooth trends after about 2020. Similarly, we assume a smooth trend rate of productivity growth after the mid 1980's. These smooth demographic and productivity trends begin to dominate the behavior of the model after the 2020's. In the decade after the year 2010 the large baby boom cohorts begin to reach retirement age. After 2010, there- fore, the behavior of the retirement income system differs greatly from its behavior before 2010. For all of these reasons, it is useful to divide the projection period into the period 1980-2010 and the period 2010-2055 and ex- amine the behavior of the economy and the retirement income system in each of those two periods. In addition to projections for the period 1980-2055, we simulated the model for the period 1970-1980. Compari- son of those simulations with actual data for that period provides one test of the validity of the model. Population Estimates The Population Model is a comprehensive system that projects the size and composition of the national popula- tion, given a set of demographic parameters. For input into the other component models of the MDM for the 1970-1979 simulations, actual population data for 1970- 1979 are used. Population projections start from a base year of 1979, because this is the latest year for which adjustment factors for Census undercounts were available at the time of this research. Table 10-1 presents the base case population projec- tions. The total U.S. population is projected to grow to 331 million in 2055, a 49 percent increase from its 1980 pro- jected value of 223 million. This growth does not occur evenly across the age groups of the population. In 2055, the 16 to 24 year old age group is projected to have 39 million people, only 4 percent more than it had in 1980. The elderly age group (65 years and older) grows by 162 percent and is projected to have 66 million persons in 2055. Table 10-1 also shows the average annual growth rates implied by the population estimates. The average annual growth rate of the total population from 1980 to 2055 is projected to be 0.5 percent. The two older age groups, those ages 55-64 and those over age 65, exhibit annual growth rates above the average for the total population. THE NATIONAL INSTITUTE ON AGING MACROECONOMIC-DEMOGRAPHIC MODEL Table 10-1 Projected U.S. Population By Age Group, 1980-2055 (Millions of Persons) Year 0-15 1624 2554 5564 G5+ Total 1980 53.9 37.6 84.7 21.2 25.1 2225 1990 58.4 31.5 103.2 20.8 30.1 244.0 2000 60.8 32.6 112.0 23.4 32.5 261.3 2010 60.2 36.0 111.9 33.2 36.1 277.4 2020 64.0 34.0 110.4 37.6 47.3 293.3 2030 64.3 37.1 113.5 319 58.7 305.5 2040 66.4 37.3 118.1 33.8 60.4 316.0 2050 68.5 38.1 119.7 36.8 62.7 325.8 2055 69.1 39.1 121.7 35.6 65.7 331.2 Rate of Growth 1980-2010 037% 0.15% 0.93% 1.51% 1.22% 0.74% 2010-2055 0.31 0.19 0.19 0.15 1.34 0.39 1980-2055 0.33 0.06 0.48 0.69 1.29 0.53 Source: Projections of the Macroeconomic-Demographic Model (Appendix Table I-1). Table 10-2 Population Distribution and Dependency Ratios, 1980-2055 Percent of Population Dependency Ratio Year 0-17 18-64 65+ 1980 27.9 60.8 11.3 1990 26.5 61.2 12.4 2000 26.3 61.3 12.4 2010 24.5 62.5 13.0 2020 24.4 59.5 16.1 2030 23.8 57.0 19.2 2040 23.6 57.3 19.1 2050 23.7 57.1 19.2 2055 235 56.7 19.8 Youth Elderly Total Labor Force 458 186 .644 .599 433 .202 .635 550 429 203 .632 513 391 .208 .599 .530 410 271 .681 .600 417 337 754 648 412 333 745 655 415 337 752 .689 414 .350 764 .708 Source: Projections of the Macroeconomic-Demographic Model (Table 1-2). The growth rate of the elderly population is particularly rapid, 1.3 percent per year. The younger age groups ex- hibit the slowest growth rates. The group age 16 to 24, which is relatively large in 1980 because in that year it contains some of the baby boom cohorts, scarcely grows at all during the projection period. Figure 10-1 illustrates the projected trends for the four adult population groups shown in Table 10-1. The young group (16-24) peaks in size in 1980. Its size falls and the size of the middle-age group (25-54) rises in the 1970's and 1980's as the baby boom cohorts successively reach their 25th birthdays. The rapid increase in the size of the middle-aged group ends in the year 2000. Shortly after the year 2000 the 55-64 years age group begins to grow. Final- ly the baby boom cohorts begin to enter their elderly years in 2010. The size of the elderly group increases rapidly from 2010 to 2030. j Table 10-2 shows the projected percentage distribution of the population from 1980 to 2055. Population groups are redefined in this table into three groups: (1) age 0 to 17, (2) age 18 to 64, and (3) age 65 and over. In 1980, elderly persons (those age 65 and over) represented about 11 percent of the population, while young persons (those under age 18) represented 28 percent of the popu- lation. After the year 2030 the relative sizes of these two groups in the population are more nearly equal. In that year, 19 percent of the population are over age 65 and about 24 percent are under age 18. The youth dependency ratio, shown in the fourth col- umn of Table 10-2, is the ratio of the number of persons under age 18 to the number of persons age 18-64 in the population. It can be calculated from the data in Table 10- 2 as the ratio of the first to the second column. The elderly dependency ratio shown in column five is the number of CHAPTER 10 Figure 10-1 Trends in U.S. Population by Age 130 016-24 120 ~ @ 25-54 | 05564 110 @ 65 + 100 90 80 70 60 50 MILLIONS OF PERSONS 40 30 20 1 { 1 | 1 1 YEARS Source: Table 10-1. 0 1970 1980 1990 2000 2010 2020 2030 2040 2050 Base Case Simulation persons age 65 and over divided by the number of per- sons age 18-64, or the ratio of the third to the second column. The total dependency ratio in column six is the sum of the youth and elderly dependency ratios. The last column of Table 10-2 shows a different depen- dency ratio concept. Using estimates from the Labor Mar- ket Model, discussed later in this chapter, this labor force dependency ratio is defined as the ratio of all persons out of the labor force to persons age 16 and older who are in the labor force. It therefore summarizes information about both demographic change and future trends in la- bor force participation. This dependency ratio increases from .6 in 1980 to .7 in 2055. This ratio falls until the year 2000, then begins to increase. It increases rapidly after 2010 when the baby boom cohorts reach retirement age. Figure 10-2 displays graphically the trends in the youth and elderly dependency ratios. The youth ratio drops sharply through the early 1980's, as the baby boom moves into adulthood. The elderly dependency ratio begins to increase rapidly in 2010, as the same baby boom popula- tion reaches retirement age. Table 10-3 compares the base case population projec- tions of the MDM to a recent Census Bureau projection and a projection by the Social Security Administration’s Office of the Actuary. These three projections are devel- oped using the same methodology, described in Chapter Table 10-3 Comparison of Projected Populations in 2040—Macroeconomic-Demographic Model, Census Bureau, and Social Security Actuary Population Percent (Millions of Persons) Distribution MDM Census SSA MDM Census SSA Male 0-15 34.0 34.8 34.61 10.7% 11.0% 10.5% 16-24 18.9 19.1 19.14 6.0 6.0 5.8 25-54 58.2 58.3 60.6 18.4 18.5 18.5 55-64 16.4 17.3 17.2 5.2 5.5 5.2 65+ 24.6 22.1 27.3 7.8 7.0 8.3 Total Male 152.1 151.6 158.8 48.1% 48.0% 48.4% Female 0-15 32.4 33.2 33.12 10.3% 10.5% 10.1% 16-24 18.4 18.5 18.44 5.8 5.9 5.6 25-54 59.9 60.1 60.3 19.0 19.0 18.3 55-64 ) 17.5 18.7 18.0 5.5 5.9 5.5 65+ 35.7 33.4 39.9 11.3 10.6 12.2 Total Female 163.9 164.0 169.7 51.9% 52.0% 51.6% Total 316.0 315.6 328.5 100.0% 100.0% 100.0% anterpolated from reported figures. Sources: Projections of the Macroeconomic-Demographic Model; Bureau of the Census, Current Population Survey, “Projections of the Population of the United States, 1977 to 2050,” Series P-25, No. 704, Table 11, (July 1977), Series II case; and Social Security Administration, Office of the Actuary, “Social Security Area Population Projections, 1981,” Actuarial Study Number 85, Table 20G, (July 1981), Alternative II case. 61 THE NATIONAL INSTITUTE ON AGING MACROECONOMIC-DEMOGRAPHIC MODEL Figure 10-2 Trends in the Dependency Ratio 0:5 0 YOUTH @® ELDERLY 0.45 0.4 0.35 0.3 0.25 FRACTION OF TOTAL POP. 0.2 0.15 1 1 1 1 1 1 | 1970 1980 1990 2000 2010 2020 2030 2040 2050 YEARS Source: Table 10-2. 2. Projections for the year 2040 are shown, because that is the latest common year presented in both the Census Bureau and SSA projections. These comparisons demon- strate that all three population projections are quite similar. Both the Census Bureau and SSA projections shown in Table 10-3 are intermediate cases. All three projections assume an ultimate completed fertility rate of 2.1 births per female. However, the SSA and MDM projections as- sume a more rapid decrease in the rate of mortality than does the Census Bureau projection. Mortality rates in 2055 are 36 percent lower (on an age-adjusted basis) than their 1978 levels in the SSA and MDM projections, while mortal- ity rates decrease only 19 percent from 1976 to 2055 in the Census Bureau projection. One other difference among the projections is the defi- nition of the population group projected. The Census Bureau and the MDM project the United States resident population and military members stationed overseas. The SSA's Office of the Actuary adds to these estimates the additional population participating in the social security system, specifically, the residents of Puerto Rico, the Vir- gin Islands, Guam, American Samoa, and U.S. civilians working abroad. The MDM base case projected population in 2040 is slightly larger, by 0.13 percent, than that of the Census Bureau projection. Absolute comparisons between the MDM and SSA projections are not meaningful because of the differences in the base year population definitions described above. Comparisons of the population distribu- 62 Figure 10-3 Trends in Major Macro Variables 3.5 0 GNP 3.0 2.5 2.0 1.5 1.0 TRILLIONS OF 1972 DOLLARS 0.5 0.0 1 1 1 1 1 1 1 1970 1980 1990 2000 2010 2020 2030 2040 2050 YEARS Source: Table 10-6. tions show that both the MDM and SSA projections have relatively more elderly persons than the Census Bureau projection, reflecting the greater decline in mortality. The SSA projection has relatively more elderly in 2040 than the MDM due to the slightly different structure of the base year population. In summary, the Population Model projects a moderate increase in the size of the United States population during the period 1980-2055. In 2055, the model projects total U.S. population to be 331 million persons. The age distri- bution of the population is projected to change substan- tially. The elderly age group grows at an average annual rate of 1.3 percent. Between 1980 and 2055 its fraction of the total population rises from 11 to 20 percent. The aging of the baby boom cohorts, those born between 1946 and 1965, influences the demographic structure of the popula- tion throughout the simulation. The growth in the elderly population is particularly rapid from 2010 to 2030 as these cohorts reach retirement age. Macroeconomic Growth Estimates This section examines the base case results for nine of the principal variables of the Macroeconomic Growth Model. It first examines four variables for aggregate output: gross national product (GNP) and three of its components— consumption, investment and government spending. It also examines two variables for inputs to aggregate na- CHAPTER 10 Base Case Simulation Table 10-4 Projected Values of Principal Macroeconomic Growth Variables, 1980-2055 (Billions of 1972 Dollars) Government Year GNP Consumption Investment? Spending Spending? 1980 $1423.4 $975.2 $162.6 $262.8 $297.5 1990 1688.1 1100.3 253.9 324.5 408.3 2000 1998.4 1223.8 389.2 381.2 516.1 2010 2299.1 1435.6 430.7 431.9 589.5 2020 2526.9 1659.2 397.4 469.3 621.4 2030 2744.1 1802.3 431.5 509.3 678.9 2040 2996.0 1919.4 514.3 561.3 785.3 2050 3195.2 2008.9 584.1 601.2 881.6 2055 3290.1 2045.9 623.1 620.1 931.6 Rate of Growth 1980-2010 1.61% 1.30% 3.30% 1.67% 2.31% 2010-2055 0.80 0.79 0.82 0.81 1.02 1980-2055 1.12 0.99 1.81 1.15 1.53 In the accounting system of the Macroeconomic Growth Model, savings includes purchases of durable goods, but investment does not. Source: Projections of the Macroeconomic-Demographic Model (Table 1-16). tional production: the size of the capital stock and hours worked by the total labor force. Finally, it examines the major determinant of the size of the capital stock, aggre- gate savings. Table 10-4 shows the trends in GNP and its major com- ponents throughout the simulation period. GNP grows from $1.4 trillion in constant 1972 dollars to $3.3 trillion in 2055. This growth represents an average annual real increase of 1.1 percent. The growth rates in the three components of GNP-consumption, investment, and_gov- ernment spending—are about equal to the growth rate of GNP itself? No large shifts in the distribution of GNP across these categories are projected. Figure 10-3 displays the trends in GNP, investment and consumption graphically. By the end of the simulation period, consumption has increased slightly as a percent- age of GNP from 61 percent (1970) to 62 percent (2055) of GNP.? The share of GNP accounted for by Federal, state, and local government purchases of goods and services (19 percent) remains constant throughout the period. (Government purchases of goods and services do not include transfer payments, such as social security, which are projected to rise sharply after 2015.) Tables I-15 and I-16 show these and other macroeconomic variables in more detail. “The fourth component of GNP—net exports—is exogenous to the mod- el and is not shown. 3The growth rates in investment between 1980-2010 are somewhat mis- leading, because the cyclical down turn in investment in 1980 produced a value that is low relative to its long term trend. Simulated investment decreases 15 percent between 1979 and 1980 then increases 20 percent between 1980 and 1985. (Actual investment fell 12.5 percent between 1979 and 1980.) A better indicator of the investment trends in the early years of the simulation is the 1970 to 2010 growth rate, 2.39 percent annually. 63 The fifth column of Table 10-4 shows the trend in aggre- gate saving® for various periods during the simulation. Aggregate saving rates are sensitive to the composition of the population. The percentage of the GNP saved in- creases as the baby boom cohorts approach retirement during the late years of the twentieth and early years of the twenty-first century. The annual rate of increase in savings exceeds the annual rate of increase in GNP from 1980 to 2010. The rate of growth in savings slows after 2010 when the baby Room cohorts enter retirement, but the propor- tion of GNP saved does not decline. From 2010 to 2055 approximately 25 to 28 percent of GNP is saved—about equal to the simulated 1970 gross saving rate of 25 percent. The growth rates of both GNP and consumption are estimated to decline in the later years of the simulation primarily because of the projected slowing of population growth. The. average growth rate in GNP from 2010-2055, 0.80 percent, is only half of the 1980-2010 growth rate, 1.61 percent. Similar decreases occur in the growth rates of all other variables displayed in Table 10-4. Table 10-5 displays the MDM estimated rates of growth in real GNP and three sets of assumed rates of growth in real GNP used in recent projections of the Board of Trust- ees of the Federal Old-Age and Survivor's Insurance (OASI) and Disability Insurance (DI) Trust Funds. The MDM estimated growth rates are comparable to the rates assumed for the Board of Trustee’s Alternative III, their “The estimate shown in column five corresponds to the saving concept defined by the Hudson-Jorgenson accounting system described in Chapter 3. It includes purchases of consumer durable goods and is not comparable to National Income and Product Accounts (NIPA) saving data. The investment variable reported in Table 10-4 corresponds to the NIPA definition, so it is not equal to the savings variable in Table 10-4. THE NATIONAL INSTITUTE ON AGING MACROECONOMIC-DEMOGRAPHIC MODEL Table 10-5 Comparisons of Selected Annual Growth Rates in Real GNP Estimated in the Macroeconomic- Demographic Model and Assumed By the Social Security Actuary, 1981-20552 Social Security Trustees Report Year MDM Alt I-A Alt II'B Alt IIT 1981 —0.2% 1.1% 1.1% 0.7% 1982 1.4 4.2 3.7 1.1 1983 23 5.0 3.5 22 1984 25 45 29 39 1985 1.6 4.2 29 3.0 1990 1.8 3.4 3.0 2.4 1995 15 2.8 2.4 23 2000 1.8 31 2.7 2.2 2055 0.6 25 21 0.9 Source: Projections of the Macroeconomic-Demographic Model; 1981 Annual Report of the Board of Trustees of the Federal Old-Age and Survivor's Insurance and Disability Insurance Trust Funds, Table 10, p. 29. aThis table reports simulations beginning in 1979. Because the MDM does not depict the cyclical behavior of the economy, these simulations do not reflect the depth of the recession that occurred in 1981-1982. Table 10-6 Projected Values of Macroeconomic Outputs and Inputs, 1980-2055 GNP : Capital Stock (Billions of GNP Per Capita Hours Worked (Billions of Year 1972 Dollars) (1972 Dollars) (Billions) 1972 Dollars) 1980 $1,423.4 $0,397.8 146.9 3,700.3 1990 1,688.1 6,919.6 170.4 4,333.4 2000 1,998.4 7,647.9 185.1 5,356.5 2010 2,299.1 8,289.2 192.3 6,527.2 2020 2,526.9 8,615.2 191.0 7,443.3 2030 2,744.1 8,983.4 191.1 8,203.5 2040 2,996.0 9,480.3 193.9 9,234.7 2050 3,195.2 9,807.6 191.3 10,476.8 2055 3,290.1 9,935.2 190.2 11,129.0 Rate of Growth 1980-2010 1.61% 0.87% 0.90% 1.91% 2010-2055 0.80 0.40 —0.02 1.19 1980-2055 1.12 0.59 0.35 1.48 Source: Projections of the Macroeconomic-Demographic Model. more “pessimistic” scenario. In the early 1980’s the MDM estimated rates of growth of real GNP do not exceed 2.5 percent. In all three of the SSA Trustee's alternatives growth rates are generally higher in these years. In the years from 2000 to 2035, the disparity between the MDM growth rates and the Alternative III growth rates is less. The MDM growth rates range from 0.6 to 1.8 percent during this period, slightly lower than the range of growth rates in Alternative III, 0.9 to 2.2 percent. Table 10-6 compares the growth in GNP shown in Table 10-4 to the growth in the two major factor inputs, labor and capital. It also shows GNP per capita. By dividing GNP by the projected population size, GNP per capita may provide a better measure of the real economic status of the population. The major determinants of long-term growth of the economy are the growth of labor and capital input and technological change. The major cause of slower growth rates after 2010 is the slower rate of increase in labor input, measured as total hours worked: in the economy. The number of hours worked actually declines slightly between 2010 and 2055. The economy manages to grow 64 CHAPTER 10 despite the declining number of hours worked, because the capital stock continues its growth and technical pro- gress increases labor productivity. Because the decrease in GNP’s rate of growth is associated with a slower rate of population growth, the percentage point drop in the growth rate of GNP per capita is smaller than the drop in the growth rate of aggregate GNP. The growth rates for GNP assumed in the social security Trustees Report, dis- played in Table 10-5, also decline after 2010. Besides comparisons with other long-run macroeco- nomic projections, the MDM can be tested by simulating the 1970 to 1980 period and comparing simulated results to actual historical experience. These comparisons are subject to two limitations. First, the model was not de- signed to simulate short term economic fluctuations. Its inability to simulate accurately year-to-year cyclical changes does not invalidate its use in simulating long-run trends. Second, the model is designed to use the best available information at each point in its solution proce- dure. When simulating years 1970 to 1979, it uses actual, not simulated values for the previous year’s variables. Thus, historical comparisons of 1970 to 1979 performance represent ten individual tests of the model's performance and not a test of its ability to forecast a ten year movement in the economy. Table 10-7 compares the simulated values of selected variables of the Macroeconomic Growth Model with actu- al values for these variables over the historical period, 1970-1979. Despite its long-term orientation, the model simulates the actual historical experience of the economy reasonably well. Unlike demand-oriented macroeconom- ic models, the MDM is not designed to simulate short-run cyclical fluctuations of the economy. Thus, the MDM esti- mates for 1975, a recession year, and 1976, a recovery year, differ from the actual values more substantially then in all other years, particularly for investment. . As one measure of forecasting reliability, the model's root mean square error’ of estimate for GNP over the period is only 1.8 percent. The model's accuracy in simu- lating consumption from 1970 to 1979 is comparable— the root mean square error of estimate is 3.1 percent. The large fluctuations in investment make this variable more difficult to simulate accurately in the short-run. The mod- el’s root mean square error of estimate for investment, 11.6 percent, is higher than for GNP and consumption, but is comparable to that of other macroeconomic simulation models. In summary, the base case simulation of the Macroe- conomic Growth Model projects a moderate rate of in- crease in real GNP over the next 75 years. In this base case, GNP grows at an average annual rate of 1.1 percent from 1980 to 2055. The growth rate of GNP is slower in the >The root mean square error is defined as the square root of the mean of the squared deviations between the simulated and actual values of any time series of results. It is approximately equal to the mean absolute error of the estimate. It is a standard indicator of the error in a simulated variable. 65 Base Case Simulation Table 10-7 Comparisons of Simulated and Actual Values of Principal Macroeconomic Variables 1970-19792 (Billions of 1972 Dollars) GNP Percent Year Simulated Actual Difference 1970 $1,068.06 $1,075.3 —0.62% 1971 1,094.3 1,107.5 -1.19 1972 1,149.2 1,171.1 -1.87 1973 1,228.8 1,235.0 —0.50 1974 1,253.8 1,217.8 2.97 1975 1,203.3 1,202.3 0.08 1976 1,224.0 1,273.0 —3.84 1977 1,317.7 1,340.5 —-1.70 1978 1,388.9 1,399.2 —0.74 1979 1,436.9 1,428.4 +0.60 Consumption 1970 $653.2 $668.9 —2.34% 1971 683.8 691.9 -117 1972 728.3 733.0 —0.64 1973 774.6 767.7 0.90 1974 782.6 760.7 2.88 1975 744.1 774.6 e393 1976 781.9 820.6 —4.71 1977 840.5 861.7 —2.46 1978 928.0 900.8 3.01 1979 972.1 924.5 5.15 Investment 1970 $166.9 $154.7 7.89% 1971 175.7 166.8 5.33 1972 189.3 188.3 0.53 1973 197.7 207.2 —4.58 1974 205.6 183.6 11.98 1975 181.6 142.6 27.35 1976 194.1 173.4 11.94 1977 220.1 200.1 10.00 1978 199.8 2143 -6.77 1979 191.3 200.8 -7.50 The actual National Income and Product Accounts data shown predate the recent revisions published by the Department of Commerce. The values shown here correspond to the data used to estimate the MDM. Sources: Projections of the Macroeconomic-Demogaphic Model and the 1981 Economic Report of the President, U.S. Government Printing Office, Washington, D.C., Table B-5. second half of the simulation period because the growth in the size of the labor force is considerably reduced in those years and there is no growth at all in hours worked. In this simulation, the MDM-estimated rates of growth in GNP are comparable to those assumed in the pessimistic scenario used by the Social Security Actuary to project trust fund balances in the OASI and DI systems. Alternative assumptions yield projections that correspond more closely to the more optimistic projections of the Actuary. THE NATIONAL INSTITUTE ON AGING MACROECONOMIC-DEMOGRAPHIC MODEL Table 10-8 Projected Values of Aggregate Labor Market Variables, 1980-2055 Labor Hours Force Employment Worked Year (Millions) (Millions) (Billions) 1980 105.5 97.4 146.9 1990 119.1 112.9 170.4 2000 132.5 125.8 185.1 2010 142.0 134.9 192.3 2020 143.4 136.7 191.0 2030 146.3 139.6 191.1 2040 150.8 144.3 1939 2050 152.3 146.1 191.3 2055 153.5 147.3 190.2 Rate of Growth 1980-2010 1.00% 1.09% 0.90% 2010-2055 0.17 0.20 —-0.02 1980-2055 0.50 0.55 0.35 Source: Projections of the Macroeconomic-Demographic Model Labor Market Estimates This section examines trends in employment and com- pensation levels for both the aggregate labor force and selected demographic groups. Because the Macroeco- nomic Growth and Labor Market Models operate simulta- neously, estimates of the values of aggregate labor input and compensation rates are determined by the simulta- neous solution of both models. However, the demogra- phic detail of the Labor Market Model is substantially greater than that of the Macroeconomic Growth Model.® Table 10-8 shows the projected trends in the aggregate variables of the Labor Market Model. The size of the total labor force grows from about 105 million in 1980 to 154 million in 2055. Ninety-three percent of this growth oc- curs by the year 2010. A similar increase occurs in employ- ment levels. The estimate of total hours worked increases less rapidly, because average hours worked per worker falls. Total hours worked actually falls slightly after 2010. Average real compensation per hour is projected to grow at a steady rate of 1.5 percent per year over the course of the simulation. Annual compensation increases less rapid- ly than hourly compensation, because the average worker works fewer hours per year. The most interesting results of the Labor Market Model are its age and sex specific projections of employment and compensation rates. Table 10-9 shows the projected trends in employment levels for males and females over and under 65 years of age. Female employment levels are Tables 1-3 through I-14 in Appendix 1 show projections of the Labor Market Model. They display trends in labor force participation, employ- ment levels, unemployment rates, and compensation for twelve age-sex groups. 