, S. bent. of i'EenEEh, Ea’ucation and Welfarfl new publication no. (5/031, #764 7é ’ / 5 PUBLIC HEALTH LIBRARY ' lands im [Lends in Pegisfered Nurse Supply Health Man power References D. C. jones P. C. Cooley A. Miedema T. D. Hartwell DHEW Publication No. (HRA) 76-15 U.S. DEPARTMENT OF HEALTH, EDUCATION, AND WELFARE PUBLIC HEALTH SERVICE I HEALTH RESOURCES ADMINISTRATION BUREAU OF HEALTH MANPOWER I DIVISION OF NURSING BETHESDA, MARYLAND 20014 DISCRIMINATION PROHIBITED—Title VI of the Civil Rights Act of 1964 states: “No person in the United States shall, on the ground of race, color, or national origin, be excluded from participation in, be denied the benefits of, or be subjected to discrimination under any program or activity receiving Federal financial assistance.” Therefore, the nurse manpower data collection program, like every program or activity receiving Federal assistance from the Department of Health, Education, and Welfare, must be operated in compli- ance with this law. The research reported in this publication was performed under Public Health Service Contract Number NOl—NU—44123 from the Division of Nursing, Health Resources Administration, and was concluded in March 1975. Division of Nursing project officer is Evelyn B. Moses, Statistician, Manpower Analysis and Resources Branch. ISSUED MARCH 1976 For sale by the Superintendent of Documents. US. Government Printing Office Washington. D.C.. 20402, Price $1.90 Stock Number 017—041—00097—5 ii T r‘v i 4 T731 FOREWORD pUBL Determination of the current and projected supply of nurses is basic to the consideration of nursing needs and to planning for effective health manpower. Certain factors present special problems in developing techniques for estimating the supply of registered nurses: the population is large; it is the largest of all the health professional occupations; and it consists primarily of women. The Division of Nursing has been active for many years in developing such techniques. The Division staff has worked with other members of the Interagency Conference on Nursing Statistics in developing methodologies currently in use. Support has been given to a number of projects which were designed to examine various aspects of the nurse supply from the viewpoints of data collection and analysis, and identification of the factors influencing the supply environment. In line with the accumulation of data on registered nurses and in the light of recently collected information, it was felt that‘a study was needed of appropriate techniques for estimating the nurse supply which would use the existing data sources and those aspects that serve as predictors. This report is the result of such a study. Mr. Dale Jones, the Project Director, and his'colleagues present a design for estimating the number of nurses that takes into account various components of the supply. The report is helpful in identify- ing an approach to the estimating procedure, as well as in defining those areas that, because of their influence on any estimates, need further study and refinement. Jessie M. Scott Assistant Surgeon General Director Division of Nursing iii ACKNOWLEDGMENTS In conducting this study, many individuals provided assistance. The authors sincerely appreciate their assistance and cooperation and are fully aware that whatever accomplishments were achieved in this study were greatly enhanced by their participation. Ms. Evelyn B. Moses, the Project Officer with the Division of Nursing, assisted in many ways. Her keen knowledge of nursing data, its sources, and its limitations, was extremely valuable to the authors. She also provided the authors many relevant references, expedited obtaining data tapes and other materials, and above all, was always willing to help the authors who frequently found need to discuss various issues with her. The authors appreciate the assistance extended by the members of the Interagency Conference on Nursing Statistics (ICONS) who provided relevant data and meaningful comments, and aided the authors through personal discussions. A critical review is essential to all studies. The authors are indebted to the sponsor and members of ICONS for their comments, as well as to Drs. D. G. Horvitz and J. T. Wakeley of the Research Triangle Institute who assisted through technical guidance and constructive reviews. The capability and patience of the secretaries was sincerely appreciated by the authors. We are indebted to their performance. The efforts of others at RTI in performing calculations, drafting the figures, and editing the report contributed significantly to the report. The authors sincerely appreciate all the assistance received; however, they take full responsibility for the content of this report. iv . CONTENTS Page Foreword .................................................. iii Acknowledgments ......................................... iv List of Tables .............................................. vi List of Figures ............................................. viii Glossary of Terms ......................................... ix 1. SUMMARY ......................................... 1 II. INTRODUCTION .................................... 5 General ........................................... 5 Objective .......................................... 5 III. DYNAMICS OF REGISTERED NURSE SUPPLY 7 Introduction ....................................... 7 Education of Registered Nurses .................... 9 Licensure of Registered Nurses ..................... 12 Employment of Registered Nurses .................. 18 IV. REGISTERED NURSE SUPPLY DATA ............. 23 Introduction ....................................... 23 Data Sources ...................................... 24 V. EVALUATION OF RELEVANT VARIABLES ....... 29 Introduction ....................................... 29 Theoretical Background ............................ 31 Previous Analyses ................................. 33 General Data Description .......................... 36 Empirical Nurse Supply Models .................... 38 Empirical Results .................................. 44 Summary .......................................... 57 VI. RTI’S NURSE SUPPLY PROJECTION MODEL ...... 59 Other Nurse Projection Models ..................... 59 RTI’s Model ....................................... 64 Model Validation ................................... 90 VII. REGISTERED NURSE SUPPLY PROJECTIONS . . . . 101 VIII. CONCLUSION AND RECOMMENDATIONS ......... 105 IX. REFERENCES ..................................... 107 99°93? mega 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. Tables Page General registered nurse supply information for 1972 . . . . 8 Age of students entering U.S. schools of nursing in 1967 . 9 Candidates passing registered nurse licensing examination 13 Newly licensed registered nurses, by age and type pro- gram .................................................. 15 . Estimate of the number of expired licenses, by year ..... 18 Employment status of registered nurses in 1972 ......... 19 Activity rate, by age—1972 ............................. 19 Percent of employed registered nurses, by age group and marital status—1972 ................................... 20 . Percent of registered nurses employed part time ......... 21 List of questions for inventory of registered nurses ...... 26 Sample of tables contained in Facts About Nursing ...... 27 Percentage distribution of employed registered nurses, by marital status and by sex ............................... 30 Registered nurse activity rates ......................... 30 Regional classification .................................. 40 Labor force participation determinants for single profes- sional nurses, 1970 ..................................... 46 Conditional labor supply relations for single professional nurses, 1970 ............................................ 47 Computed elasticities of labor force participation ......... 49 Labor force participation determinants for married profes- sional nurses, 1970 ..................................... 53 Conditional labor supply relations for married professional nurses, 1970 ............................................ 55 Means of regression variables .......................... 56 Age specific activity rates for 1966 and 1972 ............. 68 Labor force participation rates and activity rates for all females and licensed RN’s, by year ...................... 69 White female mortality rates ........................... 69 Graduations from professional nursing schools, by type of program, 1960 through 1972—73 ......................... 71 Admissions to professional nursing schools, by type of program, 1960 through 1972-73 ......................... 73 Earnings of registered nurses employed as general duty nurses in non-Federal hospitals relative to public school teachers, 1960—61 through 1972—73 ....................... 74 vi 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. Tables—(continued) Data used in fitting linear regressions to predict percent of female high school graduates for AD and BAC programs . 78 Projections of GA and W and number of female high school graduates .............................................. 79 Projected admissions to professional nursing schools by type of program 1973—74 through 1983—84 ................ 80 Program graduations from professional nursing schools by type of program, 1973—74 through 1983—85 ............... 81 Examination and first time licensure data ............... 83 Comparison of licensure ratio and estimated licensure rate 84 Projected number of licenses issued to new U.S. nursing graduates, 1972—73 through 1984~85 ..................... 84 Estimated first time U.S. licensure of foreign trained nurses ................................................. 85 Estimates of k, N(T) and C(T) for ANA inventory years . . 87 Regression coefficient estimate for linear model ......... 88 Sample of foreign trained nurses that passed nursing examination, by age .................................... 88 Licensed stock, by age—1966 ............................ 9O Licensure data ......................................... 92 Differential estimates of supply and stock ............... 92 Distribution of supply growth, by cause ................. 93 Estimated number of renewals ......................... 94 Frequency and distribution of age at graduation ........ 94 Admissions to professional nursing schools, by type of program, 1959—60 through 1964—65 ...................... 95 Relative earnings of registered nurses relative to public school teachers (W), 1956—57 through 1964—65, and number of female high school graduates, 1965—66 through 1970—71 96 Projected admissions as a percent of female high school graduates to professional nursing schools, by type of pro- gram, 1965—66 through 1970—71 .......................... 97 Projected admissions to professional nursing schools, by type of program ........................................ 97 Projected graduations from professional nursing schools, by type of program ..................................... 98 RN supply projections, by year ......................... 98 Comparison of projected 1972 values of RN stock and supply with adjusted ANA 1972 inventory data .......... 99 Stock and supply projections for selected years and as- sumptions ............................................. 101 vii P‘PWN!‘ 10. 11. 12. Figures Page Nurse supply flow ...................................... 7 Number of graduates from US. schools of nursing ....... 10 Completion rate ........................................ 11 Licenses issued to registered nurses from foreign countries 13 Number of licenses issued to registered nurses for the first time in the United States ............................... 14 Number of active renewal licenses adjusted for biennial licensing ............................................... 16 Number of licenses issued to previously licensed nurses . . 17 . Admissions to professional nursing schools as a percent of female high school graduates by type of program ........ 74 . Ratio of earnings of registered nurses to public school teachers W(t) ........................................... 75 Admissions to professional nursing schools as a percent of female high school graduates, by type of program ........ 75 Graduation from professional nursing schools, by type of program ............................................... 82 Projected trends in RN stock and supply . .- .............. 102 viii Glossary of Terms a = Activity rate—The proportion of licensed RN’s who are em- ployed. A = Annual net attrition rate. AD = Percent of female high school graduates admitted into associ- ate degree programs. AGE = Age group classification variable. BAC = Percent of female high school graduates admitted into baccalaureate degree programs. D = Mortality—Deceased registered nurses. DIP = Percent of female high school graduates admitted into di- ploma programs. E = Number of licenses issued by endorsement. f = Percent of foreign trained nurses newly licensed in the United States. F = Foreign nurse first US. license—A person trained in a non-US. school of nursing and has received the first license to practice nursing in the United States. g = US. graduates—Graduates from US. schools of nursing that prepare students for licensure as RN ’s. G = Newly licensed nurses—N umber of graduates of US. schools of nursing that are licensed in a time period. GA = Growth function representing the growth process for associ- ate degree programs. GB = Growth function representing the growth process for bacca- laureate degree programs. h = Percent of graduating nurses in an age category. H = First time licensure—Licenses issued to professional nurses for the first time in the United States. The method of licensure may be by waiver, examination, or endorsement, and includes both US. and foreign nurses. HOME = Home ownership classification variable. HR = Number of hours spent working as a registered nurse per period. i = Age subscript. k = Ratio of nurses to licenses that applies when the ratio is assumed equal for renewed and reinstated licenses. k1 = Ratio of nurses to licenses that applies for renewals. k2 = Ratio of nurses to licenses that applies for reinstatements. k3 = Ratio of nurses to licenses that applies for endorsements. KID < 18 = Number of children below 18 years of age residing in the married nurse’s household. ix Glossary—continued L = Licensed stock—The number of persons who are licensed as RN’s. LFP = Labor force participation, i.e., the proportion of trained nurses who are employed as RN’s. LH = Number of leisure hours. m = Subscript denoting married nurse. M = Mortality rate—Proportion of a specified population that die within a year. n = Maximum age. N = The number of nurses to whom licenses have been reinstated. NR = Probability a student does not retake examination after failing twice. 0 = The number of nurses who allow their licenses to expire for reasons other than death. O = Number of nurses whose licenses have expired, including for reasons of death. p = Survival probability for females. P = Average price of market goods. PFT = Probability of passing the examination the first time. PR = Probability of passing retakes of examination. PRESCHOOL = Preschool classification variable. q = Expected maximum admission rate. Q = Number of licenses renewed during a year, adjusted for bien— nial renewals. R = Number of licenses reinstated. RACE = Race classification variable. REGION = Regional classification variable. s = Subscript denoting single nurse. S = Employed registered nurse—A registered nurse that is em- ployed either part time or full time in nursing. t = Beginning of year. T = Year from January (t) to January (t + 1). TH = Number of hours available each time period. ' U = Age-specific weighting factor for reinstated nurses. V = Age-specific weighting factor for nurses whose license(s) have expired. W = Ratio of nurses’ to teachers’ wages. WRN = Hourly wage rate for registered nurse. X = Quantity of market-produced goods and services. Y = Nonearning income. Z = Test shift variable. p. = Error term. I. SUMMARY The Division of Nursing of the Bureau of Health Manpower (BHM) contracted with the Research Triangle Institute (RTI) to investigate alternative projection methodologies to determine if improved projections of the number of employed, registered nurses (RN supply) can be made. The projection methodology developed under this contract is a practical process which can be used periodically for making accurate national projections of RN supply. In conducting the study, RTI reviewed the pertinent literature, conducted analyses, developed a projection model, and made multi- ple projections of nurse supply using different assumptions. In 1972 there were roughly 1.4 million persons suitably trained for becoming an RN. Of these, about 1.1 million were licensed and .8 million were active. The number of students graduating from U.S. schools of nursing increased 36 percent between 1970 and 1973. This increase in number of graduates was due to both a larger number of admis- sions and a higher proportion of admissions that graduated (comple- tion rate). The rapidly increasing numbers of graduates of associate degree programs, where the number of graduates increased about 113 percent over the 1970—73 time period, were largely responsible for this increase. The percentage of applicants (US and foreign trained nurses) that pass their RN licensing examination the first time has slowly decreased from 85.8 in 1965 to 81.8 in 1972. The percent of those who retake the examination and pass has also decreased from 63.3 in 1965 to 52.5 in 1972. Hence, a smaller proportion of persons taking the RN licensing examination are being licensed. The proportion of foreign nurses obtaining their first license in the United States between 1962 and 1972 has fluctuated between 7 and 17 percent of all first time US. licenses issued and, hence, contributes significantly to the increase in the licensed stock of RN’s. The number of male RN’s is increasing, but males still comprise a small proportion of all RN’s (fewer than 2 percent). The most important variables influencing the decision of married nurses to participate in the labor force and the extent to which they participate are: (1) the husband’s earnings level, (2) the number of children in the household, (3) the presence of preschool children, and (4) the age of the nurse. The RN wage rate did not seem to influence married nurses’ labor force behavior in any substantial way. It appears that, as compared with other ethnic groups, black married 2 nurses are more likely to participate in the labor force and, given that they do participate, they are more likely to put in longer hours. The geographic location of the married nurse appeared to have little to do with whether or not she works. However, if she does work, her location may be a factor in determining how many hours she works. The labor force participation behavior of single nurses was influenced by different factors than those for married nurses. The most important variables influencing the decisions of single nurses to participate and for how many work hours are their nonearnings income and their age. Increases in nonearnings income appear to decrease the probability that single nurses will work, while their labor force participation tends to decline with age. Wage changes seemed unimportant in influencing either the likelihood of labor force participation or the number of hours worked. Black single nurses appeared to be a little less likely to work than whites; although there were no apparent differences in hours worked among ethnic groups of working nurses. Regional differences did appear to affect the number of hours working, single nurses provide; but they did not affect the probability of labor force participation. The projection model developed by RTI is based on the licensed stock of nurses and the flow of nurses into and out of the licensed stock resource pool. The supply of RN’s is estimated by multiplying the stock of licensed nurses by the proportion of licensed nurses that are actively employed (activity rate). The stock of licensed nurses is recursively estimated from a base year using newly licensed graduates from US. schools of nursing, foreign trained nurses that are newly licensed, nurses who have gotten their license reinstated, deaths, and the number of nurses who have allowed their license to expire. The mathematical state- ment of the model is: I] S(t) = 2a.