66 Average Hours Compensation Annual Worked Rate Co tion per Worker (1972 Dollars) (1972 Dollars) 1,508 $5.51 $8,312 1,509 6.25 9,434 1,471 7.47 10,998 1,426 8.62 12,287 1,397 9.59 13,398 1,369 10.82 14,806 1,344 12.79 17,185 1,309 15.37 20,123 1,291 16.86 21,763 —0.19% 1.50% 1.31% -0.22 1.49 1.26 —-0.21 1.50 1.29 projected to increase more rapidly than male levels, so that in 2055 female employment of 72.8 million is only slightly less than male employment of 74.5 million. This increase in female employment relative to male employ- ment does not occur in all age groups. Elderly male em- ployment levels increase more rapidly than non-elderly employment levels, reflecting the increase in elderly male population levels. Elderly female employment levels de- crease between 1980 and 2010 because of a decrease in labor force participation rates of elderly females.” The continued decrease in elderly female labor force partici- pation rates offsets much of the growth in the elderly female population from 2010 to 2055. Table 10-10 shows the trends in compensation rates for six age groups over the entire simulation period. Figure 10-4 provides a graphical presentation of the data in Table 10-10. The compensation rate of the middle-age worker (35 to 44) nearly triples over the course of the simula- tion—starting at $6.45 per hour (in 1972 dollars) and ending at $18.56 per hour. The average hourly compensa- tion rate for all workers increases at an average annual rate of 1.50 percent between 1980 and 2055. The annual rate of compensation increase is greatest for workers above the age of 65, 1.72 percent. The compensation rate of elderly workers increases relative to other groups be- cause this age group is relatively small and stable in the years 1980-2010 when macroeconomic growth is most rapid. Table 10-11 shows the hourly compensation rate of each age group relative to the compensation rate of work- ers age 35-44. The relative compensation of young work- Elderly female participation rates decrease in part because of increased future coverage of women by social security and employer pension. CHAPTER 10 Table 10-9 Projected Employment By Age and Sex, 1980-2055 (Millions) Base Case Simulation et Male Female All 65 or Sub 65 or Sub 65 or Sub Year 16-24 25-64 Over Total 16-24 25-64 Over Total 16-24 25-64 Over Total 1980 12.1 44.4 1.9 58.4 9.7 28.2 1.1 39.0 21.8 72.6 3.0 97.4 1990 10.8 53.0 2.1 65.9 9.3 36.2 1.4 47.0 20.1 89.2 35 112.9 2000 11.1 55.9 1.8 68.8 10.8 45.1 1.1 57.0 219 101.0 2.8 125.8 2010 12.5 56.7 2.1 71.3 12.2 50.6 9 63.6 24.6 107.3 3.0 134.9 2020 11.8 56.4 29 71.0 115 52.9 1.1 65.6 23.3 109.3 4.1 136.7 2030 12.8 55.4 34 71.8 12.6 53.9 13 67.8 25.4 109.3 4.8 139.6 2040 12.8 57.0 34 73.2 12.6 57.2 1.2 71.1 25.5 114.2 4.6 144.3 2050 13.0 57.1 3.7 73.9 12.8 58.0 1.3 72.2 25.8 115.1 5.1 146.1 2055 13.4 57.3 3.9 74.5 13.2 58.2 1.4 72.8 26.6 115.5 5.4 147.3 Rate of Growth 19802010 0.11% 082% 033% 067% 077% 197% —067% 164% 040% 131% 000% 1.09% 2010-2055 0.15 0.02 1.39 0.10 0.18 0.31 0.99 0.30 0.17 0.16 1.31 1.96 1980-2055 0.14 0.34 0.96 0.33 0.41 0.97 0.32 0.84 0.27 0.62 0.79 0.55 Source: Projections of the Macroeconomic-Demographic Model (Table 1-7). Figure 10-4 Average Hourly Compensation Rates 18 17 16 15 | 016-24 @ 35-44 055-64 B65 + we & 3 Q Q RN 9 — 8 - 6 5 4 3 2 1 1 1 1 L 1 1 1970 1980 1990 2000 2010 2020 2030 2040 2050 YEARS Source: Table 10-10. ers, age 16-24, remains roughly constant even as they become relatively more scarce during the period 1980- 2000. The own price elasticity of each age group is nega- tive (see Table 4-5, page 44). This would tend to increase the compensation rate of workers age 16-24 as they be- come relatively more scarce. However, the cross price elasticity of workers age 16-24 with workers age 25-54 is Cy also negative. Because the group age 25-54 is growing rapidly during the period 1980-2000, this tends to dampen wage rate growth of younger workers and offset the posi- tive effects on wage rates of the increasing scarcity of the younger age group. The relative compensation of workers age 55-64 falls as that age group increases in size during the period 2000-2020, then recovers as the relative size of the group falls in the period after 2020. Relative compen- sation of workers over age 65 follows the same pattern as the baby boom cohorts reach age 65. Table 10-12 shows the trends in average annual com- pensation per worker for four age groups. Average annual compensation for each age-sex group is equal to the prod- uct of average annual hours worked and compensation per hour. The rates of growth in annual compensation do not vary greatly across the three groups over age 25. While annual hours worked by middle-aged workers (ages 25- 54) do not fall as rapidly as do annual hours worked in other age groups, the hourly compensation rates of elder- ly workers increase more rapidly than the compensation rates for other age groups. Young workers, for whom neither annual hours worked nor compensation rates in- crease relative to other age groups, experience less rapid growth rates in annual compensation levels than the other three groups. The projections of aggregate compensation growth and of wage and salary growth of the MDM can be compared to the assumed growth in real wages used in the projec- tions for the Report of the Board of Trustees of the social security trust funds. Table 10-13 shows the average annual rate of growth in hourly compensation and in wages and salaries as estimated by the MDM. The wage and salary growth rate fluctuates between 0.8 and 2.4 percent from 1982 through 2000. These rates of earnings growth are THE NATIONAL INSTITUTE ON AGING MACROECONOMIC-DEMOGRAPHIC MODEL Table 10-10 Projected Hourly Compensation Rates By Age, 1980-2055 (1972 Dollars) . Year 16-24 25-34 35-44 45-54 55-64 65+ Total 1980 $3.11 $5.40 $6.45 $6.65 $5.97 $3.93 $5.51 1990 3.36 5.93 7.08 7.26 7.12 4.63 6.25 2000 3.86 6.86 8.25 8.53 8.78 5.92 7.47 2010 4.64 8.07 9.61 10.08 9.21 6.49 8.62 2020 © 5.28 9.12 10.71 11.33 9.74 6.94 9.59 2030 5.76 10.13 11.92 12.52 12.17 8.83 10.82 2040 6.81 11.97 14.11 14.59 14.44 10.55 12.79 2050 8.18 14.38 16.99 17.56 17.03 12.47 15.37 2055 8.89 15.68 18.56 19.19 19.25 14.11 16.86 Rate of Growth 1980-2010 1.34% 1.35% 1.34% 1.40% 1.46% 1.69% 1.50% 2010-2055 1.46 1.49 147 1.44 1.65 1.74 1.50 1980-2055 1.41 1.43 1.42 1.42 1.57 1.72 1.50 Source: Projections of the Macroeconomic-Demographic Model (Table 1-12). Table 10-11 Ratio of Hourly Compensation Rates of Selected Age Groups to the Compensation Rate of Prime Age (35-44) Workers, 1980- 2055 Year 16-24 25-34 45-54 55-64 65+ 1980 0.48 0.84 1.03 0.93 0.61 1990 0.47 0.84 1.03 1.01 0.65 2000 0.47 0.83 1.03 1.06 0.72 2010 0.48 0.84 1.05 0.96 0.68 2020 0.49 0.85 1.06 091 0.65 2030 0.48 0.85 1.05 1.02 0.74 2040 0.48 0.85 1.03 1.02 0.75 2050 0.48 0.85 1.03 1.00 0.73 2055 0.48 0.84 1.03 1.04 0.76 Source: Projections of the Macroeconomic-Demographic Model (Table 1-12). Table 10-12 Projected Values of Average Annual Compensation Per Worker By Age Group, 1980-2055 Year 16-24 25-54 55-64 65+ Total 1980 $3468 $ 9921 $ 9861 $ 4897 $ 8312 1990 3661 10773 11616 5596 9434 2000 3976 12453 13936 6774 10998 2010 4599 14182 14294 7064 12287 2020 5061 15490 14880 7263 13398 2030 5331 17108 18289 8920 14806 2040 6009 19764 21326 10168 17185 2050 6802 y 23333 24654 11339 20123 2055 7180 25284 27591 12458 21763 Rate of Growth 1980-2010 0.95% 1.20% 1.25% 1.23% 131% 2010-2055 0.99 1.29 1.47 1.27 1.28 1980-2055 0.96 1.26 1.38 1.25 1.29 Source: Projections of the Macroeconomic-Demographic Model (Table 1-14) 68 CHAPTER 10 Base Case Simulation Table 10-13 Comparisons of MDM Estimated Growth Rates in Real Compensation With Growth Rates in Real Wages Assumed by the Social Security Actuary, 1981-2000 MDM Social Security Actuary Wages and Year Compensation Salaries Alt II-A AltlI-B Alt lll 1981 =3.05 —3.07 -0.9 -09 -11 1982 0.93 0.75 1.5 02 --16 1983 1.43 1.28 2.4 0.7 0.0 1984 2.56 2.42 2.4 0.6 0.7 1985 1.04 0.97 2.4 0.7 0.4 1990 1.41 1.37 2.1 14 0.8 1995 1.82 1.75 2.0 1.5 1.0 2000 1.42 1.14 2.0 1.5 1.0 Source: Projections of the Macroeconomic-Demographic Model; 1981 Annual Report of the Board of Trustees of the Federal Old-Age and Survivor's Insurance and Disability Income Trust Funds, Table 10. Table 10-14 Comparison of MDM Estimated and Actual Labor Force Participation Rates for Selected Groups, 1970-1979 (Percent) Males 16-24 Year Olds 35-44 Year Olds 55-64 Year Olds 65 Years and Older Year Simulated Actual Simulated Actual Simulated Actual Simulated Actual 1970 73.6 69.4 96.8 96.9 85.4 83.0 29.0 26.8 1971 74.5 69.6 96.6 96.5 82.8 82.2 27.1 25.5 1972 73.8 71.3 96.4 96.4 81.2 80.5 259 24.4 1973 73.8 73.0 96.3 96.2 79.9 78.3 24.6 22.8 1974 74.0 73.8 96.1 96.0 77.8 77.4 23.2 22.4 1975 76.1 72.4 95.8 95.6 77.0 75.8 23.1 217 1976 76.3 72.9 95.7 95.4 76.2 74.5 22.6 20.3 1977 76.3 74.1 95.6 95.7 75.4 74.0 225 20.1 1978 74.6 74.9 95.6 95.7 74.5 73.5 20.8 20.5 1979 74.5 75.1 95.5 95.8 73.4 73.0 19.9 20.04 Females 16-24 Year Olds 35-44 Year Olds 55-64 Year Olds 65 Years and Older Year Simulated Actual Simulated Actual. Simulated Actual Simulated Actual 1970 49.1 51.3 49.9 51.1 40.4 43.0 9.8 9.7 1971 50.8 51.3 51.8 51.6 39.8 42.9 9.3 9.5 1972 52.4 52.9 53.0 52.0 40.1 42.1 9.0 9.3 1973 54.0 54.9 54.1 53.3 40.6 41.1 8.7 8.9 1974 55.6 56.5 55.7 54.7 39.8 40.7 8.2 8.2 1975 55.9 57.2 57.3 55.8 37.2 1.0 7.7 8.3 1976 57.2 58.0 58.2 57.8 37.4 411 7.8 8.2 1977 58.6 59.6 59.1 59.6 37.7 41.0 7.7 8.1 1978 60.3 61.8 59.3 61.6 40.0 41.4 7.8 8.4 1979 61.7 62.6 60.2 63.6 40.9 41.9 7.7 8.3 Source: Projections of the Macroeconomic-Demographic Model and the 1980 Economic and Training Report of the President, Tables A- 3 and A-4 69 THE NATIONAL INSTITUTE ON AGING MACROECONOMIC-DEMOGRAPHIC MODEL similar to the rates of growth used by the Social Security Actuary in the two intermediate scenarios, II-A and II-B. Because the compensation rate projections are compara- ble between the MDM and SSA projections, the difference between the MDM and the SSA Trustees’ GNP projections, shown in Table 10-8, is largely attributable to the greater decreases in hours worked by the labor force estimated in the MDM. Table 10-14 compares the civilian labor force participa- tion rates simulated by the model with historical data for the period 1970-1979. The model predictions for most groups are quite close to the actual values, especially for prime age males. The root mean square error for males age 35-44 is 0.2 percent. The model appears to err propor- tionally the most for the elderly. It slightly overestimates the decline in labor force participation for women age 65 and over, and it underestimates, until 1978, the decline in male labor force participation. Table 10-15 compares the simulated civilian employ- ment levels to historical data for the years 1970-1979. The trends in estimated employment levels closely follow ac- tual trends. Absolute differences are seldom more than Table 10-15 Comparison of MDM Estimates and and Actual 500,000, even in the middle age groups (age 25-54) where employment levels are relatively large. In 1979 total em- ployment levels are simulated to within 300,000 for both males and females, an error of only one-half percent. In summary, the Labor Market Model estimates that aggregate employment levels will increase from 97 mil- lion in 1980 to 147 million in 2055. Over that same period the average worker’s annual compensation grows from $8,312 to $21,763 (in constant 1972 dollars). Because of the more rapid growth in the elderly population in these years, employment increases more rapidly for elderly workers than for nonelderly workers. The trends in com- pensation levels do not vary greatly by age, although the compensation of younger workers does grow somewhat more slowly than compensation for older workers. Social Security Estimates The Social Security Model simulates the operation of two social security programs—the Old-Age and Survivors In- Civilian Employment Levels by Age and Sex, 1970-1979 (Millions of Persons) Males 16-24 Year Olds 25-54 Year Olds 55-64 Year Olds 65 Years and Older Total Year Simulated Actual Simulated Actual Simulated Actual Simulated Actual Simulated Actual 1970 8.6 8.6 31.6 313 7.4 6.9 24 2.1 50.1 49.0 1971 9.3 9.0 31.6 31.3 7.2 6.9 2.2 2.0 50.3 49.2 1972 9.8 9.8 325 31.9 7.2 6.9 2.2 1.9 51.6 50.6 1973 10.3 10.6 33.2 32.7 7.1 6.8 2.1 19 52.7 52.0 1974 10.5 10.7 33.4 33.1 7.0 6.8 2.0 1.9 53.0 52.5 1975 10.5 10.1 33.1 32.6 7.0 6.7 20 1.8 52.7 51.2 1976 10.8 10.6 33.8 33.3 7.1 6.7 2.0 1.7 53.7 52.4 1977 11.1 11.2 345 34.2 7.1 6.8 21 1.7 54.8 53.9 1978 11.4 11.6 33.5 35.1 7.1 6.9 2.0 1.8° 56.0 55.5 1979 11.5 11.8 36.2 35.9 7.1 6.9 1.9 19 56.8 56.5 Females 16-24 Year Olds 25-54 Year Olds 55-64 Year Olds 65 Years and Older Total Year Simulated Actual Simulated Actual Simulated Actual Simulated Actual Simulated Actual 1970 6.9 7.2 17.1 17.4 39 4.0 1.1 1.0 29.0 29.7 1971 7.3 7.3 17.7 17.5 3.9 4.1 1.1 1.0 29.9 29.9 1972 7.7 7.8 18.5 18.2 39 4.1 1.1 1.0 31.2 31.1 1973 8.2 8.3 19.3 19.0 4.0 4.1 1.1 1.0 32.5 32.4 1974 8.5 8.6 20.0 19.8 4.0 4.0 1.0 1.0 33.4 334 1975 8.5 8.5 20.6 20.0 3.8 4.0 1.0 1.0 33.7 33.6 1976 8.8 89 21.3 21.1 3.8 4.1 1.0 1.0 34.8 33.6 1977 9.1 93 22.1 22.2 39 4.2 1.0 1.0 36.0 35.41 1978 9.7 9.9 22.7 23.6 4.2 4.3 1.1 1.1 37.6 36.7 1979 10.0 10.1 23.8 24.8 43 4.4 1.1 1.1 39.0 38.9 Source: Projections of the Macroeconomic-Demographic Model; “Age Distribution of Male Military Personnel,” Selected Manpower Statistics, 1978, Department of Defense, page 39; and Employ- ment and Training Report of the President, 1980, Table A-15. 70 CHAPTER 10 Table 10-16 1983 Projected Numbers of Beneficiaries and the Base Case Simulation Levels of Benefit Payments by the OASI and DI Systems, 1980-2055 OASI DI Average Average Average Average Total Primary ~~ Secondary Total Total Primary Secondary Total Years Beneficiaries ~~ Benefit Benefit . Payments Beneficiaries ~~ Benefit Benefit ~~ Payments (Millions) (1972 $) (1972 $) (Billions of (Millions) (1972 §) (1972 §) (Billions of 1972 §) 1972 §) 1980 29.7 $1,949 $1,301 $ 50.5 4.1 $2,238 $1,343 $76 1990 34.3 2,257 1,521 69.0 4.3 2,933 1,760 10.6 2000 36.5 2,705 1,822 88.7 5.1 3,577 2,146 151 2010 40.1 3,209 2,155 116.8 6.4 4,056 2,434 21.7 2020 50.0 3,671 2,455 169.0 6.9 4,458 2,675 25.6 2030 59.0 4,007 2,678 220.3 6.4 4,892 2,935 25.9 2040 59.9 4,386 2,941 246.2 6.5 5,501 3,301 299 2050 62.2 5,013 3,357 292.9 7.0 6,341 3,805 36.8 2055 64.9 5,407 3,615 330.1 6.9 6,810 4,086 39.2 Rate of Growth 1980-2010 1.01% 1.68% 1.70% 2.83% 1.45% 2.00% 2.00% 3.56% 2010-2030 1.95 1.12 1.09 3.22 0.00 0.94 0.94 0.89 2030-2055 0.38 1.21 1.21 1.63 0.30 1.33 1.33 1.67 1980-2055 1.05 1.37 1.37 253 0.69 1.49 1.49 2.21 These projections do not reflect changes brought about by the Social Security Amendments of 1983. Source: Projections of the Macroeconomic-Demographic Model. surance (OASI) program and the Disability Insurance (DI) program. The third social security program—Medi- care—is simulated by a separate MDM model. This section focuses particularly on the projected long-term funding status of the OASI and OASDI systems as of 1982.8 Like the Population and Labor Market Models, the Social Security Model contains extensive demographic detail. This sec- tion, however, describes only the aggregate projections of the numbers of beneficiaries and the level of benefits of the Social Security Model. Table 10-16 shows the 1982 projected numbers of beneficiaries and levels of benefit payments by the OASI and DI programs. Total OASI and DI beneficiaries were projected to be 65 million and 7 million respectively by 2055. Total benefit payments were projected to increase even more rapidly than beneficiaries. The benefit pay- ments of the OASI system were projected to be $330 billion (in 1972 dollars) in 2055. The DI system was pro- jected to pay $39 billion in benefits in the same year. 8The version of the Social Security Model described in this monograph was completed in 1981. It does not reflect changes in the Social Security Act contained in the Social Security Amendments of 1983, enacted April 20, 1983. These Amendments brought about fundamental changes in OASDI coverage, benefit levels, scheduled tax rates, interaction of OASI benefits and public employee pension benefits, and the normal retire- ment age after the vear 2000. These changes will significantly affect the future revenues, expenditures, and fund balances of the OASDI system. The estimates reported in this chapter reflect funding problems that were corrected in whole or in part by the 1983 Amendments. 71 Figure 10-5 Trends in SOC SEC Beneficiaries by Type 60 O PRI OASI @ SEC OASI O PRI DI ® SEC DI 40 30 BILLIONS OF PERSONS 20 10 1 1970 1980 1990 2000 2010 2020 2030 2040 2050 YEARS 1 | 0 1 Source: Table 10-16. THE NATIONAL INSTITUTE ON AGING MACROECONOMIC-DEMOGRAPHIC MODEL Table 10-16 shows the estimated average benefit pay- ment to primary beneficiaries and to secondary beneficia- ries (including survivors). Because the benefit payment levels of most secondary beneficiaries are determined directly from the primary benefit amount, the average primary benefit is the most useful indicator of the trend in individual payments made by the system. Average benefit levels were projected to increase approximately 1.4 per- cent annually, so that the average primary OASI beneficia- ry receives $5,407 per year in 2055 (in 1972 dollars). The average primary DI beneficiary receives $6,810 per year in 2055. Figure 10-5 shows the variation in the growth rates in the numbers of beneficiaries of various types. The rate of growth is particularly great for primary OASI beneficiaries between 2010 and 2030—two percent per year. This inter- val coincides with the movement of the baby boom into the retirement years. However, between 2035 and 2045 the primary OASI beneficiary population declines. During this period the number of deaths of members of the baby boom cohorts exceed the number of retirements from the much smaller cohorts born during the late 1960’s and the 1970's. Table 10-17 1982 Projected Tax Revenues and Expenditures of the OASI and Combined OASDI Systems, Selected Years, 1990-2055% (in 1972 §) The number of secondary beneficiaries increases more slowly than the number of primary beneficiaries for two main reasons. First, with increasing labor force participa- tion of women, the number of elderly females with a primary benefit is expected to increase and the number with a secondary (spouse’s) benefit is expected to fall. Second, lower fertility rates are expected to reduce the number of dependent children beneficiaries relative to the number of primary beneficiaries. Under current law, social security benefits are financed by a payroll tax levied on earnings up to a taxable maxi- mum.’ This base case simulation assumes this system will continue and projects the level of tax revenues under the 1982 statutory tax rates. The total compensation amounts projected by the Labor Market Model are used to estimate the total payroll base subject to social security taxes.” In 1982, the payroll tax rates for OASI and DI were 4.575 and 0.825 percent respectively, and they were scheduled to increase to 5.1 and 1.1 percent in 1990."! Both the employ- er and the employee must contribute at these rates on all earnings up to the maximum taxable earnings amount. The MDM applies the statutory contribution rates to pro- OASI OASDI Payroll Payroll Tax Tax Year Revenues Expenditures Balance Revenues Expenditures Balance 1990 78.9 69.0 9.9 95.9 79.6 16.3 2000 100.6 88.7 11.9 122.2 103.7 18.5 2010 117.9 116.8 1.1 143.4 138.5 4.9 2020 127.7 169.0 —413 155.2 194.6 —394 2030 141.3 2203 —-79.0 171.7 246.2 —74.5 2040 166.1 246.2 —80.1 201.9 276.1 —74.2 2050 193.0 292.8 —99.8 234.6 329.6 -95.0 2055 208.3 330.1 —121.8 253.3 369.2 -1159 Rate of Growth 1990-2010 2.03% 2.67% —10.40% 2.03% 2.80% —5.83% 2010-2030 0.91 3.22 NA. 0.90 2.92 NA. 2030-2055 1.56 1.63 NA. 1.57 1.63 NA. 1990-2055 1.50 2.44 NA. 1.51 2.39 NA. “These projections do not reflect changes brought about by the Social Security Amendments of 1983. Source: Projections of the Macroeconomic-Demographic Model (Ta- bles 1-18 and 1-19). 72 The 1983 Amendments provided that revenues from taxing one-half the social security benefits of individuals with incomes above a specified level would be paid to the social security system. 1%pages 51 through 55 describe this procedure. "This was changed by the 1983 Amendments. CHAPTER 10 Table 10-18 Base Case Simulation 1982 Projected Cost Rates and Tax Rates of the OASI and - Combined OASDI Systems, Selected Years, 1990-20552 OASI OASDI Cost Tax Cost Tax Year Rate Raté® Difference’ Rate Rate’ Difference’ 1990 .089 102 013 103 124 021 2000 .090 102 012 105 124 019 2010+ 101 102 .001 120 124 .004 2020 135 102 —.033 155 124 —.031 2030 159 102 —.057 178 124 —.054 2040 151 102 —.049 170 124 —.046 2050 155 102 —.053 174 124 —.050 2055 162 102 —.060 181 124 —.057 These projections do not reflect changes brought about by the Social Security Amendments of 1983. Sum of employer and employee tax rates. “Tax rate less cost rate. Negative number indicates deficit. Source: Projections of the Macroeconomic-Demographic Model (Table 1-19). Figure 10-6 Trends in OASI Taxes and Payments 300 280 260 240 220 200 180 160 140 120 100 80 60 40 20 l ad | l 1 l 1 1970 1980 1990 2000 2010 2020 2030 2040 2050 YEARS | OTAXES @ PAYMENTS BILLIONS OF 1972 DOLLARS Source: Table 10-17. jected payroll tax bases to estimate tax collections in each system through 2055. Table 10-17 displays the 1982 base case projections (in constant 1972 dollars) of tax revenue, expenditures, and the surplus or deficit for OASI and for the combined OASDI systems. By 2055, the OASDI system is projected to receive $253 billion in taxes annually—$208 billion in OASI and $45 billion in DI. With expenditures of $369 73 billion, OASDI is projected to have a deficit of $116 bil- lion. A deficit of this size would be equal to 3%: percent of projected GNP in 2055. A second measure of the funding status of each system is the cost rate, defined as the ratio of benefit payments to the taxable payroll base. Table 10-18 shows 1982 projec- tions of the estimated cost rates of the OASI and OASDI systems and compares them to the 1982 statutory sched- uled tax rates. Cost rates of the OASI system alone and the combined OASI and DI systems nearly double over the course of this simulation. The rates of increase in these cost rates are more rapid after 2010 when the baby boom cohorts pass into the elderly population. Figure 10-6 shows the upward trends in both OASI payroll tax collections and OASI benefit payments (from Table 10-17). The steady upward trend in tax collections coincides with the expansion of the economy in general and the accompanying increase in earned income. The path of total social security benefit payments is less steady. OASI benefit payments, which grow about one percent annually between 1980 and 2010, rise sharply from 2010 to 2030. The rate of increase slows between 2030 and 2045 but accelerates again in the final ten years of the simula- tion. Table 10-19 compares 1982 MDM estimates of future OASI cost rates with those projected under the two inter- mediate alternatives prepared by the Social Security Actu- ary for the 1981 Social Security Trustees Report. The patterns of changes in the cost rates are similar in all three sets of projections. The MDM cost rates are slightly lower between 1981 and 1990 and slightly larger after 1990 than the Alternative II-B cost rates. In the MDM, the maximum cost rate is 16.2 percent in 2055. In the SSA Actuary’s projections, the maximum cost rates are lower, 14.3 per- cent in Alternative II-A and 15.2 percent in Alternative II-B, THE NATIONAL INSTITUTE ON AGING MACROECONOMIC-DEMOGRAPHIC MODEL Figure 10-7 Comparison of OASI Cost Rates 20 19 -0 MDM 18 -@® ALT IIB = Z i Q j= ml a = 7H 6 - 5 l- 4+ 3 - 2 | 1 — 0 1 1 1 1 1 1 1980 1990 2000 2010 2020 2030 2040 2050 YEARS Source: Table 10-19. Table 10-20 Table 10-19 Comparisons of OASI Cost Rates Projected by the Macroeconomic-Demographic Model and the Social Security Actuary, 1981-2055 Social Security Actuary Year MDM Alt. II'A Alt. II-B 1981 9.4 9.9 9.9 1985 8.7 9.9 10.4 1990 8.9 9.6 10.6 2000 9.0 9.0 9.9 2010 10.1 9.3 10.0 2020 13.5 12.0 12.7 2030 15.9 14.3 15.2 2040 15.1 14.3 - 15.2 2050 15.5 14.1 15.1 2055 16.2 14.2 15.2 “These projections do not reflect changes brought about by the Social Security Amendments of 1983. Sources: Projections of the Macroeconomic-Demographic Model and Table 26, 1981 Annual Report of the Board of Trustees of the Old-Age and Survivors Insurance and Disability Insurance Trust Funds. Comparisons of Simulated and Actual OASI Beneficiary Populations by Benefit Type, 1970-19792 (Millions of Persons) Primary Beneficiaries Secondary Beneficiaries Percent Percent Year Simulated Actual Difference Simulated Actual Difference 1970 13.3 13.3 0 10.2 10.2 0 1971 13.8 13.9 -0.7 10.4 10.4 0 1972 143 14.6 =21 10.6 10.6 0 1973 14.8 15.4 -39 10.8 10.9 -0.9 1974 15.4 16.0 -38 11.0 11.0 0 1975 15.9 16.2 -19 11.1 10.8 2.8 1976 16.4 16.8 —-24 11.2 11.0 1.8 1977 17.0 “174 -23 11.2 11.1 0.9 1978 17.4 17.9 —2.8 11.2 11.2 0 1979 17.9 18.6 -38 11.3 11.2 0.9 “These projections do not reflect changes brought about by the Social + Security Amendments of 1983. Sources: Projections of the Macroeconomic-Demographic Model and Table 52, Department of Health and Human Services, Social Security Administration, Social Security Bulletin, Annual Sta- tistical Supplement, 1977-1979. and occur in earlier years. Figure 10-7 presents a graphical comparison of the MDM and Alternative II-B cost rates. The biggest difference between the projection series is the rate of increase in the cost rate—the MDM projects more rapid growth in the cost rate between 1985 and 2030 74 than the SSA Actuary. The primary contributor to the more rapid growth in cost rates in the MDM projections is a lower rate of earnings growth. While the MDM estimate of average real wage growth, 1.5 percent per year, is about the same as the SSA Actuary’s intermediate assumption, CHAPTER 10 average annual compensation and GNP grow somewhat less in the MDM than the Actuary assumes. The reason for this is that the MDM projects a long-term decline in the average number of hours worked annually per worker. Because social security taxes and benefits are based on annual earnings, the long-term decline in annual hours worked projected by the MDM has the same effect on the payroll tax base as a reduction in the rate of productivity growth. To test the reliability of the Social Security Model, it was used to simulate the number of beneficiaries from 1970 to 1979. Table 10-20 compares these simulated values to actual values. The Social Security Model generates benefi- ciary estimates which replicate quite closely the actual numbers of beneficiaries during the 1970-1979 period. The number of primary beneficiaries of the social security system increased considerably between 1970 and 1979. The Social Security Model estimates slightly slower growth in primary beneficiaries than actually occurred. Although the estimated number of beneficiaries is simu- lated exactly in 1970, a slight underestimate emerges in 1971. This underestimate increases gradually throughout the historical period (1970-1979). In summary, in the 1982 version of the Social Security Model, OASI and DI benefit payments were projected to increase substantially more rapidly than the payroll tax base. This is consistent with the Social Security Actuary’s projections using similar assumptions. As a result, under the 1982 law, OASDI expenditures as a proportion of taxable earnings (the cost rate) were projected to rise above the legislated tax rate of 12.4 percent of taxable payroll. The cost rate was projected to be 18 percent in 2030, about one percentage point greater than the cost rate projected by the Social Security Actuary in the Alter- native II-B projections. Both the MDM and the Social Secu- rity Actuary projected large OASDI deficits beginning in the year 2015, which last through the end of the projection period. Private Pension Estimates This section describes the aggregate results of the projec- tions of the Private Pension Model. Tables 1-20 through I- 28 in Appendix I contain a more complete set of results of the base case private pension simulation. The Private Pen- sion Model forecasts the trends in the number of partici- pants, contributions, beneficiaries, average and total benefit payments, and fund balances of the three major types of private pension plans—defined benefit, defined contribution, and individual accounts.'? 2This version of the Private Pension Model was completed before the enactment of the Economic Recovery Tax Act of 1981 (ERTA). ERTA significantly increased the number of persons eligible to establish an Individual Retirement Account (IRA) and is expected to result in a large increase in the number of persons with IRAs. Therefore, the estimates reported in this monograph probably under-estimate the number of persons with IRAs. 75 \ Base Case Simulation Table 10-21 shows the projected numbers of active gross participants’? in the private pension system by age and sex, assuming that age-sex specific coverage and par- ticipation rates remain constant.'* Projected changes in pension participation thus reflect changes in the size and composition of the labor force. The number of gross par- ticipants is projected to increase from 48 million in 1980 to 72 million in 2055. Table 10-22 shows the projected trends in the number of beneficiaries and the level of benefit payments in em- ployer-based private pension plans. Trends in both de- fined benefit and defined contribution plans are displayed.'> In both types of plans the growth in beneficia- ries is projected to be rapid during the next forty years. The number of beneficiaries nearly triples between 1980 and 2010 in both defined benefit and defined contribu- tion plans. This rapid growth in the future number of beneficiaries results from the rapid growth in private pen- sion plan participation over the past thirty years. Growth in the number of beneficiaries accelerates between 2010 and 2020, as the baby boom cohorts retire. The growth in the number of beneficiaries slows after 2020 and remains roughly constant after 2030. Because of the growth in the levels of annual compen- sation, real average retirement benefit levels also rise more than two percent annually over the 1980 to 2010 period.'® Consequently, the growth rate in total benefit payments by, private pension plans is very large. Total benefit payments are projected to increase 81 percent in real terms between 1980 and 1990, then increase another 70 percent between 1990 and 2000. By 2055, defined benefit and defined contribution plans pay more than $220 billion (in 1972 dollars) in benefits. 13The number of gross participants equals the total number of partici- pants in all types of pension plans. An individual participating in more than one plan (e.g., a defined benefit plan plus a defined contribution plan) is counted more than once. Because many workers participate in more than one plan, the number of gross participants significantly exceeds the number of net participants, which equals the number of persons participating in one or more pension plans. “Because of uncertainty regarding future trends in private pension coverage and participation rates, for this base case simulation we assumed that age-sex specific coverage and participation rates would remain constant at their 1979 values. Changes in the numbers of cov- ered workers and participants, therefore, reflect only changes in the size and composition of the labor force. The model can easily simulate alternative trends in pension coverage and participation. "In this base case simulation, we assumed that the distribution of pen- sion plan participants and retirees between defined benefit and de- fined contribution plans remains constant after 1979. Consequently, the numbers of participants and retirees in both types of plans grow at approximately the same rate. Other estimates of future private pension benefit levels have been published which project less growth in average benefits. The differ- ences from the MDM projections may largely be due to the fact that the other estimates project more growth in future private pension cover- age and participation than this base case simulation assumes. Increas- ing participation may tend to reduce growth in average benefits, because many of the additional participants may be relatively low paid workers. See, for example, American Council of Life Insurance, “Pen- sion Coverage and Expected Retirement Benefits,” October 1982. THE NATIONAL INSTITUTE ON AGING MACROECONOMIC-DEMOGRAPHIC MODEL Table 10-21 Projected Numbers of Active Gross Participants in the Private Pension System by Age and Sex, 1980-2055 (Millions) Year 16-24 1980 34 1990 3.0 2000 3.1 2010 5.5 2020 23 2030 3.6 2040 3.6 2050 3.6 2055 38 Year 16-24 1980 21 1990 2.0 2000 23 2010 2.6 2020 24 2030 2:7 2040 2.7 2050 2.7 2055 2.8 25-34 9.8 25-34 4.7 5.8 6.2 7.2 7.8 7.4 8.0 8.0 8.0 Male 35-44 45-54 7.8 6.5 11.2 7.3 12.5 10.1 10.4 11.1 10.9 9.3 11.7 9.6 11.1 10.3 119 9.6 12.0 9.9 Female 35-44 45-54 34 3.0 5.1 3.7 6.6 5.6 6.4 6.7 7.4 5.9 8.0 6.5 7.6 7.8 8.2 7.5 83 7.7 55-64 4.8 4.7 4.7 BV GVW X= O00 ~ ® 55-64 19 2.0 2.4 3.0 32 29 VE NO =O Gross participants equal the sum of the total number of participants in each type of pension plan. Workers participating in more than one plan are counted once for each plan in which they participate. Source: Projections of the Macroeconomic-Demographic Model (Table 1-22). Table 10-22 65+ 0.4 0.5 0.4 0.5 0.6 0.7 0.7 0.8 0.9 65+ 0.3 0.3 0.3 0.2 0.3 0.3 0.3 0.3 0.3 Total 32.7 38.0 40.2 41.2 40.8 40.7 41.6 41.9 42.1 Total 15.3 18.9 233 26.1 27.1 27.8 29.4 29.8 30.0 Projected Beneficiaries and Benefit Payments in Private Pension Plans by Type of Plan, 1980-2055 Year 1980 1990 2000 2010 2020 2030 2040 2050 2055 Rate of Growth 1980-2010 2010-2055 1980-2055 Defined Benefit Plans Defined Contribution Plans Beneficiaries (Millions) 5.90 9.31 11.35 15.73 22.38 25.15 23.03 24.04 25.37 3.32% 1.07 1.96 Average Benefits (1972 %) $1,981 2,263 3,091 3,820 4272 4,788 5,533 6,380 6,826 2.21% 1.30 1.66 Total Payments (Billions of 1972 §) $ 11.69 21.06 35.08 60.10 95.59 120.39 127.43 153.40 173.19 5.60% 2.38 3.60 Source: Projections of the Macroeconomic-Demographic Mode] (Ta- bles I-25 and 1-27). 76 Beneficiaries (Millions) 3.01 4.73 5.77 7.99 11.37 12.78 11.70 12.22 12.89 3.31% 1.07 1.96 Average Benefits (1972 §) $1,263 1,483 2,101 2,524 2,728 2,946 3263 3,638 3,830 2.33% 0.93 1.49 Total Payments (Billions of 1972 $) $ 3.79 7.02 12.11 20.17 31.02 37.65 38.18 44.45 49.38 5.73% 2.01 3.48 CHAPTER 10 Base Case Simulation Table 10-23 Projected Private Pension Fund Balances By Type of Plan, 1980-2055 (Billions of 1972 Dollars) Defined Benefit Plans Defined Contribution Plans Total Total Fund Total Total Fund Year Payments Contributions Balances Payments Contributions Balances 1980 $11.69 $33.92 $191.62 $3.79 $13.25 $92.71 1990 21.06 68.86 647.51 7.02 18.20 200.78 2000 35.08 92.12 1,226.93 12.11 25.83 341.58 2010 60.10 110.69 1,860.52 20.17 30.13 486.44 2020 95.59 110.18 2,230.78 31.02 32.21 561.70 2030 120.39 126.79 2,414.57 37.65 37.00 583.39 2040 127.43 155.92 2,719.87 38.18 43.91 639.91 2050 153.40 185.69 3,192.18 44.45 52.22 748.68 2055 173.19 205.05 3,431.73 49.38 57.12 806.12 Rate of Growth 1980-2010 5.61% 4.02% 7.87% 5.73% 2.78% 5.68% 2010-2055 2.38 1.38 1.37 2.01 1.43 1.13 1980-2055 3.66 2.43 3.92 3.48 1.97 2.93 Source: Projections of the Macroeconomic-Demographic Model (Ta- bles 1-27 and 1-28). Table 10-24 Comparisons of Simulated and Actual Participants in and Contributions to the Private Pension System, 1970-1977 Total Gross Participants Total Contributions (Millions) (Billions of 1972 Dollars) Percent Percent Year Simulated Actual Difference Simulated Actual Difference 1970 33.4 33.5 —.3% $12.7 $16.1 —21.4% 1971 35.5 35.9 —-1.1 15.2 NA. NA. 1972 38.4 36.8 4.3 18.7 NA. NA. 1973 41.4 38.7 7.0 223 NA. NA. 1974 43.9 40.6 8.1 25.7 225 14.3 1975 43.8 41.3 6.1 25.6 253 1.2 1977 45.8 43.2 6.0 35.6 33.7 5.6 Sources: Projections of the Macroeconomic-Demographic Model, data for 1970 through 1975 from American Council of Life Insur- ance, 1980 Pension Facts, and data for 1977 derived from Daniel Beller, “Preliminary Estimates of Participant and Finan- cial Characteristics of Private Pension Plans, 1977,” U.S. Depart- ment of Labor, (1981). Data for 1976 not available. During the period 1980-2010, while the private pension system is maturing, total private pension benefit payments (including individual accounts) are projected to increase at an annual average rate of 5.6 percent per year,'” while total social security retirement benefits (OASI) increase at an average rate of 2.7 percent per year. After the year 2010, the private pension system will essentially have reached a mature stage, and growth of total benefit payments will primarily reflect growth in the elderly population and real income growth. From 2010 to 2030, total private pension "See Appendix Table 1-28 for total private pension benefit payments. 77 benefits increase at an average annual rate of 3.4 percent, while OASI benefits increase at an average annual rate of 3.2 percent. From 2030 to 2055, total private pension benefits increase at an average annual rate of 1.4 percent, while OASI benefits increase 1.6 percent per year. Table 10-23 shows the projected long-term trends in private pension fund balances. After 1980, total pension contributions and benefit payments are estimated to in- crease rapidly. Annual contributions in defined benefit plans increase from $34 billion (1972 dollars) in 1980 to $111 billion in 2010, an annual real rate of growth of four percent. The rate of growth in aggregate contributions to THE NATIONAL INSTITUTE ON AGING MACROECONOMIC-DEMOGRAPHIC MODEL defined contribution plans is lower, 2.8 percent annual- ly.!® Annual contributions exceed annual benefit pay- ments in defined contribution plans until 2020 and in defined benefit plans until about 2025. When the baby boom cohorts reach retirement age, benefit payments be- gin to approach the sum of contributions and investment earnings, and the growth in fund balances begins to slow. However, after 2040 total contributions exceed benefit payments in every year through the end of the simulation period, with both growing at a rate of about two percent per year. Figure 10-8 displays graphically the projected long- term trends in private pension fund balances that are shown in Table 10-23. The fund balances of all three pro- totype private pension plans are projected to grow rapidly through 2015. After that time, growth in the fund balances of the defined benefit and defined contribution plans de- creases. The balance for IRA and Keogh accounts contin- ues to increase steadily in all years of the simulation, although its rate of increase slows somewhat after 2015.'° To test the reliability of the model, we simulated the 1970-1979 period using data through 1970. The decade of the 1970’s was a period of tremendous growth and change for the private pension system. Historical simulations of the Private Pension Model capture much of this change. Table 10-24 compares simulated and actual trends in pen- sion plan participants and contributions from 1970-1977. The simulated number of gross participants is almost exactly equal to the actual number in 1970. However, in 1972 and 1973 the model simulates greater growth than apparently occurred. From 1973 through 1977 the model simulation is about six to eight percent greater than the available data. Some of the overestimates of the number of participants is attributable to the base case simulation of the Labor Market Model. That model estimates the total number of workers well, but it overestimates the growth in the number of male workers and underestimates the growth in the number of female workers during the 1970- 1979 period. This by itself accounts for an overestimate in private pension plan participants of about 1.5 percent. There does not exist a consistent series of historical data for contributions to private pension plans. One wide- ly cited estimate of contributions in 1977 is the study done for the Department of Labor by Daniel Beller (1981). The total contribution estimates of the Private Pension Model, displayed in Table 10-24, are six percent higher than Beller’s estimate for 1977. Because plan participant esti- mates are also six percent high in 1977, the estimated contribution rates for that year may be accurate. Contributions to defined benefit plans grow more rapidly than contri- butions to defined contribution plans during the period 1980-2010 because we assume that during that period defined benefit plans amor- tize the estimated unfunded liabilities existing in 1980. (See pages 83 and 84.) No new unfunded liabilities are forecast after 1980. Defined contribution plans have no unfunded liabilities. “These projections do not include any assumptions about increasing use of IRA accounts due to the passage of the Economic Recovery and Tax Act of 1981. 78 Figure 10-8 Trends in Private Fund Balances by Type of Plan 3.2 | oDB @DC — OIRA 2.8 2.6 2.4 2.2 2 1.8 1.6 1.4 1.2 1+ 0.8 0.6 0.4 0.2 TRILLIONS OF 1972 DOLLARS 1 1 1 1 1980 1990 2000 2010 2020 2030 2040 2050 YEARS 0 1970 Source: Table 10-23. The Private Pension Model simulation is very close to the actual growth in the number of private pension bene- fit recipients during the 1970’s (see Table 10-25). The 83 percent increase in recipients between 1970 and 1979, from 4.75 million to 8.71 million beneficiaries, is simulat- ed to within three percent. Average real private pension benefits remained nearly constant over this period. (The values in Table 10-25 are in constant 1972 dollars.) The model's simulated average benefits also remain nearly constant in real terms for the same period. The simulated average benefit levels range from zero to about six per- cent greater than the estimated actual average benefit lev- els during the period 1970-1979. In summary, the Private Pension Model projects a rapid growth in the number of beneficiaries and level of bene- fits in the private pension system through 2020. Although the growth in contributions is also rapid, its growth rate is less than the growth rate of benefits. The growth in total private pension benefits exceeds the growth in social se- curity benefits between 1980 and 2010, but after 2010, the growth in the private pension system's aggregate benefits is comparable to that of the social security system. Public Employee Pension Estimates The public sector pension systems are also expected to expand over the 1980 to 2055 period. Public employer systems include Federal, state, and local government pen- CHAPTER 10 Base Case Simulation Table 10-25 Comparisons of Simulated and Actual Beneficiaries and Benefit Payments From the Private Pension System, 1970-1979 Total Retirees Average Benefits (Millions) (1972 Dollars) Percent Percent Year Simulated Actual Difference Simulated Actual Difference 1970 4.74 4.75 —~.2) $1791 $1677 6.80% 1971 5.12 5.19 =1.35 1799 1717 4.78 1972 5.53 5.56 154 1800 1801 —0.06 1973 5.96 6.10 2.50 1796 1763 1.87 1974 6.41 6.41 0 1788 1740 2.76 1975 6.89 7.12 =325 1774 1667 6.42 1976 7.38 7.61 -3.02 1759 1673 5.14 1977 7.89 8.00 —1.38 1744 1683 3.62 1978 8.42 8.36 72 1729 1696 1.95 1979 8.97 8.71 2.99 1713 1668 2.76 Sources: Projections of the Macroeconomic-Demographic Model. Actu- al data from Alfred M. Skolnik, “Private Pension Plans, 1950- 1974,” Social Security Bulletin, Vol. 39, No. 6 (1976); American Council of Life Insurance, Pension Facts (Washington, D.C., 1977), pp. 30-31, 36; Employee Benefit Research Institute, Re- tirement Income Opportunities in an Aging America: Income Levels and Adequacy (Washington, D.C., 1982), p. 82. Table 10-26 Projected Beneficiaries and Total Benefit Payments in Public Pension Plans By Type of Plan, 1980-2055 Total Beneficiaries Total Benefit Payments (Millions) (Billions of 1972 Dollars) Civil State Civil State Year Service Military & Local Service Military & Local 1980 1.41 1.19 3.09 $0.72 $6.27 $9.34 1990 1.96 1.35 3.83 10.78 8.27 13.54 2000 2.31 1.45 3.80 13.54 10.20 15.47 2010 2.35 1.43 4.16 16.43 11.73 20.07 2020 2.44 1.43 4.27 18.29 13.28 22.96 2030 2.52 1.43 5.25 20.94 15.00 28.53 2040 2.61 1.43 5.45 25.81 17.12 35.45 2050 2.70 1.43 5.12 30.90 19.62 36.32 2055 2.75 1.43 5.07 33.79 21.10 36.42 Rate of Growth 1980-2010 1.72% 0.61% 1.00% 3.02% 2.11% 2.58% 2010-2055 0.34 0.0 0.44 1.62 1.31 1.33 1980-2055 0.89 0.46 0.66 2.18 1.63 1.83 Source: Projections of the Macroeconomic-Demographic Model (Ta- bles 1-29 and 1-32). 79 THE NATIONAL INSTITUTE ON AGING MACROECONOMIC-DEMOGRAPHIC MODEL sion programs. The two major Federal programs cover Civil Service workers and members of the military. Al- though these systems are generally older and more ma- ture than private plans, many have been less well funded. As a result, contributions and fund balances can be expect- ed to increase over the projection period.’ Table 10-26 shows the MDM base case projections of beneficiaries and benefit payments of three public sector systems. The two Federal plans, those for Civil Service and military employees, are single integrated systems.?! The estimates for state and local employees reported in Table 10-26 are an aggregation of the estimates of the four pro- totype state and local plans, which in turn were developed to represent the universe of over 6000 state and local plans.?? The numbers of beneficiaries increase in all three sys- tems. The Civil Service retired population nearly doubles from 1980 to 2055. The growth in the state and local retired population is smaller, but state and local retirees still outnumber Federal retirees at the end of the simula- tion period. The number of military retirees increases only slightly, because the MDM assumes the size of the armed forces remains unchanged after 1980. The MDM assumes that the compensation rates of pub- lic sector employees in each age-sex group change at the same rate as those of private sector workers. Thus, public sector average benefit levels increase at about the same rate as that of private sector defined benefit plans. (All public sector plans are assumed to be defined benefit plans.) Differences between the growth in total benefit payments among the plans principally reflect differences in the growth of the respective beneficiary population. By 2055, the three public sector plans pay out about $91 billion (1972 dollars) in benefits. Table 10-27 shows the average pension benefits paid by the three main public sector pension systems. The growth in average benefits is determined largely by the growth in compensation. Average benefits for the Federal Civil Ser- vice plan grow at an average annual rate of 1.3 percent from 1980 to 2055. Average benefits from the military retirement system grow 1.4 percent per year. The average benefits of state and local plans are scheduled to grow about 1.2 percent per year. Average benefits from the public employer plans remain higher than private pen- sion benefits throughout the simulation period. 2The Social Security Amendments of 1983 specify that all new Federal employeees after January 1, 1984, will be covered by the social security system. It is likely that a new supplementary pension system will be developed for those employees. The simulations reported in this sec- tion assume that all Federal employees continue to participate in the existing Civil Service Retirement System. 2!Because of their different average age levels and retention behavior in the armed forces, the MDM simulates enlisted and officer personnel separately. 22The MDM simulates four different prototype state and local plans—a plan for hazardous duty workers, a plan for state educators, a plan for local educators, and a plan for all other state and local employees. Chapter 7 describes the public employee pension model. 80 Table 10-28 shows the projected trends in the fund balances of public sector systems. Since the military and hazardous duty local government employees participate in pay-as-you-go systems, no simulations of fund balances for these systems are performed. The projections present- ed here do not forecast a funding crisis for either the state and local or Federal Civil Service systems. In the MDM, contributions to public employee plans are simulated on a normal cost basis—contributions are set to fully fund all pension plan obligations. If we had developed projections based on actual rather than normal cost contributions, our projected fund balances would have been substantially smaller. The average annual rate of growth in total contributions | from 1980 to 2055, 1.0 percent for civil servants and 0.4 percent for state and local employees, is less than the average annual growth in private pension plan contribu- tions, which is 2.2 percent over the same period. The growth in total benefits paid by public sector plans is also much less rapid than the growth in total private sector benefits, because the public sector beneficiary population grows much more slowly from 1980 to 2010. Average annual increases in total benefit payments of 2.2 percent (Federal Civil Service) and 1.8 percent (state/local) be- tween 1980 and 2055 far exceed the growth rates in con- tributions to those plans. Consequently, fund balances increase until 2030 for Federal Civil Service and state/local plans, then decrease. Figure 10-9 shows graphically the trends in these two fund balances. Figure 10-9 Trends in Pension Fund Balances by Type of Plan 600 0 CIVIL SERVICE @ STATE/LOCAL nN = oS - [= S BILLIONS OF 1972 DOLLARS 100 0 1 1 1 1 1 1 1 1970 1980 1990 2000 2010 2020 2030 2040 2050 YEARS Source: Table 10-28. CHAPTER 10 Base Case Simulation Table 10-27 Projected Average Benefits Paid By Public Employee Pension Plans By Type of Plan, 1980-2055 (1972 Dollars) Year Civil Service Military State and Local Total 1980 $ 4,902 $ 5,259 $3,022 $3,949 1990 5,608 6,127 3,539 4,592 2000 5,967 7,038 4,073 5,217 2010 7,116 8,187 4,823 6,103 2020 7,645 9,272 5,381 6,739 2030 8,452 10,475 5,435 7,040 2040 10,059 11,954 6,510 8,300 2050 11,649 13,701 7,100 9,439 2055 12,536 14,735 7,190 9,933 Rate of Growth 1980-2010 1.25% 1.49% 1.57% 1.46% 2010-2055 1.27 1.31 0.89 1.09 1980-2055 1.26 1.38 1.16 1.24 Source: Projections of the Macroeconomic-Demographic Model (Table 1-32). Table 10-28 Projected Public Pension Fund Balances By Type of Plan, 1980-2055 (Billions of 1972 Dollars) Federal Civil Service State and Local Total Total Fund Total Total Fund Year Payments Contributions Balances Payments Contributions Balances 1980 $6.72 $11.27 $102.54 $9.34 $15.21 $172.50 1990 10.78 , 12.09 151.75 13.54 14.75 249.31 2000 13.54 13.35 189.91 15.47 17.62 328.24 2010 16.43 14.33 214.08 20.07 16.52 418.51 2020 18.29 15.28 227.73 22.96 16.99 471.20 2030 20.94 16.98 235.97 28.53 21.26 525.24 2040 25.81 19.08 225.30 35.45 18.66 514.26 2050 30.90 21.68 181.20 36.32 20.00 467.59 2055 33.79 23.37 146.70 36.42 20.74 448.52 Rate of © Growth 1980-2010 3.02% 0.80% 2.48% 2.58% 0.28% 3.00% 2010-2055 1.62 1.09 —5.78 1.33 0.51 0.15 1980-2055 2.18 0.98 0.48 1.83 0.41 1.28 Source: Projections of the Macroeconomic-Demographic Model (Ta- bles 1-31 and 1-32). 