(t)L.(t> |=1 where Li(t) = pi_1Li_,(t—1)+ Gi_1(T-—1) + Fi_1(T—1) + Ni—1(T_1)_ Oi—1(T— 1) the number of RN’s employed in the nursing profession at time t, ai(t) = the activity rate (the ratio of employed to licensed RN’s) at time t in age category i, S(t) 3 the licensed stock of RN’s that hold licenses at time t in age category i, pi = the female survival probability for the ith age category, Gi(T) = the number of graduates of US. schools of nursing licensed during year T in age category i, Fi(T) = the number of RN’s trained in non-U.S. schools of nursing that have received a US. RN license dur- ing year T in age category i, the number of nurses that are reinstated during year T in age category i, Oi(T) = the number of living nurses that allow all their licenses to expire during year T in age category i, t = January lst of year T, and T = the calendar year beginning at time t. Li(t) II N i(T) Submodels have been developed for most of the variables included in the model. The submodels were developed for the purpose of projecting the variable or for relating the variable to existing data. In checking the model, January 1972 estimates of nurse supply were made using data from the 1966 American Nurses’ Association (ANA) Inventory of RN’s and from annual licensure data. Two estimates were used: one assuming a constant activity rate based on 1966 data and the other assuming the activity rate increased at the 1963 to 1966 growth rate. The estimated values were compared with the 1972 ANA Inventory of RN’s adjusted to January 1972. Using the constant activity rate, the RTI model’s estimate of supply was 2.8 percent lower than the observed value; assuming the increasing activity rate, the estimate of supply was 2.8 percent higher than the observed supply. The 1980 projected supply of registered nurses under four differ- ent sets of assumptions ranged from 941,000 to 1,034,000. The 1985 projections ranged from 964,000 to 1,168,000 for the same four sets of assumptions. II. INTRODUCTION General The Division of Nursing, Bureau of Health Manpower (BHM), in conjunction with other agencies and organizations, has historically provided projections of the supply of nurses in the United States. The Bureau has used the same projection technique for several years in making these estimates. Partly because factors that influ- ence nurse supply are continually changing, and partly because other, more sophisticated nurse supply models have been proposed, the adequacy of this projection technique methodology has been questioned. To determine if a better projection methodology is available to make periodic estimates of nurse supply, the Division of Nursing contracted with the Research Triangle Institute (RTI) to conduct this study. The focus of this study is twofold: first, to provide projections of nurse supply; and second, to develop a practical projection method- ology that can be routinely utilized over time. The type of projection methodology developed in this study consists of: (1) a model that uses periodically collected national data and (2) procedures that update the values of the model parameters as more recent data become available. , Under this contract, RTI reviewed the appropriate literature, developed a projection methodology, and made projections of regis— tered nurse supply. The projection model developed by RTI is described in Section VI and the projections for several different assumptions are presented in Section VII. The variables that affect the supply of registered nurses are analyzed in Section V with pertinent data sources indicated in Section IV. The dynamics or flow of registered nurses in and out of the active supply are discussed in Section III. Objective The objective of the study is to develop a practical process which can be used periodically for making accurate national projections of registered nurse supply. This objective specifies a “practical process.” Often complex pro- jection methodologies are developed that are technically good but are seldom, if ever, used because of lack of data and/or because few people know how to implement the model. In this study the 6 endeavor was to keep the model conceptually simple and its imple- mentation practical while still providing accurate projections of nurse supply. This objective also specifies periodic model application. For a model to be used periodically, suitable data must be available periodically. To the degree possible, the model was designed to use data from sources that are expected to exist in the future. The model was designed to provide national projections of regis— tered nurses. State projections of RN’s and national and State projections of licensed practical nurses are not included in the scope of this study. III. DYNAMICS OF REGISTERED NURSE SUPPLY Introduction This section describes the flow of nurses into, and out of, the active nurse supply and identifies the factors that affect the flow. The flow of nurses affecting the size of the trained, licensed and active populations is depicted in figure 1. Figure 1.—Nurse supply flow U.S. FOREIGN NURSE TRAINED GRADUATES NURSES TRAINED LICENSED STOCK FIRST TIME LICENSES j—‘r— REINSTATED U.S. FOREIGN T _‘+—‘ r LICENSED EMPLOYED STOCK NE_ REENTERED ACTIVE SUPPLY 1 / INACTIVE I LICENSE REVOKED, SUSPENDED OR EXPIRED I DEATHS OUT MIGRATION The trained population can be increased through graduates from US. schools of nursing and by foreign trained nurses entering the United States. This population can be reduced through deaths and out-migration. The licensed population is a subset of the trained population, since not all graduates obtain a license, and some previously licensed nurses are no longer licensed. The size of the licensed population isincreased through first time licensure and reinstate- ment of previously licensed nurses, while license expiration, suspen- sion, and revocation, and death of nurses decrease the size of the hcensed stock. Not all persons that are licensed are actively employed in nurs- ing. The number of employed nurses is increased through newly Table 1.—General registered nurse supply information for 1972 1972 value Data (1,000) source ' Size of trained stock , . . . 1,400 Estimated from data (1, p. 146) Size of licensed stock , . , _. 1.130 (2,1972) Size of active supply ,, ,, _, , .4. 795 (2,1972) Number of nursing students graduating during year , , 52 (3,1972—73) Number of foreign trained nurses entering the United States during year .. ,. 2 Number of first time licenses issued dur- ing year—United States .. .- 48 (3, 1972—73) Number of first time licenses issued dur— ing year—foreign .,. . . . 7 Estimated from data (3, 1972—73) Number of reinstatements during year . 13 (3, 1972—73) Number 01 revocation, suspension. expira- tion during year ., , Number of nurses employed for the first time during year _ . . , ,. 3 Number of nurses reentering nursing dur- ing year .. 2 Number of nurses becoming inactive dur- ing year . ,., . . , . 2 Number of deaths during year _ ,, . 4 Number of nurses migrating out of the United States during year ,.. n . 2 ‘ llallC numbers in parentheses refer to literature Clied in reierence IISl. page 107 2 No known available data. 5‘ No known available data at the time this report was prepared. However, a study is underway for the DIVISIon oi Nursmg to obtain this information on a sample of newly licensed nurses. 'Could be computed lor the licensed population usmg the age distribution 01 the licensed population contained in Statistical Abstracts (4) and published age-sex-spemfic mortality rates. 9 licensed nurses entering the labor force and by previously employed nurses becoming active. The active nurse supply is decreased through nurses choosing to be inactive for a variety of reasons. The magnitude of all the three specified populations and the flow factors in figure 1 would enhance the description of registered nurse supply; however, data limitations prevent a complete quanti— tative description. Table 1 presents available data, or estimates, for the year 1972. The sources of the information are also provided. In 1972 roughly 80 percent of the trained stock of nurses were licensed and 70 percent of those licensed were actively employed in nursing. Thus, only about 60 percent of the trained stock of nurses were actually employed. The inflow to the licensed stock (first time licenses and reinstatements) was approximately 5 percent of the licensed stock. Education of Registered Nurses Nurses receive their training either from one or a combination of three type of US. schools of nursing or from a foreign country. Training of foreign nurses is not discussed in this report; however, a few comments are given on how foreign trained nurses enter the United States and become a part of the trained nurse stock. Graduates of us. Schools of Nursing Nursing programs in the United States have typically been classified as an associate degree program, a diploma program, or a baccalaureate degree program. An associate degree program is commonly characterized as a 2-year community college program and baccalaureate degree program as a 4—year college program. A Table 2.—Age of students entering U.S. schools of nursing in 1967 (5) Number of entering students Baccalau- Associate reate Age Diploma degree degree Total 19—20 ,,,,,,,,,,,,,, 895 549 802 2,246 21-22 ______________ 2,856 2,524 3,130 8,510 23-32 ______________ 341 1,617 547 2,505 33—42 ______________ 44 704 26 774 43— ________________ 15 488 15 518 10 Figure 2.—Number of graduates from US. schools of nursing TOTAL 40 _. NUMBER OF GRADUATES, (1,000) DIPLOMA U D 20 10 BACCALAUREATE DEGREE ASSOCIATE DEGREE I l I I I gl 1962 64 66 68 70 72 74 SCHOOL YEAR ENDING ON STATED YEAR Source: National League for Nursnng data In Facts About Nursing. (3) diploma program is often assumed to be a 3-year hospital based program. As with most classification schemes, exceptions do exist. The importance of these three programs in providing graduates over the past decade has been shifting as indicated in figure 2. The number of graduates from associate degree programs has increased rapidly, from baccalaureate or higher degree programs, moderately, while the numbers of graduates of diploma programs has decreased. The age distribution of the students entering the three types of programs is indicated in table 2. The students in the associate degree program tend to be older than those in the other two types of programs. Obviously, a significant number of students are not entering a nursing program immediately after graduating from high school. This is particularly true for those entering associate degree programs. No information was available on the age of the students at graduation. However, by assuming an average program enrollment duration and that attrition is equally distributed across all ages, the student age at graduation can be estimated. 11 80 , Figure 3.—Completion rate /DIPLOMA COMPLETION 7° RATE, PERCENT 60 '\ 50 I i I I 1964 65 68 70 72 SCHOOL YEAR ENDING ON STATED YEAR ASSOCIATE DEGREE l Source: Computed from National League for Nursing data In Facts About Nursing. (3) The completion rates for the three types of programs have also been changing over the last decade as indicated in figure 3. The diploma program has the highest completion rate. However, for the school year 1972—73, the baccalaureate degree program’s completion rate was equally high. The completion rate for both the baccalau- reate'and associate degree programs has tended to increase while the diploma program has fluctuated but has remained relatively constant. Men are constituting a larger share of the total number of nursing school admissions. In the 1971—72 school year, 6.0 percent of all persons admitted to schools of nursing were men. (6, [1. 1,7)1 This figure was up from 3.5 percent in 1968-69 and 1.8 percent from 1965-66 school years. (7) Presently, men are a small part of nurse graduates but, if the trend continues, they will provide an increas- ing share of the graduates. In summary, the flow of graduates from US. schools of nursing into the trained population is increasing. The rapid growth of the associate degree program and the higher completion rates have contributed significantly to the increasing number of graduates from US. schools of nursing. Foreign Trained Nurses Entering the United States Nurses trained in foreign countries and entering the United 1 Italic numbers in parentheses refer to literature cited in reference list, page 107. 12 States add to the stock of trained nurses. These foreign trained nurses may enter the United States through a variety of arrange- ments, from living in Canada or Mexico and working in the United States, to immigrating to the United States with the intent of becoming a US. citizen. The Immigration and Naturalization Service reports the number of nurses entering the United States in fiscal year 1969 was 5,466, or about 13 percent of all additions to the stock of trained nurses in that year. (3, 1969, 8)2 At least one hospital occasionally recruits outside the United States. (9) Other foreign trained nurses enter the United States as spouses of US. citizens. The total number of foreign trained nurses in the United States is unknown but appar- ently constitutes a significant portion of the nurse stock. Licensure of Registered Nurses General A trained nurse must be licensed in order to be eligible for employment as a registered nurse. Licenses must initially be obtained for graduates of US. schools of nursing through examina- tion, and for foreign trained nurses by examination or endorsement. Previously licensed nurses who have allowed their licenses to expire or who have had their licenses revoked or suspended may be able to get their licenses reinstated after being reviewed by the licensing agency. Nurses with current licenses are licensed for 1 year in most States and 2 years in the remaining States. These licenses must be renewed at expiration. Nurses who move to another State must obtain a license to practice in that State. This is usually accom- plished through a review of the nurse’s credentials and the subse- quent issuance of a license by endorsement. Licensing is a State responsibility, hence, each State collects its own licensure data. Since no State requires termination of license when a nurse moves out of the State, the number of valid licenses exceed the number of licensed nurses in each State. Normally, the license would expire in 1 or 2 years; however, some nurses renew their license even though they no longer reside in that State. For this reason and others discussed later, there is a significant differ- ence between the number of licenses and the number of licensed nurses. For example, in 1972 it was estimated that there were 2 Altman’s terminology of activity rate and labor force participation rate will be used in this report. Labor force activity rate is the number of active nurses to all licensed nurses where participation rate is based on all potential nurses; i.e., trained but not necessarily licensed. 13 Table 3.—Candidates passing registered nurse licensing examination (3) Percent passed examination Year First time Subsequently 1972 ,,,,,,,,,,,,,,,,,,,,,,,,,,,, 81.8 52.5 1971 __,.,W ' ' 1970 H 83.2 53.8 1969 ,_ MM. M, M, 84.3 57.7 1968 84.6 58.3 1967 _- 85.3 58.2 1966 ,,,,,,,,,,,,,,,,,,,, . 