81 THE NATIONAL INSTITUTE ON AGING MACROECONOMIC-DEMOGRAPHIC MODEL In summary, the Public Employee Pension Model illus- trates the behavior of mature pension systems and pro- jects a less rapid growth in the public sector pension system than for either the private pension system or the social security system. The trends in projected fund bal- ances of both the Federal Civil Service and state and local systems reflect this behavior. Benefit payments exceed contributions from 2000 through 2055 in the Federal Civil Service system and from 2010 through 2055 in the proto- typical state and local system. Neither system is projected to become insolvent. However, these projections assume contributions are adequate to fully fund both systems. Supplemental Security Income Estimates The Supplemental Security Income (SSI) program repre- sents an important part of the income support system for low income and disabled retirees. Total Federally admin- istered SSI benefits were 6.9 billion dollars in 1979. Al- though SSI is smaller than any other pension system simulated by the Macroeconomic-Demographic Model (except IRA/Keogh accounts), it is a major source of sup- port for about two million low income elderly. The Supplemental Security Income (SSI) system pro- vides three types of means tested benefits: (1) benefits to the aged, (2) benefits to the blind, and (3) benefits to the disabled. The first benefit type is simulated with a simple model of the income distribution. The other two are pro- jected by assuming constant rates of participation and constant real average benefits by age and sex group. Table 10-29 shows projected trends in the number of Table 10-29 beneficiaries and total benefit payments of the programs in the SSI system. The number of low income aged benefi- ciaries is projected to fall from over two million in 1980 to about 570,000 in 2055. This reduction occurs because the SSI Model assumes a constant absolute standard of pover- ty, i.e., that the income eligibility standard remains con- stant in real terms. As average income increases, fewer aged individuals fall below this poverty standard. While the number of low income aged SSI beneficiaries decreases, the number of disabled beneficiaries increases at about 0.6 percent annual rate of growth over the period. Disabled beneficiaries increase because: (1) the size of the general population increases, and (2) the age groups most likely to become eligible for SSI disability benefits increase relative to other age groups. The rate of increase in the number of disabled beneficiaries slows in the sec- ond half of the projection period. Because the relatively small number of blind beneficia- ries grows little in absolute numbers, the total number of SSI beneficiaries remains relatively constant from 1980 to 2055. However, the distribution of SSI beneficiaries shifts from an even split between the aged and disability pro- grams to a concentration in the disability program. Trends in the level of total benefit payments are similar to trends in the number of beneficiaries. Total benefit payments increase slightly from $4.7 billion (1972 dol- lars) in 1980 to $5.1 billion in 2055. The growth in benefit payments to disabled individuals offsets the sharp decline in benefit payments to the aged over the period. Figure 10-10 shows the trends in benefit payments graphically for each of these three SSI programs. Table 10-30 compares simulated beneficiary estimates for 1974 through 1979 with actual beneficiary data for each of the three programs in the SSI system. For 1979, Projected Beneficiaries and Benefit Payments of the Supplemental Security Income System, 1980-2055 Beneficiaries Benefit Payments (Millions) (Billions of 1972 Dollars) Year Aged Blind Disabled Total Aged Blind Disabled Total 1980 2.15 0.08 2.19 4.42 $1.82 $0.11 $2.75 $4.68 1990 1.88 0.09 2.39 4.360 1.59 0.13 3.01 4.73 2000 1.33 0.10 2.60 4.03 1.13 0.14 3.27 4.54 2010 1.57 0.11 3.03 4.71 1.33 0.16 3.81 5.30 2020 2.02 0.12 3.29 5.43 1.72 0.18 4.14 6.03 2030 1.44 0.13 3.26 4.83 1.22 0.19 4.09 5.50 2040 1.02 0.14 3.26 4.42 0.86 0.20 4.10 5.17 2050 0.75 0.14 3.45 4.34 0.63 0.21 4.34 5.18 2055 0.57 0.15 3.50 4.22 0.48 0.22 4.39 5.09 Rate of Growth 1980-2010 —1.04% 1.07% 1.09% 0.21% — 1.04% 1.26% 1.09% 0.42% 2010-2055 -2.22 0.69 0.32 —0.24 —-224 0.71 0.32 -0.09 1980-2055 -1.75 0.84 0.63 —0.06 -1.76 0.93 0.63 0.11 Source: Projections of the Macroeconomic-Demographic Model (Table 1-33). OND CHAPTER 10 total beneficiaries in the aged program is over estimated by eight percent while total beneficiaries in the blind and in the disability programs are underestimated by zero and 1.6 percent, respectively. In summary, the nature of the SSI program is projected to change significantly over the next 75 years. This pro- gram, which currently is an important source of income to the low income aged population, is projected to concen- trate increasingly on low income disability beneficiaries. These projections have assumed a constant absolute stan- dard of poverty. The relative shifts projected here would be smaller if increasing real poverty standards were to be ‘assumed instead. Table 10-30 Comparisons of Simulated and Actual Beneficiaries of Supplemental Security Income, 1974-1979 (Thousands of Persons) The Aged Year Simulated Actual Percent Difference 1974 2,313 2,286 1.2% 1975 2,649 2,307 14.8 1976 2,534 2,147 18.0 1977 2,517 2,050 279 1978 2,158 1,968 9.7 1979 2,019 1,872 7.9 The Blind Year Simulated Actual Percent Difference 1974 71 75 —5.3% 1975 72 74 -27 1976 73 76 -39 1977 74 77 =39 1978 75 77 -2.6 1979 77 77 0 The Disabled Year Simulated Actual Percent Difference 1974 2,035 1,636 +24.4% 1975 2,065 1,933 + 68 1976 2,087 2,012 + 37 1977 2117 2,109 + 0.4 1978 2,139 2,172 - 15 1979 2,165 2,201 - 16 Sources: Projections of the Macroeconomic-Demographic Model; De- partment of Health and Human Services, Social Security Ad- ministration, Social Security Bulletin, Annual Statistical Supplement, 1977-1979, Table 148. Medicare Estimates Medicare pays for a portion of inpatient, outpatient, and home health care for the elderly. The total costs of the Medicare program increased almost six-fold from 1970 ($7.5 billion) to 1981 ($44.8 billion). In the base case forecasts reported in this section, the Medicare projec- tions are based upon an assumption of constant real age 2A Medicaid model is now being added to the MDM. 83 Base Case Simulation and sex specific Medicare payments for each category of service. Although recent history suggests that this assump- tion may understate expected increases in Medicare pro- gram costs, it provides a useful benchmark for identifying increases in medical costs due solely to demographic ef- fects. The estimates presented here do not include Medic- aid expenditures, a substantial portion of which are devoted to nursing home care for the elderly.?? In Table 10-31 we present estimates of Medicare expen- ditures for each of five categories of health services from 1980 through 2050 (in 1972 dollars). The growth rate in total real Medicare expenditures, 1.4 percent per year from 1980 to 2050, is much faster than the growth in the size of the adult population, 0.6 percent per year over the same period. The faster rise in Medicare expenditures reflects the increasing share of the adult population that is above age 65 after 2010. As has been the case for many variables discussed earlier, the most rapid period of ex- penditure escalation occurs from 2010 through 2030. During these twenty years the annual growth rate in Medi- care expenditures is 2.2 percent. From 2030 through 2050, Medicare expenditure growth is quite modest— about 0.5 percent annually. Figure 10-10 Trends in SSI Benefit Payments by Benefit Type 4.5 0 AGED 4 ® BLIND O DISABLED 2 3.5 3 o 3 a & 2.5 [59 © wn 2F z S ZS 15F a 1+ 0.5 0 1980 1990 2000 2010 2020 2030 2040 2050 YEARS Source: Table 10-29. Even assuming no change in the real per capita costs of medical care for the elderly, the expenditures of the Medi- care program are projected to increase substantially dur- ing the next eighty years—over two and one-half times— due to demographic change alone. The actual costs in- THE NATIONAL INSTITUTE ON AGING MACROECONOMIC-DEMOGRAPHIC MODEL Table 10-31 Projections of Future Medicare Expenditures for Selected Programs, 1980- 2050 : (Billions of 1972 Dollars) Home Inpatient ~~ Nursing Outpatient Health Year Hospital Care? Physician Services Care® Total 1980 11.56 0.25 3.70 0.82 0.30 16.63 1990 13.86 0.31 4.42 0.95 0.36 19.91 2000 15.51 0.37 4.88 1.05 0.41 22.22 2010 17.28 0.41 5.43 1.18 0.46 24.76 2020 21.62 0.49 6.89 1.42 0.56 30.98 2030 26.89 0.64 8.50 1.64 0.72 38.39 2040 29.05 0.76 8.97 1.68 0.81 41.27 2050 30.06 0.81 9.23 1.74 0.84 42.70 Rates of Growth 1980-2010 1.35% 1.66% 1.29% 1.22% 1.44% 1.34% 2010-2050 1.39 1.72 1.34 0.98 1.52 1.37 1980-2055 137 1.71 1.31 1.08 1.48 1.36 2Excludes Medicaid. PIncludes both Parts A and B. Source: Projections of the Macroeconomic-Demographic Model. Table 10-32 Total Benefit Payments From the Retirement Income System By Source, 1980-2050 (Billions of 1972 Dollars) Public Private Employer Year OASI Pensions Pensions SSI Medicare Total 1980 $50.5 © $16.2 $22.3 $1.8 $16.6 $107.4 1990 69.0 29.5 32.6 1.6 19.9 152.6 2000 88.7 49.4 39.2 1.1 22.2 200.6 2010 116.8 83.8 48.2 13 24.8 274.9 2020 169.0 131.8 54.5 1.7 31.0 387.6 2030 220.3 163.8 64.5 1.2 38.4 488.2 2040 246.2 170.9 78.4 0.8 413 537.6 2050 292.8 203.3 86.8 0.6 42.7 626.2 Rate of Growth 1980-2010 2.83% 5.63% 2.60% — 1.08% 1.35% 3.18% 2010-2050 2.32 2.24 1.48 -191 1.37 2.08 1980-2050 . 2.54 3.68 1.96 —153 1.36 2.55 “Aged program only. Source: Projections of the Macroeconomic-Demographic Model. QA CHAPTER 10 Base Case Simulation Table 10-33 Index of Average Benefit Payments From Various Components of the Retirement Income System, 1980-2050 (OASI Average Benefit in 1980 Equals 100) Federal Private Civil State OAS! Pension Service Military and Local 1980 100.0 87.1 252.7 269.4 156.1 1990 1159 101.0 291.1 313.9 184.3 2000 138.8 135.9 306.5 359.6 209.3 2010 164.6 166.2 365.7 418.4 248.7 2020 188.4 186.4 405.1 475.8 286.2 2030 206.0 206.6 451.5 537.2 292.2 2040 225.2 232.0 524.0 610.0 341.2 2050 256.9 265.2 608.3 700.0 371.5 Rate of Growth 1980-2010 1.68% 2.18% 1.24% 1.48% 1.56% 2010-2050 1.12 1.18 1.28 1.29 1.01 1980-2050 1.36 1.60 1.26 1.37 1.25 Source: Projections of the Macroeconomic-Demographic Model. Table 10-34 Total Income Received By the Elderly As Percent of the GNP, 1980-2050 (Billions of 1972 Dollars) Elderly Income Retirement Savings and Total Income as Percent Year Benefits" Investments Earnings Income GNP of GNP 1980 $ 73.7 $14.7 $12.2 $100.6 $1423.4 7.1% 1990 109.8 19.6 17.2 146.6 1688.1 8.7 2000 145.8 19.0 17.2 182.0 1988.4 9.1 2010 197.4 212 20.5 239.1 2299.1 10.4 2020 293.1 29.8 27.3 350.2 2526.9 13.9 2030 386.9 42.8 ~ 35.2 464.9 2744.1 16.9 2040 429.6 46.8 38.0 514.4 2996.0 17.2 2050 492.0 57.8 40.7 590.5 3195.2 18.5 Rate of Growth 1980-2010 3.34% 1.23% 1.75% 2.93% 1.61% 1.28% 2010-2050 231 2.54 1.73 2.29 0.83 1.45 1980-2050 2.75 1.98 1.74 2.56 1.16 1.38 “Benefits from social security retirement (OASI), private pensions, pub- lic employee pensions, SSI, and Medicare received by persons age 65 and older. Source: Projections of the Macroeconomic-Demographic Model. 85 THE NATIONAL INSTITUTE ON AGING MACROECONOMIC-DEMOGRAPHIC MODEL curred by the Medicare program will undoubtedly be greater than these projections, as increases in the price of medical care continue to outpace the increases in most other prices. Also, these projections have excluded Medic- aid expenses for nursing home care, which will also rise rapidly as the number of elderly over age 70 continues to rise. Summary This chapter has presented base case estimates of the Macroeconomic-Demographic Model. By projecting the interaction of the population, labor force, and the econo- my with the various retirement income programs, the model provides policymakers with a useful analytical tool for assessing alternative policies. In addition, it provides a useful framework for evaluating retirement programs together. Table 10-32 summarizes the projected real expendi- tures of each of the five retirement income systems dis- cussed above and their total over the next 70 years. The projected increase is a consequence of the increase in the relative size of the elderly population over the period. Table 10-32 also shows the relative increase in importance of private pensions over the period. While social security and public employer pension expenditures grow at an average annual rate of 2.5 and 2.0 percent, respectively, over the period, private pension expenditures grow at a rate of 4.1 percent. In 1980 we estimate that only 15 per- cent of the total expenditures of these five systems was received in the form of private pensions. In 2050, approxi- mately 32 percent of the total is projected to be from private pensions. Social security remains the mainstay of the retirement income system, however, providing 47 percent of the total of these five programs in 2050. This projection may be a conservative estimate of the possible increase in the relative size of private pensions. We have not assumed that current participation rates in private pension plans will increase, although increases over the next 80 years are likely. The increase in the 86 relative importance of private pensions occurs because of the maturing of the private pension system. The initiation of private pension plans is essentially a post-World War 11 phenomenon. The older social security and public em- ployer systems have already experienced a larger share of their growth. Table 10-33 illustrates another dimension of this growth. Table 10-33 shows the average real benefit paid by each retirement program relative to the 1980 OASI average benefit. Thus, the 2050 value of 256.9 for the OASI program implies the average OASI benefit grows 156.9 percent from 1980 to 2050. Because average benefit levels in each system are linked to the growth in wages, the average benefits pro- vided by all three major retirement income systems are estimated to increase at approximately the same rates throughout the entire simulation period. The Federal em- ployee programs, the Civil Service and military retirement systems, are projected to continue to pay relatively higher benefits than other programs throughout the 1980 to 2050 period. Table 10-34 shows projections of the aggregate income situation of the elderly population (age 65 and over) from 1980 through 2050. In Table 10-34, the aggregate earnings of elderly workers as simulated by the Labor Market Mod- el are added to the benefits of the five retirement income programs that are received by persons age 65 and older. An estimate of the income from savings and investments held by elderly individuals, based on MDM projections, is added to the other income estimates. The MDM projects a major increase in the share of national income received by elderly individuals from earnings plus these five major programs. In 1980, we estimate that about 7.1 percent of all income went to elderly individuals from these sources. Between 1980 and 2010 we estimate that this percentage will increase to 10.4 percent. Over the following 20 years, coincident with the entrance of the baby boom generation into their elderly years, the percentage will increase an- other 7 percentage points. By 2050 elderly incomes from these sources alone are estimated to be about 19 percent of total GNP. Appendix A Equations of the Macroeconomic Growth Model The Macroeconomic Growth Model of the Macroecono- mic-Demographic Model was described in Chapter 3. The Macroeconomic Growth Model forecasts the status of the aggregate national economy. This appendix lists the equa- tions used to generate these forecasts. 87 List of Equations of the Growth Model 1. Household Model Full wealth: WF = (1 + NW) * W(-1) + LDA1 * (EL — HR — RT) + LDA2 * PL * LH Full consumption: PF*F = WF * (AA1 * Rl + AA2 *R2 + AA3 *R3 + AA4 * R4 + AAS *RS5 + AAG * RO + AA7, *R7 + AA8 * RS + BFT * t)/(-1 + BNT * 1) Cong pprion of goods and services; re = AC + BCC * In + BC] * In PF*F PF*F PF*F Consumption expenditure identity: PC*C + PL*1J = PF*F Full consumption price index; In Ph = 05+ F Cc - PG(=1“C(=1) +n PC iE + 05+ Bed go, B= b Ij ¢ 2) * In Poh 2 PF * F FP(—1) * F(—1) PL(—1) Time resources: L+1= 2. Production Model So gmprion goods output: Kio = AD + BDD * In PCS + BDI * In PIS + BDK * In PKD + BDT * t PAD *1D pment goods output: $I = Al + BID * In PCS + BII * In PIS + BIK * In PKD + BIT * t PLID * Copal i services input: er 2 AR 4+ BKD *Inn PCS + BRI * In PIS + BRK * In PKD PLD * LD + BRKT *{ Aggregate index of production efficiency: A = AT + BTD *InPCS + BTI*InPIS + BTK *In PKD + BTT *t Rate of change of production efficiency: 05* (A + AC-1)) = PLD Gy : hel Cy | rondo) PCS PLD * LD MD(—) ID(—1) PCS(—1 —05+( PBS PIS HH = 3 brs PLD * LD PLDC 1) * LD] =D) +05 (240 KD * he ye R= i )* In bkD PLD * LD PLD(—1) * LD(— 1) PKD(— 1) Inputs equal outputs: PKD*KD «+ PID * 1D = PCS 2 CS 4 PIS * 1S 88 3. Market Balance Equations Value of consumption: (1 + TCI" PLS *CS + PCE*CE = PC*C + PCG *CG + PCR*CR Value of investment: (1 +TDH*PIS*IS =Pl*I + PIG *IG + PIR * IR Value of capital services: (1 — TK) * (PKD * KD — TP * PKLG * K(-1)) = N*PK(-1) *K(-1) + U*PKLG * K(-1) — (PKLG PK(-1)) * K(-1) Value of labor services: (1 — TL) * (PLD * LD + PLE * LE + PLG *LG + PLR*LR) = PL*L Quantity of consumption: CS + CE =C+ CG + CR Quantity of investment: IS=1+ 1G + IR Quantity of capital services: KD = AKD * K(-1) Accumulation of capital: K = AKI *I + (1. U) * K(-1) Quantity of labor services: LU = (RLU UYQ) *L * .01 L=1LD + LE + LG + LR + LU 4. Price and Financial Equations Capital Prices: PK = AK * PI PKLG = AKL * PI Saving, income, and investment: S=Y+E-HR-RT-PC*C S=PI*I + PG * (G G(-1)) + PR * (RR(-1)) Y=NW*W(-1) + D—-V + PL*L Depreciation: D = U * PKLG * K(-1) Capital gains: V = (PKLG PK(-1)) * K(-1) + (PG PG(-1)) * G(-1) + (PR — PR(-1)) * R(-1) Rate of return on wealth: NW * W(-1) — V = N * PK(-1) * K(-1) — (PKLG — PK(-1)) * K(-1) + (1-TV) * (EI+VR) + NRE + EJ RV TW * PWIG * QWL Accumulation of wealth: W=W-1)+S—-D+V Real wealth: QW Il In _os+ (PKK | PK(-DK(-D,, K QWL Ww W(-1 K(—1) + 05+ (FG G + PG(—1) S-n., G oY W(-1) Ge=1) + 05 *( R + PR(—1) RCD, Ww W(—1) R(—1) 89 Price deflator for wealth: PW = W/QW Price deflator for wealth, lagged: PWLG = (PKLG *K(-1) + PG *G(-1) + PR*R(-1))/QW/(-1) 5. Government and Rest of World Tax revenues: RC = TC * PCS * CS Rl =TI*PIS*IS RP = TP * PKLG * K(-1) RK = TK * (PKD * KD — RP) + TV * (EI + VR) RL=TL*(PLD*LD + PLE*LE + PLG*LG + PLR*LR) RW = TW * PWLG * QW(-1) Government purchases: E=PCG*CG + PIG*XIG + AIG *1G Government deficits and debt: RE = PCE * CE PLE * LE DG = (E + EI + EIR + EJ + EL + ER) — (RC + RE + RI + RK + RL + RP + RT + RV + RW) G = (DG + ET)/PG + G(-1) Rest of world deficit and debt: DR = PCR*CR + PIR*IR + PLR*IR + VR + NRE — EIR — ER — HR R = (DRET)/PR + R(-1) 6. Investment Composition Real investment components: CDIQ = A1*R1 + A2*R2 + A3*R3 + A4*R4 + A5*RS + AG *R6 + A7*R7 + AB*R8 + BRB3 * 1 IRSQ = B1*R1 + B2*R2 + B3*R3 + B4*R4 + BS*RS + B6*R6 + B7*R7 + B8* R8 + BRB6 * 1 IPEQ = I — IRSQ — CIIQ Prices of investment components: CDIP = BRA4 + BRB4 * PI PIRS = BRAS + BRBS * PI PIPE = IPE/IPEQ Expenditures on investment components CDI = CDIP * CDIQ IRS = PIRS * IRSQ IPE =PI*1 — CDI — IRS 7. Bridge to National Income and Product Accounts Imputed services from consumer durables: CDIMQ = BRAl + BRBI * KD CDIMP = BRA2 + BRB2 * PKD CDIM = CDIMQ * CDIMP 90 Purchases of consumer durables: CDIQ = A1*R1 + A2*R2 + A3*R3 + A4*R4 + AS*R5 + AG *R6 + A7*R7 + A8 *R8 + BRB3 * I CDIP = BRA4 + BRB4 * PI CDI = CDIQ * CDIP Personal consumption expenditure: CNIA = PC * C CDIM + CDI CNIAQ = C — CDIMQ + CDIQ + CYQ PCNIA = CNIA/CNIAQ Gross private domestic investment: INIA = PI * I CDI INIAQ = I — CDIQ + IYQ PINIA = INIA/INIAQ Government purchases: GNIA = PCG * CG + PIG * IG + PLG * LG GNIAQ = CG + IG + LG + GYQ PGNIA = GNIA/GNIAQ Net exports: XNIA = PCR* CR + PIR*IR + PLR*LR + VR XNIAQ = CR + IR + LR + VRQ + XYQ Gross National Product: GNP = CNIA + INIA + GNIA + XNIA GNPQ = CNIAQ + INIAQ + GNIAQ + XNIAQ PGNP = GNP/GNPQ 8. Coefficient Values AC = —.221308 AAl = —.0710942 AD = 121491 AA2 = —.0859844 AF = '— 0775777 AA3 = —.0859844 Al = 505149 AA4 = — 0546291 AT = 0147425 AAS = —.0405804 i AAG = —.0238521 il gy ns AA7 = —.0968787 BDD = .989827 MAS = 1057902 BDI = —.733767 Al = 111.850 BDK = —.256061 A2 = 458577 BDT = .000596963 A3 = —94.626 BFT = —.00205210 Ad = 501.240 BI = .566331 AS = —579.768 BIK = .167436 A6 = —996.320 BIT = .00235925 A7 = 173.740 BKT = —.00295622 A8 = 778.168 BNT = —.0287785 BTT = —.000133813 Bl == a LDAl = 11.8688 B3 = 389.349 LDA2 = LDAl B4 = —66.168 BRAl = —17.3958 BS = 150.463 BRA2 = 0542482 B6 = 881.922 BRAS = —.121340 B8 = 420.134 BRB1 = 318465 BRB2 = .943804 BRB3 = .226268 BRB4 = 859542 BRBS = 1.14764 BRB6 = 225258 91 List and Notation of Endogenous Variables A Cc CDI CDIM CDIMP CDIMQ CDIP CDIQ CNIA CNIAQ cs DG DR "ri GNIA GNIAQ GNP GNPQ INIA INIAQ IPE IPEQ IRS IRSQ IS = fr Index of the change in aggregate production efficiency. Real personal consumption expenditure, includ- ing services of consumer durables. Value of purchases of consumer durables. Imputed value of services from the stock of con- sumer durables. Price of imputed value of services from consumer durables. Quantity of imputed services from consumer durables. Price of purchases of consumer durables. Quantity of purchases of consumer durables. Consumption expenditure, NIPA definition. Real consumption expenditure, NIPA definition. Real output of consumption goods and services by the private sector. Depreciation on private, domestic, assets. tangible Government deficit, excluding social insurance funds Net foreign investment, or U.S. surplus with the rest of the world. Government purchases of goods and services. Quantity of full consumption. Net claims on government, excluding claims on social insurance funds. Government purchases of goods and services, NIPA definition. Real government purchases of goods and service, NIPA definition. Gross national product, NIPA definition. Real gross national product, NIPA definition. Real gross private domestic investment, including purchases of consumer durables. Gross private domestic investment expenditure, NIPA definition. Real gross private ‘domestic investment, NIPA definition. Value of investment in capital goods for production. Real investment in capital goods for production. Value of investment in residential structures. Real investment in residential structures. Real output of investment goods by the private sector. Real private domestic capital stock. Real private domestic capital services. Quantity of labor services supplied by the house- hold sector. LD y LU N NW PC PCNIA PCS PF PGNIA PGNP PI PINIA PIPE PIRS PIS PK PKD PKLG PLD PWLG QW zg 22 8 % 92 Quantity of labor services purchased by the pri- vate domestic sector. Quantity of leisure time. Quantity of labor services unemployed. Nominal rate of return on private domestic tangi- ble assets. Nominal rate of return on private national wealth. Price of personal consumption expenditure, in- cluding services of consumer durables. Price of personal consumption expenditure, NIPA definition. Price of output of consumption goods and ser- vices by the private domestic sector. Price of full consumption. Price of government purchases, NIPA definition. Price deflator, gross national product, NIPA definition. Price of gross private domestic investment, in- cluding purchases of consumer durables. Price of gross private domestic investment, NIPA definition. Price of investment in capital goods for production. Price of investment in residential structures. Price of output of investment goods by the private domestic sector. Price deflator for the private domestic capital stock. Price deflator for private domestic capital services. Current price index for lagged capital stock. Price of labor services purchased by the private domestic sector. Price deflator for private national wealth. Price index for lagged wealth, using current weights. Real private national wealth. Net claims on the rest of the world. Revenue from excise and sales taxes, less subsi- dies, allocated to consumption goods and services. Current surplus of government enterprises. Revenue from excise and sales taxes, less subsi- dies, allocated to investment goods. Corportate profits tax accruals and personal in- come taxes allocated to property compensation. Revenue from personal income taxes,. less re- funds, allocated to labor compensation. Revenue from property taxes. Revenue from estate, death, and gift taxes. C(SEX, AGE) * L; (SEX, AGE)/(C(AGG) Lo(SEX, AGE) A SEX where L, is the new level of labor supply, L, is the prior level of labor supply and C(AGE) is compensation aggregated over 2 sex and 11 age categories into 3 age categories. Divisia index for price changes: INDQ(AGG) = 3 H(JAGG) * (INDQ(AGG)) J where parameters H are shown in Table C-2. Adjustment of prices: P; (SEX, AGE) = EXP(INDP(AGG)) * P, (SEX, AGE) where P; is the new wage for an age/sex group, Py is the old wage for the same age/sex group and AGG is the aggregate group containing the particular group in question. 57 Appendix C Modeling the Substitution Between Age Groups This appendix summarizes the derivation of the form of the production function used in the Labor Market Model. This pro- duction function forms the core of the demand sector of the model and is used to estimate the elasticities of price with re- spect to quantity changes which adjust wages to their equilibri- um values during any execution of the Labor Market Model. Most of the derivation presented in this appendix follows closely the mathematics of the unpublished paper by Joseph Anderson, “Substitution among Age Groups in the United States Labor Force.” ~ 98 I Derivation Assume that aggregate production can be represented with a . single production function of the following type: Q = f(K,Y,M,0) (C-1) where Q is aggregate output, K is capital services, Y is labor services obtained from young workers, M is labor services obtained from middle-aged workers, and O is labor services obtained from older workers. Approximate f as follows: InQ =a +anK+alnY+a,InM+ a2InO Vib INK? + aby InKInY + Y% bey InK In M Vabgo INK In O + % by InY In K + % by In Y? VabyyInYInM + 2 bygInYInO + Y2by InM In K Vabyy InMInY + V2 by In M> + % bye In Mn O Vabok INO InK + ¥2boy InO InY Va boy In O In M + % bog In O? + + + + +0 2 (C-2) An efficient producer in a competitive marketplace must set marginal revenue product equal to product price so that dln Q dQ dInK dK K K Q Q The same relations must hold true for inputs Y, M, and O. There- fore we have four factor share equations like the following: Pg Sk P C3 Sk = ax + bln K + V2 (by + by) InY + V2 (bgy + by) In M + V2 (bo + boy) In O (C-4) Sy = ay + ¥ (by + bg) In K + by InY + 2 (by + by) In M + 2 (by, + boy) nO (C5) Su = ay + Ya(bgy + by) InK + ¥% (by + by) InY + byy In M + ¥ (byo + boy) In O (C-6) So = ag + V2 (bgo + bog) In K + ¥2 (byg + boy) In Y + V2 (byo + boy) In M + bog In O (C7) Note that this system has 20 parameters. If we impose symmetry on this system, then bxy = V2 (by + by) (C-8) by = ¥2 (byw + bug) (C9) bxo = ¥2 (bxo + box) (C-10) byy = V2 (byy + byy) (C11) byo = ¥2 (byo + boy) (C-12) byo = 2 (byo + bow) (C-13) The imposition of symmetry reduces the number of parameters to be estimated to 14 and considerably simplifies the estimated set of share equations. Sk = ag + bk InK + by InY + by InM + by InO (C-14) Sy = ay + bx InK + byyInY + byyInM + bygInO (C-15) 9 Su = ay + byk InK + byyInY + by InM + byoInO (C-16) So = ag + bog InK + boyInY + boy InM + bog InO (C17) If we further assume that this production function exhibits con- stant returns to scale, then we can derive another 9 restrictions, only 5 of which are independent. ag + ay + ay +a =1 (C-18) bko = —bk — bxk — bxu (C-19) byo = —by = by — by (C-20) bywo = — bu = bwy — bum (C-21) boo = —bok = boy — bom (C-22) bok = —by — bux — bk (C-23) boy = —byy — buy — byy (C-24) bow = —bxm = bym — buy (C-25) boo = —bxo = byo = bmo (C-26) These restrictions reduce the system of share equations to Sk = ag + bg(In K—1n O) + bln Y=In O) + byy(ln M—In O) (C-27) Sy = ay + bln K—1n O) + by(ln Y=In O) + byy(In M—In O) (C-28) Su = ay + bgy(In K—=1In O) + byy(In Y=In O) + byy(ln M—In O) (C-29) This system has only 9 parameters. We can use restrictions C-19 through C-26 to derive the share equation C-17 if necessary. Before we consider the next set of restrictions, let us define a Hessian matrix for this system. bXK + aK(aK-1) bKY + aK a¥ bXM + aK aM bKO 4 gK 20 _bXY 4282 bY 4 a¥(@’-1) b™ + a a" b© + a¥ a© THEM 4 gK gM pYM | gY gM pMM 4 gM(gM.1) hMO 4 gM 40 bKO + a¥ 2° bYC + a¥ a© bMO + aM a0 OO + 30(a0-1) (C30) Concavity applies if and only if H is negative semi-definite. We test this by using a Cholesky factorization of H. Define the follow- ing two matrices: 1 0 0 0 L= 21 1 G0 (C31) 831 832 1 0 841 842 843 1 d, 0 0 0 D= 0 d, 0 0 (C32) 0 0 d, 0 0 0 0 d, Then we can rewrite Has L D L’ and H is negative semi-definite if and only if all d’s in D are nonpositive. If we carry out the matrix multiplication we find that bx = d; =~ dag (ag-1) (C-33) byy = gd; — a ay (C-34) bxv = 81 di — ag ay (C-35) by = gi di + dy — ay (a1) (C-36) bymy = 82181 di + gad — ay ay (C37) byw = a1 di + g32dx + ds — ay (ay1) (C38) Thus we can rewrite the system C-27 through C-29 in terms of 9 new parameters— a, ay, ay, di, dy, ds, 821, 831, and ga». Although no new absolute restrictions are imposed, we must partially restrict each of the d parameters if concavity is to be satisfied. Thus, d <0 (C-39) dh <0 (C-40) ds <0 (C41) One final set of restrictions remains before the system can be estimated. The production system of the Labor Model must be consistent with the production system of the Macroeconomic Growth Model. Since the latter has only two inputs—labor and capital—and the former has four as described above, we must ensure that each of three labor subgroups has the same relation- ship with the capital input as the aggregate labor group does in the two factor system. Formally then we impose weak homogen- eous separability on our system. This is equivalent to requiring that bxyy = r ay (C42) b= 1.0 (C-43) by = r (ax-1) (C-44) These three new restrictions enable us to eliminate two param- eters (r is an added parameter). If we choose to eliminate the parameters g*' and g3!, we obtain bx = di — a (a1) (C45) bxy = ay di/(ag-1) — ag ay (C-46) by = ay di/(ag-1) — ag ay (C47) by = ag di/(ag-1)* + d; — ay (ay-1) (C-48) bom = ay ay diag 1)? + gpdy — ay ay (C-49) bum = ay di/(ag-1)? + g3, dy + ds — ay (ayl) (C-50) Therefore we estimate the following equations Sk = ag + [dy ax (ag-D(In K — In O) +[a, dy/(ag1) — a al(lnY — In 0) + [aydi/(ag1) — a@yl(InM —1n0) (C51) Sy = ay + [ay di/(agl) — aga) (In K — In O) +[a,di/(a1) -d; —ay(@-D]UAnY — In O) + [ay ay dy/(ag-1] + gpd — aya, (InM — In O) {C:52) Su = ay + [aydi/(ag-1) — agay)(In K - In O) + [aaydi/(a1) — gd, — ay ayn — In O) + [ay di/(a¢l) + gs dy + dz — ay (ay-1)] (InM — In O) (C-53) where dj, d,, and dz are constrained to be less than or equal to 0. II . Results Equations C-51, C-52, and C-53 are estimated with three stage least squares on annual data for the period 1947 through 1978. These data were drawn from the Gollop-Jorgenson data base described in Chapter 4. The constraints C-39 through C-41 are imposed by setting dP = -v* for each d;. Results of this estimation are: Table C-1 Estimates of the Parameters of a Translog Production Function Coefficient t ay 402 198.49 ay .039 64.47 fy 452 274.84 vi 460 62.44 Va 0.000 0.00 vs 170 10.78 2 —1.405 0.00 Test statistics (Chi-Square values) Test for weak homogenous separability 20.16 Test for concavity 123.68 Unfortunately we reject the hypothesis that our restrictions hold in both cases. Despite this we can use these results to derive both Hick’s elasticities of complemetarity and demand elasticities of price with respect to quantity. If these restrictions are not im- posed, we could obtain unreasonable results, such as positive own price elasticities. Hicks elasticities are defined as follows: (C54) (C-55) Using C-45 through C-50 and C-54 and C-55 we obtain the following estimates of the Hicks elasticities. Cij Cii 1.4 biy/(a;a)) (b;,— a;)/a in Table C-2 Hicks Flasticity of Complementarity Estimates K y M O K -1.31 14 14 14 (3212) (2852) (2852) (2852) y 39 <5 59* (-38) (3.05) (73) M -73 01 (-17.17) ( 04) o 3.14 (6.50) (t statistics in parentheses) * Artifact of the results since v* = 0. Finally, we obtain the required elasticities of price with re- spect to quantity by noting that e; = C;S;. We use 1972 actual values for S; to obtain the elasticity values listed in Table C-3. It is these elasticity values which comprise the core of the demand sector of the Labor Market Model and which determine price adjustment in each of its iterations. 100 Table A-3 Estimated Values of the Elasticities of Price with Respect to Quantity K K 496 b/ 037) Y 062 b/ (.002) M 062 b/ (.002) oO 062 b/ (.002) ¥ .010 b/ (.0003) -175 b/ (.084) -.044 b/ (.044) -.035 (.049) M .060 b/ (.002) -257 b/ (.083) -270 b/ (.050) -144 (.050) Estimates assume 1972 factor share values. bSignificant at the 1 percent level. Standard errors in parentheses. o 015 b/ (.001) -050 (.083) -.035 (.050) -261 b/ (.050) 101 Aoi - Appendix D Equations of the Social Security Model The Social Security Model of the Macroeconomic-Demo- graphic Model was described in Chapter 5. The Social Security Model projects financial and benefit information for the OASI and DI systems by age and sex. This appendix lists the equations and variables used in the Social Securi- ty Model. 103 Variables of the Social Security Model A. Dimension Descriptions In general a dimension of length 2 is sex and length 14 is benefit type. Age dimensions are variously 11, 72, and 85 as described below. Sex: 1=Male and 2=Female. Age: Aggregate age categories: 1=16-17 5=45-54 9=065-67 2=1824 6=5558 10=08-71 3=2534 7=59-61 11=72 + 4=35-44 8=062-64 Single year of age categories: through age 85 for all beneficiaries through age 72 for new primary beneficiaries Benefit type: Old age: (1) Primary retirees C2) Aged spouse (3) Child ( 4) Spouse with child Survivors: (5) Aged widow(er) (6) Disabled widow(er) {7 Child (8) Widow(er) with child (9 Parent Disabled: (10) Disabled primary (11) Child (12) Spouse with child (13) Aged spouse (14) ~~ Lump sum death benefit ($255) B. List of Variables WGNDX(105) = A vector created each simulation year used to index wages. Equals average annual earnings in simulation year 2 divided by average earnings in year I. Its first value represents 1951; its last 2055. COMHIS(2, 11, 105) = From the labor model, we get average per person annual earnings for each of the 11 age groups and 2 sexes. COMHIS stores these values for each year of simulation. OLDBEN(2, 85) = Average benefits for present beneficiaries. 104 SSYR = The year of the simulation, 1951 = 1. The simulation begins in Year 20 (1970). YEAR = Actual calendar year of simulation. CPI$ = CPI, lagged one year. YRWG(2, 11) = Average annual compensation per worker, from the labor model. WORKRS(2, 11) = Number of full and part-time work- ers in the current simulation year, by age-sex group. Obtained from the labor model. : AVEWAG(105) = Average annual earnings (excludin, fringe), for all workers, by year (1 = 1951). MINMB = Minimum monthly benefit. EARNTS = Earnings test amount. TOTBEN(14) = Total benefit payments in the current period, by benefit type. TOTCST(14) = Total benefit payments inthe current period, by benefit type. TBENFT(2) = Total cost of benefits 1 = OASI 2 = DI. NPCTS = Number of percents presently in old PIA table (this will vary with time, but freeze at 22 in 1989). AVERET = Average benefit paid to primary retirees this year. AVEDIS = Average benefit paid to primary disableds this year. AVRET$ = AVERET, lagged one year. NORET = Number of retired primaries. NODIS = Number of retired disableds. SURBEN = Average survivor's benefit. CPIINC(105) = Change in CPI. Ratio of this year’s CPI to last year’s. (Always set to 1) NEWPOP(2, 72) = Population of new primary benefi- ciaries, by age and sex. OLDPOP(2, 85) = Population of primary beneficiaries that have been claiming benefits since prior to the current simulaton year. Goes to 85, as this actually 83+ NEWBEN(2, 72) = Average annual benefit for new pri- mary beneficiaries, by sex and age. Actually, the earliest possible age to be a new primary is 31. These bene- fits have been adjusted for early/late retirement. Note that all financial variables are in constant 1972 dol- lars unless otherwise specified. C. List of Parameters NPIABD(2) = Bend points in new PIA formula (1977 Law) OPIAPT(22) = Percentages for bend points in old PIA Table. 22 are required, as one will be added for each year that the tax max is increased, and the last update will be in 1989. (This is based on an assumption that the youngest people needing an old PIA table are born in 1917, and re- tire at very latest at age 72.) PIA78(11) = The percentages for the 1978 PIA table, retained for use in the transitional people. NPIAPT(3) = New PIA percents. OPIABD(21) = Old PIA bend points. They correspond to OPIAPT in that OPIAPT(I) is the per- cent used for the interval from OPIABD(I-1) to OPIABD(I). Therefore, there is no OPIABD(22). TAXMAX = Taxable wage base for the current simula- tion year. TWB(105) = TAXMAX for previous years, by year (1951 = 1). NAGE = Number of age groups. Presently 11. LOWAGE = Lowest age permissible to be a new prima- ry beneficiary. Presently age 31 for disabled. HIAGE = Highest age maintained by model, a catch-all for it and higher ages. Presently 85. PCENT(14) = Percent that beneficiary group gets of its primary beneficiary's PIA. Equations of the Social Security Model Factor to convert from compensation to earnings: F1 = 841 X EXP (-.002 X (NYR-7)) Taxes collected: SSTAX = TCOMP X F1 X F2 X F3 X F4 X RATE Index year: SSYR = NYR + 22 (1951 equals 1) Increase in wages: DELWG = AVEWAG(SSYR 2) / AVEWAG(SSYR 3) 105 Wage indices: WGNDX(YEAR) = AVEWAG(SSYR) / AVEWAG(YEAR) WGNDX(YEAR) = 1 (if YEAR SSYR-2) (otherwise) Present Population: OLDPOP(SEX,NUMAGE) = OLDPOP(SEX,NUMAGE-1) X (1-MORE(SEX,NUMAGE-1)) New award population: NEWPOP(SEX,NUMAGE) = IPOP(SEX,NUMAGE) X PNB(SEX,NUMAGE) Laws used in each PIA determination: New law used if born after 1922 Old law used in born before 1917 New law used for disability benefits if born after 1916 Transitional formula used in all other cases Years used in AMW and AIME calculations: N = difference between the year applicant attained the age of 61 and the greater of 1956 or the year the person turned 26. Maximum of 35. Old law calculation: 1. Pick the highest N years’ earnings from 1951 to present. Take their average and call the Average Monthly Wage(AMW) 2. Use the PIA table to find PIA corresponding to this AMW. New law calculation: 1. For all since 1950 or age 26 (whichever is greater) choose the lesser of the earnings that year or the taxable maximum. 2. Multiply these earnings by the corresponding wage indices 3. Choose the high N of these and average. This results in an Average Indexed Monthly Earnings (AIME) value. 4. Using the simplified PIA table find the PIA. Transitional calculation: 1. Determine a PIA under the new law. 2. Determine a PIA under the old law, except: a. Use the 1978 PIA table. b. Increase the resultant benefit by the increase in the CPI from 1978 to the time of retirement. c. Choose the higher of a or b. Final benefit calculation: Adjust the PIA values for early/late retirement, earnings test, and maximum family benefit limits to determine the actual benefit paid. Appendix E Equations of the Private Pension Model The Private Pension Model of the Macroeconomic-Demographic Model was described in Chapter 6. It projects the number of participants and amounts of pension contributions, as well as the number of retirees and the amounts of retirement benefits paid for each of four pension statuses. All projections are further disaggregated by age and sex. This appendix lists all variables and equations of the Private Pension Model. 107 Variables of the Private Pension Model A. Dimension Descriptions In general, a dimension of length 2 is sex, length 10 is age, and length 4 is pension plan job type. Sex: 1 = Male 2 = Female Age: 1= 1617 6 = 55-61 2 = 18-24 7 = 62-64 3 = 25-34 8 = 65-67 4 = 35-44 9 = 68-71 5 = 45-54 10 =72 + Pension Plan Type: 1 = Defined Benefit 2 = Defined Contribution 3 = Individual 4 = None B. List of Variables W(10,2,4) = COVB(10,2,4) = PARW(10,2,4) = VESW(10,2,4) = R(72,2,4,18) = TB(4) m= DBCC(10,2,4) = UL = C(10,2) = CB(4) = YRPMTS(Plan, = Numerical Age, Sex, (Year of Retirement 1946) CB$(4) = ACCT(Plan, Number of Years Worked to Date, Sex, Numerical Age) Number of workers in each age/sex group in each type of private job in thousands. Number of persons covered by a plan, by age, sex, and plan type. Number of persons participating in a plan by age, sex, and plan type. Persons vested by age, sex, and plan type. Number of persons retired, by (in order) age, sex, plan type and number of years retired. Total benefits paid by status type 1-4. Defined benefit per worker normal costs for each specific age/sex job group. Costs of amortizing unfunded liabilities for each funded plan. Annual compensation by age and sex. Cash balance, by plan type, for funded plans. It is used to determine level of cash yields. A table which will supply the coverage annual benefit paid to all individuals of a certain age and sex who retired in a par- ticular year. Last year’s fund balances by plan type. Account information (described in docu- mentation) used to calculate retirement benefits for YRPMTS. C. List of Parameters PIC(10,2,4) = Probability that a worker is covered in a particular plan given he/she is in a par- ticular age/sex group. PIP(10,2,4) = Probability that a worker participates in a private pension plan given that he/she is in a particular age/sex group and works at an establishment with a particular type of plan. PIV(10,2,4) = Probability that a worker is vested in a private pension plan given that he/she is in a particular age/sex group and partici- pates in a particular type of plan. BR = Benefit rate, for our defined benefit plan. IR = Whatever interest rate we end up using : for vield on our pension fund assets. Pres- ently, it equals a historical rate of 2.1% for return on pension assets. MPT(10) = The median age of each age group. FUNC(72) = An array which, given a numerical age, will return the number of the age group to which it belongs. M(Age,Sex) = Probability that an individual will die this vear, given numerical age and sex. From the UP-1984 tables. PIR(Numerical ~~ = Probability that an individual will retire age, Sex, this year, given age, sex and job/plan type. Job Type) DCR = Defined contribution rate; rate used for all defined contribution plans. ANNUT(Numer- = Cost of an annuity guaranteed to pay $1 ical Age, Sex) annually for an individual's life, given nu- merical age and sex. Assumes life expec- tancies given in Wyatt Company's UP-1984 table. DNRAAA(Numer- ical Age, Sex) Probability that an individual will remain with a job to normal retirement age, given numerical age and sex. Equations of the Private Pension Model A. Coverage, Participation and Vesting: Number of workers covered and in a particular plan: 1) COVB;jk = PIC; X Wij i=1...10 (Age) j=1...2(Sex) k=1...4(Pan) Number of workers participating in each plan: (2) PARW; = COVBj X PIP Same dimensions as above. Number of vested workers in each plan: (3) VESWj = PARWj, X PIV Same as in (1) and (2). In all above, we are simply applying prevalence rates. 108 B. Retirement: Number of retirees: (4) Ruspy = Number of retirees of numerical age a, sex s, plan p, retired y years. = Ws X PIR, ify = 1and a = 54; = {R, -issiy=1) X M,15) ify =1and 72 = a 55 and y-a = 54 = (Ry15py-1) X M,12) + (Rypspy) X Mop) ifa = 72 and ya = 55 The first section of this equation refers to new retireds; it is the number of workers in a particular age group times the probabil- ity that they will retire; this is an expected value of retireds. The second moves already-retireds up one year year, decrementing for mortality. The third is the same as the second, but, as the “age 72” category is actually those age 72 and older, requires the extra term. The subscript “A” in “W,s", is different than the subscript “a”, in the other variables. “A” is the corresponding cohort number to a numerical age a. C. Contributions and Benefits 1. Defined benefit normal costs: (5) DBCC, = C., X DNRAAA,, X ANNUT,s X BR Under the accrued benefit cost method, normal cost equals annual benefit times probability that an individual will stay on the job to age 65 times present value of the cost of an annuity at that time to cover the benefit. 2. Lifetime Payroll Accounts: (6) ACCTpy = DCR X Cys ify=1andp =2o0rp =3 (First time persons in defined contribution plans get same percent of their salaries put into their accounts.) ACCT, = Cys (First time persons in defined benefit plans have an amount equal to the year’s salary, as the submodel maintains career averages.) ACCTpysa = ACCT (y-1)5,(a1) x {1 #4 IR) + DCR. % Cas ify =1landp =2o0rp =3 (Add to the value of the individual's retirement fund, compound interest, for defined contribution.) ACCTpysa == ACCT (y-1)5(a1) + Cas (Same, for defined benefit career salary sums.) ify=1andp = 1 3. Total Contributions and Benefits: 2 10 Gy TG = 3 > (DBBC,s s=1 a=1 X PARW,s;) + UL 2 10 (8) TC; = : 2 (PARW 5p X Cys s=1 A=1 v5 pop) p=23 72 2 17 (9) TB, = 3 3 3 a=5s5 s=1 NYRRET = 1 YRPMTS, s nyrrer X Ruspy p=l3 4. Accounts of Retiree’s Yearly Benefits: YRPMTS contains information disaggregated by age, sex, year of retirement, and plan type on annual benefits paid for defined benefit. (10) YRPMIS,, = BR % ACCT 4;., foralls,a =54 Here, we assume 21 year service on those retiring with defined benefit plans. (11) YRPMTS,,,,, = weighted average of lump sum annuity values, amortized over the life expectancy S. Cash Balances: (12) CB,=CBy-1) X (1 +R) + (TC, — TB,) X (1 + IR/>) Here, we compound interest on the lagged assets, and determine the change in assets. The change is in- creased by one-half the interest rate as a continuity correction. 109 Appendix F Equations of the Public Employee Pension Model The Public Employee Pension Model was described in Chapter 7. It projects the number of participants and amounts of contributions in the public pension system and the number of retirees and amount of retirement benefits paid for each of eight public pension systems. All projections are further disaggregated by age and sex. This appendix lists all variables and equations of the Public Employee Pension Model. Appendix G contains further information about the development of the equations for estimating public employment growth within this model. 110 Variables of the Public Employee Pension Model A. Dimension Descriptions In general, a dimension of length 2 is sex, length 10 is age, and length 8 is pension plan job type. Sex: Age: NBR N= N= Male Female 16-17. 6 = 55-61 18-24 7 = 62-64 25-34 8 = 65-67 35-44 9 = 68-71 45-54 10 = 72 + Pension Plan Type: x 4 NR = Federal civil service Military enlistees Military officers State and local hazardous duty State and local general and administrative workers State educators Local educators State and local employees not covered by any plan B. List of Variables W(10,2,8) COVB(10,2,8) PARW(10,2,8) VESW(10,2,8) R(34,2,8,34) TB(8) DBCC(10,2,8) UL(8) ©(10,2) CB(8) YRPMTS(Plan, Numerical Age, Sex, (Year of Retirement 1946)) CB$(8) ACCT(Plan, Number of Years Worked to Date, Sex, Numerical Age) Number of workers in each age/sex group in each type of public job in thousands. Number of persons covered by a plan, by age, sex, and plan type. Number of persons participating in a plan by age, sex and plan type. Persons vested by age, sex, and plan type. Number of persons retired, by (in order) age, sex, plan type and number of years retired. Total benefits paid by fund type 1-4 for this year. Defined benefit per worker normal costs for each specific age/sex job group. Costs of amortizing unfunded liabilities, for each funded plan. Annual compensation by age and sex. Cash balance, by plan type, for funded plans. It is used to determine level of cash yields. A table which will supply the coverage annual benefit paid to all individuals of a certain age and sex who retired in a par- ticular year. Last year’s fund balances by plan type. Account information (described in docu- mentation) used to calculate retirement benefits for YRPMTS. STED LED HAZT SLGA FCST Total number of state educators, in thou- sands of persons. Total number of local educators, in thou- sands of persons. Total number of hazardous duty workers, in thousands of persons. Total number of state and local general and administrative workers, in thousands of persons. Total number of federal civil service workers, in thousands of persons. C. List of Parameters PIC(10,2,8) PIP(10,2,8) PIV(10,2,8) BR(8) IR MPT(10) FUNC(72) M(Age,Sex) PIR (Numerical Age, Sex, Job Type) DCR ANNUT(Numer- ical Age, Sex) DNRAAA(Numer- ical Age, Sex) Probability that a worker is covered in a particular plan given he/she is in a par- ticular age/sex group. Probability that a worker participates in a public pension plan given that he/she is in a particular age/sex group and works in a particular sector. Probability that a worker is vested in a public pension plan given that he/she is in a particular age/sex group and partici- pates in a particular type of plan. Benefit rate by plan type. Presently, it equals a historical rate of 1.8% for return on pension assets. The median age of each age group. An array which, given a numerical age, will return the number of the age group to which it belongs. Probability that an individual will die this year, given numerical age and sex. From the UP-1984 tables. Probability that an individual will retire this year, given age, sex and job/plan type. Defined contribution rate; rate used for all defined contribution plans. Cost of an annuity guaranteed to pay $1 annually for an individual's life, given nu- merical age and sex. Probability that an individual will remain with a job to normal retirement age, given numerical age and sex. Equations of the Public Employee Pension Model A. Estimation of Public Employees Total number of federal civil servants, in thousands: (1 in FCST = 2800 + 32 X (NYR 14) State and local hazardous duty, in thousands: 2) HAZT = 84.882 + .726451 X DPI State and local general and administrative, in thousands: 3) SLGA = 306.767 + 4.582 X DPI State educators, in thousands: 4) STED = -577.755 + .0000554 X CAP Local educators, in thousands: (5) LED = -47389 + .15325E — 3 X SAC if SAC = SAC(-1) = 2748.99 + 76.4168 X (NYR — 14) else Shares of the labor force: 6) SHR, =W,/ 3 W, AS A = 1 10 (age groups) S=1 2 (sex) All totals determined previously are apportioned using the shares determined above. B. Coverage, Participation and Vesting Number of workers covered and in a particular plan: 7) COVB;j = PICij X Wik i=1 10 (Age) j =1 2 (Sex) k =1 8 (Plan) Number of workers participating in each plan: (8) PARWjji = COVBijjk X PIP, Same dimensions as above. Number of vested workers in each plan: 9) VESWjj = PARW X A Same as in (1) and (2). In all above, we are simply applying prevalence rates. C. Retirement Number of retirees: (10) R,»y = Number of retirees of numerical age a, sex s, plan p, retired y years. = Wy % PIR, ify = 1and a = 39; = (Ra1spy-1) X (My) ify=1and 72 =a = 39and ya < 16 = (Ra1spy-1) X M,15) = (Ryzspy) X M0) ifa = 72 and y-a < 55 The first section of this equation refers to new retireds; it is the number of workers in a particular age group times the probabil- ity that they will retire; this is an expected value of retireds. The second moves already-retireds up one year, decrementing for 112 mortality. The third is the same as the second, but, as the “age 72” category is actually those age 72 and older, requires the extra term. The subscript “A” in “W,", is different than the subscript ‘a’, in the other variables. “A” is the corresponding age group number to a numerical age a. D. Contributions and Benefits Defined benefit normal costs for all plans that are funded: (11) DBCC,sp = Cp X DNRAAA,, X ANNUT,, X BR, Under the accrued benefit cost method, normal cost equals annual benefit times probability that an individual will stay on the job to age 65 times present value of the cost of an annuity at that time to cover the benefit. E. Lifetime Payroll Accounts (12) ACCT, = ACCT (y.1y5(a1) + Cas Cas ify =1 ify =1 These are the accounts maintained of career salary history, used in determining benefit levels. F. Total Contributions and Benefits To 1B). 1G.= DBCCyg ( ) P g=1 A= ( ‘ASP X PARW,s) + UL, P=1567 72 2 34 (14) 1B; = 3 3 ~ a=39 s=1 NYRRET = 1 YRPMTS, nvrerr X Raspy p=17 G. Accounts of Retiree’s Yearly Benefits YRPMTS contains information disaggregated by age, sex, year of retirement, and plan type on annual benefits paid for defined benefit. (15) YRPMTSp,,, = BRp X ACCT) 5, for all s, a 39 Here, we assume 21 year service on those retiring with defined benefit plans. H. Cash Balances (16) CB, = CB,(-1) X (1 + IR) + (TC, TB,) xX (1 + IR/2) P= 1567 Here, we compound interest on the lagged assets, and determine the change in assets. The change is increased by one-half the interest rate as a continuity: correction. Appendix G Modeling Public Employment Growth The Public Employee Pension Model divides public employ- ment into seven sectors: Federal civil service, military enlisted personnel, military officers, state and local hazardous duty per- sonnel, state and local general and administrative workers, state educators, and local educators. The core macroeconomic/ labor market model does not provide such detailed information as number of persons by job type. Consequently, it must be esti- mated in the Public Employee Pension Model. Each year the total number of workers in each sector of the public employee work- force is estimated, then that number is distributed among the 20 age-sex groups according to the distribution of civilian private sector workers forecast by the Labor Market Model. This appen- dix explains the techniques used to estimate total workers in each sector of public employment. 113 1. Federal Civil Service Federal civilian employment has grown very slowly in the past twenty years—less than one percent annually from 1960 to 1968, and about two-tenths of a percent annually from 1968 to 1980. In order to project Federal civilian employment, equations specify- ing that Federal employment is a function of total civilian em- ployment, real GNP, and a time trend, and of a time trend alone, were estimated by ordinary least squares. A simple time trend alone provided the best equation. This equation projects that the size of the Federal civil service will increase by about 66 percent during the simulation period to 5.8 people. The equation employed was: FCST = 2800 + 32 * TIME (136.7) (47.2) E08 (t statistics in parentheses below coefficients.) (TIME = 1in 1960, 2 in 1961, etc.) 2. Military An examination of military manpower levels over the past three decades shows that the major force generating changes in the size of the armed forces is the presence or absence of U.S. military conflict. Removing years of active military conflict, the size of the armed forces has been relatively constant at about 2.3 million. We used that level for the entire simulation period. The Department of Defense's Office of the Actuary provides informa- tion on the agesex composition of the military forces.! We as- sumed that the demographic composition of the armed forces will remain the same in all future years as it was in 1980. 