7 85.6 58.5 1965 _- 85.8 63.3 1964 - 85.7 62.8 ' Not available approximately 197,000 more licenses than nurses (2, 1972, p. 123), or about 15 percent of the licenses were due to multiple licensing. First Time Licensure in the United States Graduates of U.S. schools of nursing must pass} an examination in order to obtain their first license. Table 3 indicates the percentage of candidates (United States and foreign) that passed the exam the first time, and those that passed subsequently. The number of U.S. graduates eventually obtaining licenses is not directly obtainable from the data; however, estimates derived later in this report indicate the percentage is in the middle to high nineties. Foreign trained nurses may obtain licenses either by endorse- ment or examination. Figure 4 shows that the number of foreign Figure 4.—Licenses issued to registered nurses from foreign countries (3) 6 r' NUMBER OF FOREIGN 5 - NURSES LICENSED, (1,000) FIRST TIME ENDORSEMENT 1954 56 58 60 62 64 66 68 70 72 74 V E AR Note: A U.S. license obtained by a foreign trained nurse by examination may not be the first U.S. licensei 14 Figure 5.—Number of licenses issued to registered nurses for the first time in the United States (3) 60' (1,000) NUMBER OF LICENSES, 1962 64 66 68 70 72 Y EAR trained nurses obtaining licenses by examination is increasing rapidly. Apparently, States are requiring more of the foreign trained nurses to take the examination, with some States making the examination mandatory for all non-US. trained applicants. Nurses who were previously issued a US. license in one State, based on a license or certificate from a foreign country, may still have to take the examination if they move into a State in which the examination is mandatory. Hence, a US. license obtained by foreign trained nurses by examination may not be their first US. license. Figure 5 shows the number of licenses issued to registered nurses (both US. and foreign trained) for the first time in the United States, by year. This figure shows that the number of first time licenses has been steadily increasing over this time period. Table 4 indicates the age distribution of those persons newly licensed in the United States by type of program. The table indicates that a large percentage of the newly licensed nurses are fewer than 25 years old; however, a significant percent are 25 years or older. Of particular interest are the nurses who received training in the rapidly growing associate degree program in which one-half of the nurses were 25 or more years old when they received their licenses. Renewed Licenses States require nurses to renew licenses either annually or bienni- 15 Table 4.-—Newly licensed registered nurses by age and type program1 (10) Age distribution, percentage Baccalau- Associate reate Age Diploma degree degree <20 , .2 1.2 0.0 20 WW, 6.8 16.3 .1 21 ,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,, 60.4 13.1 4.6 22 ,,,,,,,,,,,,,,,,,,,,,,,,,,,, W , 15.7 8.1 61.7 23 4.9 5.8 19.9 24 ._ W ,,,,,,,,,,,,, 2.6 4.2 5.3 25 -- 1.3 5.4 2.4 26 W, W 1.7 2.6 1.8 27 ,,,,,,,,,,,,,,,,,,,, .4 2.0 .4 28 .6 2.8 .6 29 - _ _,_ W-.. , .8 2.8 .3 30—34 _ . 1.5 10.4 1.0 35—39 W 1.2 8.2 .6 40—44 .6 6.8 .3 45and over __ W. __W_ .8 9.1 .1 Other ,,,,,,,,,,,,,,,,,,,,,,,,,,,,, .6 1.2 .7 Total ,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,, 2100.1 100.0 299.8 ‘ Based on a sample of approxumately 4.760 newly licensed registered nurses. 1 Error due to rounding. ally. Some nurses maintain licenses in more than one State; hence, the number of renewed licenses exceed the number of nurses renewing their licenses. The number of active renewed licenses ‘adjusted for biennial licensing is shown in figure 6 to be steadily increasing. Licenses Issued to Previously Licensed Nurses Licenses issued to registered nurses previously licensed in the United States are categorized by the ANA (3) as reinstatements, endorsements, and examinations. In general, reinstatement of a license refers to reinstating a license in the same State in which the nurse was previously licensed, whereas an endorsement would be based on a license obtained in another State. Some nurses are required to take the examination before they are reissued a license. Because States differ in how they classify the licenses issued to previously licensed nurses, the accuracy of the reported number of licenses issued by the type license is questionable. Apparently, the published values of reinstatements are not a good indicator of the 16 Figure 6.—Number of active renewal licenses adjusted for biennial licensing 1A, NUMBER OF RENEWED LICENSES, (1,000,000) .9 I l l l J 1962 64 66 68 70 72 YEAR Source: Computed from data in Facts About Nursmg. (3) number of nurses who have allowed their license to expire and then later have it reinstated. The number of licenses reported as being issued by reinstatment, endorsement, and examination are shown in figure 7 as a function of time. Most of the licenses issued to previously licensed nurses are reported to be by endorsement. The number of licenses reported as being reinstated is about one-third of those obtained through endorsement. The number of nurses who were required to take an examination before being reissued a license was reported as being fewer than 400 per year during the time period 1962 through 1970. Hence, the number of licenses issued based on examination is small in comparison with endorsements and reinstatements. Examina- tions were combined with endorsements in figure 7 because the two were combined by the ANA for the years 1971 and 1972. (3) 17 Figure 7.—Number of licenses issued to previously licensed nurses (3) 60 l- NUMBER OF LICENSES, (1,000) 50 — TOTAL 40 - ENDORSEMENTS AND EXAMINATIONS 30 20 — 10 - REINSTATEMENTS I l I l l 1962 64 66 58 70 72 YEAR Note: The accuracy of the reported number of reinstated licenses Is questionable because of State differences in classifying licenses. Many of the nurses who obtained licensure through endorsement have an active license in another State. The extent of multiple licensing among those nurses who have recently moved and have obtained a license through endorsement is not known, but is believed to be high. Reinstatement of a previous license is considered a renewal in several States and, therefore, these are counted as renewals. Most States, however, do reinstate licenses. Nurses who have their licenses reinstated may also have an active license in another State, hence, the possibility of multiple licensure. Expired, Revoked, and Suspended Licenses The number of licenses and the number of licensed nurses are 18 Table 5.—Estimate of the number of expired licenses by year First time Previously Expired licensed in licensed in Renewals ‘ Renewals ' licenses in Year T year T—1 year T—1 in year T-1 in year T year T 2 1963 ,, __ . ,,, . ,, 35,312 40,429 985,400 1,010,300 50,800 1964 , ,, ,, , 32,780 45,804 1,010,300 1,042,500 46,400 1965 _ 36,088 42,266 1,042,500 1,064,500 56,400 1966 , , . ,, A ,, , . 36,432 45,406 1,064,500 1,101,100 45,200 1967 W,” 37,466 48,310 1,101,100 1,138,500 48,400 1968 ,, _ . ,, , , 42,499 51,272 1,138,500 1,184,000 48,300 1969 A ._ , ,, 46,456 53,017 1,184,000 1,225,300 58,200 1970 ,_ _, , ,, , 45,708 53,878 1,225,300 1,261,400 63,500 1971 _ _, . , _ . . , . 47,010 54,890 1,261 ,400 1,317,400 45,900 1972 ,,, ,, 50,171 55,513 1,317,400 1,345,700 77,384 ‘Adjusted for biennial licensing, 2 The number of expired licenses IS not equal to the number of nurses who no longer have an active license, because many nurses have multiple licenses. Source: Computed trom American Nurses' Assocnatlon data. (3) decreased by revocation, suspension, and expiration. Almost all the reductions of licenses are due to expirations and few result from suspension or revocation. Because licenses are issued for a 1- to 2- year period, the license stays valid for that time period (unless revoked) regardless of whether the nurse is living, residing in the State of licensure, or has any intention of becoming an active nurse. The license expires when the nurse fails to renew the license. Presently, no record is kept on expired licenses; however, they can be estimated using licensure data. The number of nurses who have allowed all their licenses to expire is not determinable from licen- sure data. The estimate of expired licenses given in table 5 was computed using the following relationship: Licenses expired during year T equals the sum of the first time licenses, licenses issued to those previously licensed, and renewed licenses issued in year T—1, minus renewed licenses in year T. Employment of Registered Nurses Many licensed nurses are not actively employed, however, those that are may be employed either full time or part time. Table 6 provides the number and percentage of licensed nurses, full-time and part—time employed nurses, and licensed nurses not employed in nursing for the year 1972. Though some of the characteristics of the active and inactive licensed nurses are known, very little is known about the flow in and out of the active supply. 19 Table 6.—Employment status of registered nurses in 1972 (2) Type employment Number Percent Full time_,,, M, _, 505,201 44.8 Regular part time _,_. W, 159,609 14.1 Irregular part time m, ,, 78,591 7.0 Full time or part time not known but employed in nursing ,,,,,,,.-,,e_,,_ , _ 35,069 3.1 Notemployed in nursing ,.-,,_,,,_ V. 316,611 28.1 Employment status not known ,..,, ,,_ 32,576 2.9 Total , , , 1,127,657 100.0 Table 7.-Activity rate by age—1972 (2) Number of Number of Active inactive active Total nurses licensed licensed licensed to total, Age group nurses nurses nurses percent < 25 ,,,,,,,,,,,,,, 6,223 66,991 73,214 91.5 25 to 29 ,,,,,,,,,,,, 37,329 126,869 164,198 77.3 30 to 34 _- ,_ ., ,, We, 48,766 94,243 143,009 65.9 35 to 39 H”, , "W. _ 43,144 84,937 128,081 66.3 40 to 44 33,705 82,517 116,222 71.0 45 to 49 ,,,,,,,,,,,, 32,204 91,609 123,813 74.0 50 to 54 ,,,,,,,,,,, 25,622 73,530 99,152 74.2 55 to 59 ,,,,,,,,,,,,, 21,355 54,900 76,255 72.0 60 to 64 ,_ ,_ -. m .2, 21,896 43,413 65,304 66.5 65 and over ........ 33,055 25,248 58,303 43.3 Some preliminary data from a 1972 national sample survey concerned with employment opportunities for newly licensed nurses indicated that 95 percent, 90 percent, and 94 percent of the gradu- ates of baccalaureate, associate degree, and diploma programs, respectively, were employed as nurses. (11) The percentages of newly licensed nurses not seeking work were 4, 5, and 3 percent, respectively. The remainder of this group not employed as nurses were working outside nursing or seeking employment. Altman (1) indicates that young children in a family had a negative influence on a married nurse participating in the labor force. Bognanno (12) also found that the presence (but not the number) of children in the household had an effect on the labor force participation decision. The dip in activity rate1 during the child-rearing ages (reported in table 7) also indicates the influence 20 of children. A Missouri study (13) on inactive registered nurses found that 45.8 percent of the inactive nurses reported family responsibility as the primary reason for choosing to be a home- maker. This, coupled with the reported information that 90.9 per- cent of the inactive nurses are married, and that 84.2 percent of the married nurses have children, further indicates the importance of children to the employment status of nurses. Benham (1,) also found that the presence of children under 5 had an impact on participa- tion rates as computed from 1950 and 1960 census data. Another factor that appears to influence employment is marital status. A single person is less likely to have alternative means of support, hence, the more likely the need for self-reliance and employment. The percentage of nurses employed in 1972 is shown in table 8 by marital status and age. For all age groups, a higher percentage of single nurses are employed than are married nurses. Though some of this difference can be explained by the presence of children, it appears other factors are also involved. Wages are always considered important to the employment sta- tus of nurses. Altman (1) reports that his study and some others suggest that more nurses work when wages are higher; however, he also states that not all researchers share this View, including Bognanno. (12) In general, Altman and others indicate that the single nurse is either not affected or negatively affected by a wage increase; i.e., nursing hours provided by single nurses would stay the same or decrease as wages increase. With the activity rate for licensed single nurses already high (see table 8), a smaller percent- age than their married counterparts are available to reenter the nursing field. The only other way to increase the hours provided by the single nurse is to increase the hours worked per week. With only 4.3 percent of the single nurses working on a regular part-time Table 8.—Percent of employed registered nurses by age group and marital status—1972 (2) Age group Single Married <25 97.2 87.1 25to 29 .. . . . .. . . . 95.3 71.7 3010 34 . . _ . .. . 93.4 60.5 35t039 .. .. _ . . 92.9 61.9 40to44 .. ... _ .. . _ . 91.7 67.0 45 to 49 - . _. . .. . _ 90.1 70.4 50t054 . .. . _ .. . 87.4 69.9 55t059 ..... . . _ 83.5 66.1 6010 64 .. . . .. . .. . 77.8 58.1 65 and over _ . . , . . 48.1 35.0 21 basis (15), any significant change in the number of hours worked per week must come from full—time employees working longer workweeks. Here, the single nurse would have to decide which is of more value—leisure time or money. (16') For the married nurse, the husband’s earnings were found to be a relevant variable. The relationship is that as the husband’s earn- ings increase, the nurse’s participation in the labor force will decrease. (1, 12, 11,) Part-time nurses have increased from 17.9 percent of the active registered nurses’ labor force in 1960 to 29.1 percent in 1973 (see table 9). Of the 238,200 part-time nurses employed in 1972, 67 percent were regular part-time employees, while the remaining 33 percent included both those who were employed part time irregu- larly and those whose part-time work pattern could not be deter- mined. (2, p. 18) Altman in his projections estimated that the proportion of part- time nurses would increase with time (1); however, as shown in table 9, the trend of increasing use of part-time nurses has leveled off since 1970. Bognanno (12) found that the amount of time a married nurse devoted to nursing was not a function of the presence, number, or age of her children nor of her husband’s income. Apparently, little is known as to what factors influence a nurse to work part time. Table 9.—Percent of registered nurses employed part time Part-time Part-time nurses. nurses, Year percent Year percent 1973 _,, . 29.1 1967 A”, ,.,,.,,,. 26.0 1972 ,___ 29.7 1966 n-.. 25.0 1971 . 28.8 1964 ,,. 22.7 1970 ., 28.7 1962 a” 21.3 1969 ._,_ .,_ 27.5 1960 W, 17.9 1968 ,,__,._,_.r,_ , 26.7 1958 _ ,,,,,,,,,,,, 16.5 Source: Computed from American Nurses' Assocnation data. (3) 23 IV. REGISTERED NURSE SUPPLY DATA Introduction This section of the report identifies existing national data sources which are potentially useful in estimating the trend in the supply of registered nurses. Non-national data sources could also be useful in estimating nurse supply, but are too numerous and varied to discuss in this report. One of the most important considerations in the development of any model is to determine if a more sophisticated model will be obstructed by inaccurate or incomplete data. Each additional factor (variable) incorporated in the model definition imposes additional data requirements. These additional requirements may undermine the overall reliability of the data used in the projections. Specifi— cally, the inclusion of more variables in the model increases the likelihood that multiple, perhaps partially inappropriate, data sources will be required. These multiple sources are likely to be inconsistent with respect to both quality and the definition of individual variables. For instance, two different surveys on inactive registered nurse attitudes could very easily define inactive regis- tered nurses differently; e.g., as inactive but licensed nurses or as all trained nurses that are inactive, including those that are unlicensed. Certain characteristics of the available data critically affect projections of nurse supplies. Four particularly important factors are discussed in the following paragraphs. The Critical Variables Used by the Model The model’s ability to predict accurately the aggregate supply of nurses is influenced by the capacity of the critical variables to perform as leading indicators of change in the supply of nurses. Thus, if the variables are inappropriate or if data for the appropri- ate variables are unavailable, the projections will be hampered. Therefore, the critical exogenous variables which “drive” the pro- jections model must themselves manifest certain characteristics to make the modeling exercise worthwhile. These variables must reflect some or all of the following characteristics: (1) they must be well controlled policy variables, (2) they must be currently available lagged variables, or (3) they must be more predictable directly than the endogenous output variables; i.e., the supply projections. 24 The Quality of the Data Used in the Study If the correct variables have been included in the model, but the available data contain significant errors, the exact relationship between supply and the variables used to predict supply will likely be impossible to derive. The Frequency of the Data Collection The time intervals between data observations or surveys deter- mines the kind of projection model that can be developed. If data are collected and new supply estimates are made annually, then it may be inappropriate to include variables to predict some of the supply components when the components can be measured directly. However, if the data collection period is every 10 years, then variables that predict short-term supply behavior as well as long- term supply behavior need to be included in the model. Data Consistency Data drawn from more than one source may embody obvious discrepancies that impede its utility and may lead to faulty conclu— sions. For instance, licensure data collected from the States may differ due to dissimilar operations and definitions and, when com- bined, may not truly represent national data. The remaining part of section IV describes the different national data sources. The discussion will concentrate on the following topics for each data source: (a) What data are available? (b) How often are the data collected? (c) How does the quality of the data affect the potential reliabil- ity of the projection models? Data Sources U.S. Census Data The U.S. census data are collected decennially by the Bureau of the Census. A great variety of information is available from this source.l Examples of RN related census data that have been cited in other studies include: (a) Total number of registered nurses by age. (b) RN income. 1 See U.S. Bureau of the Census, 1.970 Census User’s Guide, Part I. U.S. Government Printing Office, Washington, D.C., 1970. pp. 12—18, for a listing of the specific questions asked in the 1970 census questionnaire. 25 (c) Labor force participation rates by age. (d) Income of RN substitute labor. (e) Income of male heads of household. (f) Number of children under age 5. (g) Number of practicing doctors. (h) Percent of unmarried nurses. (i) Number of inactive RN’s. (j) Supply of practical nurses. In addition, a wealth of other data relevant to projecting nurse , supply is also available. The principal asset of census data is that it constitutes a comprehensive sample which is valid for national aggregates. Consequently, sample information is available to infer characteristics of all nurses, including unlicensed nurses. Also, all of these data can be stratified by State; hence, a great deal of explanatory analysis is possible. Studies using these data include Altman’s, Benham’s, and Yett’s. (I, I4, 17) One suspected problem with the census data is that the number of reported RN’s includes many student nurses. (1) It is also known that there is a tendency for census respondents to overstate their actual level of professional achievement or for the professional level to be misinterpreted during census data acquisition. It is possible that the census estimates of RN’s are from 15 to 25 percent above the true value. In developing a model to predict the supply of RN’s, there are other problems. Principal among these are the following: (1) the 10- year period between consecutive data collection periods is probably too long to provide the desired accuracy; (2) census data are not available for analysis until months after the collection period; and (3) additional information is needed before nurses can be accurately classified. These deficiencies in the census data encouraged the development of a more complete inventory which is now described. The American Nurses’ Association (ANA) Inventories The ANA inventories were taken in 1972, 1966, 1962, 1956—58, 1951, and 1949. The quality of the inventories has improved signifi- cantly over time. One of the principal reasons for establishing the inventories was to provide more complete descriptive information about licensed nurses; i.e., the employment status, the marital status, age, education background, type of position, employer, etc. Because of the regulatory authority of State licensing agencies, the inventories can be regarded as a complete census of licensed RN’s. This contrasts with the less complete public use sample that is based on the decennial census data which rely on statistical inferen- tial procedures to develop national characteristics. A complete list 26 of the data that are available from the 1972 ANA Inventory is shown in table 10. (2) The principal deficiency of the ANA inventories with respect to forecasting nurse supply is the absence of data on certain variables which have a demonstrated causal relationship to nurse attrition, reentry into the labor market, and the degree to which part-time nurses participated in the labor force. Most of this information would directly or indirectly relate to economic characteristics such as the RN’s wage, husband’s income, number of children, and job satisfaction. These variables are good candidates for expanding the inventories to make them suitable for predicting RN supply (for which the inventories are not currently designed). Nurse Career-Patte rn Study The National League of Nursing initiated in 1962 a longitudinal study to obtain definitive information about the biographical char- acteristics of nursing students, their occupational goals, and their reasons for choosing nursing as a career. (5) The purpose of the Nurse Career-Pattern Study is to obtain definitive information about: the characteristics of students in the four types of nursing educational programs; their stated reasons for choosing nursing in the type of nursing preparation; their contribution in the health field after graduation; and, for those students who did not complete the program, their reasons for withdrawal and their activities afterward. The study includes students entering a selected sample of State— approved schools of nursing in the fall of 1962, 1965, and 1967. Data are collected on the students when they enter school, when they Table 10.—List of questions for inventory of registered nurses (2) Questions 1. Sex 2. Marital status 3. Year of birth 4. Social security number 5. Active, inactive, part-time status 6. Average yearly hours of part—time employment 7. Residency location 8. Employment location 9. Employer category (i.e., hospital, public health, etc.) 10. Position held (i.e., instructor, general duty nurse, etc.) 11. Major clinical teaching or practice area (geriatric, medical, etc.) 12. Education preparation—type of nursing school 13. Highest degree held 27 drop out of school, just before graduation, and at intervals after graduation. About 45,000 students are included in the longitudinal study. The type of information obtained from this study is to help explain: (1) the relationship between type of nursing education program and subsequent employment; family responsibility and nurses’ work life; type of nursing education program and occupa- tional goals; and (2) characteristics of students who withdraw from nursing education programs. The sampling methodology used in this study was well planned and the response rate was high. Therefore, it appears that the data from this study are reliable and fill a gap in the previously described data sources. Other National Sources Many organizations collect a variety of data related to nursing. Much of these data are accumulated into the document, Facts About Nursing, 21 series of documents originating in 1935 and Table 11.—Sample of tables contained in Facts About Nursing (3) Sample of tables Nurse Manpower Number of RN's in relation to population Number of employed RN's by field of employment Number of FtN’s by highest educational preparation Number of RN’s by employment status Number of RN's by type position Number of RN's by area of clinical practice Number of RN's by age and sex Licensure for Practice Number of students taking the licensure examination Number of licenses renewed for RN's Number of licenses issued to RN’s previously licensed Number of licenses issued to RN‘s for the first time Number of licenses issued to RN's from foreign countries Nurse Education Admissions of students to initial programs Graduations of students to initial programs Enrollment in initial programs Economic Status of Registered Nurses Starting salary in short~term general hospitals Salaries in hospitals and medical schools Median annual salaries of registered nurses in selected public health nursing services Percent distribution of annual salaries of school nurses 28 published by the American Nurses’ Association. Facts About Nurs- ing includes information on nurse manpower, licensure for practice, nurse education, economic status of registered nurses, and other information related to nursing. Some of the tables in Facts About Nursing that could be useful in projecting the trend in nurse supply are listed in table 11. Caution must be exercised in using much of the data because of inconsisten- cies among States, schools, and other organizations in the reporting of the data. 29 V. EVALUATION OF RELEVANT VARIABLES Introduction Nurse supply models can be classified as one of two types: (1) econometric models that explain nursing labor market interrela- tionships, and (2) projection models that attempt to synthesize the results of the econometric studies and the behavioral relationships among nurse supply related variables. To some extent, the second type of model depends on the former, because the variables that are identified in the econometric models as important determinants of the future supply are the driving variables in the projection models. In this section previous analyses of the determinants of the labor force behavior of registered nurses are discussed, and additional empirical work by RTI using 1970 census data is presented. The purpose of this section is to determine the appropriate variables to include in RTI’s projection models and, secondly, to determine the appropriate variables to include in future ANA inventories or in other censuses of registered nurses. The studies that are reviewed in this section analyze the labor force participation decision of the nursing profession. That is, RTI is analyzing the determinants of movement between the trained' pool of registered nurses and the active stock of nurses. Table 12 shows the percentage distribution of employed regis- tered nurses by both marital status and sex. Married females are an important proportion of the total RN labor force, since more than 98 percent of all employed nurses are females and 68.7 percent of all employed nurses are married. Table 13 shows the activity rates for both single licensed and married licensed nurses, by year, for 3 inventory years: 1962, 1966, and 1972. The table shows a general trend that has been occurring in the Nation over the past several years; i.e., that the activity rate of married nurses is increasing in general, but increasing particularly in the middle-age range. Clearly, one of the most important problems in making accurate projections of the number of active registered nurses is the difficulty of predicting those factors which will influence activity rates. Although in the projections framework of this study, RTI does not analyze the number of hours of nursing services supplied by those nurses who are working, it is important to recognize the significance of that variable as an additional factor in the overall level of nursing services supplied. Consequently, this section pre- sents an analysis of the determinants of both the working propor- 30 Table 12.—Percentage distribution of employed registered nurses by marital status and by sex Year Classification 1962 1966 1972 Marital status: Married .. . . . . . 61.0 63.5 68.7 Single .. . .. . . , . .. 25,6 22.4 18.4 Other' . .. .. 11.0 11.6 12.0 Not reported .. . . . . 2.4 2.5 0.9 Sex: Female .. .. . . . . 98.9 98.9 98.5 Male .. .- .. _ .. 0.9 1.1 1.4 Not reported ....... .. -. 0.1 2 0.0 0.1 'Widowed. divorced, or separated. -' Less than 0.05. Source: Computed from American Nurses' Assomation data. (2) Table 13.—Registered nurse activity rates Age range Marital status Year 320 30—39 40—49 50—59 260 Single _ , 1962 93.5 92.6 91.7 89.2 68.7 1966 95.5 92.9 91.3 87.9 66.7 1972 96.2 93.2 90.9 85.5 60.7 Married ............ 1962 59.0 50.6 61.6 65.3 48.7 1966 64.5 52.7 63.5 67.7 52.1 1972 75.6 61.2 68.7 68.4 49.5 Source: Computed from American Nurses' Assomation data. (2) tion of the trained stock of professional nurses and the number of hours supplied by those nurses who are working. In the following section a brief description is given of the theoreti- cal background that has been developed in previous economic analyses of the determinants of labor force behavior. The next section is a brief review of three previous studies which have looked at either the labor force participation rate of nurses or number of hours supplied or both. These studies were conducted by Benham, Bognanno, and Altman. Then a general description of the data which were used in the present study is given. Data consist largely of that of the US. census. Sections discuss the empirical nurse supply models that were developed for RTI’s analysis of labor force behavioral determinants and give the empirical results of RTI’s 31 study and compares those results with previous analyses. Finally, the results of the RTI analyses, as well as the results of previous studies, are summarized and their relevance to the construction of an appropriate projections model is indicated. Theoretical Background In this analysis it is hypothesized that separate theories of market work behavior exist for the single and the married profes— sional nurse. Both theories are extensions of the standard market work—leisure choice model. (1.9, 20) The basic hypotheses of the orthodox model can be stated as follows. The nurse will seek to maximize her personal welfare for a given set of tastes, Z, by maximizing her consumption of leisure hours, L, and market- produced goods and services, X.l However, this maximization is subject to two constraints: (1) the sum of the hours, HR, spent working and the hours, LH, spent as leisure must equal the fixed number of hours, TH, available each time period and (2) disregarding dissavings, the money spent on market goods cannot exceed the sum of current earnings (wage income), WRN, and nonearnings income, Y.2 Under the assumption that the second constraint is just met, the primary conditions for the nurse’s simultaneous optimization of her rates of market goods and leisure consumption yield predicted equilibrium values. The secondary conditions for this optimization take the analysis one step further by predicting the direction of the effects of the independent variables which affect the equilibrium values. (21) The most important of these results, for the purposes of this study, is that this theory provides a sound foundation on which to base our empirical estimation equations. In particular, the most critical variable in this study, viz, the number of hours of nursing services offered, is one of the predicted equilibrium values that emerges from the foregoing analysis. The rate at which nursing ‘ Leisure is interpreted here as all nonmarket uses of time. Hence, the term should not connote indolence so much as it should the use of time in home production activities, such as child-rearing, homemaking, etc. 2 The first constraint may be stated as: TH : HR + LB (0 Wage income is the product of the wage rate, WRN, and the number of hours worked, HR. The total expenditure on market goods is the product of their price, P, and the quantity, X. Therefore, the second constraint may be stated as: PX S WRN~ HR + Y (ii) 32 services are provided will, given a fixed set of tastes, Z, be a function of the nurse’s wage rate, the price of market goods, and her nonearnings income. HR = H(WRN,P,Y,Z) (1) HR = Number of hours spent working as a registered nurse per period. WRN = Hourly wage rate for registered nurses. P Average price of market goods. Y N onearnings income. Z = Taste shift variables. The major predictions of the orthodox model of work-leisure choice can be summarized under the assumption that tastes are fixed. First, as the registered nurse’s wage rate rises, so does the price of leisure. This “price effect” will have a tendency to cause the nurse to substitute away from leisure and toward purchased goods. On the other hand, as the nurse’s wage rate rises, so does her income, at any given number of hours worked. This “income effect” would tend to reduce the number of hours she works, because it is expected that the consumption of all goods, including leisure, would increase with income. The net increment in hours worked at different wage rates may then be positive or negative depending on the extent to which either the “price effect” or the “income effect” dominates. The phenomenon of the “backward bending” labor supply curve—which implies that at relatively low wage rates an increase in pay will increase hours worked, and at relatively high wage rates an increase will decrease hours worked—is in fact attributed to differences in the relative strength of these two effects over different wage ranges. The second major prediction of the theory is that an increase in the average price of market goods will decrease the number of hours worked, because it reduces the relative price of leisure. Finally, an increase in nonearnings income is expected to increase the consumption of both leisure and market goods resulting in a smaller number of hours worked. The proper statement of an operational model which can be used to predict whether a registered nurse will participate in the labor force at all (and, if so, how much) depends basically on one’s insight regarding the taste shift variables, Z, that ought to be specified. Before discussing the empirical nurse labor supply models and supporting data that were employed in the present study, a brief description and critique of the major existing statistical studies of nurse labor supply relationships is presented. 