3. State and Local Governments State and local governments have expanded rapidly in the past 20 years. It is unlikely that these rates of growth will continue through 2055.2 For this reason we disaggregated state and local government employment into several sectors. The approach was also necessitated by the variation among the pension plans of different state and local government sectors. Historical data from 1960 to 1979 was obtained from the Bureau of the Census, Government's Division. The data were broken down into the four sectors of state and local government employment, and analysis indicated modeling of separate components would bet- ter explain growth of state and local systems. Equations explaining the level of state and local general and administrative workers, hazardous duty workers, state educa- tors, and local educators were each estimated in order to deter- mine suitable independent variables. All data are from the Department of Labor’s Employment and Earnings series. Varia- bles tested for usefulness as independent variables included time, real GNP, real disposable personal income, total popula- tion, civilian employment, school age children, and college age persons. Results of the regressions for each component follow: e The level of real disposable personal income was highly correlated with the number of general and administrative workers. The estimated equation was: "Unpublished data, Department of Defense, op. cit. At the current rates of growth, fifty million persons would work in state and local governments by 2055, 42 percent of the labor force. 114 ADM = 307 + 4582.9 (DPI) (3.0) (34.9) R* = 985 (DPI is in billions of 1972 dollars.) As real personal income rises, services demanded of the state and local government sector also rise, creating employment growth in this component. e Hazardous duty workers (police and firefighters) also showed a high correlation with real disposable personal income. The estimated equation was: AZD = 84884.8 + 726.4 (DPI) 41) (261) R? = 974 Although hazardous duty workers represent only two percent of state and local employment, they were treated separately be- cause of the unique features of their pension plans. For example, state general and administrative workers usually have 5 or 10 year cliff vesting, while hazardous duty officers are more often in pension plans with 20 or 25 year cliff vesting, creating apprecia- bly different retirement and benefit patterns. Most state general and administrative workers plans are fully or partially funded. Most hazardous duty workers plans are not. e The number of state educators could be best explained by the size of the college age population (CAP), as would be expected, since state educators are usually state university personnel. SED = —577.8 + .0055 (CAP) {=173) (393) R? = 989 State educators were distinguished from local educators be- cause their number is related to different variables. State educa- tors are, for the most part, concerned with the college and university population who are generally ages 18-24. Educators on the local level are involved with the elementary and secon- dary school population, ages 5-17, since local educators are mainly primary and secondary school instructors. However, the number of employed instructors was not correlated with the size of the school age population when that population declined. Consequently, an equation specifying that the number of local educators is a function of the size of the population age 5 through 17 was estimated only for years when that population was growing, from the beginning of the period to the early seventies. LED = —473.9 + 015 (8AC) (-84) (134 R? = 952. After 1971, the number of school age children (5-17) declined. One might have expected to find teaching staffs diminishing in response to the declining enrollment. However, the number of local educators continued to grow from 1971 to 1979. This pat- tern was best represented by a linear time trend. LED = 2749 + 7.6 (Time) (273) (123) R = 9 Appendix H Report Generating Capabilities of the Macroeconomic-Demographic Model The Macroeconomic-Demographic Model simulates over 300,000 variables for each of the 86 years from 1970 to 2055. As a result, the preparation of a report of a single simulation is a major task. Careful attention to the design of these simulation reports is warranted for two reasons. Not only are these reports essential for describing the results of any simulation to interest- ed policymakers and fellow researchers, but they serve as useful guides to the development of the model during the research process. We developed most of the report writing routines in conjunc- tion with each of the major models of the Macroeconomic- Demographic Model. These reports can be produced at the conclusion of any single simulation year. In addition to the reporting functions within each model we have developed a number of specialized routines which provide summary tables for the entire model at the conclusion of a full simulation. Below we first describe the report generating capabilities of the Labor Market Model, Social Security Model, Private Pension Model, Public Employee Pension Model and the Medicare Model. We then describe each of three summary report writers. The Labor Market Model simulates the outcomes of the labor market in each year of the simulation—compensation rates, em- ployment levels, hours worked, and participation rates for each of twenty-two demographic groups. An example of an annual report of the Labor Market Model is shown in Table H-1. The report first displays the estimates of all major Labor Market variables for all twenty-two demographic groups—compensa- tion rates (the price of labor), unemployment rates, hours worked annually, participation rates, the number in the labor force (in millions of workers), the quantity of hours worked, the share of total labor input, and the total compensation received by each group. Next, the report shows aggregate values for some major variables in three groups—young, middle-aged, and old workers. These aggregate values are calculated through a Divisia aggregation and are the major determinants of movements in compensation rates over time. The aggregate table displays only compensation rates, quantities (in terms of hours worked), share of hours worked in the group, and total compensation earned by the group. The last section of the Labor Market report displays those variables of primary interest to the pension mod- els—population, employment, and annual compensation in twenty age-sex groups. Despite the complexity of the Social Security Model, the re- sults are reported in a relatively simple format. Information is generated for both the revenue and payment sides of the system as shown in Table H-2. Primary beneficiaries are distinguished by age, sex, and status—disabled and retired. Secondary benefi- ciaries are distinguished by type—spouse, widow, child, etc.— and status. Both average and total benefits paid are displayed for retirees, survivors, and disabled beneficiaries. Finally, an esti- mate of total Social Security payroll taxes received is presented. Analogously to the Social Security Model, the Private Pension Model simulates both revenues and payments to the three differ- ent prototypical private pension plans. Table H-3 shows some 115 typical output from the report generated annually by this model. First the retiree population is disaggregated by age and plan type—defined benefit, defined contribution, or IRA/Keogh. Next the replacement rates for new retirees in that simulation year are shown—again by age and plan type. The replacement rate is defined here as the ratio of the average benefit level in that age- plan type class to last year’s average earnings in that same age class. Next, financial data—total contributions, total payments, and total assets held—are shown. Finally, workers either cov- ered, participating, or vested in a plan are distinguished by age and plan type. The Public Employee Pension Model really is an amalgam of the simulations of six public sector pension systems. Table H-4 shows a typical year’s summary output. The output structure is very similar to that used in the Private Pension Model. First, the report displays the distribution of retirees by age and plan type. Next, financial data are shown for all plans. In this subtable the State and Local Other category includes State and Local General plans, State Educator plans, and Local Educator plans. The Mili- tary and Hazardous Duty plans receive no contributions and have no fund balances because they are pay-as-you-go plans. Finally, covered, participating, and vested workers are distin- guished by age and plan type. Table H-5 shows a sample report from the Medicare Model. The Medicare Model generates estimates of recipients and ex- penditures in considerable demographic detail—twenty age-sex groups are presented. Six different categories of health services are also presented. In addition to these annual reports of each of the major mod- els, summary reports covering the entire simulation period can be produced at the conclusion of a full run. Table H-6 shows a full report for the principal macrovariable in the Macroecono- mic Growth Model, GNP. Annual changes are calculated and the simulated path of the variable is compared to its base values. Base values are historical values where available. For future years, base values are those contained in the most recent Base Case simulation. Thus, we can easily compare the influence of any policy-induced effects on the trend of any macrovariable at the conclusion of a simulation. At the bottom of this report, the growth rates of the macrovariable are calculated in ten year and full period intervals. An important feature of the Macroeconomic-Demographic Model is its ability to generate thirty special reports at the con- clusion of a simulation. These reports cover every aspect of the simulation including: Population trends by age and sex Major macroeconomic trends Major labor variables by age and sex Social security participants, beneficiaries, and fund status Private pension system participants, beneficiaries and fund status ® Public pension system participants, beneficiaries and fund status Each of these reports presents simulated values at five year intervals over the course of the simulation. These reports are designed to provide a reasonable amount of trend information for a set of linked variables so that possible correlations in the behavior of major variables in the model can be observed. Ap- pendix I provides a complete listing of all these reports from the Base Case simulation. These thirty special summary tables are also important be- cause they serve as the basis for the operation of other report generating routines. At the conclusion of any important simula- tion, these tables are saved on disk and used as inputs to two Table H-1 report generating programs which run separately from the rest of the models of the Macroeconomic-Demographic Model. The more versatile of these routines compares all tables gen- erated by any two different simulation runs. We have. found examination of tables such as these at the end of a policy run to be the best way to understand the operation of the Macroecono- mic-Demographic Model and from which to formulate judg- ments about the actual effects of any policy change. A second report generating program provides the ability to portray these changes graphically. The final report generating routine performs a highly special- Simulation of the Labor Market in 1980 Age-Sex Group Values are Price Urate Hours Prate Labor Force Quantity Share Compensation Male 16-17 1.29 20.88 706.40 0.493 1.146 0.0078 1.479 18-24 3.30 14.16 1256.72 0.819 12.154 13.111 0.0893 45.901 25-34 6.04 5.84 1752.74 0.949 17.066 28.165 0.1919 170.040 35-44 7.60 4.24 1867.94 0.953 11.994 21.454 0.1461 162.971 45-54 7.82 3.57 1894.76 0.908 10.030 18.325 0.1248 143.283 55-58 7.2 2.49 1791.55 0.836 3.681 6.431 0.0438 46.341 59-61 21 3.62 1791.55 0.747 2.264 3.909 0.0266 28.168 62-64 7.21 375 1791.55 0.536 1.397 2.408 0.0164 17.332 65-67 4.74 7.25% 1318.84 0.307 0.751 0.917 0.0062 4.346 68-71 4.74 4.29 1318.84 0.235 0.648 0.818 0.0056 3.877 72+ 4.74 2.89 1318.84 0.122 0.612 0.784 0.0053 3.713 Female 16-17 1.07 22.46 705.42 0.427 1.706 0.933 0.0064 1.003 18-24 3.08 13.79 1085.57 0.668 9.759 9.134 0.0622 28.143 25-34 3.95 8.74 1278.44 0.626 11.399 13.299 0.0906 52.487 35-44 3.92 6.16 1354.64 0.608 8.019 10.194 0.0694 39.011 45-54 3.99 5.60 1421.65 0.537 6.274 8.420 0.0574 33.609 55-58 3.68 3.98 1419.30 0.484 2.329 3.174 0.0216 11.669 59-61 3.68 4.46 1419.30 0.404 1.358 1.841 0.0125 6.768 62-64 3.68 4.47 1419.30 0.274 0.821 1.114 0.0076 4.094 65-67 2.45 6.36 1116.01 0.157 0.469 0.490 0.0033 1.200 68-71 2.45 3.47 1116.01 0.101 0.366 0.305 0.0027 0.966 723 2.45 2.58 1116.01 0.038 0.317 0.345 0.0023 0.844 Aggregate Group Values are Price Quantity Share Compensation Young 3.1359 24.323 0.1657 76.526 Middle 6.0231 99.858 0.6802 602.300 old 5.6960 22.625 0.1541 129.339 Total 146.806 1.0000 808.164 Special Data Check for Input Numbers to Pension Submodels Males Females Population Workers Compensation Population Workers Compensation 16-17 4161 1622 912 3998 1323 758 18-24 14843 10433 4400 14610 8414 3345 25-34 17984 16069 10582 18204 10402 5046 35-44 12581 11486 14189 13191 7325 5304 45-54 11042 9672 14815 11684 5923 5674 55-61 7433 5772 12909 8173 3533 5218 62-64 2606 1344 12909 2996 785 5218 65-67 2442 695 6249 2987 439 2734 68-71 2754 620 6249 3612 354 2734 72+ 5010 594 6249 8342 309 2734 116 Table H-2 Projections for the Social Security System for 1980 Primary Beneficiaries, in Thousands Disabled Retired Age Male Female Age Male Female 25-34 22 7 62-64 769 858 35-44 165 53 65-74 5725 4871 45-54 398 157 75-84 2455 2410 55+ 1015 440 85+ 640 630 Secondary Beneficiaries, in Thousands OASI: DIL: Aged Spouses: 2799 77 Children: 646 1413 Spouses with Children 194 394 Aged Widow(er)s: 4226 Disabled Widow(er)s: 129 Surviving Children: 2645 Widow(er)s with Children: 566 Parents: 15 Special Age-72: 94 Average Retirement Benfit: 1950.90 Average Survivor's Benefit: 1432.85 Average Disability Benefit: 2237.49 Total OASI Benefits Paid: 50542992 Total DI Benefits Paid: 7580326 Average New OASI Award: 2221 Total Primary Beneficiaries: 18357 Total Primary Benefits Paid: 35812896 Total Secondary Population: 11314 Total Secondary Benefits Paid: 14735338 Total New OASI Primary Beneficiaries: 1520 Total DI Primary Beneficiaries: 2257 Total DI New Primary Beneficiaries: 352 SS Tax This Year in Billions: 59.30 ized but useful function—it develops reports which track the experience of a single cohort over its adult lifetime. Since one of the models strongest features is its demographic detail, we often find it informative to observe how groups who are demographi- cally advantaged, i.e. small relative to prior and following co- horts, and disadvantaged, ie. large relative to prior and following cohorts, fare over their lifetime. An example of a report showing the earnings of the cohorts born at ten year intervals from 1950 through 2030 is shown in Table H-7. If we look along the diagonal, we see the successive entry compensation rate (at age 20) experienced by each group in 1972 dollars. The process of economic growth causes the steady increase in the entry compensation rate. As each group ages, its compensation rates increase as its labor market exper- ience grows, until age 60. After age 60, group compensation rates may or may not fall depending on the relative balance of economic wide productivity increases and productivity de- creases due to aging. 117 Table H-3 Private Pension Projections for 1980 Number of Retirees, by Age & Plan Plan Type Defined Defined IRA/TSA Age Benefit Cont. /KEOGH None 55-61 I 616896 313461 80836 1011975 1 62-64 1 1090438 554089 142893 1920407 I 65:67 1 1516502 770895 193410 2863446 1 68-71 1 1523489 775119 181519 3259001 I 72+ 1 1062365 544415 61367 2071756 1 Total I 5809690 2957979 660025 11126585 I Replacement Rates for New Retirees Defined Defined Benefit Cont. Other None Male Female Male Female Male Female Male Female 55-61 0.23 021 0.13 0.11 0.11 009 00 00 62-64 022 021 015 013 013 011 00 00 65-67 043 041 034 027 029 023 00 00 68-71 044 043 042 035 035 029 00 00 72+ 045 044 046 039 038 032 00 00 Financial Aggregates, by Plan Type (In Billions of Dollars) Defined Defined IRA/TSA Benefit Cont. /[KEOGH Total Contributions 33.95 13.23 3.90 Total Benefits Paid 11.71 3.80 0.74 Average Benefit 2015.90 1285.40 1115.30 Fund Balance 191.78 92.59 27.74 Numbers of Covered, Participating, and Vested Workers Covered Participating Vested Defined Defined IRA/TSA Defined Defined IRA/TSA Defined Defined IRA/TSA Benefit Cont. /KEOGH Benefit Cont. /KEOGH Benefit Cont. /IKEOGH 16-17 610979 310497 80128 428207 228571 80128 21187 11309 80128 18-24 3909969 1987033 512782 2742591 1463965 512782 1051302 561174 512782 25-34 7683828 3904895 1007714 8778169 4685689 1007714 4803563 2564088 1007714 35-44 5538493 2814644 726359 6778885 3618495 726359 5371258 2867119 726359 45-54 4532469 2303385 594421 5785523 3088248 594421 5227074 2790154 594421 55-61 2583444 1312897 338811 3349801 1788087 338811 3435863 1834025 338811 62-64 593056 301388 77777 769680 410845 77777 661564 353134 77777 65-67 154063 78294 20204 155017 82745 20204 213732 114085 20204 68-71 132208 67187 17338 132742 70854 17338 120018 64062 17338 72+ 122500 62253 16065 122763 65528 16065 71697 38270 16065 118 Table H-4 Public Pension Projections for 1970 Number of Retirees, By Age & Plan Job Type Federal Military State & Civil Enlisted Military Local SEL State Local Age Service Persons Officers Hazardous General Educators Educators None 35-44 10871 97057 10991 0 0 0 0 0 45-54 54030 201548 94728 1570 9630 3030 5300 0 55-61 112099 132234 80795 13341 82037 20223 62377 18625 62-64 73035 49713 29050 13046 83159 17391 71170 27809 65-67 75144 16289 27054 19053 124966 23873 108990 43348 68-71 116112 5673 12570 20301 140788 23114 117682 46965 72+ 251162 1518 8565 24273 184330 12680 101620 64926 Total 692453 504032 263753 91584 624910 100311 467139 201673 Financial Aggregates, By Job Type (In Billions of Dollars) Federal State & State & Civil Local Local Service Military Hazardous Other Total Contributions 5.77 0.00 0.00 14.05 Percent of Payroll 25.26 0.00 0.00 34.37 Total Benefits Paid 2.55 3.58 0.30 3.21 Percent of Payroll 11.16 16.70 8.20 7.86 Fund Balance 3.25 0.00 0.00 10.94 Covered Workers, By Age and Job (In Thousands) Job Type Federal Military State & Civil Enlisted Military Local SEL State Local Age Service Persons Officers Hazardous General Educators Educators None 16-17 2 51 0 32 195 62 107 386 18-24 139 1519 S0 108 649 206 358 224 25-34 667 745 198 126 757 240 418 137 35-44 565 308 128 119 719 227 396 115 45-54 736 36 23 88 536 169 295 112 55-61 384 0 0 38 232 72 127 79 62-64 89 0 0 12 81 25 44 30 65-67 23 0 0 7 51 16 28 19 68-71 16 0 0 S 41 12 22 16 72+ 0 0 0 0 0 LY 0 65 119 Table H-4 (Cont'd.) Participants, By Age and Job (In Thousands) Job Type Federal Military : State & Civil ~~ Enlisted Military Local SEL State Local Age Service Persons Officers Hazardous General Educators Educators None 16-17 2 51 0 3 24 7 12 0 18-24 139 1519 50 61 371 117 204 0 25-34 667 745 198 100 604 191 333 0 35-44 565 308 128 99 602 190 331 0 45-54 736 36 23 73 450 141 247 0 55-61 384 0 0 27 173 53 94 0 62-64 89 0 0 8 60 18 32 0 65-67 23 0 0 5 37 11 20 0 68-71 16 0 0 3 30 8 16 0 72+ 0 0 0 0 0 0 0 0 Vested Workers, By Age and Job (In Thousands) Job Type Federal Military State & Civil ~~ Enlisted Military Local SEL State Local Age Service Persons Officers Hazardous General Educators Educators None 16-17 0 0 0 0 0 0 0 0 18-24 41 0 0 17 107 33 58 0 25-34 520 0 0 51 311 97 171 0 35-44 518 153 12 62 382 119 209 0 45-54 0698 36 23 50 315 98 172 0 55-61 376 0 0 17 113 34 61 0 62-64 87 0 0 5 39 11 20 0 65-67 21 - 0 0 3 24 7 13 0 68-71 14 0 0 1 18 4 10 0 72+ 0 0 0 0 0 0 0 0 120 Table H-5 Simulation of the Medicare System in 1980 Recipients By Type of Program and Age Males Inpatient Nursing Home—A Physician <35 34887 277 980 68376 35-44 41998 352 1313 81115 45-54 79514 830 3223 149910 53-50 78390 1140 4015 148982 60-64 124638 2516 7685 232625 65-69 730000 14752 36828 1742504 70-74 667479 19962 47103 1504846 73-59 500083 21930 43024 1017333 80-84 316676 19516 35061 587484 Over 85 258504 24656 28188 440475 Total 2832166 105931 207419 5973650 Females Inpatient Nursing Home—A Physician <35 21507 140 798 43799 35-44 25709 251 1293 52525 45-54 55753 776 3421 115143 55-59 61746 1263 4615 133522 60-64 87968 2281 7600 196785 65-69 760697 17285 50107 2340882 70-74 771937 27999 71099 2230521 7359 676750 43125 75570 1695017 80-84 548142 51064 67850 1221334 Over83 511977 66613 63500 1050496 Total 3522184 210797 345852 9080024 Total Inpatient Nursing Home—A Physician <25 56395 417 1778 112176 35-44 67707 603 2606 133641 45-54 135267 1605 6643 265053 55-50 140136 2403 8630 282504 60-64 212605 4797 15285 429410 65-69 1490697 32037 86935 4083386 70-74 1439415 47961 118202 3735367 75-9 1176833 65055 118593 2712350 80-84 864818 70580 102910 1808818 Over 85 770481 91269 91687 1490971 Total 6354350 316728 553271 15053674 Expenditures By Type of Program and Age (In Millions of 1972 Dollars) Males Inpatient Nursing Home—A Physician <35 81292 318 454 22541 35-44 84238 394 608 24825 45-54 150724 775 1342 443062 53-99 149653 1002 1572 44078 60-64 241947 2196 3039 68105 65-69 1290606 11966 14900 451374 70-74 1232838 16031 18673 415588 7579 950729 16643 15787 289480 80-84 602441 15069 12415 168060 Over 85 481187 17104 10216 119877 Total 5265653 81499 79007 1648289 Outpatient Home—B 44056 48652 81148 73435 108362 694189 595385 391762 219316 162400 2418703 Outpatient 26712 29791 62465 67587 92072 917170 843036 623221 436010 377511 3475573 Outpatient 70769 78443 143614 141022 200433 1611358 1438421 1014983 655326 539911 5894276 Outpatient 36756 30913 38338 24566 29599 80847 68258 38737 19025 13410 380450 121 . 830 987 1775 1841 3439 13403 18057 17059 13206 14960 85557 Home—B 708 978 2200 2545 3906 22245 33122 34589 35298 40773 176363 Home—B 1538 1965 3975 4385 7345 35648 51179 51648 48503 55733 261920 Home—B 352 410 665 624 1144 4261 5426 5607 3693 4550 26731 Total 149408 174418 316400 307803 479264 3231674 2852829 1991191 1191258 929183 11623425 Total 936064 110547 239758 271277 390612 4108384 3977713 3148270 2359695 2110868 16810784 Total 243072 284965 556158 579081 869876 7340059 6830543 5139462 3550954 3040050 28434192 Total 141713 141387 236206 221496 346029 1853951 1756811 1316983 820702 646345 7481626 Table H-5 (Cont'd.) Females Inpatient Nursing Home—A Physician Outpatient Home—B Total <35 52479 182 441 16181 25963 288 95534 35-44 55861 284 644 17899 23776 434 98899 45-54 115845 776 1654 37280 36333 899 192786 55-59 127690 1208 2149 39927 25904 939 197817 60-64 177978 2019 3290 53226 26940 1333 264785 65-69 1253830 14467 19710 479657 93413 6655 1867731 70-74 1340780 23532 28063 485254 83698 9984 1971308 75-59 1235127 34745 30041 391504 52957 10485 1754858 80-84 1003443 ~~ 40116 26571 287769 35806 10612 1404317 Over 85 934824 52353 24672 244576 31112 12980 1300513 Total 6297855 169682 137235 2053273 435902 54609 9148555 Total Inpatient Nursing Home—A Physician Outpatient Home—B Total <25 133771 500 896 38722 62719 639 237248 35-44 140099 677 1252 42724 54690 844 240286 45-54 266568 1551 2996 81642 74671 1564 428992 55-50 277344 2210 3721 84005. 50470 1563 419313 60-64 419924 4214 6329 121331 56539 2477 610815 65-69 2544436 26433 34611 931031 174260 10916 3721684 70-74 2573618 39563 46737 900842 151956 15410 3728122 75:30 2185856 51389 45828 680985 91695 16092 3071842 80-84 1605884 55186 38986 455829 54831 14305 2225016 Over 85 1416010 69457 34887 364453 44522 17530 1946858 Total 11563508 251181 216242 3701562 816352 81340 16630183 122 Table H-6 GNPQ Simulated Base % Di ji ; 1970 10869580 ( 08) 10753000 ( —03 We ms 328100 C4 13) doraree 1 ® 0 nl BI a i Nr ln: ID pie num 1972 11955520 ( 9.7) 11711001 ( 5.7) 209 2027 35112693 ( 123) fea ( iD -15.18 1974 12827148 ( 3.0) 12178000 ( —14) 533 2029 36049126 ( 13) 4302.4570 ¢ i By 1975 10883665 ( 152) 12023000 ( —13) —948 2030 36523799 ( 13) 4376.8945 bod) ole 1976 12835645 (17.9) 12730000 ( 59) 083 031 37050102 ( 13) 4699090 ( 21) 1977 12947771 (09) 13405000 ( 53) —341 2032 37589204 (14) 45608594 ( 21 ~1701 178 1854051 { 33) 13993000 ( 41) 355 035 Jmoosel ( 14) dos9e0d4 ( a» ied 1070 13903009 { 93) 10830 ( 21 4B 0m pease ( 13 dosed « ob le 1900 1325.09 { —114) 14093900 { ~1%) -601 203 Paes ( 13 ears ( a1) los 199) 16052301 ( 61 14504961 ( 32) 339 2036 1701030 ( 13) 9923987 ( 31) —loas 1982 14499187 ( 33) 15111970 ( 40) —405 2037 40248535 ( 14) 50553555 ( 213 -19.83 1983 15010625 ( 35) 15531970 ( 28) —336 2038 40753267 ( 13) 5160.5078 $30 Tre 1984 1564.6805 ( 42) 1587.6970 ( 22) —145 2039 41295078 ( 13) 5267.8477 { 30 1955 10355680 { 43) 1647901 ( 33-073 0d 41a (1m sds ( oh) A 1986 16874583 (32) 17168950 ( 42) —171 2041 42328555 ( 12) 54822812 ¢ 2) 108 17352041 (27) 17803960 ( 37) 265 2042 4811893 ( 11) 81836 ( 19) 2340 1988 17839907 ( 29) 18191960 ( 22) —194 2043 43310664 ( 12) 5698.1758 LA 3 1089 18425898 ( 33) 18562971 ( 20) —074 2044 43796406 ( 1.1) 5809.253 £4 By 1990 1907.8025 ( 35) 19085559 ( 28) —004 2045 44256758 ( 1.1) boi ln 29 1991 19543274 (24) 19669570 ( 3.1) —0.64 2046 44727500 ( 1.1) 6038.0273 ¢ 20 Bz 1992 19909558 ( 19) 20042549 ( 19) —066 2047 45197578 ( 11) 61557656 {3% a 1993 20350518 (22) 20468550 ( 21) —058 2048 4569.9805 ( 1.1) 62758047 (10) TE oS Ts. AIH) iE et WMT I) Cos 1995 21367378 ( 24) 21654500 ( 30) —133 2050 46731914 ( 11) 65229414 dE ae 1996 21868425 ( 23) 22246499 ( 27) —170 2051 47230078 ( 11) 6650.1211 TE 1997 22424834 (2.5) 22859050 ( 28) 190 2052 47766328 ( 11) 67998437 (19) -2898 1908 22910430 ( 22) 23438000 ( 26) —233 203 455194 ( 12) 6912058 ( 19) —3000 1999 23435601 ( 23) 24086970 (27) -270 2054 28860703 (1.1) 70468398 19) 23066 : 2 -5 : : 2001 24436587 ( 22) 25283089 { 2 323 2055 49418281 (11) 71842578 ( 19) -31.21 2002 24869712 (18) 25862991 ( 23) —384 Mean: 29255950 3477.2549 ~~ —15.86 2003 25348445 (19) 26455229 ( 23) —4.18 2004 25830055 ( 19) 27061040 ( 23) —455 Oromth Rates: 0.03 226 2005 26318711 (19) 27680730 ( 23) —492 1970-1975 0.03 3.20 2006 2680.4187 ( 18) 28314009 ( 23) —533 Ie 433 3.20 2007 27281248 (18) 28962410 ( 23) —580 1985-1900 3.13 2.98 2008 27714910 (1.6) 29625640 ( 23) —645 1990-1995 229 2.56 2009 28122124 (15) 30303411 ( 23) —7.20 1995-2000 2.28 2.68 2010 28542114 ( 15) 30997339 ( 23) —7.92 20002003 193 229 2011 2890.7458 ( 13) 31539751 ( 17) —835 2005-2010 1.64 2.29 2012 2927.3406 ( 13) 32091699 ( 17) —878 2010-2015 1.26 1.75 2013 29644814 ( 13) 32653201 ( 17) —9.21 2015-2020 117 175 2014 30029780 ( 13) 33224719 ( 17) —962 2020-2025 123 173 2015 30379175 (12) 33806111 ( 17) —10.14 2025-2030 131 173 2016 30730078 (12) 34397700 ( 17) —1066 20a0-2033 1.42 2.08 2017 31087546 ( 12) 34999641 ( 17) —1118 2035204 1.30 208 2018 31455913 (12) 35612120 ( 17) 1167 000-2043 1.14 1.95 2019 31820264 ( 12) 36235320 ( 17) —1218 205.200 1.09 195 2000 32195422 (12) 36869419 ( 17) —1268 s930a0ee 1.12 1.95 2021 32583457 (12) Fsomdl (17) 1 Irons: 1.80 226 6709 (13) 38156101 ( 1.7) =1350 Root Mean Squared Difference 840.14 2023 3339.6648 ( 12) 3881.6189 17) -13. SO 9% Aco] Met 347 2024 3380.0227 ( 3 3048.7690 : I “13% RNsD Baca Veas 24.1612 123 Table H-7 Average Hourly Wage, In 1972 Dollars 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 2055 1950 3.10 5.41 5.61 7.15 7.50 8.08 8.73 10.46 11.20 2.75 8.49 9.55 10.74 12.07 13.28 14.57 16.27 0.0 1960 3.41 5.83 6.12 8.01 8.65 9.58 10.51 12.08 1325 0.55 10.74 12.07 13.28 14.57 16.27 18.29 Male For Persons Born in Year: 1970 3.78 6.53 7.06 9.49 10.41 11.54 12.63 14.90 16.76 12.07 13.28 14.57 16.27 18.29 1980 4.42 7.74 8.49 11.43 12.51 13.73 14.94 18.84 20.74 14.57 16.27 18.29 124 1990 5.26 9.33 10.21 13.60 14.80 16.28 17.88 22.75 25.41 18.29 2000 6.31 11.10 12.07 16.13 17.71 19.68 21.67 28.56 2010 7.55 13.16 14.45 19.50 21.47 23.87 2020 9.10 159 17.52 23.65 2030 11.05 19.29 Appendix 1 Summary Tables for the Base Case Simulation The tables of this appendix show the results of the Base Case simulation described in Chapter 10. The many as- sumptions and parameters used in the preparation of this Base Case are discussed in the description of the Macroe- conomic-Demographic Model presented in Volume One. The tables are presented in the following order: MDM Model Table Numbers Population I-1 to I-2 Labor Market I-3 to I-14 Macroeconomic Growth I-15 to I-16 Social Security I-17 to I-19 Private Pension 1-20 to 1-28 Public Employer Pension 1-29 to 1-32 Supplemental Security Income 1-33 to 1-34 Macro and Actuarial 1-35 to 1-36 At present, no summary reports are generated for the Medicare Model. 