33 Previous Analyses This section briefly reviews three major statistical studies of the labor force response of professional nurses. These studies, like RTI’s empirical work reported in this section, are not actually parts of projections models, but rather they are analyses of the determi- nants of whether a professional nurse will participate in a labor force and the extent to which she will participate given that the nurse is already in the labor force. Three major existing studies are those reported by Benham (11;, 22), Bognanno (12, 19), and Altman. (1) The Benham Model Benham (14) developed an econometric model of the labor market for registered nurses. The model is an explanatory model which analyzes the critical variables that influence the number of em- ployed registered nurses and the extent to which they participate in the labor market. The model incorporates three simultaneous equations which de- fine the various (both endogenous or predicted and exogenous or determinant) determining movements in the three endogenous variables. These three variables are: (1) RN wage rates; (2) RN labor force participation rates; and (3) the stock of RN’s. The equations which describe movements in these three variables are defined as the demand equation, the labor force participation equation, and the geographical distribution (stock) equation, respec- tively. Benham first developed the appropriate theoretical determi- nants of each of the endogenous variables. He then defined empiri- cal counterparts of those determinant (independent) variables and endogenous variables; finally, he econometrically estimated the coefficients of these equations via three-stage least squares. In the theoretical demand equation, it was hypothesized that as either the stock of nurses or the labor force participation rate increase (both resulting in an increased supply of active RN’s) the wages paid to RN’s would fall. Second, it was hypothesized that as the wages of nurse substitutes, e.g., hospital attendants or practical nurses, fall, so do RN wages. Finally, based on previous positive estimates of the income elasticity of demand for medical care, it is argued that as personal income rises, so does the demand for RN’s, and, hence, wages rise. Since the labor force participation rate of RN’s times the stock of RN’s equals the total supply of RN’s, the labor force participation equation and the geographic distribution (stock) equation are sim- 34 ply the two components of supply. Hence, Benham argues that both of these variables—i.e., the participation rate and the stock—will rise as wages rise. As the husband’s income rises, the labor force participation rate is expected to fall because of the family income effect on the wife’s demand for leisure (or household production). Similarly, as the number of small children in the household rises, the labor force participation rate is expected to fall. The per capita stock of nurses in a given geographic region is assumed to be positively associated with the regional income level. The rationale behind this somewhat unsubstantiated argument is that: (1) married nurses’ husbands tend to work in higher income levels than other women’s husbands, or that (2) unmarried nurses tend to congregate in higher income regions because “a more affluent group of potential husbands live there.” (11;, p. 247) Finally, the current stock of nurses is assumed positively related to the number of nursing school graduates, lagged an arbitrary 10 years. The empirical counterparts to each of the above mentioned theoretical variables were drawn from both 1950 and 1960 census cross—sectional (State) data; hence, State averages were the units of observation. Model estimates were made twice, once for each census year. The validity of the above relationships was confirmed by the data in that all predicted signs were as hypothesized. However, the magnitude of several of the regression coefficients varied signifi- cantly between census years. The results of these estimates had several interesting implica- tions for nurse supply projections models and nurse-related data collection efforts. First, the analysis indicated that the wage for nurse substitutes significantly affected the wages and the supply (both the stock and the participation) of RN’s. Specifically, as the wages of nurse substitutes increase, so do the wages of RN’s. Second, the RN labor force participation rate falls, but the stock of RN’s rises as the husbands’ incomes rise. Third, the presence of small children significantly lower RN participation rates. Finally, it appears that nurses are increasingly responsive over time to higher wages; i.e., they migrate readily. This suggests local nurse short- ages are not likely to be ameliorated by providing more local training facilities. The Bognanno Model The most important difference between Bognanno’s work (12, 1.9) and that done by Benham (11,, 22), is that Bognanno assumes that individual nurses are facing a perfectly elastic demand schedule. This assumption is valid because of the way in which his data were gathered and analyzed. His observations were based on a random 35 sample of married nurses in the State of Iowa during 1968. His sample excluded nurses who were more than 65 years of age or those who were members of religious orders. Because the observa- tions were taken on individuals and their personal characteristics, it was valid to assume that the wage rate facing each individual was a constant. Bognanno analyzed several different models, but they were all variants of the two basic models that are pursued in this chapter as well. First, he attempts to explain differences in labor force partici- pation rates (the percentage of nurses who are working) in response to several variables. Secondly, he tries to project or estimate the numbers of hours of labor services provided by those nurses who are working. Particularly, Bognanno hypothesizes: (1) that both labor force participation rates and the number of hours worked are functions of the nurse’s nonearnings income or of her husband’s income; (2) that labor supply is a function of the nurse’s wage rate; (3) that it is a function of the number of children in the household and the presence or absence of preschool children; and (4) that it is a function of age, as well as the status of home ownership; i.e., renting a dwelling versus present or prospective home ownership. The hypothesized responses of labor force participation to each of these variables are as follows. Based on the classical labor-leisure choice model, Bognanno hypothesizes that an increase in the husbands’ earnings will decrease the female’s extent of labor force participation, that an increase in her wage will increase labor force participation, that large numbers of children and the presence of preschool children will reduce her labor force participation, and that her labor force activity will follow the ordinary age-labor force participation profile. That is, in early years the labor force partici- pation rate will be high or the numbers of hours provided will be high, that in the early middle ages, labor force participation rate will fall and then rise again in the late middle age-years and that, nearing retirement, the labor force participation rate will fall again. The inclusion of classification variables to indicate whether or not the nurse owns or rents or is buying her dwelling was intended to standardize for differences in taste which result from home owner- ship. One of the most significant results that Bognanno found was that the probability of labor force participation by married nurses was not significantly related to her wage rate. This constrasts with the results found by Benham. The second significant result was that the wage rate elicited a significant response in hours worked among those nurses who are active in the labor market. Another important result was that the husband’s earnings was a very important variable in determining both the labor force participation rate and the number of hours worked. But Bognanno’s study 36 determined that the magnitude of the response to husband’s earn- ings was much larger for labor force participation rate than for the number of hours supplied by those in the labor force. Also, the number of children in the household and the presence of preschool children were both quite important in reducing the number of hours supplied by the married nurses. The results that are presented in this section are similar not only in specification but also in response magnitudes to those given in the RTI study. This is because both studies involved an analysis of individual nurses; however, the data sources used in the two studies were significantly different. RTI’s data source was the 1970 census information, whereas Bognanno’s were developed in an independ- ent State survey. RTI’s results Will be compared with Bognanno’s, where magnitudes of the estimated parameters of RTI’s models are presented. The Altman Model Altman attempted to derive a nursing labor supply function using statistical analysis. (1, pp. 117—133) Altman’s data relied primarily on information obtained in a 1962 ANA inventory of professional nurses and on associated data from the 1960 census of population. Unlike Bognanno or Benham, he relied on a combination of data sources rather than strictly on one or another to develop estimates of his labor supply function. Data on the economic variables were mainly derived from the census, because the ANA inventory sup- plies very little information on several of the variables. Like Benham’s model, Altman’s model did involve cross-sectional analy- sis using State aggregates. Therefore, he used a two-stage analysis in which he estimated both the supply and demand functions. The estimated wage rate which came from the demand model was substituted into the second stage analysis of the number of hours of nursing service provided or supplied. Altman’s model provided some very tenable results. His estimates of the elasticity of labor supply with respect to changes in the wage rate developed at the means were 0.89 for married nurses, and 0.64 for all nurses. Although his analysis is based on sketchy and piecemeal data, the results are, in general, consistent with those developed by both Benham and Bognanno. In the following sections, the analysis conducted at RTI, which is a logical extension of the three studies reported here, is discussed. The analysis incorporates data on individuals as noted above and follows the spirit of Bognanno’s model. General Data Description The data used in this statistical analysis consisted of disaggre- 37 gated data about registered nurses from the 1970 Public Use Samples available through the Bureau of the Census. (23) Specifi- cally, the data comprised several parameters concerning the regis- tered nurse’s current work status, family structure, family finances, location, race, and other related variables. Two of the one-in-a- thousand Public Use Samples were accessed. These samples were derived by the Bureau of Census by randomly selecting the appro- priate number of observations from two basic US. data sets: (1) a 5 percent State sample and (2) a 15 percent sample of all individuals in the Nation. Both of these samples were developed from two separately administered long-form questionnaires used in the 1970 Census of Population and Housing. (24, pp. 12—18) Therefore, the two one-in-a-thousand samples from which the data on registered nurses were obtained consisted of roughly 1/50 and 1/150 of all basic records in the 5 percent and in the 15 percent samples, respectively. In either case the total number of observations included approxi- mately 280,000 persons. (23, pp. 2—11) From each of these two basic data sets three alternative one-in-a-thousand Public Use Samples have been created: (1) State, (2) county, and (3) neighborhood characteristics Public Use Samples. The first type was selected because of interest in regional differences where regional classifica- tions consisted of State groupings. In any event, the inferences drawn from these three alternative subsamples should not differ substantively, since they are randomly derived from the same basic data set for either the 5 or 15 percent sample. In selecting individual observations from the Public Use Samples, an attempt was made to eliminate all observations for which any of the regression variables were obscurely defined or altogether unde— fined. For example, if the nurse’s earnings were not reported or if her reported earnings exceeded the total reported income for her household, that observation was deleted. Several other computa- tional conditions were also imposed on the other variables as well, in effect, to remove partial nonresponses or inaccurate responses from the data on which the analysis performed. However, it should be recognized that despite these efforts, the census data on regis- tered nurses contain several errors and deficiencies.3 Some at- tempts are currently underway to partially correct some of these problems“ Nonetheless, it is felt that the data are sufficiently valid 3 See page 29 for explanation of error. 4 Several months after this analysis was completed, preliminary results of a “post-enumeration" Census survey were made available to the Division of Nurs— ing. This survey reviewed the data on those respondents who were classified as registered nurses in the 1970 census. The results indicated that roughly 22 percent of those individuals were not registered nurses. They may have been licensed practical nurses, or other misclassifled respondents. 38 to regard the analysis presented below as an important increment to the knowledge about nurse labor supply. Empirical Nurse Supply Models The empirical analysis of the extent to which nurses supply labor to the market was conducted separately for single and married nurses. The basic reason for this dichotomization is that the taste shift variables referred to in equation 1 must be fundamentally different for married as compared to single professional nurses. Most importantly, these variables would differ with respect to the presence of a husband and children in the household. Nonetheless, the empirical models posited here derive from the same basic theoretical model specified in equation 1; the appeal to that model is used to justify the inclusion of wage and income variables in the estimation model. Two separate measures of the extent to which professional nurses provide labor to the market were employed, resulting in two distinct empirical models. The first measure was employment on either a part-time or full-time basis, implying a binary dependent variable. This measure resulted in a linear probability model which provided estimates of the extent to which the various exogenous factors influence the probability that a professional nurse will participate in the labor force. The second measure of labor services was an estimate of the number of hours worked per week, implying a continuous dependent variable. This variable implied a conditional labor supply model—the inclusion of individual nurses in the analy- sis was conditional on their current participation in the labor force. Single Nurses The linear probability model (Model 1) and the conditional labor supply model (Model 2) that were developed for single professional nurses are given in equations 2 and 3. Model 1 4 LFPS = a10 + all Y + al2 WRN + 2 b” AGE. (2) i=1 i + {V N H c” RACE». + d1 HOME + 2 e” REGION. 1—1 +# Model 2 39 4 HR, = a20 + .2121 Y + a22 WRN + 2 b2, AGE, (3) + v.1 II + 3: LFPs HR. WRN AGE, RACE, HOME RE GION, a,b,c,d,e u 1% N j=l X (:2, RACE, + d2 HOME + 2 e2i REGION, 1:1 Binary dependent variable whose value is unity if the observed single nurse works at all, and zero otherwise. Number of hours the single nurse works each week estimated as the mean of the range of indicated hours worked last week. (24, p. 17) All nonearnings income in 1969, computed as the difference between total household income and the nurse’s earnings income. (24, p. 18) Registered nurse’s expected wage rate in her State, estimated by the median income of all registered nurses who worked full time during 1969 in the State where the nurse is located. (25, table 176') Age group classification variables. AGE1 is unity if the nurse’s age is 30—39 years, and zero otherwise. AGEZ,4 are similarly defined for the age ranges 40— 49, 50—59, and 60 or more years, respectively. The base classification group is 29 or fewer years of age. Race classification variables. RACEx is unity if the nurse is neither white nor black, and zero other- wise. RACE.2 is similarly defined for blacks. The base classification group is white. Home ownership classification variable. HOME is unity if the nurse is not a renter. The base classifi- cation group consists of rent-paying nurses. Regional classification variable. REGION, is unity if the nurse is located in Region I, and zero other- wise. REGION2_8 are similarly defined for Regions II—VIII. The base classification group lives in Re- gion IX. (See table 14 for a listing of the State groupings in each region.) Regression coefficients. Error term. The variable which is defined above as nonearnings income, Y, can be regarded as the empirical counterpart to the same variable in the theoretical equation 1. This variable would normally measure 40 the level of transfer payments—social security, unemployment benefits, etc.~plus income from financial aSsets for the single nurse. It is argued here, based on the theoretical analysis in the section on the Bognanno model, that the partial effect of nonearn- ings income will be negatively related to the supply of labor. Table 14.—Regional classifications Reg ion States I Pacific . ., _,, , Alaska, California, Hawaii, Oregon, and Washington , , , , . Arizona, Colorado. Idaho, Montana, Nevada, New Mexico, Utah, and Wyoming III West North Central . _, , Iowa, Kansas, Minnesota, Missouri, Ne- braska. North Dakota, and South Dakota , , Illinois, Indiana, Michigan, Ohio, and Wis- ” Mountain IV East North Central consm V West South Central ,, ,, _ Arkansas, Louisiana, Oklahoma. and Texas VI East South Central , , Alabama. Kentucky, Mississippi. and Ten- nessee VII South Atlantic , . , _ ,_ , Delaware. District of Columbia, Florida, Georgia. Maryland, North Carolina, South Carolina, Virginia, and West Virginia VIII Middle Atlantic . , New Jersey, New York, and Pennsylvania IX New England , , , , ,, , Connecticut, Maine, Massachusetts, New Hampshire, Rhode Island, and Vermont ' Source: Facts About Nursmg. (3, 1972—73, p. 20) As mentioned above, theexpected effect of increased wages, WRN, is uncertain. Only if the “substitution effect” dominates will the probability of labor force participation increase or the number of hours supplied increase. The other price variable in the theoretical model of equation 1 is the average price, P, of purchased goods. Under ideal conditions every estimate of the wage rate would be deflated by the price of the average market basket in each State. However, the simpler approach of assuming minimal cross sectional variation in the average price of market goods was employed. Therefore, it is implicitly assumed that the expected wage variable, WRN, is actually the ratio of the price of leisure to the price of all other goods, where the latter is assumed constant, cross sectionally. The remainder of the variables in the models of equations 1 and 2 are classification variables for which theoretical justification is provided by the taste shift variables, Z, in equation 1. The dummy variables technique was used to incorporate these variables. (26, p. 221) This approach simply involves the specification of several 41 binary variables to identify the partial effect of attributes on the individual’s labor force behavior. The regression coefficients—the b, c, d, and e coefficients of equations 2 and 3—are interpreted as the difference in the magnitude of the dependent vam'able—LFPS or HR,—that one could expect to observe between the base classifica- tion group and the group which shares the attribute identified by a particular dummy variable. The taste shift variables that were hypothesized in the single nurse models were age, race, home ownership, and regional classifi— cation variables. It is argued here that each of these factors could alone account for differences in the preferences of individual nurses. It was expected that both the probability of labor force participation and the number of hours worked will follow the traditional female labor force participation with extensive participa- tion in young age groups, slightly falling participation in the early middle-age groups, and rising again in the late middle—age groups, before falling off substantially among those in their sixties and seventies. Similarly, cultural differences may be expected in labor force behavior as between ethnic groups. The base classification group comprises whites, whose labor force participation is compared with blacks and all other ethnic groups combined. The home ownership classification variable was suggested by Bognanno. (12, p. 81,) Although it cannot be argued that home ownership necessarily implies either a positive or a negative effect on labor force participation, its inclusion is theoretically justified. On the one hand, it can be argued that families who own their own homes receive a substantial flow of nonmarket rental income from their eq uity, and hence, that their labor supply would be lower than renters. This is an especially important consideration in an infla- tionary economy in which home mortgages are denominated in nominal (not real) dollars. In these cases, home ownership not only implies additional income from the real value of the equity in the home, but also implies windfall gains in the real value of the equity itself. On the other hand, however, there is a contrary tendency, especially among young homeowners, for the female to work more because of the debt burden imposed by home ownership. On balance one cannot predict which of these effects will dominate. Finally, the regional classification variable is included for several reasons. First, these dummy variables would standardize for in- terregional cost of living differences. Hence, the wage ratio can be regarded as an intraregional relative wage, since the denominator of the ratio would have been adjusted for regional effects. Also, the regional variables adjust for differences in the degree of urbaniza- 42 tion and the accessibility Of potential employers, for differences in labor force mobility, for differences in hospital and clinical employ— ment practices, and monopsony market power, and for cultural differences. Married Nurses The linear probability model (Model 3) and the conditional labor supply model (Model 4) that were developed for married professional nurses are given in equations 4 and 5. Model 3 LFP333 _ a33 + a33 Y + a32 WRN + a3. KID<18 (4) + a3. PRESCHOOL + 23,103., AGE + \ c33 RACE3 i=1 i=1 8 + d3 HOME + 2 e33 REGION-3 + 3; i=1 Model 4 HRm = am + a33 Y + a3 WRN + 23 KID<18 (5) + a3, PRESCHOOL + \ b33 AGE3 + \ C33 RACE3 j=1 3:1 X + d, HOME + )3 e33 REGION, + ,1 i=1 LFP333 Binary dependent variable whose value is unity if the observed married nurse works, and zero other- wise. HR333 Number of hours the married nurse works each week, estimated as the mean of the range of indi- cated hours worked last week, if the nurse worked. (24, p. 17) KID<18 Number of children below 18 years of age residing in the married nurse’s household. PRE— Preschooler classification variable. PRESCHOOL is SCHOOL unity if the married nurse has any children under 6 years of age, and zero otherwise. All other variables are as defined in equations 2 and 3. The two models that are hypothesized here for married profes- sional nurses account for the major differences in tastes that one can expect between married and single nurses. These differences mainly relate to the interpretation of the nurse’s nonearnings income and t0 the number and presence of children in her house- 43 hold. Otherwise, the married nurse models are very similar in structure to those hypothesized for single nurses. The nonearnings variable, Y, as computed in the empirical analy- sis must, for married nurses, be interpreted as consisting mainly of the income earned by the nurse’s husband. As the husband’s contribution to family income increases, the nurse’s provision of labor services is expected to decrease, resulting in her increased consumption of leisure (or equivalently nonmarket goods). For young nurses the “leisure” may involve home production activities (as, for example, child-rearing) for which purchased substitutes are either very imperfect (as, for example, day nurseries) or relatively inaccessible. In the longer run, however, it is expected that an increase in family income through an increase in the husband’s income would result in more leisure for the nurse, i.e., fewer hours of work at home or in the market. This increased leisure would be generated by using fewer home—produced goods and by using fewer hours of housework to produce the remaining home-produced goods. These two options are well—illustrated by two examples: (1) more meals purchased away from home reduces the level of meal produc- tion at home, and (2) more TV dinners and pre-prepared frozen foods reduces the time requirement for the remaining meals that are produced at home. Another important factor that discourages the nurse’s market work when her husband’s income increases is progressive income tax schedules—both Federal and State. Since nurses are often secondary workers in the family, their potential income is viewed as a supplement to their husband’s income. Therefore, in making the decision regarding work, the family will ordinarily compute the after—income-tax wage that the nurse is earning. Obviously under a progressive income tax system the nurse’s after-tax wage will fall as her husband’s income increases. Hence, in addition to the classic income effects associated with increases in the husband’s income, the income tax laws create an additional work disincentive by reducing the after-tax wage of the nurse as a secondary worker. The number, KID<18, of children in the nurse’s household who are under 18 years of age was introduced into the model to give additional control over variation in tastes for home-produced goods. The argument is that the demand for the nurse’s time in the household increases with the number of children. It is also some- times argued that, as the number of children in the household increases, older children can substitute for the mother in producing home goods, so that once a certain family size is reached, the nurse’s supply of market labor increases. (12, p. 81,, 27, pp. 92—99) Hence, there is some justification for including both a linear and quadratic term in the variable KID< 18. RTI included a quadratic 44 term in some preliminary analysis of one of the subsamples and found that only the linear term was important. KID<18 was also converted to a classification variable using the dummy variables technique. The observed pattern of coefficients in that analysis also supported the conclusion that only the linear term was important. Hence, it was concluded that the appropriate specification should include only the linear term in the variable KID<18. Because our preliminary analysis was conducted using only one of our subsam- ples, potential specification bias was not likely to have been as important as it might otherwise have been as a result of this exploratory analysis. Finally, a classification variable, PRESCHOOL, was included to account for the presence of preschool children in the household. If the nurse has preschool children one would expect her tastes and her value in the production of household services to be significantly different from mothers with older children and from married nurses who have no children. In particular, we would expect the presence of preschoolers in the household to reduce the level of nursing services offered by the married nurse. Empirical Results In addition to the specific functional forms reported in the previous section, RTI investigated several other specifications, e.g., double logarithmic and semilogarithmic models, with transformed values of the same variables in the argument list of each model. The results in all cases were either quite similar to those reported here or statistically much less valid as judged by the significance of computed F-statistic of the overall regressions. In no case were the F-values of alternative specifications more significant than those of the linear specification. Therefore, this section reports only the results of the linear models that were posited above. The tables in this section report three sets of regression coeffi- cients for each of the four models identified in the previous section. The first set corresponds to the observations from the one-in-a— thousand Public Use Samples based on the 5 percent census sample (Sample I); the second set corresponds with that based on the 15 percent census sample (Sample II); and the final set is based on the combined observations (Pooled Sample) of the first two. Each of the four models were tested to determine whether it was statistically valid to pool the two sets of observations. Constraining the coeffi- cients in each of the two samples to be equal allows the computa- tion of a test statistic based on the difference in the regression sum of squares in the regressions based on Samples I and II and on the Pooled Sample. (26, pp. 136—7) This test statistic, referred to as the 45 Chow Statistic at the bottom of each table, is compared against tabulated values on the F-distribution to determine the validity of the constraint. For Model 1 the value of the test statistic was 1.64. At the .025 significance level one cannot reject the hypothesis that pooling the two samples is a valid approach. For all of the other three models the evidence for pooling is even more conclusive—one cannot reject the pooling hypothesis at the .10 level in any case. As a result, the discussion and interpretation of the empirical results will focus on the reported coefficients for the pooled samples as shown in the right column of each table. However, it is instructive to observe the disparity between the coefficients on some variables in the subsample results. The extent of the differences gives some insight regarding the extent of “averaging” represented by the pooled sample results. It should be noted that all the estimates presented in the following tables are derived by ordinary least squared (OLS) regres- sion analysis. In the examination of the determinants of the labor force participation rate, it is quite apparent that OLS estimates are inefficient, because the binary dependent variable—the nurse does or does not work—implies an irregular error structure commonly referred to as heteroscedastic. Consequently, some authors apply generalized least squared (GLS) techniques to alleviate this prob- lem. (12, p. 85) However, since Goldfeld and Quandt (28, p. 134), using Monte Carlo Analysis, have found that “. .. if anything, OLS is somewhat to be favored” over GLS, we have not attempted to use GLS techniques. As a consequence, the inferences drawn from the OLS calculated statistics are more conservative than under GLS, since the latter doesn’t change the OLS estimated coefficients but only increases their statistical significance. One final general comment regarding the interpretation of the effects of classification variables such as age, race, region, etc. All of the regression results presented here show the partial effects of each classification variable in explaining variation in the dependent variable which represents the extent of labor force activity. These regression results are, in general, not likely to coincide with the inferences one would draw from simple frequency tabulations by classification group, e.g., age, because the effects of other variables are not held constant in those simple tabulations. This considera- tion accounts for what may initially seem apparent discrepancies in the analysis of classification effects. Single Nurses Tables 15 and 16 show the results of the analysis of labor force participation determinants and of conditional labor supply rela— 46 Table 15.—Labor force participation determinants for single professional nurses, 1970 1 Pooled Independent Sample Sample variables | || sample Y - - , —0.0048 —0.0035 2—0.0054 (1.17) (0.82) (1.82) WRN ,,,,,,, - 0.0317 A0.1066 0.0078 (0.52) (1.32) (0.16) Age ranges (base: s29 years) 30—39 . —0.0428 0.0908 0.0080 (0.62) (0.96) (0.14) 40—49 0.0739 70.1169 —0.0056 (0.88) (1.04) (0.08) 50—59 _. .. —0.1069 —0.1737 3—0.1717 (1.14) (1.62) (2.46) 260 __ 4—0.6500 4—0.3420 '—0.4749 (7.67) (3.84) (7.96) Race (base: while) Other _ 0.2069 70.2751 ~0.0001 (1.29) (1.27) (0.00) Black —0.1870 2 —0.1732 3 —0.1946 (1.31) (1.72) (2.49) Rent (base: renters) Homeowner .. , ~0.0267 0.0263 0.0228 (0.39) (0.35) (0.46) Region (base: region IX) I , ,- - 0.1163 2—0.2872 ~0.0006 (0.97) (1.84) (0.01) II - _ - -, —0.0030 —0.1554 0.0223 (0.03) (1.31) (0.31) III -- _ , “0.2701 3—0.3863 0.0248 (2.28) (2.34) (0.25) IV , , -- —O.1146 -" —0.7163 —0.2786 (0.50) (2.54) (1.56) V - _ . - - 0.1925 —0.3083 0.0901 (1.08) (1.26) (0.62) VI -- _ _ - 0.0741 270.2537 0.0001 (0.76) (1.85) (0.00) VII -. . 0.0313 —0.2940 —0.0356 (0.21) (1.57) (0.31) VIII , 0.1737 2 —0.3889 —0.0574 (1.08) (1.65) (0.43) Intercept .. _ 0.6723 1.9150 “0.9046 (1.36) (2.93) (2.31) See Ioolnotes at end 01 table 47 Table 15.—Labor force participation determinants for single professional nurses, 1970 1—continued Independent Sample Sample Pooled variables I ll sample Fl2 W .37 .23 _ .23 F-statistic _ W W . 44.96 4 2.27 t 4.97 Chow statistic W W, _ _ 1.64 Number of observations .. 159. 148. 307. ‘All parentheSIzed values below the reported regressmn coeffICIents are the absolute values 01 the t-statistics. The reported Significance levels are assomated With either a one-tailed or a two-tailed t-test depending on whether a Sign was hypothesued. 2 Significant at the .10 level. -‘ Significant at the .05 level. ‘Signiticant at the .01 level. Table 16.—Conditional labor supply relations for single professional nurses, 19701 Independent Sample Sample Pooled variables I II sample Y —0.0619 0.1399 0.0146 (0.57) (1.38) (0.20) WRN .................... 0.3269 —5.1923 —1.8020 (0.20) (2.62) (1.49) Age Ranges (base: s29 years) 30—39 W 0.0697 0.8030 0.5340 (0.04) (0.38) (0.39) 40—49 W . W . W 3.2704 4,1828 24.1758 (1.52) (1.36) (2.45) 50—59 W 2.3821 0.4337 1.0712 (0.93) (0,16) (0.61) 260 W 27.4867 1.6246 2.5372 (2.11) (0.69) (1.32) Ftace (base: white) Other W .W .W —2.0885 1.8108 —0.5789 (0.53) (0.23) (0.17) Black W —0.6036 1.3018 —O.9812 (0.14) (0.53) (0.48) Rent (base: renters) Homeowner W . W . W 0.8256 2 ~3.8697 —1.6156 (0.45) (2.09) (1.29) Region (base: region IX) l - 74.8569 —2.6664 ~3.6934 (1.45) (0.69) (1.50) See lootnotes at end of table 48 Table 16.—Conditional labor supply relations for single professional nurses, 1970 1—continued Independent Sample Sample Pooled variables I ll ‘ sample II .. 0.5233 —1.4080 —0.0332 (0.21) (0.50) (0.02) III 0.1780 —6.3724 —0.9942 (0.06) (1.54) (0.41) IV .. - , —2.1280 ”—38.6806 343.4831 (0.34) (4.48) (2.66) V , .. ,. —3.4125 —9.0494 —5.4134 (0.74) (1.53) (1.53) VI - —3.4049 2—8.2818 2—5.0081 (1.32) (2.30) (2.44) VII _ _ . _ __ . _ 3.8337 —6.4889 —0.7492 (0.92) (1.46) (0.26) W“ —5.0948 —6.2810 —4.0804 (1.17) (1.10) (1.19) Intercept ._ 237.7390 “80.3585 3 54.1525 (2.81) (5.02) (5.53) R2 . .16 .29 .13 F-statistic .. .. . __ 1.24 32.23 21.93 ChowStatistic _ _ _ 1.38 Number of observations .. 131. 112. 243. ‘ All parenthesrzed values below the reported regreSSIon coefficrents are the absolute values of the t-slatistics. The reported Significance levels are assocmted With either a one-tailed or a two-tailed Hest depending on whether a Sign was hypothesnzed. 3 Significant at the 05 level "Significant at the .01 level. tions, respectively, for single nurses. The variables that appear to be most important in influencing the labor force participation rates of single nurses include the nonearnings income variable, the age classification variable, and the ethnic classification variable. Specifi- cally, the estimated coefficient on the nonearnings income variable for single nurses was —.0054. This coefficient was significant at the 10 percent level. Table 17 shows the corresponding values of the elasticity associated with that coefficient. The value of that elastic- ity is — .038. This value implies that a 1 percent increase in nonearn- ings income for a single professional nurse will result in a .038 percent decrease in the single nurse’s labor force participation rate. In the conditional labor supply model shown in table 16, the coefficient on nonearnings income was not significantly different from zero. In neither of the estimation models for single nurses was the estimated coefficient on the wage variable significant. In fact, in the 49 Table 17.