125 Table I-1 US Population by Age and Sex (in thousands) Male Female All 16-24 25-54 55-64 65+ Total 16-24 25-54 55-64 65+ Total 16-24 25-54 55-64 65+ Total 1970 16389. 35088. 8827. 8407. 68711. 16066. 36656. 9838. 11681. 74241. 32455. 71744. 18665. 20088. 204881. 1975 18203. 37997. 9344. 9184. 74728. 17842. 39530. 10432. 13237. 81041. 36045. 77527. 19776. 22421. 213640. 1980 19004. 41607. 10039. 10206. 80856. 18608. 43079. 11169. 14941. 87797. 37612. 84686. 21208. 25147. 222483. 1985 17615. 46148. 10345. 11129. 85237. 17291. 47689. 11420. 16410. 92810. 349006. 93837. 21765. 27539. 233259. 1990 15926. 50734. 9975. 12147. 88782. 15613. 52468. 10874. 17991. 96946. 31539.103202. 20849. 30138. 243958. 1995 15222. 53765. 9907. 12839. 91733. 14874. 55664. 10796. 19046.100380. 30096.109429. 20703. 31885. 253341. 2000 16501. 55014. 11215. 13069. 95799. 16093. 57007. 12210. 19456.104766. 32594.112021. 23425. 32525. 261299. 2005 18162. 54805. 13719. 13415. 100101. 17708. 56796. 14835. 19998.109337. 35870.111601. 28554. 33413. 269040. 2010 18213. 55000. 15993. 14642. 103848. 17755. 56930. 17221. 21506.113412. 35968.111930. 33214. 36148. 277361. 2015 17527. 54710. 17593. 16945. 106775. 17084. 56540. 18910. 24309.116843. 34611.111250. 36503. 41254. 285752. 2020 17233. 54309. 18131. 19625. 109298. 16787. 56055. 19508. 27711.120061. 34020.110364. 37639. 47336. 293305. 2025 17903. 54719. 16881. 22390. 111893. 17428. 56425. 18201. 31414.123468. 35331.111144. 35082. 53804. 299751. 2030 18802. 55877. 15350. 24372. 114401. 18295. 57584. 16532. 34344.126755. 37097.113461. 31882. 58716. 305464. 2035 19097. 57516. 15006. 24849. 116468. 18577. 59230. 16108. 35524.129439. 37674.116746. 31114. 60373. 310866. 2040 18923. 58205. 16359. 24638. 118125. 18403. 59891. 17481. 35727.131502. 37326.118096. 33840. 60365. 316021. 2045 18904. 58536. 17797. 24537. 119774. _ 18376. 60196. 18980. 35813.133365. 37280.118732. 36777. 60350. 320913. 2050 19322. 59022. 17802. 25631. 121777. 18774. 60674. 18971. 37074.135493. 38096.119696. 36773. 62705. 325787. 2055 19885. 60024. 17239. 27025. 124173. 19312. 61664. 18360. 38683.138019. 39197.121688. 35599. 65708. 331155. Table I-2 Dependency Ratios Elderly Youth Employment 1970 0.236 0.691 0.649 1975 0.241 0.612 0.621 1980 0.245 0.527 0.599 1985 0.253 0.493 0.582 1990 0.260 0.487 0.559 1995 0.261 0.494 0.537 2000 0.256 0.483 0.513 2005 0.259 0.458 0.508 2010 0.280 0.447 0.530 2015 0.317 0.455 0.564 2020 0.365 0.472 0.600 2025 0.409 0.483 0.630 2030 0.426 0.479 0.648 2035 0.425 0.472 0.651 : 2040 0.412 0.468 0.655 2045 0.416 0.472 0.670 2050 0.431 0.477 0.689 2055 0.442 0.476 0.708 126 Table I-3 Total Labor Force by Sex (Thousands) Male Female 16-24 25-34 35-44 45-54 55-64 65+ Total 16-24 25-34 35-44 45-54 55-64 65+ Total 1970 12063. 12093. 10958. 10594. 7536. 2434. 55679. 7901. 5741. 5902. 6343. 3971. 1146. 31004 1975 13848. 14614. 10681. 10514. 7193. 2117. 58967. 9968. 9111. 6690. 6459. 3877. 1023. 37128 1980 14254. 17067. 11994. 10032. 7345. 2027. 62720. 11463. 11357. 8013. 6345. 4459. 1140. 42776 1985 13318. 18847. 14679. 9949. 7654. 2230. 66677. 10973. 12574. 9838. 6551. 4661. 1292. 45889 1990 12052. 19380. 17077. 11136. 7145. 2180. 68970. 10448. 13888. 11933. 7726. 4703. 1427. 50125. 1995 11398. 17922. 18669. 13453. 6767. 1980. 70190. 10565. 14218. 13874. 9895. 4934. 1354. 54839. 2000 12327. 16098. 19118. 15467. 7222. 1856. 72088. 12038. 14689. 15412. 11635. 5559. 1108. 60441. 2005 13706. 15549. 17653. 16750. 8246. 1940. 73843. 13379. 15726. 15503. 13058. 6430. 933. 65028. 2010 13792. 16793. 15875. 17060. 9076. 2219. 74815. 13458. 16938. 14968. 13823. 7097. 941. 67225. 2015 13215. 18207. 15348. 15728. 9615. 2644. 74757. 12893. 18335. 15288. 13166. 7487. 1050. 68219 2020 12914. 18118. 16574. 14148. 9532. 3058. 74343. 12592. 18235. 17235. 2199. 7591. 1185. 69037. 2025 13397 17429. 17970. 13664. 8594. 3419. 74473. 13055. 17535. 18671. 12060. 7290. 1314. 69926 2030 14089. 17199. 17885. 14717. 7682. 3582. 75154. 13722. 17299. 18574. 13401. 6803. 1365. 71164 2035 14273. 17832. 17195. 15873. 7356. 3530. 76059. 13897. 17921. 17866. 15235. 6613. 1321. 72853. 2040 14039. 18569. 16937. 15708. 7793. 3498. 76544. 13664. 18645. 17603. 16035. 7065. 1277. 74289. 2045 13915. 18709. 17515. 15003. 8078. 3583. 76803. 13537. 18772. 18192. 15686. 7339. 1292. 74817 2050 14164. 18446. 18200. 14675. 7859. 3866. 77212. 13774. 18498. 18887. 15341. 7237. 1399. 75136 2055 14554. 18373. 18321. 15075. 7444. 4092. 77859. 14145. 18413. 18998. 15738. 6857. 1457. 75608 Table 1-4 Total Labor Force by Age (Millions) } Total 16-24 25-34 35-44 45-54 55-64 65+ TOTAL 1970 19964 17834 16860 16937. 11508. 3581. 86683 1975 23816. 23725. 17370. 16974. 11070. 3140. 96094. 1980 25717. 28424. 20007. 16377. 11804. 3167. 105497 1985 24291. 31421. 24517. 16500. 12315. 3521. 112566. 1990 22500. 33268. 29010. 18862. 11848. 3607. 119095 1995 21964. 32139. 32543. 23348. 11701. 3334. 125029. 2000 243065. 30787. 34530. 27102. 12782. 2964. 132529 2005 27084. 31274. 33156. 29808. 14675. 2873. 138871. 2010 27250. 33731. 30844. 30883. 16173. 3160. 142040. 2015 26108. 36543. 30636. 28894. 17102. 3694. 142976 2020 25506. 36353. 33809. 26346. 17123. 4243. 143380. 2025 26452. 34964. 36641. 25725. 15884. 4733. 144398. 2030 27811. 34498. 36459. 28118. 14485. 4947. 146318 2035 28170. 35753. 35061. 31107. 13969. 4851. 148912 2040 .27703. 37214, - 34540. 31743. 14858. 4775. 150833. 2045 27452. 37481. 35707. 30688. 15417. 4875. 151620 2050 27938. 36944. 37087. 30017. 15097. 5265. 152347 2055 36786. "37319. 30813. 14301. 5549. 153468 28700. 127 Table I-5 Male Labor Force Participation Rates by Age 16-17 18-24 25-34 35-44 45-54 55-57 5861 062-64 065-67 06871 72+ 1970 47.6 81.9 96.6 96.8 94.2 91.5 86.0 74.4 51.1 355 15.1 1975 47.1 85.0 95.2 95.8 91.5 86.6 78.3 59.9 37.1 27.1 13.7 1980 49.3 82.2 94.9 953 90.9 83.7 74.7 53.6 31.0 23.7 123 1985 52.8 81.4 95.1 95.4 91.0 84.8 76.2 57 33.0 25.4 112 1990 54.9 81.0 94.8 95.1 90.3 83.6 74.3 52.1 29.0 23.0 10.3 1995 55.9 80.5 94.4 94.7 89.3 80.8 70.8 46.4 25.0 19.1 9.8 2000 56.5 80.6 93.9 94.1 88.1 76.8 65.9 39.5 25.0 15.3 99 2005 58.4 80.5 93.3 93.6 86.8 72.7 60.7 35.0 25.0 15.0 10.4 2010 60.0 80.1 92.9 93.2 85.8 69.9 56.3 35.0 25.0 15.0 10.8 2015 60.0 79.6 92.5 92.8 85.0 67.6 53.2 35.0 25.0 15.0 11.2 2020 60.0 79.3 92.2 92.5 84.5 606.1 50.9 35.0 25.0 15.0 11.4 2025 60.0 79.3 92.0 92.3 839 . 649 49.1 35.0 25.0 15.0 11.1 2030 60.0 79.3 91.7 92.0 833 63.6 47.2 35.0 25.0 15.0 10.8 2035 60.0 78.9 91.4 91.7 82.5 613 45.0 35.0 25.0 15.0 10.7 2040 60.0 78.2 91.0 91.3 81.7 58.6 42.0 35.0 25.0 15.0 11.1 2045 60.0 77.5 90.5 90.8 80.7 56.1 40.0 35.0 25.0 15.0 11.5 2050 60.0 77.2 90.1 © 90.4 79.8 53.8 40.0 35.0 25.0 15.0 11.8 2055 60.0 77.0 89.6 89.9 78.8 51.5 40.0 35.0 25.0 15.0 12.0 Table I-6 Female Labor Force Participation Rates by Age 16-17 18-24 25-34 35-44 45-54 55-57 5861 062-64 065-67 6871 72+ 1970 31.0 54.8 44.9 49.9 52.6 47.9 38.8 30.1 18.9 13.0 4.9 1975 34.2 62.4 58.5 57.3 52.6 46.2 35.1 25.4 14.9 9.9 4.0 1980 42.4 66.9 62.4 60.7 54.3 48.1 39.7 27.0 155 10.0 3.8 1985 46.1 67.8 62.7 61.3 56.6 49.2 40.8 29.3 16.6 102 4.0 1990 51.0 70.9 67.1 63.8 59.1 51.4 45.0 30.8 17.1 10.6 3.9 1995 55.9 75.4 73.8 67.6 623 52.8 49.6 31.1. 16.6 9.3 3.6 2000 56.5 80.6 84.4 72.9 63.0 52.1 50.2 28.8 14.5 6.8 3.1 2005 58.4 80.5 933 787 64.5 50.2 48.0 25.7 12.1 5.0 2.6 2010 60.0 80.1 92.9 84.1 66.2 49.1 459 233 10.4 5.0 23 2015 60.0 79.6 92.5 88.6 67.6 48.0 44.1 21.4 10.0 5.0 2.0 2020 © 60.0 79.3 92.2 92.5 69.2 48.1 44.3 20.6 10.0 5.0 2.0 2025 60.0 79.3 92.0 92.3 70.5 49.8 47.8 20.9 10.0 5.0 2.0 2030 60.0 79.3 91.7 92.0 72.3 51.9 47.2 21.2 10.0 5.0 2.0 2035 60.0 78.9 91.4 91.7 75.7 52.8 450 , 209 10.0 5.0 2.0 2040 60.0 78.2 910 + 913 79.7 52.1 420 19.1 10.0 5.0 2.0 2045 60.0 77.5 90.5 - 90.8 80.7 52.0 40.0 17.8 10.0 5.0 2.0 2050 60.0 77.2 90.1 90.4. 79.8 52.6 400 - 16.7 10.0 5.0 2.0 2055 60.0 77.0 89.6 89.9 78.8 51.5 40.0 15.9 10.0 5.0 2.0 128 Table I-7 Total Private Non-Agricultural Employment by Age and Sex (Millions) Male Female 16-24 25-34 35-44 45-54 55-64 65+ Toul 16-24 25-34 35-44 45-54 55-64 65+ Total 1970 10,7 117-107 103 7.4 2.4 53.2 6.9 5.4 5.6 6.1 3.9 11 29.0 1975 117 °:138 102 100 7.0 20 54.8 85 83 6.2 6.1 38 1.0 33.8 1980 12.1 16.1 1S 9.7 7d 19 58.4 9.7 104 25 6.0 4.3 11 39.0 1985 11.8 181 14.2 9.7 75 21 63.5 96 ‘11.7 9.4 6.3 4.5 1.2 42.7 1990 108 186 165 ..109 7.0 21 65.9 93 129 114 7.4 4.5 1.4 47.0 1995 103 172 "181 13.1 6.6 19 67.1 94 133 13.3 9.5 4.7 13 51.6 2000 111 154 185 15.0 7.0 1.8 68.8 108 “138: 148 11.2 5.3 11 57.0 2005 124 149 17.1 16.2 79 1.8 70.4 120 "148 - 149 127 6.1 09 61.4 2010 125.161 154 165 8.7 21 713 12,2 16.0 144 134 6.8 0.9 63.6 2015 120 178 149 133 9.2 23 71.4 117° 173 © 148 128 71 1.0 64.7 2020 11.8 174 16.1 13.8 9.1 29 71.0 115.0472 «167 118 72 11 65.6 2025 12.2 16.7. 174 133 82 33 71.4 11.9 16.6 18.1 11.7 6.9 13 66.6 2030 128-165 173 143 73 3.4 71.8 126 164 180 13.0 6.5 15 67.8 2035 130. 171 167 -154 741 3.4 72.7 128.170 + 174.148 6.3 13 69.6 2040 128 179. 164 153 7.4 3.4 73.2 126. 178. 171 15.6 6.7 13 71.1 2045 128 180 170 146 77 3.4 73.5 126.179 . 17.7: 153 70 12 71.8 2050 130° 177 16 143 75 3.7 73.9 128 17.7 184 150 6.9 13 72.2 2055 134 17.7 178 147 7.1 39 74.5 13.2 177 186 154 6.5 14 728 Table I-8 Total Private Non-Agricultural Employment by Age (Millions) 16-24 25-34 35-44 45-54 55-64 65+ Total 1970 17.7 17.1 16.3 16.4 113 25 82.3 1975 202 22.1 16.4 16.1 10.8 3.0 88.6 1980 21.8 26.4 19.0 15.7 11.4 3.0 97.4 1985 215 29.8 23.6 16.0 12.0 3.4 106.2 1990 20.1 31.6 28.0 18.3 11.5 35 1129 +1995 19.7 30.5 31.4 226 11.3 3.2 118.7 2000 219 29.2 33.4 26:2 123 2.8 125.8 2005 244 29.7 22.1 28.9 14.0 27 131.8 2010 24.6 321 29.9 299 -. 15.4 3.0 134.9 2015 23.7 _ 348 29.7 28.1 16.3 35 136.1 2020 233 34.6 32.8 25.6 16.3 4.1 136.7 - 2025 24.1 333 355 250 15.1 4.5 137.7! 2030 254 329 35.4 274 13.8 48 139.6 2035 25.8 34.2 34.0 30.3 i tL) 4.7 LE A42.2. 2040 25.5 35.6 33.6 30.9. 14.2 4.6 © 144.3 2045 25.3 “- 35.9 34.7 29.9 14.7 47 145.3 2050 - 25.8 35.5 36.1 29.3 14.4 51 146.1 : 5.4 147.3 2055 26.6 35% 363 301 136 129 Table I-9 Unemployment Rates by Age and Sex Male Female 16-24 25-34 35-44 45-54 55-64 G5+ Total 16-24 25-34 35-44 45-54 55-64 65+ Total 1970 0.110 0.033 0.024 0.024 0.019 0.028 0.044 0.122 0.061 0.049 0.038 0.023 0.026 0.064 1975 0.154 0.058 0.045 0.045 0.021 0.046 0.071 0.152 0.088 0.068 0.056 0.032 0.045 0.089 1980 0.151 0.058 0.042 0.036 0.029 0.050 0.069 0.151 0.087 0.061 0.056 0.041 0.044 0.089 1985 0.111 0.039 0031 0.023 0.023 0.042 0.048 0.123 0.069 0.046 0.043 0.035 0.039 0.069 1990 0.102 0.039 0.031 0.025 0.023 0.046 0.045 0.113 0.068 0.042 0.040 0.041 0.032 0.063 1995 0.098 0.040 0.032 0.028 0.021 0.051 0.043 0.107 0.064 0.039 0.036 0.048 0.039 0.059 2000 0.099 0.042 0.031 0.030 0.033 0.052 0.046 0.106 0.061 0.037 0.033 0.049 0.050 0.057 2005 0.099 0.042 0.031 0.031 0.038 0.051 0.047 0.103 0.059 0.036 0.031 0.049 0.050 0.056 2010 0.095 0.041 0.030 0.031 0.046 0.047 0.047 0.097 0.057 0.035 0.030 0.049 0.051 0.054 2015 0.091 0.040 0.029 0.029 0.046 0.043 0.045 0.091 0.056 0.034 0.029 0.049 0.052 0.052 2020 0.090 0.040 0.030 0.026 0.046 0.041 0.045 0.087 0.055 0.032 0.029 0.048 0.047 0.049 2025 0.091 0.040 0.030 0.026 0.046 0.038 0.045 0.086 0.053 0.030 0.029 0.048 0.041 0.048 2030 0.091 0.040 0.030 0.027 0.046 0.038 0.045 0.085 0.051 0.029 0.027 0.049 0.041 0.047 2035 0.088 0.039 0.030 0.029 0.042 0.039 0.044 0.081 0.049 0.029 0.025 0.049 0.044 0.045 2040 0.085 0.039 0.030 0.028 0.044 0.040 0.044 0.076 0.047 0.027 0.024 0.049 0.047 0.043 2045 0.083 0.038 0.030 0.027 0.046 0.038 0.043 0.071 0.045 0.026 0.023 0.049 0.042 0.041 2050 0.082 0.038 0.030 0.026 0.046 0.036 0.043 0.068 0.043 0.025 0.023 0.049 0.037 0.039 2055 0.081 0.038 0.030 0.027 0.046 0.036 0.043 0.066 0.041 0.023 0.022 0.049 0.035 0.038 Table I-10 Unemployment Rates by Age 16-24 25-34 35-44 45-54 55-64 65+ Total 1970 0.115 0.042 0.033 0.030 0.020 0.027 0.051 1975 0.153 0.070 0.054 0.049 0.024 0.045 0.078 1980 0.151 0.070 0.050 0.043 0.034 0.048 0.077 1985 0.116 0.051 0.037 0.031 0.028 0.041 0.056 1990 0.107 0.051 0.036 0.031 0.030 0.040 0.052 1995 0.102 0.051 0.035 0.031 0.033 0.046 0.050 2000 0.103 0.051 0.034 0.031 0.040 0.051 0.051 2005 0.101 0.050 0.033 0.031 0.043 0.051 0.051 2010 0.096 0.049 0.032 0.031 0.047 0.048 0.050 2015 0.091 0.048 0.031 0.029 0.047 0.046 0.048 2020 0.088 0.047 0.031 0.028 0.047 0.042 0.047 2025 0.089 0.046 0.030 0.027 0.047 0.039 0.046 2030 0.088 0.045 0.030 0.027 0.047 0.039 0.046 2035 0.085 0.044 0.029 0.027 0.045 0.040 0.045 2040 0.081 0.043 0.029 0.026 0.046 0.042 0.043 2045 0.077 0.042 0.028 0.025 0.047 0.039 0.042 2050 0.075 0.040 0.027 0.024 0.047 0.036 0.041 2055 0.073 0.039 0.027 0.024 0.047 0.036 0.040 130 Table I-11 Average Hourly Wage By Age and Sex (1972 Dollars) . Male . Female 16-24 25-34 35-44 45-54 55-64 65+ Total 16-24 25-34 35-44 45-54 55-64 65+ Total 1970 2.58 40... 5765 307 329 484 224 295 295.298 30% - 279 . 184 2.78 1975 2.71 S00. 619 638 557.39 522 239 325.320. 323. :297. 49] 3.00 1980 328.607 768. 787 709 466 6.44 2.86 397 394 402°. 362 24) 3.66 1985 2.35 627. 789. 812 741. 487 6,72. .293 410 406 414 378. 252 3.81 1990 3.56 672° 845 870: 854 56] 738 3.10 439 436 444 435 290 4.14 1995 3.83 736: 926" 953 992 693 8.28 3.34 48): 477 486. 506 337 439 2000 4.10 8.00 1006. 1036 1077 708 9.06 3.59 533: 519. S529 990: 366 499 2005 4.51 378 11.05.1337 11.12 + 73] 9.85 3.96 874-2569. S80 867 "378 . 544 2010 4.93 9.85 -:12.0Y 1236. 11.55 7.47 10.57 4.33 624. 61% 631.579 386 5.88 2015 31..:1022: 128% 1323-1151... 757 11.18 4.68 668: 662: 675. 887 391 6.27 2020 $61 1082 1362 1401 1204 792 11.81 4.94 707. 7202. .7215 614.489 6.64 2025 884 1130 1433 14.75: 1339 --881 12.53 5.14 7.45 739... 7.55. .68%: 435 7.02 2030 612 “1206 1518-1562. 1520 "1006 13.40 5.39 785. 782. 397 80: 320: 749 2035 658 13.04 1641 1689 17.17 11.29 14.58 5.80 855. ..846,. 862 876 584 :.815 2040 7.24. 1431 1800 1853. 1821 1198 1593 6.38 935 .928 946. 929.619 - 893 2045 798 15.77 19.84 2043 19.54 12.85 1781 703:2103) 1023 1043 997 664 983 2050 870 1728 21.73 2237: 21.56 14.18 1920 766 11.29 1120 1142: 1100 733 1077 2055 9.45 1889 23.76 24.46 2438 16.04 21.10 -832. 1235 1225 1249 1244 829 1181 Table I-12 Unemployment Rates by Age 16-24 25-34 35-44 45-54 55-64 65+ Total 1970 2.463 4.280 4.978 4.923 4.407 2.883 4.248 1975 2.588 4.473 5.278 5.408 4.801 3.369 4.521 1980 3.106 5.402 6.451 6.645 5.969 3932 5.509 1985 3.176 5.572 6.649 6.822 6.240 4.097 3.735 1990 3.361 5.925 7.080 7.256 7.118 4.634 6.249 1995 3.609 6.419 7.685 7.881 8.171 5352 6.920 2000 3.861 6.865 8.250 8.534 8.784 5.917 7.474 2005 4.244 7.438 8.938 9.318 9.038 6.266 8.053 2010 4.638 8.070 9.608 10.075 9.214 6.492 8.622 2015 4.995 8.622 10.188 10.735 9.342 6.626 9.103 2020 2.279 9.119 10.711 11.328 9.736 6.942 9.587 2025 5.495 9.584 11.266 11.884 10.726 7.722 10.138 2030 5.759 10.133 11.922 12.522 12.172 8.825 10.816 2035 6.197 10.937 12.879 13.429 13.640 9.922 11.733 2040 6.810 11.969 14.108 14.589 14.438 10.546 12.791 2045 7.505 13.165 15.536 16.028 15.472 11.316 14.040 2050 8.180 14.381 16.994 17.555 17.034 12.471 13.371 2055 8.887 15.682 18.558 19.193 19.252 14.112 16.857 131 Table I-13 Labor Input by Age and Sex (in thousands) Male Female All 16-24 25-54 55-64 65+ Total 16-24 25-54 55-64 65+ Total 16-24 25-54 55-64 65+ Total 1970 13.915 62.549 13.686 3.432 93581 7.268 233505 5.643 1341 37.757 21.183. 86.053 19.329 4.773 131.338 1978 141363 63.173 12.775 2760 930069 8822 27865 3378 1.122 43187 - 23.185 91.038 18.151 13.882 136.256 1980 14.309 67.952 12.776 2.541 97.578 10.068 31.961 6.069 1.216 49.314 24.377 99.913 18.845 3.757 146.892 1985 13.907 76.549 13.372 2.799 106.626 9.921 36.539 6.371 1.378 54.209 23.828 113.088 19.742 4.177 160.835 1990 12.409 83.154 12.393 2.670 110.627 9.471 42.403 6.358 1.509 59.741 21.880 125.557 18.751 4.179 170.367 1995 11.370 86.653 11.654 2.345 112.022 9.490 48.256 6.567 1.391 65.704 20.860 134.909 18.220 3.736 177.726 2000 11.812 86.774 12.143 2.123 112.852 10.698 53.082 7.334 1.096 72.210 22.510 139.856 19.476 3.219 185.061 2005 12.745 84.346 13.613 2.143 112.847 11.855 56.240 8.404 0.899 77.397 24.600 140.585 22.017 3.042 190.243 2010 12.507 82.718 14.707 2.389 112.320 11.921 57.930 9.204 0.884 79.939 24.428 140.648 23.911 3.273 192.259 2015 11.751 80.853 15.446 2.788 110.838 11.425 59.071 9.648 0968 81.112 23.176 139.924 25.094’ 3.756 191.951 2020 11.257 79.382 15.204 3.168 109.012 11.036 60.109 9.741 1.083 81.970 22.294 139.491 24.946 4.251 190.981 2025 11.456 79.227 13.619 3.488 107.790 11.240 60.863 9.314 1.191 82.608 22.695 140.090 22.934 4.680 190.399 2030 11.847 79.890 12.094 3.588 107.418 11.629 62.153 8.643 1.220 83.644 23.476 142.042 20.737 4.808 191.062 2035 11.780 80.824 11.540 3.449 107.592 11.579 64.333 8.350 1.157 85.418 23.359 145.157 19.889 4.605 193.011 2040 11.323 80.218 12.073 3.321 106.936 11.151 65.836 8.855 1.091 86.934 22.475 146.054 20.928 4.413 193.870 2045 10.922 79.097 12.355 3.304 105.677 10.775 66.180 9.127 1.084 87.165 21.696 145.277 21.482 4.388 192.842 2050 10.807 78.198 11.885 3.461 104.351 10.677 66.156 8929 1.153 86.916 21.485 144.354 20.814 4.614 191.267 2055 10.801 77.864 11.131 3.549 103.345 10.682 66.566 8392 1.175 86.815 21.483 144.431 19.523 4.723 190.160 Table I-14 ’ Average Annual Compensation by Age and Sex (1972 Dollars) Male Female All 16-24 25-54 55-64 65+ Total 16-24 25-54 55-64 65+ Total 16-24 25-54 55-64 65+ Total 1970 3339. 10283. 9390. 4772. 8313. 2352. 4108. 4061. 2212. 3609. 2952.. "8163. 7557. 3051. 6782. 1975 3323. 10741. 10101. 5409. 8876. 2493. 4384. 4254. 2196. 3834. 2975. 8340. 8068. 4361. 6951. 1980 3880. 12874.12697. 6146. 10767. 2086... 5318, 5132. 2689. 4634. 3468. 19921. 9861. 4897. 8312. 1985 3941. 13241. 13255. 6389. 11278. 3016. © 5472. 5357. 2797. .4829 3526. 10178. 10289. 50068. 8684. 1990 4080. 14169. 15159. 7207. 12398. 3171. 5886. 6137. 3160. 3%274 3661. 10773. 11616. 5596. 9434. 1995 4238. 15583. 17453. 8147. 13823. 3362. 6419. 7083. 3606. 5849. 3819. 11662. 13152. 6289. 10358. 2000 4364. 16927.18717. 8545. 14867. 3575. 6962. . 7620. 3814. © 6325, 3976. 12453. 13936. 6774. 10998. 2005 4651. 18382. 19078. 8510. 15791. 3912. 7618. 7793. 3832. 6856. 4287. 13346. 14165. 6990. 11628. 2010 4938. 19591. 19280. 8431. 16657. 4251. 8255. 7898. 3824. 7390. 4599. 14182. 14294. 7064. 12287. 2015 5192. 20516. 19376. 8338. 17359. 4556. 8799. 7955. 3805. 7861. 4877. 14835. 14385. 7058. 12842. 2020 5370. 21496. 20126. 8550. 18117. 4745. 9290. 8279. 3921. 8290. 5061. 15490. 14880. 7263. 13398." 2025 5491. 22549. 22247. 9340. 18983. 4848. 9758. 9166. 4302. 8713. 5173.1G225. 16249. 7944. 14019. 2030 5660. 23796. 25241. 10468. 20066, 4995. 10322. 10419. 4844. 9240. 5331. 17108. 18289. 8920. 148006. 2035 5962. 25471. 28105. 11480. 21582: 5261. 11146. 11628. 5344. 10004. 5615. 18308. 20336. 9815. 15920. 2040 6381: 27485.29517. 11841. 23271. 5630. 12193. 12246. 5551. 10919. 6009. 19764. 21326. 10168. 17185. 2045 6828. 29847.31319. 12315: 25183. 6022. 13397. 13035. , 5818. 11939. 6428. 21509. 22629. 10599. 18640. 2050 7228. 32298, 34186. 13172. 27115. 6371. 14621. 14273. -6272. 12965. 6802. 23333.. 24654. 11339. 20123. 2055 7631. 34914. 38227. 14429. 29250. 6723. 15933.. 16012; 6927. 7180. 25284. 27591. 12458. 21763. 132 14093. Table I-15 Macroeconomic Inputs (1972 Dollars) Capital Labor Capital Inputs (Q) Inputs (Q) Inputs (V) 1970 409.298 616.647 412.709 1975 519.730 669.045 484.285 1980 600.448 748.483 501.270 1985 661.766 824.455 566.185 1990 760.490 878.568 654.552 1908.1 878.668 922.024 739.988 2000 1026.752 965.852 847.800 2005 1196.205 998.865 950.432 2010 1373.254 1015.516 1043.418 2015 1553.349 1019.981 1124.091 2020 1723.526 1020.931 1197.964 2025 1897.329 1023.935 1276.702 2030 2079.091 1033.678 1366.383 2035 2298.663 1050.496 1483.795 2040 2552.989 1061.517 1609.279 2045 2848.817 1062.235 1741.179 2050 3177.070 1059.894 1877.455 2055 3529.660 1060.092 2025.580 Table I-16 GNP and Its Components (Billions of 1972 Dollars) Government Net GNP Consumption Investment Purchase Exports 1970 1068.603 653.246 166.923 247.034 1.400 1975 1203278 744.134 181.580 254.965 22.600 1980 1423398 975.211 162.571 262.817 22.800 1985 1541.648 1039.697 194.439 292.873 14.640 1990 1688.110 1100.281 253.939 324.490 9.400 1995 1835.094 1156.889 321.672 ~~ 350.253 6.280 2000 1998431 1223.826 ~~ 389.234 ~~ 381.172 4.200 2005 2159.291 1323.312 425.623 408306 2.050 2010 2299.139 1435558 430.650 431.931 1.000 2015 2420.196 1558.533 410.501 450.162 1.000 2020 2526.910 1659.240 397.371 469.299 1.000 2025 2631.446 1737.844 404.083 488.520 1.000 2030 2744.105 1802.305 431528 509.272 1.000 2035 2873.577 1859.358 476.820 536.400 1.000 2040 2995.977 1919.432 514.261 561.284 1.000 2045 3099.783 1965.143 = 549.972 583.669 1.000 2050 3195.197 2008.879 584.072 .601.246 1.000 2055 3290.145 2045.943 623.081 620.121 1.000 Inputs (V) 133 Labor 584.421 662.839 870.895 969.173 1113.447 1280.708 1441.858 1599.061 1727.884 1817.872 1901.499 2001.475 2139.627 2333.284 2542.968 2761.801 2982.049 3231.356 Capital Share 0.414 0.422 0.365 0.369 0.370 0.366 0.370 0.373 0.377 0.382 0.387 0.389 0.390 0.389 0.388 0.387 0.386 0.385 Labor Share 0.586 0.578 0.635 0.631 0.630 0.634 0.630 0.627 0.623 0.618 0.613 0.611 0.610 0.611 0.612 0.613 0.614 0.615 Table I-17 Beneficiaries of the Social Security System by Benefit Type (Millions) Retired Disabled Total New Total Total New Total Total Primary Primary Secondary Total Primary Primary Secondary Total OASDI 1970 13.254 1.302 10.216 23.470 1.490 0.319 1.131 2.621 26.091 1975 15.898 1.474 11.145 27.043 1.986 0.334 1.761 3.747 30.790 1980 18.357 1.520 11.314 29.671 2.257 0.352 1.884 4.141 33.813 1985 20.687 1.706 11.361 32.048 2.428 0.367 1.730 4.158 36.206 1990 22915 1.722 11.359 34.274 2.510 0.379 1.818 4.328 38.602 1995 24.395 1.669 11.396 35.791 2.649 0.402 1.876 4.525 40.317 2000 25.136 1.648 11.345 36.481 2.959 0.448 2.096 5.055 41.535 2005 26.227 1.925 11.250 37.477 3.392 0.509 2.360 5.752 43.229 2010 28.771 2.329 11.362 40.133 3.770 0.559 2.614 6.384 46.518 2015 32.970 2.705 11.608 44.578 3.957 0.583 2.793 6.750 51.328 2020 38.080 3.043 11.890 49.970 4.003 0.586 2.900 6.903 56.873 2025 43.277 3.132 12.089 55.366 3.848 0.562 2.883 6.731 62.097 2030 46.901 2.852 12.068 58.969 3.639 0.537 2.762 6.401 65.370 2035 48.352 2.693 11.917 60.269 3.632 0.539 2.713 6.345 66.614 2040 48.323 2.561 11.624 59.947 3.783 0.567 2.763 6.546 66.493 2045 48.713 2916 11.543 60.256 4.020 0.597 2.871 6.891 67.147 2050 50.683 3.160 11.539 62.222 4.044 0.598 2.928 6.972 69.193 2055 53.296 3.158 11.596 64.892 3.994 0.592 2931 6.925 71.817 Table I-18 Benefits Paid by the Social Security System by Benefit Type (In 1972 Dollars and Billions of 1972 Dollars) OASI DI Total Average Average Total Average Average Total OASDI New Award Benefit Payments New Award Benefit Payments Payments 1970 1468.471 1514.088 29.886 1645.119 1738.841 3.771 33.657 1975 1936.973 1897.968 44.092 2168.684 2167.968 6.596 50.688 1980 2216.326 1949.149 50.498 2339.446 2237925 7.582 58.080 1985 2471.860 2087.965 59.132 2923.348 2596.939 9.002 68.134 1990 2702.495 2257.060 69.001 3191.806 2932.670 10.559 79.560 1995 3018.749 2469.093 79.186 3575.669 3268.841 12.339 91.525 2000 3268.008 2704.863 88.662 3808.064 3576.613 15.080 103.741 2005 3494.299 2949.689 99.696 4055.395 3820.278 18.370 118.065 2010 3726.082 3209.326 116.825 4299.187 4055.998 21.654 138.479 2015 3929.459 3462.665 141.091 4502.078 4265.064 24.028 165.119 2020 4096.391 3671.447 168.998 4699.363 4458.109 25.602 194.601 2025 4286.914 3845.168 197.457 4910.301 46606.719 26.030 223.487 2030 4531.531 4007.134 220.260 5157.730 4892.293 25.909 246.170 2035 4840.926 4176.230 235.249 55006.812 5167.824 27.183 262.432 2040 5181.773 4386.598 246.162 5909.727 5501.168 29.931 276.093 2045 5554.984 4658.441 262.986 6386.711 5893.266 33.845 296.831 2050 5999.910 5012.605 292.785 6864.355 6340.988 36.781 329.566 2055 6474.422 54006.668 330.070 7380.219 6809.781 39.177 369.246 134 Table I-19 Current Actuarial Balance of the OASI and OASDI Systems OASDI DI Current Current Taxes Actuarial Actuarial Taxes Actuarial Actuarial Paid Ratio Balance Paid Ratio Balance 1970 31.69 0.089 —0.005 27.54 0.079 —0.006 1975 42.11 0.119 —0.020 37.22 0.104 -0.016 1980 59.38 0.099 0.002 50.62 0.086 0.000 1985 77.08 0.101 0.013 64.24 0.087 0.008 1990 95.88 0.103 0.021 78.87 0.089 0.013 1995 109.86 0.103 0.021 90.37 0.089 0.013 2000 122.24 0.105 0.019 100.55 0.090 0.012 2005 133.93 0.109 0.015 110.17 0.092 0.010 2010 143.38 0.120 0.004 117.94 0.101 0.001 2015 149.61 0.137 —0.013 123.06 0.117 -0.015 2020 155.23 0.155 —0.031 127.69 0.135 —0.033 2025 162.06 0.171 —0.047 133.30 0.151 —0.049 2030 171.71 0.178 —0.054 141.25 0.159 —-0.057 2035 186.23 0.175 —0.051 153.19 0.157 —0.055 2040 201.86 0.170 —0.046 166.05 0.151 —0.049 2045 21823 0.169 —-0.045 179.51 0.149 —0.047 2050 234.62 0.174 —0.050 192.99 0.155 —0.053 2055 253.25 0.181 -0.057 208.31 0.162 —0.060 Table I-20 Number of Covered Workers in the Private Pension System by Age and Sex (Millions) Male Female 16-24 25-34 35-44 45-54 55-64 65+ Total 16-24 25-34 35-44 45-54 55-64 65+ Total 1970 3.6 5.8 5.6 5.3 3.8 0.5 24.6 2.4 24 23 2.6 1.4 03 113 1975 4.0 6.9 53 5.1 3.6 0.4 25.3 2.9 3.7 2.6 2.6 14 0.2 13.2 1980 4.1 8.0 6.0 4.9 3.6 0.4 27.1 33 4.6 3.1 25 1.6 0.2 153 1985 4.0 9.0 7.4 5.0 3.8 0.5 29.7 33 5.2 38 2.6 1.7 0.3 16.8 1990 3.7 9.3 8.6 5.6 3.6 0.5 31.2 3.2 5.7 4.7 3.1 1.7 0.3 18.6 1995 35 8.6 9.4 6.7 3.4 0.4 32.0 32 5.9 5.5 4.0 1.7 0.3 205 2000 3.8 7.7 9.7 7.7 3.6 0.4 32.8 3.7 6.1 6.1 4.7 1.9 02 227 2005 4.2 7.4 8.9 8.3 4.0 0.4 33.3 4.1 6.5 6.1 5.3 22 0.2 24.5 2010 4.2 8.0 8.0 85 4.4 0.5 33.6 4.1 7.0 5.9 5.6 25 0.2 25.4 2015 4.1 8.7 7.8 7.8 4.7 0.6 33.6 4.0 7.6 6.0 5.4 2.6 0.2 25.9 2020 4.0 8.7 8.4 7.0 4.6 0.6 33.4 39 7.6 6.8 5.0 27 0.3 26.2 2025 4.1 83 9.1 6.8 4.2 0.7 33.3 4.1 7.3 74 49 25 03 26.5 2030 4.3 8.2 91 73 3.7 0.8 33.5 43 7.2 7.4 5.5 2.4 03 27.0 2035 4.4 8.5 8.7 79. 