—Computed elasticities of labor force participation Married nurse models Single nurse models Independent variables LFP Hours LFP Hours Y($1000/yr) w"... . 2e0.236 2—0.087 3—0.038 0.002 WRN($1000/yr) , . —0.056 0.171 0.066 —0.321 K|D<18 , —0.027 2~O.058 Preschoolers (base: no preschoolers) Preschool ._ W--. H-.. ._ _, “—0.164 2—O.037 Age ranges (base: s 29 yrs.) 30—39 .-,- 0.021 0.000 0.002 0.002 40—49 _ - 10.027 0.002 ~0.001 40.012 50—59 .-._ , . —0.003 0.006 '—0.020 0.002 260 - n-.. W--- 2~0.032 —0.003 270.082 0.005 Rent (base: renters) Homeowner , ,_. —0.002 —0.029 0.013 —0.019 'AII elasticmes are computed at the means. The Significance levels are those assomated With the regreSSion coellncuents. “ Significant at the .01 level. “ Significant at the .10 Level. 'Significant at the .05 level. conditional labor supply model, the estimated coefficient had the wrong sign. This evidence indicates that the elasticity of labor supply response among single nurses in a response to an increase in the wage rate is very small or nonexistent. Rather, alternative taste variables are more influential in determining whether or not the nurse participates in the labor force; and if the nurse partici- pates, to what extent. Compared with single nurses under 30 years of age, those in the age range of 30—39 and 40—49 do not have significantly different labor force participation rates. On the other hand, those in the 50—59-year-old age group and those older than 59 both have significantly lower rates of labor force participation. In particular, those in the 50—59—year—old age group have a labor force participation rate which is about .17 lower than that of 29 year olds or younger. For those nurses older than 59 years of age, the labor force participation rate is .47 lower than for single nurses under 30 years of age. On the other hand, for those single nurses who are in the labor force, the age classification variables are substantially less important. It appears, as shown in table 16, that only the 40—49— year-old group provides a statistically significantly different num- ber of hours of labor services compared to those provided by nurses under 30 years of age. On the average, those in the 40—49-year—old age group would provide about 4 hours per week more labor than 50 the younger nurses. Those in the other groups do not appear to provide significantly different numbers of hours of nursing services. On the other hand, the regional classification variables seem to be somewhat more important in determining the labor supply of nurses who are working. In particular, those nurses who work in Region IV and Region VI appear to offer somewhat lower number of hours in the nurse labor supply market. The elasticities of labor supply responses as shown in table 17 indicate the effect on each dependent variable of percentage changes on the independent variables that are listed. All the elasticity estimates that are associated with statistically significant regression coefficients are indicated by footnotes in the table. Those which do not have footnotes are associated with nonsignificant regression coefficients. For example, since the regression coefficient associated with the 40- to 49-year-old age group in model 2 (a model that predicts the number of hours single nurses will work each week) is statistically significant at the .05 level, one can interpret the elasticity coeffi- cient as follows: a 1 percent increase in the proportion of single professional nurses that are in the 40- to 49-year-old age group will cause a .012 percent increase in the number of hours that these single nurses work each week. Similar interpretations would be applied to the other variables that are listed in that table. Although the elasticity parameters that are shown in table 17 for the classification variables—Le, for the preschool, the age range, and the rent classification variables, are quite small in absolute value. Since these groups may involve very large numbers of individuals, the changes in labor supply associated with small percentage changes in the number of people in each classification can be very substantial. Married Nurses The empirical results for the analysis of labor force behavior of married nurses were, overall, much more significant than those associated with our analysis of single professional nurses. In partic— ular, the values of the F—statistics associated with the overall regressions were, in most cases, at least double those associated with the single professional nurse models. The variables that were by far the most important in explaining the labor force behavior of married nurses were the husband’s income variable, the number of children, the presence of preschool children in the household, race, and finally the age group in which the married nurse was classified. The results obtained by RTI on the importance of husband’s income (both in direction and in magnitude) were very similar to those derived by Bognanno. (12) Bognanno found that changes in 51 the husband’s earnings were much more important in affecting whether or not the nurse works than in affecting the extent to which she works, given that she is in the labor force. In particular, he estimated that the elasticity of labor supply lies between —.08 and —.12 for married nurses. Our corresponding estimate shown in table 17 is —.087. On the other hand, Bognanno estimated that the elasticity of labor force participation with respect to changes in the husband’s earnings rate was —.28. Our corresponding estimate shown in table 17 was —.236. The obvious implication is that the major effect of increases in the husband’s earning rate will be in the wife’s discontinuing labor force activity, rather than in her reduc- tion of number of hours worked, even though professional nursing does allow part-time activity to a much greater extent than alterna- tive professions. The wage rate appeared to be unimportant in determining either the labor force participation rate or the number of hours of the labor supplied by married professional nurses in this study. These results contrast with both those found by Bognanno (12) and by Benham. (14) Although alternative specifications on the wage variable and alternative specifications of both the labor force participation and the hours model were experimented with, in no case was a significant and positive coefficient on the wage variable found. However, it should be noted in table 19 that the t—statistic associated with the wage coefficient in the pooled sample is 1.11, which falls short of statistical significance in the test of the hypothesis that the regression coefficient is different from zero. This evidence may be construed as supporting a positive labor supply response as a result of wage increases among married nurses. It is interesting to note the extent to which RTI’s results parallel those derived by Bognanno. (12) In his study, he found that the wage rate had a very negligible effect on the probability of labor force participation, but that it had a much more significant effect on the extent of labor services supplied by those nurses already working in the market. RTI’s analyses seem to confirm Bognanno’s results, as can be noted in table 19. The regression coefficient on the wage rate for the labor force participation model was of the wrong sign and very close to zero. On the other hand, as previously indicated, the coefficient in the wage rate in the conditional labor supply model was much larger and was near statistical significance. As indicated previously, an alternative specification of the varia- ble which indicates the number of children under the age of 18 in the family, KID< 18, was tried. The specification finally chosen was that which incorporated only a linear term. The number of children under the age of 18 seemed to have negligible effect on whether or 52 not a registered nurse will enter the labor force. Again, the estimated coefficient is of the proper sign and its value is very close to zero. However, the t-statistic is 1.14 and is close to statistical significance. In the conditional labor supply model, on the other hand, the estimated coefficients were highly significant in all cases. These results indicate that, on the average, a 1 percent increase in the number of children under the age of 18 in a family will result in a .058 percent decrease in number of hours she supplies in the market (see table 1'0. In both the labor force participation rate model and the condi- tional labor supply model for registered nurses, the presence of preschool children in the household was highly statistically signifi- cant. As expected, the effect of preschool children in the household is to reduce both the number of nurses who work and the number of hours that they work each week. In particular, the estimated coefficient shown in table 18 indicates that the proportion of nurses with preschool children participating in the labor force 1s 0.28 lower than for those who do not have preschool children. The associated elasticity estimate shown in table 17 implies that a 1 percent increase in the proportion of women who have preschool children will result in a 0.037 percent decrease in the number of hours worked and will result in a 0.164 percent decrease in the proportion of nurses working in the labor force. As in the models that were developed for single nurses, the inclusion of age classification variables was quite important. In particular, it appears that those married nurses in the 40—49-year- old age group have a significantly higher labor force participation rate than those who are under 30 years of age; and that those who are older than 59 years of age have a much lower labor force participation rate. Among those nurses working, only those older than 59 years of age, have significantly lower labor supply rate than those nurses under 30 years of age. The ethnic variables were important in both models. The results indicate that black nurses have a higher labor force participation rate. Specifically, it is estimated that black nurses’ labor force participation rate is 0.07 higher than white nurses. Nurses of other ethnic backgrounds are not significantly different from white nurses in terms of their labor force participation rates. Similarly, it appears that black nurses will supply about 3 hours more per week to the labor market than will white nurses and other nurses—ie, nurses of other ethnic backgrounds will not supply a significantly different amount of labor from the amount supplied by white nurses. The regional classification variables appear to be unimportant in 53 determining the rates of labor force participation among married nurses. However, there appeared to be more significant regional effects in determining the extent to which nurses supplied labor, given they are in the market. In particular, in Region I it appears that nurses supply on the average of about 5 hours less work per week; in Region V, four hours more per week; and in Region VIII, about 4 hours more per week. Table 20 shows the means of all the observed variables that were used in the pooled regression analysis. All the means of classifica- tion variables——preschool, age, race, rent, and region—are the pro— portions of the total set of observations that shared the specified characteristic. For example, 14 percent of all married nurses ana- lyzed with the LFP model were 50—59 years of age. Overall, these results indicate the significance of the nonearnings or husband’s earnings variable; of the variable which indicates the number of children or the presence of preschool children in the household; and of age classification variables. These are variables that are sufficiently significant to deserve consideration for inclu- sion in any census of registered nurses or any model which tends to explain labor force participation or the labor supply behavior of registered nurses. Table 18.—Labor force participation determinants for married professional nurses, 1970 1 Independent Pooled variables Sample I Sample ll sample Y ..,_.-._._ .-, 2—0.0151 2—0.0121 2—0.0137 (7.16) (5.23) (8.88) WRN .-,,.., .-,- ,- .. .- 0.0414 —0.0666 ~0.0052 (1.16) (1.76) (0.20) KID <18 _- *00182 —0.0016 —0.0110 (1.30) (0.12) (1.14) Preschoolers (base: no Preschoolers) Preschool ............... 2—0.2564 L0.3139 2—O.2831 (6.05) (7.45) (9.53) Age Ranges (base: S 29 years) 30.39 77777 77 77 0.0445 0.0337 0.0447 (0.93) (0.68) (1.31) 40—49 _--___,-.__-..-.._, 30.0887 0.0611 10.0779 (1.65) (1.07) (2.01) 50—59 - .................. —0.0290 «0.0103 —0.0160 (0.50) (0.17) (0.38) 2 60 ........ . ........... L0.3395 2—O.2973 2$03204 (4.52) (4.06) (6.18) See loolnoles a1 end 01 table 54 Table 18.—Labor force participation determinants for married professional nurses, 1970 1—continued Independent Pooled variables Sample | Sample ll sample Race (base: white) Other . 70.1350 0.1004 —0.0267 (0.90) (0.77) (0.27) Black ,W “0.1201 20.2256 20.1723 (1.66) (3.41) (3.54) Rent (base: renters) Homeowner ,,,,,,,,,,, 0.0091 3 0.2256 —0.0021 (0.22) (3.41) (0.07) Region (base: region IX) l ....................... 0.0165 " 0.1551 0.0875 (0.22) (2.06) (1.63) Il ,,,,,,,,,,,,,,,,,,,,,, 0.0355 0.0523 0.0494 (0.59) (0.86) (1.16) “I .......... . ,,,,,,,,,,,,,, 0.0730 —0.0388 0.0240 (0.97) (0.52) (0.45) W ,,,,,,,,,,,,,,,,,,,,,,,,, 0.0741 0.1236 0.0927 (0.67) (0.99) (1.13) V ........................ 0.1660 #00863 0.0586 (1.60) (0.81) (0.80) VI ...................... 0.0858 —0.0055 0.0426 (1.30) (0.08) (0.92) Vll ,,,,,,,,,,,,,,,,,,,,, 0.0717 0.0243 0.0528 (0.83) (0.28) (0.85) VIll ,,,,,,,,,,,,,,,,,,,, 0.1275 —0.0216 0.0569 (1.23) (0.23) (0.81) Intercept ................. " 0.5669 2 1.3338 2 0.8964 (1.98) (4.42) (4.34) R2 ___________________________ .17 .18 .16 F-statistic ................ '1 8.39 2 9.53 216.79 Chow Statistic ..,_--,.._--..- — — 1.07 Number of observations ., 817. 819. 1636. ‘AII parenthesized values below the reported regression coeffiments are the absolute values of the t-statistics. The reported Significance levels are assomated with either a one-tailed or a two-tailed t-test depending on whether a Stgn was hypothesized. “ Significant at the .01 level. =‘ Significant at the .10 level. ‘Significant at the .05 level. Table 19.—Conditional labor supply relations for married professional nurses, 1970 1 See footnotes at end of table. Independent Sample I Sample II Pooled variables sample _________________________ 2—O.2062 2—O.3810 2—0.2900 (2.61) (4.56) (5.11) WRN ______________________ 0.4335 1.0005 0.8222 (0.43) (0.91) (1.11) KID <18 _________________ 2—1.4102 2—1.3671 2—1.3432 (3.31) (3.39) (4.64) Preschoolers (base: no Preschoolers) Preschool _______________ 2—4.4752 2~4.6049 2—4.61 10 (3.56) (3.59) (5.19) Age Ranges (base: S 29 years) 30—39 __________________ —0.2352 0.3263 0.0460 (0.16) (0.21) (0.04) 40—49 __________________ 0.8903 —0.5457 0.3198 (0.59) (0.33) (0.29) 50—59 __________________ 3 2.7929 —0.2489 1.2980 (1.77) (0.14) (1.13) 2 60 .................. 4—6.1031 —0.4838 3—2.7194 (2.56) (0.21) (1.65) Race (base: white) Other .................. 3.2301 3.2645 3.7487 (0.64) (0.86) (1.26) Black ___________________ 3 3.5293 3 3.2106 2 3.1455 (1.93) (1.90) (2.56) Rent (base: renters) Homeowner ____________ ‘~2.4294 —0.0370 —1.2604 (2.11) (0.03) (1.48) Region (base: region IX) I ______________________ 2—7.6896 —2.8005 2—4.9401 (3.45) (1.27) (3.20) It ______________________ —2.3273 —0.6093 —1.6876 (1.36) (0.33) (1.35) III ______________________ 0.3135 1.2692 0.9940 (0.15) (0.56) (0.65) IV ______________________ 3.0977 3.0315 3.1624 (1.00) (0.89) (1.38) V ______________________ 2.8329 3.9781 4 4.0671 (1.00) (1.26) (1.97) VI ______________________ —1.2141 0.2618 —0.5453 (0.65) (0.13) (0.40) VII ____________________ —0.2828 —1.4113 —1.2744 (0.11) (0.55) (0.72) VIII _____________________ 1.7161 3 5.2390 “ 3.9296 (0.62) (1.88) (2.02) 56 Table 19.—Continued Independent Pooled variables Sample | Sample l| sample Intercept ________________ 2 37.2437 2 32.5160 2 34.0522 (4.65) (3.76) (5.82) R2 ________________________ .22 .19 .19 F-statistic ________________ 2 7.15 2 6.05 212.42 Chow §tatistic ____________ — — 0.773 Number of observations __ 502. 511. 1013. ' All parentheSIzed values below the reported regression coelficients are the absolute values 01 the t-statistics. The reported significance levels are associated With either a one-tailed or a two-tailed t-test depending on whether a Sign was hypothesized. 2 Significant at the .01 level. '-‘ Significant at the .10 level. ‘ Significant at the .05 level. Table 20.—-Means of regression variables 1 Married nurse models Single nurse models Variable LFP Hours LFP Hours Y($1000/yr) ,,,,,,,,,,,,,,,,,,,,,, 1 1.094 9.878 5.795 5.660 WRN($1000/yr) __________________ 6.876 6.845 7.041 7.046 KID <18 ______________________ 1.575 1.416 Preschoolers (base: no preschoolers) Preschool ____________________ 0.374 0.269 Age ranges (base: s 29 years) 30—39 _________________________ 0.298 0.274 0.156 0.173 40—49 _________________________ 0.225 0.262 0.111 0.107 50—59 ________________________ 0.140 0.160 0.098 0.095 2 60 ________________________ 0.065 0.048 0.143 0.078 Race (base: white) Other _________________________ 0.013 0.012 0.023 0.021 Black _________________________ 0.056 0.075 0.078 0.070 Rent (base: renters) Homeowner ___________________ 0.775 0.763 0.472 0.457 Region (base: region IX) l ____________________________ 0.079 0.082 0.081 0.078 ll ,,,,,,,,,,,,,,,,,,,,,,,,,,,,,, 0.208 0.211 0.270 0.267 lll ____________________________ 0.156 0.157 0.121 0.123 W ________________________ 0.038 0.043 0.020 0.012 V ,,,,,,,,,,,,,,,,,,,,,,,,,,,, 0.072 0.077 0.046 0.053 Vl ____________________________ 0.180 0.179 0.205 0.206 Vll ___________________________ 0.100 0.094 0.075 0.078 Vlll ___________________________ 0.038 0.041 0.036 0.033 Regressand ..................... 