3.6 0.7 33.9 4.4 7.5 7.1 6.2 23 03 27.8. 2040 4.4 89 8.6 7.8 3.8 0.7 34.2 4.3 7.8 7.0 6.6 25 03 28.4 2045 4.3 9.0 89 75 3.9 0.8 34.3 4.3 7.9 7.2 6.4 2.6 0.3 28.7 2050 4.4 8.9 9.2 7.3 3.8 u.8 34.4 4.4 7.8 7.5 6.3 25 03 28.8 2055 4.5 8.8 9.3 7.5 3.6 0.9 34.6 4.5 7.8 7.6 6.5 2.4 0.3 29.0 135 Table 1-21 Number of Covered Workers in the Private Pension System by Age (Millions) 16-24 25-34 3544 45-54 55-64 65+ Total 1970 6.0 8.2 7.9 7.8 5.2 0.8 35.9 1975 6.9 10.5 7.9 7.7 5.0 0.7 38.6 1980 7.4 12.6 9.1 7.5 5.2 0.7 42.4 1985 7.3 14.2 113 7.6 5.5 0.8 46.6 1990 6.8 15.0 13.3 8.7 5.2 0.8 49.8 1995 6.7 14.4 14.9 10.7 5.1 0.7 32.3 2000 7.4 13.8 15.7 12.4 5.5 0.6 55.5 2005 83 13.9 15.1 13.6 6.3 0.6 57.8 2010 8.4 15.1 14.0 14.1 6.9 0.7 59.0 2015 8.1 16.3 13.8 13.2 7.3 0.8 59.5 2020 7.9 163 . 15.2 12.0 73 - 0.9 59.6 2025 82 15.7 16.5 11.7 6.7 1.0 59.8 2030 8.6 15.5 16.4 12.8 6.1 1.1 60.5 2035 8.8 16.0 15.8 14.1 5.9 1.0. 61.7 2040 8.7 16.7 15.6 14.4 6.3 1.0 62.6 2045 8.6 16.9 16.1 13.9 6.5 1.0 63.0 2050 8.8 16.6 16.8 13.6 6.3 1.1 63.3 2055 9.0 16.6 16.9 14.0 6.0 12 63.7 Table I-22 Number of Participants in the Private Pension System by Age and Sex (Millions) Male Female 16-24 25-34 35-44 45-54 55-64 65+ Total 16-24 25-34 35-44 45-54 55-64 65+ Total 1970 2.4 5.8 5.9 5.6 4.0 0.4 241 12 2.0 2.0 25 1.4 0.2 9.3 1975 33 8.4 6.9 6.8 4.8 0.4 30.5 1.8 37 2.8 3.0 1.7 0.2 13.3 1980 3.4 9.8 7.8 6.5 4.8 0.4 32.7 21 4.7 3.4 3.0 19 0.3 15.3 1985 33 11.0. 9.6 6.5 5.0 0.5 36.0 © 20 5.3 4.2 3.1 2.0 0.3 16.9 1990 3.0 113 11.2 7.3 4.7 0.5 380 ~~ 20 5.8 5.1 37 2.0 0.3 18.9 1995 29 10.5 12.2 8.8 45 0.4 393 . 20 6.0 5.9 4.8 2.1 03 21.1 2000 3.1 9.4 1255 10.1 4.7 0.4 40.2 23 6.2 6.6 5.6 2.4 0.3 233 2005 35 9.1 11.6 10.9 5.4 0.4 40.8 2.6 6.7 6.7 6.3 2.7 0.2 25.1 2010 39 9.8 10.4 11.1 5.8 0.5 41.2 2.6 7.2 6.4 6.7 3.0 0.2 26.1 2015 3.4 10.6 10.1 10.3 6.2 0.6 41.1 25 7.8 06 - 64 32 0.2 26.7 2020 33 10.6 10.9 9.3 6.1 0.6 40.8 24 7.8 7.4 5.9 32 0.3 27.1 2025 3.4 10.2 11.8 9.0 5.5 0.7 40.6 25 7.5 8.1 5.8 3.1 0.3 273 2030 3.6 10.1 11.7 9.6 4.9 0.7 40.7 27 7.4 80 - 65 29 0.3 27.8 2035 3:7 10.4 11.3 10.4 4.8 0.7 413 27 . 77 7.7 7.4 2.8 0.3 28.7 2040 3.6 10.9 11.1 10.3 5.0 0.7 41.6 2.7 8.0 7.6 7.8 3.0 0.3 29.4 2045 3.6 11.0 11.5 9.8 5.2 0.7 418 - 27 8.1 7.9 7.6 31 0.3 29.7 2050 3.6 10.8 11.9 9.6 5.1 0.8 41.9 2.7 8.0 8.2 75 3.1 0.3 29.8 2035 38 10.8 12.0 9.9 4.8 0.9 42.1 2.8 8.0 83 7.7 29 0.3 30.0 - 136 Table 1-23 Number of Participants in Employer Pension Systems by Age (Millions) 16-24 25-34 35-44 45-54 1970 3.6 7.7 7.9 8.1 1975 51 12.1 9.7 9.8 1980 5.5 14.5 11.1 9.5 1985 5.4 16.3 13.8 9.7 1990 5.0 17.2 16.3 11.0 1995 4.9 16.5 18.2 13.6 2000 5.4 15.6 19.1 15.7 2005 6.0 15.7 18.2 17.2 2010 6.1 17.0 16.9 17.8 2015 59 18.4 16.7 16.7 2020 57 18.4 183 15.2 2025 6.0 17.7 19.9 14.8 2030 63 17.5 19.8 16.1 2035 6.4 18.1 19.0 17.8 2040 6.3 18.9 18.7 18.1 2045 63 19.0 19.4 17.5 2050 6.4 18.8 20.1 17.1 2055 6.6 18.7 203 17.6 Table I-24 Overall Participation and Coverage Rates by Sex Participation Coverage Male Female Total Male Female Total 1970 0.454 0319 0.406 0.463 0.388 0.437 1975 0557 0392 0494 0463 0392 0.435 1980 0.560 0.391 0.493 0.465 0392 0.436 1985 0567 0397 0.499 0468 0394 0.438 1990 0577 0404 0505 0473 0396 0.441 1995 0.585 0409 0509 0476 0398 0.442 2000 0.585 0.410 0.506 0.476 0.398 0.441 2005 0580 0409 0500 0473 0399 0.439 2010 0578 0411 0.499 0.472 0.399 0.438 2015 0.576 0.412 0.498 0.471 0.400 0.437 2020 0575 0412 0497 0470 0399 0436 2025 0571 0411 0494 0468 - 0399 0.434 2030 0.568 0.410 0.491 0.466 0.398 0.433 2035 0568 0412 0.491 0.466 0399 0434 2040 0.569 0.414 0.492 0.467 0.400 0.434 . 2045 0569 0414 0.492 0.467 0.400 0.434 2050 0567 0413 0.491 0.466 0399 0433 2055 0.564 0412 0489 0465 0399 0432 55-64 5.4 6.4 6.7 71 6.7 6.6 74 8.1 8.9 9.4 9.4 8.6 7.8 7.6 8.0 83 8.1 77 137 65+ 0.6 0.7 07 0.8 0.8 0.7 Total 334 43.8 48.0 53.0 57.0 60.4 63.6 65.9 67.3 67.8 67.9 68.0 68.5 69.9 71.1 71.5 71.7 72.0 Table I-25 Retirees in the Private Pension System by Plan Type (Millions) New Retirees All Retirees DB DEC IRA Total DB DC IRA © Total 1970 0.412 0.209 0.054 0.675 3.091 1.392 0.054 4.736 1975 0.511 0.260 0.067 0.838 4.332 2.215 0.344 6.891 1980 0.623 0317 0.082 1.022 5.901 3.005 0.660 9.565 1985 0.729 0.370 0.096 1.195 7.743 3.936 0.988 12.667 1990 0.783 0.398 0.103 1.283 9.308 4.730 1.220 15.257 1995 0.819 0.416 0.107 1.342 10.407 5.288 1.364 17.059 + 2000 0.927 0.471 0.122 1.520 11.349 5.767 1.487 18.603 2005 1.199 0.609 0.157 1.965 12.990 6.601 1.702 21.293 2010 1.445 0.734 0.190 2.369 15731 7.994 2.062 25.787 2015 1.647 0.837 0.216 2.700 19.098 9.705 2.503 31.307 2020 1.787 0.908 0.234 2.930 22.375 11.370 2.933 36.678 2025 1.749 0.889 0.229 2.867 24.627 12.515 3.229 40.371 2030 1.597 0.812 0.209 2.618 25.146 12.778 3.297 41.220 2035 1.484 0.754 0.195 2.433 24.304 12.330 3.186 39.840 2040 1.513 0.769 0.198 2.480 23.052 11.704 3.019 37.75% 2045 1.704 0.866 0.223 2.793 22.915 11.645 3.004 37.564 2050 1.774 0.902 0.233 2.908 24.042 12.217 3.152 39.411 2055 1:733 0.881 0.227 2.841 23373 12.894 3.326 41.593 Table I-26 Retirees in the Private Pension System by Age and Sex (Millions) Male Female 55-57 58-61 62-64 65-67 68-71 Tdi 55-57 58-61 62-64 65-67 068-71 72+ 1970 0.058 0.300 0.469 0.632 0.700 0.847 0.040 0.243 0.315 0.377 0.333 0.403 1975 0.027 0.369 0.741 0.835 0.791 0.830 0.030 0.402 0.814 0.925 0.641 0.485 1980 0.033 0.445 0.829 1.129 1.110 0.940 0.036 0.496 0938 | 1.352 1.370 0.866 1985 0.036 0.505 1.041 1.370 1.575 1.512 0.038 0.561 1.196 1.679 1.784 1.770 1990 0.037 0.539 1.130 1.586 1.724 1.696 0.039 0.585 1.280 1.943 2.241 2.457 1995. 0.043 0.556 1.174 1.716 1.914 2.120 0.046 0.604 1.311 2.048 2.471 3.057 2000 0.059 0.666 1.304 1.758 2.037 2.415 0.063 0.726 1.469 2.090 2.567 3.449 2005 - 0.077 0.877 1.728 2.066 2.166 2.591 0.082 0.950 1.927 2.468 2.733 3.626 2010 0.086 1.096 2.214 2.725 2.670 2.804 0.091 1.185 2.447 3.228 3.33% 3.886 2015 0.093 1.231 2.473 3.348 3.478 3.489 0.098 1.321 5 3.942 4.319 4.780 2020 0.089 1.296 2.749 3.748 4.119 4.537 0.095 1.390 3.029 , 4.377 5.099 6.148 2025 0.078 1.212 2.778 4.029 4.652 5.431 0.083 1.302 3.061 4.698 5.710 7.339 2030 0.071 1.119 2.406 3.883 4.865 6.170 0.076 1.201 2.657 4.524 5.957 8.290 2035 0.076 1.005 2.336 3.483 4.446 6,533 0.081 1.075 2.568 4.049 5.440 8.749 2040 0.089 1.123 2.236 3.176 4.245 6.126 0.093 1.198 2.449 3.673 5.160 8.188 2045 0.090 1.261 2.628 3.454 3.89 5713 0.095 1.341 2.865 3.976 4.697 7.552 2050 0.087 1.267 2.761 3975 4.46; 5.437 0.092 1.346 3.001 4.550 3.351 7.085 2055 0.084 1.217 2713 4.048 4.913 6.051 0.088 1.291 2.944 4.616 5.863 7.765 138 Table I-27 Private Pension Benefits Average Benefit Total Benefit Replacement Rates (1972 Dollars) (Billions 1972 Dollars) (At Age 65) Defined Defined Defined Defined Defined Defined Benefit Contribution IRA All Benefit Contribution IRA All Benefit Contribution IRA All 1976. -1955.59 1490.46 1226.94 1790.99 6.04 237 0.07 8.48 0.25 0.19 0.16 0.23 1975 2028.28 1367.91 1184.11 1773.92 8.70 3.03 041 12.22 0.25 0.17 0.14 0.22 1980 1981.26 1262.86 1112.60 1695.65 11.69 3.79 673. 1622 0.20 0.13 0.11:0.17 1985 2026.22 1276.08 1094.50 1720.49 15.69 5.02 "1.08 21.79 0.20 0.12 011.017 1990 2262.79 1483.22 1191.59 1935.51 21.06 7.02 145 29.53 0.20 0.13 0.10. 0.17 1995 2632.05 1784.51 1336.50 2265.76 27.39 9.44 1.82 38.65 0.20 0.14 010-017 2000 3091.23 2100.81 1494.95 2656.59 35.08 12.11 222 49.42 0.22 0.15 0.11 0.19 2005 3498.05 2349.35 1621.47 2991.93 45.44 15.51 276 63.71 0.25 0.17 0.12021 2010 3820.17 2523.58 1706.59 3249.23 60.10 20.17 352 83.79 0.26 017 012 022 2015 4056.54 2637.58 1747.33 3432.01 77.47 25.60 437 107.45 0.27 0.18 0.12 023 2020 4272.23 2728.04 1755.00 3592.23 95.59 31.02 5.15 131.76 0.28 0.18 011 023 2025 4511.13 2828.01 1751.94 3768.70 111.10 35.39 5.66 152.15 0.26 0.17 0.10022 2030 4787.87 2946.22 1749.97 3974.01 120.39 37.65 5.77 163.81 0.25 0.15 0.09 0.21 2035 5130.18 3090.48 1753.17 4227.80 124.68 38.17 5.59 168.44 0.24 0.14 0.08 0.20 2040 5532.82 3262.50 1757.68 4527.12 127.43 38.18 531 170.92 0.24 0.14 0.08 0.20 2045 5962.62 3451.03 1751.03 4847.22 136.63 40.19 5.26 182.08 0.24 0.14 0.07 0.20 2050 6380.45 3638.43 1728.35 515838 153.40 44.45 5.45 203.30 0.23 0.13 0.06 0.19 2055 6825.60 3829.99 1699.33 5487.00 173.19 49.38 5.65 228.22 0.22 0.12 0.06 0.18 Table I-28 Contributions and Assets of the Private Pension System (Billions of 1972 Dollars) Total Contributions Fund Balances Defined Defined Defined Defined Benefit Contrib IRA All Benefit Contrib IRA All 1970 6.82 3.65 219 12.66 120.08 40.71 2.10 162.89 1975 14.80 22 2.97 25.56 133.91 55.84 14.32 204.07 1980 33.92 13.25 3.91 51.08 191.62 92.71 27.78 212.11 1985 60.07 15.24 4.47 79.79 407.15 143.87 44.82 595.84 1990 68.86 18.20 5.18 92.24 647.51 200.78 63.74 912.03 1995 80.71 22 309 108.91 922.21 267.16 85.51 1274.89 2000 92.12 25.83 6.66 124.61 1226.93 341.58 109.73 1678.23 2005 102.65 28.55 6.93 138.13 1549.07 418.16 134.25 2101.47 2010 110.69 30.13 7.08 147.91 1860.52 486.44 156.97 2503.93 2015 105.82 30.80 7.14 143.76 2077.04 534.58 176.53 2788.15 2020 110.18 32.21 7.15 149.54 2230.78 561.70 192.55 2985.03 2025 116.57 34.34 7.18 158.09 2330.18 574.28 205.90 3110.36 2030 126.79 37.00 7.26 171.05 2414.57 583.39 218.47 3216.43 2035 141.24 40.43 7.40 189.07 2536.87 603.16 23237 3372.40 2040 155.92 43.91 7.51 207.34 2719.87 639.91 248.57 3608.36 2045 170.28 47.89 7.56 225.74 2952.08 691.98 266.41 3910.47 2050 185.69 52.22 7.59 245.50 3192.18 748.68 284.28 422513 2055 205.05 57.12 7.64 269.81 3431.73 806.12 301.71 139 4539.56 Table I-29 Retirees in the Public Pension System by Plan Type (Millions) New Retirees All Retirees vi2 Civil State & Civil State & Service Military Local Total Service Military Local Total 1970 0.093 0.072 0.186 0.351 0.707 0.772 1.377 2.856 1975 0.095 0.053 0.213 0.361 1.084 1.012 2.294 4.390 1980 0.097 0.051 0.221 0.368 1.409 1.193 3.092 5.694 1985 0.100 0.051 0.161 0312 1.711 1.280 3.679 6.670 1990 0.102 0.051 0.161 0.313 1.962 1.350 3.825 7.137 1995 0.103 0.051 0.150 0.304 2.158 1.405 3.820 7.383 2000 0.105 0.051 0.206 0.362 231 1.450 3.797 7.558 2005 0.107 0.051 0.227 0.385 2.308 1.447 3.971 7.726 2010 0.109 0.051 0.170 0.330 2.353 1.433 4.162 7.948 2015 0.110 0.051 0.194 0.355 2.396 1.432 4.144 7.971 2020 0.112 0.051 0.217 0.381 2.437 1.432 4.268 8.137 2025 0.114 0.051 0.225 0.391 2.477 1.432 4.446 8.355 2030 0117 0.051 0.350 0.518 2.524 1.432 5.249 9.205 2035 0.119 0.051 0.209 0.379 2570 1.432 5.635 9.637 2040 0.120 0.051 0213 0.384 2.614 1.432 5.445 9.490 2045 0.122 0.051 0.224 0.398 2.656 1.432 5.247 9335 2050 0.124 0.051 0.236 0.412 2.703 1.432 5.116 9.251 2055 0.126 0.051 0.241 0.417 2.746 1.432 5.065 9.243 Table I-30 Retirees in the Public Pension System by Age and Sex (Millions) Male Female 55-57 58-61 62-64 65-67 68-71 72+ 55-57 58-61 62-64 6567 68-71 72+ 1970 0.14 0.28 0.27 0.29 0.35 0.54 0.03 0.08 0.10 0.17 0.16 0.19 1975 0.16 0.40 0.44 0.39 0.50 1.02 0.03 0.09 0.13 0.15 0.28 0.46 1980 0.19 0.44 0.49 0.52 0.66 1.53 0.03 0.09 0.13 0.17 0.29 0.79 1985 0.19 0.45 0.52 0.56 0.76 2.06 0.03 0.10 0.13 0.17 0.29 1.04 1990 0.19 0.43 0.47 0.54 0.77 251 0.03 0.08 0:12 0.16 0.27 h17 1995 0.18 0.42 0.47 0.50 0.72 2.82 0.03 0.08 0.11 - 0.15 0.26 1.23 2000 0.20 0.46 0.48 0.50 0.69 292 0.03 0.09 0.11 0.14 0.26 1.25 2005 0.20 9.52 0.55 0.55 0.71 2.80 0.03 0.09 0.12 0.15 0.28 126 2010 0.20 0.52 0.60 0.63 0.80 275 0.04 0.10 0.13 0.16 0.29 1.30 2015 0.20 0.48 0.54 0.62 0.85 2.84 0.04 0.09 0.13 0.16 0.28 1.30 2020 0.20 0.48 0.54 0.59 0.83 3.01 0.04 0.10 0.14 0.17 0.30 131 2025 0.20 0.47 0.54 0.60 0.83 2.11 0.04 0.10 0.14 0.19 0.33 1.38 2030 0.21 033 0.60 0.65 0.93 3.26 0.04 012 0.18 0.24 0.46 157 2035 0.20 0.48 0.59 0.67 0.95 3.47 0.04 0.11 0.16 0.23 0.47 1.87 2040 0.20 0.45 0.51 0.59 0.90 3.55 0.04 0.10 0.14 0.19 0.41 2.02 2045 0.20 0.46 0.51 0.56 0.79 3.52 0.04 0.10 0.14 0.17 0.38 2.04 2050 0.20 0.46 0.52 0.58 0.80 3.38 0.04 0.11 0.15 0.19 0.38 2.02 2055 0.20 0.45 0.52 0.58 0.84 3.31 0.04 0.11 0.15 0.20 0.41 2.01 140 Table I-31 Contributions and Assets of the Public Pension System (Billions of 1972 Dollars) Total Contributions Fund Balances Civil State & Civil State & Service Local Total Service Local Total 1970 10.046 13.825 23.871 31.321 64.991 96.312 1975 10.370 14.458 24.829 69.020 121.465 190.485 1980 11.273 15.213 26.486 102.540 172.495 275.035 . 1985 11.523 14.292 25.815 129.810 214.598 344.408 1990 12.089 14.746 26.835 151.749 249.310 401.059 1995 12.790 15.357 28.147 172.111 285.177 457.287 2000 13.348 17.619 30.967 189.911 328.239 518.150 2005 13.855 18.494 32.349 203.825 376.517 580.342 2010 14.327 16.524 ~~ 30.851 214.080 418.510 632.590 2015 14.750 16.732 31.482 221.609 446.273 667.882 2020 15.283 16.990 32.273 227.728 471.199 698.927 2025 16.024 17.259 33.283 232.932 494.605 727.537 2030 16.979 21.258 38.237 235.969 525.243 761.213 2035 18.093 18.104 36.196 234.618 531.882 766.500 2040 19.079 18.657 37.736 225.301 514.255 739.555 2045 20.252 19.310 39.562 207.355 491.213 698.568 2050 21.676 20.002 41.678 181.198 467.587 648.785 2055 23.369 20.735 44.104 146.698 448.516 595.213 Table I-32 Benefits Paid by the Public Pension System by Plan Type Average Benefit . Total Benefit Replacement Rates (1972 Dollars) (Billions 1972 Dollars) (At Age 65) 1970 3878.85 4608.86 2711.64 3509.50 2.63 3.56 373 992 0.50 0.60 035 046 1975 4343.21 4893.27 2779.27 3646.26 4.53 4.95 6.38 15.80 0.53 0.60 034 045 1980 4902.48 5259.13 3022.07 3949.71 6.72 6.27 934 22.34 0.49 0.53 030 040 1985 5304.52 5699.06 3278.52 4256.61 8.87 729 1206 28.22 0.52 0.55 032 041 1990 5608.35 6127.02 3539.03 4591.72 10.78 827 13.54 3258 0.49 0.53 031 040 1995 5633.18 6589.78 3786.33 4855.30 11.93 9.26 14.46 35.65 0.43 0.51 029 037 2000 5967.36 7037.57 4072.67 5216.54 13.54 10.20 1547 39.21 0.43 0.51 030 038 2005 6667.87 7574.10 4388.19 5660.17 15.10 10.96 | 17.43 43.49 0.48 0.54 031 040 2010 7115.71 8187.41 4823.12 6102.97 16.43 11.73 20.07 48.24 0.49 0.57 033 042 2015 741880 872555 5186.57 6487.97 17.44 12.49 21.49 51.43 0.50 0.59 035 044 2020 7645.37 9272.32 5380.76 6738.82 18.29 13.28 2296 54.53 0.50 0.60 035 044 2025 7965.28 9852.29 5395.89 6915.66 19.36 1411 2399 57.46 0.47 0.58 032 041 2030 8452.20 10474.95 543549 7039.52 20.94 15.00 28.53 64.47 0.44 0.54 0.28 0.36 2035 9200.51 11183.38 5858.71 7532.94 23.21 16.01 33.01 72.24 0.43 0.52 027 035 2040 10059.27 11953.89 6510.35 8300.22 25.81 17.12 3545 7837 0.44 0.52 0.28 0.36 2045 10881.59 12789.68 (6938.08 8947.62 28.37 1831 36.40 83.09 0.44 0.52 0.28 0.36 2050 11648.77 13701.27 7099.57 9438.76 30.90 19.62 36.32 806.84 0.43 0.50 026 035 2055 12536.02 14734.69 7190.10 9932.85 33.79 21.10 36.42 91.31 0.41 0.48 0.23 032 141 Table 1-33 Recipients of and Benefits Paid by the SSI System (Millions) Recipients Benefits Aged Disabled Blind Aged Disabled Blind 1970 0.0 0.0 0.0 0.0 0.0 0.0 1975 2.649 2.065 0.072 2246352 2593.640 105.505 1980 2.148 2.186 0.078 1821.504 2745.616 113.864 1985 2.214 2.299 0.083 1877.472 2887.544 121.299 1990 1.876 2.394 0.087 1590.848 ~~ 3006.804 128.095 1995 1.473 2.473 0.091 1249.104 3106.088 133.940 2000 1.332 2.601 0.096 1129.536 3266.856 140.746 . 2005 1.398 2.801 0.102 1185504 3518.056 149.551 2010 1.565 3.033 0.109 1327.120 ~~ 3809.448 159.249 2015 1.860 3.211 0.115 1577.280 4033.016 168.161 2020 2.023 3.294 0.120 1715.504 4137.262 176.195 2025 1.824 3.307 0.126 1546.752 4153.590 184.379 2030 1.435 3.259 0.131 1216880 4093.304 191.786 2035 1.115 3.231 0.135 945.520 4058.136 198.325 2040 1.019 3.262 0.139 864.112 4097.070 203.952 2045 0.905 3.357 0.142 767.440 4216391 208.829 2050 0.748 3453 ~~ 0.145 634.304 4336.965 212.780 2055 0.568 3.495 0.148 481.604 4389.719 216.770 Table I-34 Ratio of Selected Average Benefits to the Average OASI Benefit in 1980 (Percent) Private Federal State/ OASI Pension Civil Srv. Military Local 1970 77.7 91.8 199.9 236.4 139.1 1975 97.4 91.0 2229 251.1 142.6 1980 100.0 87.0 251.5 269.9 155.0 1985 107.1 88.2 272.1 292.4 168.2 1990 115.8 99.3 287.7 314.4 181.6 1995 126.7 1163 289.0 338.1 194.2 2000 138.7 136.3 306.2 361.1 208.9 2005 151.3 153.5 342.1 388.6 225.1 2010 164.7 166.7 365.1 420.1 2475 2015 177.6 176.1 380.6 447.7 266.1 2020 188.4 184.3 392.3 475.7 276.0 2025 197.3 193.3 408.7 505.5 276.8 2030 205.6 203.9 433.6 537.4 2789 2035 214.2 216.9 472.0 573.8 300.5 2040 225.0 2323 516.1 613.3 334.0 2045 239.0 248.7 558.3 656.2 356.0 2050 257.0 264.6 597.6 703.0 364.2 2055 277.4 281.5 643.2 755.9 368.9 142 Total 0.0 4945.496 4680.980 48806.312 4725.805 4489.129 4537.137 4853.109 5295.816 5778.453 6028.961 5884.719 5501.969 5201.980 5165.133 5192.656 5184.051 5088.152 Table I-35 Macroeconomic Variables (Billions of 1972 Dollars) N 1970 0.047 1975 0.034 +1980 0.029 1985 0.029 1990 0.031 1995 0.023 2000 0.017 2005 0.012 2010 0.005 2015 0.002 2020 0.004 2025 0.008 2030 0.015 2035 0.012 2040 0.010 2045 0.003 2050 —0.001 2055 —0.003 Table I-36 0.036 0.018 0.028 0.033 0.034 0.025 0.019 0.015 0.008 0.005 0.007 0.012 0.020 0.016 0.014 0.006 0.000 —0.003 Long Term Actuarial Balances 1981-2005 2006-2030 2031-2055 1981-2055 OASDI 1.44 -227 -493 -192 OAS! 1.04 —2.53 -—5.25 -2.25 Savings 266.230 333.358 297.497 353.818 408.287 452.734 516.123 558.828 589.464 607.197 621.449 643.643 678.888 732.680 785.283 834.768 881.618 931.616 143 Bibliography 1 10. 4 if 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23, 24. 25, 26. American Council of Life Insurance, Pension Facts 1980, American Council of Life Insurance, Wash- ington, D.C., 1980. Anderson, Joseph M., “An Economic-Demographic Model of the United States Labor Market”, unpub- lished Ph.D. dissertation, Harvard University, 1977. Barro, Robert, “Are Government Bonds Net Wealth?”, Journal of Political Economy, November 1974. Beller, Daniel, “Patterns of Worker Coverage by Private Pension Plans”, Department of Labor, Labor Management Services Administration, 1980. — “Preliminary Estimates of Participant and Financial Characteristics of Private Pension Plans, 1977”, U.S. Department of Labor, 1981. Berndt, Ernst and Laurits Christensen, “The Translog Function and the Substitution of Equipment, Structures, and Labor in U.S. Manufacturing, 1929-68”, Journal of Econometrics, 1973. Board of Actuaries of the Civil Service Retirement System, Annual Reports, Numbers 48 through 57, U. S. Government Printing Office, Washington, D.C. Board of Trustees, Federal Old-Age and Survivor's Insurance and Disability Insurance Trust Funds, 1981 Annual Report, U. S. Government Printing Office, Washington, D.C. Bowen, William and Aldrich Finnegan, The Economics of Labor Force Participation, Princeton Univer- sity Press, 1969. Cain, Glenn and Harold Watts, Income Maintenance and Labor Supply, Rand McNally Publishing Company, Chicago, 1973. Cartwright, William, S., “Review: Social Security, Pensions and Savings”, Chapter 31 in the Appendix to the Final Report of the President's Commission on Pension Policy, 1981. Christensen, Laurits and Dale Jorgenson, “U.S. Real Product and Real Factor Input, 1929-1967", Review of Income and Wealth, Series 16, No. 1, March 1970. —— , and Lawrence Lau, “Transcendental Logarithmic Production Frontiers”, Review of Economics and Statistics, February 1973. Cooper, Sophia and Denis Johnston, “Labor Force Projections for 1970-80", Monthly Labor Review, February 1965. Crocker, Thomas et. al., The Actuary’s Pension Handbook, Pension Publications, Los Angeles, 1955. Dernburg, Thomas and Kenneth Strand, “Hidden Unemployment 1953-62: A Quantitative Analysis by Age and Sex”, American Economic Review, March 1966. Economic Report of the President, 1981, U.S. Government Printing Office, Washington, D.C. Employment and Training Report of the President, 1980, U.S. Government Printing Office, Washing- ton, D.C. Federal Council on Aging, The Need for Long Term Care: Information and Issues, Department of Health and Human Services, page 50. Feldstein, Martin, “Social Security, Induced Retirement, and Aggregate Capital Accumulation”, Journal of Political Economy, September/October 1974. Fellers, William and Paul Jackson, “Non-Insured Pension Mortality: The UP-1984 Table”, paper pre- sented at the joint program conference of actuaries in public practice, Society of Actuaries, May 1975. Fisher, Charles R., “Difference by Age Groups in Health Care Spending”, Health Care Financing Review, Spring 1980. Gollop, Frank and Dale Jorgenson, “U.S. Productivity Growth by Industry, 1947-1973”, paper present- ed at the conference on New Developments in Productivity Measurement, November 1975. Greenough, William and Francis King, Pension Plans and Public Policy, Columbia University Press, New York, 1976. Hewitt Associates, Investment Performance Report. House Pension Task Force, Report on Public Employee Retirement Systems, presented to the Commit- tee on Education and Labor, House of Representatives, Ninety-Fifth Congress, U. S. Government Printing Office, Washington, D.C., March 1978. 145 27. 28. 29. 30. 31. 32. 33 35. 36. 37: 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 31, S52. 33. * U.S. GOVERNMENT PRINTING OFFICE: 1984- 421-132:4733 146 Hudson, Edward and Dale Jorgenson, “U.S. Energy Policy and Economic Growth, 1975-2000”, The Bell Journal of Economics and Management Science, Vol. 5, No. 2, 1974. —, Christian Kimbal, and Paul Maguire, “U.S. Econometric Growth Model and Projections to 2055”, Dale Jorgenson Associates, Cambridge, Massachusetts, February 1981. Ibbotson, Roger and Rex Sinquefield, “Stocks, Bonds, Bills and Inflation: Historical Returns (1926- 1978)”, Financial Analyst's Research Foundation, Charlottesville, Virginia, 1979. ICF Incorporated, “Analysis of the Potential Effects of a Minimum Universal Pension System (MUPS)”, Chapter 36 in the Appendix to the Final Report of the President’s Commission on Pension Policy, 1981. , A Private Pension Forecasting Model, Washington, D.C., 1979. — “Retirement Income Opportunities in an Aging America: Income Levels and Adequacy”, August 1981. Johnson and Higgins, “1979 Executive Report: Funding Costs and Liabilities on Large Corporate Pension Plans”, 1979. Kochar, Satya, “Representative Payments under the SSI Program, August 1977", Research and Statistics Note, No. 9, U.S. Department of Health and Human Services, September 1980. Kurz, Mordecai, “The Effects of Social Security and Private Pensions on Family Savings”, Chapter 35, Part II in the Appendix to the Final Report of the President's Commission on Pension Policy, 1981. Kurz, Mordecai and Robert Spiegelman, “The Effects of Pensions on Capital Formation: A Framework for Sample Analysis”, Chapter 35, Part I in the Appendix to the Final Report of the President's Commission on Pension Policy, 1981. McKay, Steven, “Computing a Social Security Benefit after the 1977 Amendments”, U.S. Department of » Health, Education and Welfare, Social Security Administration, January 1979. , Long Range Cost Estimates for Old Age, Survivors and Disability Insurance System, 1980, Actuarial Study, No. 83, U.S. Department of Health and Human Services, Social Security Administra- tion, September 1980. Mincer, Jacob, “Labor Force Participation and Unemployment: A Review of Recent Evidence”, in Robert Gordon and Margaret Gordon (eds.) Prosperity and Unemployment, John Wiley and Sons, 1966. Munnell, Alicia, The Future of Social Security, The Brookings Institution, Washington, D.C., 1977. Office of the Actuary, Defense Manpower Data Center, DoD Statistical Report on the Military Retire- ment System, FY 1979 and FY1980, Defense Manpower Data Center, Alexandria, Virginia. President's Commission on Pension Policy, Coming of Age: Toward a National Retirement Income . Policy, 1981. Rogers, Gayle Thompson, “Pension Coverage and Vesting among Private Wage and Salary Workers, 1979”, Office of Research and Statistics, Social Security Administration, Paper Number 16, June 1980. Russell, R. Robert, “The Estimation of Covered Real Wages”, U.S. Social Security Administration, 1977. Schieber, Sylvester and Patricia George, “Retirement Income Opportunities in an Aging America: Coverage and Benefit Entitlement”, Employee Benefit Research Institute, Washington, D.C, 1981. Social Security Administration, Social Security Bulletin, Statistical Supplements, 1970-1979. Social Security Administration, Termination Experience of Disabled-Worker Beneficiaries under OASDI, Office of the Actuary, Actuarial Study Number 65, March 1979. Stein, Bruno, Social Security and Pensions in Transition, Free Press, New York, 1980. Tella, Alfred, “The Relation of Labor Force to Employment”, Industrial and Labor Relations Review, April 1964. ; , “Labor Force Sensitivity to Employment by Age, Sex”, Industrial Relations, February 1965. The Urban Institute, Winklevoss and Associates, Government Finance Research Center, Dr. Bernard Jump, Jr., “The Future of State and Local Pensions”, April 1981. " Wachtel, Paul, “The Impact of Demographic Changes on Household Savings, 1950-2050", Chapter 30 in the Appendix to the Final Report of the President's Commission on Pension Policy, 1981. Winkelvoss, Howard, Pension Mathematics, Pension Research Council, Richard D. Irwin, Inc. Homewood, Illinois, 1976. - 430 GENERAL LIBRARY - U.C. BERKELEY BOOO50237y