0.644 32.952 0.831 39.459 Number of observations ________ 1636. 1013. 307. 243. ' Means from pooled samples. 57 Summary RTI’s analysis of the 1970 census data on individual registered nurses confirms many of the results derived in previous studies. The most important variables influencing the decision of married nurses to participate in the labor force and, if they do, the extent to which they participate are: (1) the husband’s earnings level, (2) the number of children in the household, (3) the presence of preschool children, and (4) the age of the nurse. The wage rate did not seem to influence married nurses’ labor force behavior in any substantial way. However, wage increases seemed close to inducing increases in hours worked among working nurses. It appears that, as compared to other ethnic groups, black married nurses are more likely to participate in the labor force, and they are more likely to put in longer hours. Owning a house, as opposed to renting a dwelling, appeared inconclusive in affecting the labor force behavior of married nurses. The opposing effects of the nonpecuniary income from home ownership and of the in- creased debt burden associated with it seemed to outweigh each other. Finally, the regional location of the married nurse appeared to have little to do with whether or not she works. However, if she is working, her location may be a factor in determining how many hours she works. Overall, the ability to explain the labor force behavior of single nurses was much less encouraging than for the married nurses. The evidence indicates that the most important variables influencing their working decisions are their nonearnings income and their age; increases in nonearnings income appear to decrease the probability that they will work and because labor force participation rates tend to decline with age. Of those already working, however, nonearn— ings income seemed unimportant in determining how much they worked. Also, age seemed to have an effect only in increasing the hours supplied by those 40—49 years of age. Wage changes seemed unimportant in influencing either the likelihood of labor force participation or the number of hours worked. The opposing substi- tution and income effects of wage variations seemed to outweigh each other. Black single nurses appeared to be a little less likely to work than whites; although there were no apparent differences in hours worked among ethnic groups of already working nurses. Home ownership was not a significant determinant of single nurses’ labor force participation. Finally, regional differences did appear to affect the number of hours of work provided by single nurses; but did not affect the probability of labor force participation. 58 Overall, it appears from the statistical analysis in this and other studies that concentration on several price, income, and family characteristic variables should better explain the supply of regis- tered nurses from the already existing stock. This analysis is, however, a short-run approach since the issues associated with life cycle models in which many variables, such as the number of children and choice of career, are endogenous. Also, a study was not made of the demand for leisure in a consumer demand framework, where the coefficients on both the price and income variables are derived in a general model that is consistent with the requirements of consumer demand theory studied. (29) However, the analysis presented here does point the way for appropriate specification of the projections model, given current data constraints; and in the longer run, it provides a sound basis for the choice of new variables to be collected in future inventories of trained professional nurses. 59 VI. RTI’s NURSE SUPPLY PROJECTION MODEL In this section of the report, the RTI projection model is dis- cussed. The development of this model follows naturally from the previous section in which some of the determinants of nurse supply were established. To facilitate model development, RTI attempted to build on existing models. Two models in particular were exam- ined. These models are discussed in the following section. Other Nurse Projection Models Division of Nursing Model DESCRIPTION: This model is presented 'in Health Manpower Source Book, Section 2, Nursing Personnel. (30) The model is based on a formula developed by the National League for Nursing (31) and can be described by the following formula: S(t + 1) = S(t) + g(T) — AS(t) where S(t) = the supply of active RN’s at the beginning of year T g(T) = RN graduates during the year T. A the annual net attrition rate. II The baseline data used to initialize this model were developed from trends of historical data. Thus, the supply of active RN’s, S(t) was estimated to be 680,000 in 1969. Similarly, the net annual attrition rate was estimated at 3 percent based on 1963—1966 ANA inventory data. These numbers were revised as later data became available. The baseline projections of graduates, g(T), was developed from a combination of several parameters. First, current and future co- horts of 17—year-old girls were derived from census data. These cohorts constituted the population from which new entrants into nursing schools (who were not currently nursing school students) were assumed to be forthcoming. Similarly, data were gathered on nursing school student populations from which new graduates were generated. Given these populations, several parameters were postu- lated: the percentage of 17-year-old girls expected to attend some nursing school; the proportion of those admitted girls expected to 60 select diploma programs, baccalaureate programs, and associate degree programs; and the completion rates from these programs. Given these parameters and the lengths of training—2, 3, and 4 years for the associate, diploma, and baccalaureate programs, respectively—the model simply operates on these nursing school student populations to produce time-series estimates of new gradu- ates by type of degree program. The model was employed to produce both baseline and alternative active nurse supply projections. Individual parameters were varied within “reasonable” ranges to develop alternative projections under ceteris pam‘bus conditions. Specifically, the Division of Nursing model was used to provide 15 separate estimates of the number of nursing school graduates and of the supply of active registered nurses. These 15 estimates were produced by combining alterna- tive variations in four basic parameters: (1) proportions of new students enrolled, by program type; (2) completion rates, by pro- gram types; (3) admission rates; and (4) attrition rates. CRITIQUE: This model is designed to use accurately obtained RN supply information and then project future supply on the basis of four critical variables. No attempt is made to incorporate known data on mortality rates, foreign nurse supply, or to account for the influence of age on attrition. Economic variables such as nurses’ earnings, husbands’ earnings or nurse substitute earnings, that have a demonstrated relationship to labor force participation rates, are not explicitly incorporated in the model. Finally, no time series data core is incorporated in order to estimate historical trends in this rate even though the sensitivity analyses demonstrated that the nurse supply function was extremely sensitive to changes in the attrition rate parameter. ’ This model demonstrated a potential for producing a wide range of supply estimates, even though many behavioral relationships were ignored. Thus, no insight is acquired about the relative accuracy between any time set of projections made by the model. Hence, the need for expeditious handling of the data tended to undermine the quality and usefulness of the model. The model does provide useful output, because it identifies some of the specific behavioral parameters that are critical for making supply projec- tions. Finally, it should be noted that empirical evidence indicates that this model did not respond accurately to the critical variables. The above model estimated approximately 740,000 nurses would be employed in 1972; however, the ANA inventories indicated about 780,000 employed nurses—an underestimate of about 5 percent. 61 The Altman Model DESCRIPTION: Altman’s model (1) consists of five subcompo- nents: (1) a nursing school admission’s component, (2) a nursing school graduation component, (3) RN stock survival component, (4) a foreign nurse supply component, and (5) a labor force participation rate component. The model is also age specific. The dynamic context of this model is similar to the Division of Nursing (DN) model. However, the subcomponents, particularly the first identified above, rely on econometric estimation techniques. The admissions model, described more completely below, yields serial estimates of training program specific admission rate; the variables produced by this admission’s model, though different, because they are estimated in a much more sophisticated framework, are the same series that are produced by the DN model. The nursing school completion rates are applied to the serial estimates of admissions to generate graduation rates and in turn new RN graduates. Age-specific survival rates and labor force participation rates, also discussed below, are calcu- lated and applied on an age-specific basis to the current and projected stock of nurses. The linking of these four subcomponents is similar to the se- quence of the DN model. However, this model is more complex. The admissions to nursing programs, completion rates, and labor force participation rates are based on econometric submodels. Hence, the section below discusses the behavioral relationships developed in these submodels. The admissions submodel is comprised of four equations. Three of these equations represent the admissions behavior associated with each of the nurse training programs, i.e., the associate, diploma, and baccalaureate programs. The fourth equation is merely an identity; it combines the sum of three program-specific estimates into a total admission estimate. The equations representing the associate and diploma program allow for substitution between these two pro- grams. No dependence is hypothesized between enrollment in these two programs and the baccalaureate program. Each of these equa- tions is described in the following paragraphs. The admission rates of each training program were defined as the ratio of the number of the new matriculants to each nursing program in year T to the number of 17-year-old graduates in year T—l. The equation for the associate degree admission rate is ex- pressed as a function of the average wage rate of beginning average nurses divided by the beginning average wage for school teachers (the principal alternative to nursing as a career selection for 17-year—olds), the diploma admissions rate, and a term represent- 62 ing the growth process observed in this training program. This latter term can be regarded as a nonlinear time effect. In the diploma admission equation, the rate is expressed as a function of the same relative wage variable, of the admission rate to associate programs, and of the declining growth (i.e., nonlinear) characteristics observed in this training program. The baccalaureate admission rate was expressed as function of the same relative wage variable and of the growth process of baccalaureate degree nursing programs. For this program the growth was assumed to be constant, i.e., linear. The completion rates for students of the three nursing programs were found by Altman to manifest no discernible trend behavior and to have only small variance components. Therefore, only the average completion rates based on 1963—1968 data were used for making projections. The age-specific survival rates (probabilities of surviving a speci- fied number of additional years, by age) are based on Bureau of Census data. There was insufficient evidence for varying these survival rates systematically over time. Hence, these rates were also assumed to be fixed. The flow of immigrant nurses was recognized by Altman as a significant component of additions to the stock. However, because of the lack of data that might enable one to observe the significant behavioral relationships affecting this supply component, a simple model was developed expressing the ratio of immigrant RN’s to domestic graduates as an increasing linear function of time. Altman’s projection of age-specific labor force participation rates was complicated somewhat by the unavailability of 1970 age-specific census data. Although he had at his disposal only the 1960 rates, he used an approximation equation (discussed below) to obtain 1970 rates on an age-specific basis. then, in projecting future nurse supplies, he used two projection procedures: (1) he assumed that the participation rates remained constant at the estimated 1970 rates, and (2) he assumed that half the rate of growth exhibited by white female age groups between 1960 and the estimated 1970 rates would be distributed evenly throughout the 1970’s. Although the 1970 labor force participation rates are now available to check Altman’s estimates, the procedure he used is worth reviewing, because it is a critical item in the supply projections, and because it may provide insight on methods of handling these rates for future years. Altman started with the participation rates based on the 1960 census data. He then noted that the 1960 census estimate for active RN’s was 14.1 percent higher than the 1960 ANA estimate and that the ANA 1970 estimate of active nurses was approximately 700,000. 63 He concluded that the 1970 census figure for active RN ’s would be 700,000 X 1.141 = 798,500. He then found that the census count would have been 731,000 if the age-specific participation rates had remained constant. He, therefore, concluded that the additional 70,000 (798,500 — 731,000) RN ’5 were attributable to an increase in the stock of RN’s and/or a change in their age composition. The other 31,000 (731,000— 700,000) were assumed to result from an increase in the proportion working. It appears that this estimation technique may be inaccurate. It seems more reasonable that the 31,000 increase in RN’s was due to an increase in the stock; similarly, the 70,000 increase was more likely due to a change in age composition and/or an increase in the participation rates. However, the central point is that the participa- tion rates were changing and that the behavioral relationships of the participation rates based on the 1970 census data could not be developed. Altman then established a new labor force participation rate by assuming that the change in the age—specific rates for nurses was the same as the participation rate for all white females. CR1 TQUE: Altman’s model provides estimates of the number of employed RN ’s by predicting the additive components of the follow- ing estimates: 0 total 17-year-old females — non-RN career selec- tion = admission to nursing schools, admission — dropouts = domestic graduates, graduates + new foreign supply = total new supply, old stock + new supply — death = new stock, 0 new stock — unemployed = employed RN’s. This is analogous to the logic flow exhibited in the DN model. However, Altman’s model also employs an econometric analysis of the behavioral relationship between the components identified above and the critical variables influencing these components. An application of this model using 1970 census data and 1972 ANA inventory data would likely improve its projections. These were not available for analysis and, hence, could not be incorpo- rated by Altman. This modification could result in substantial changes in the labor force participation rate projections. The model has several positive features, but one limitation is that it required data from multiple sources. Census data, ANA inventory data, Bureau of Labor Statistics data, and others were all used in the model. In fact, differences in the census estimates and ANA estimates may have contributed to inaccurate labor force participa- tion rate analysis. Also, nurses’ earnings and the earnings of public school teachers and secretaries are based on a 13 SMSA sample. The specific data derived from this sample were the general nurse 64 and teacher wage rates. These were used to form the ratio which represented the relative attractiveness of choosing nursing as a career with respect to an alternative career. Since the samples are drawn from SMSA’s, no rural component of wages is included in the wage estimates. Studies have shown that there are significant differences between urban and rural nurse earnings. This disparity, of course, is true for teachers’ wages as well; whether it is of the same proportion is unknown. However, there is a more significant discrepancy in the Department of Labor data. The nurse samples are drawn from hospitals, which constitute the largest employer of nurses. Yet, there are a significant number of nonhospital nurses who earn substantially higher wages than those in the employ of a hospital. (18) Though the nurse earnings data are biased, they were the best data available at that time. RTI was also forced to use data limited to general duty nurses in non-Federal short-term metropoli— tan area hospitals. Though the data used by RTI is more general than the data used by Altman, it is not representative of RN’s earnings in the United States. However, Altman and RTI felt it was better to use these limited data than ignoring their existence. In Altman’s model the labor force participation rate estimates were based on census data. The RN specific census data are known to include a significant percent of non-RN’s. To what extent this data error alters the rate estimates is not known. Because the census is conducted only every 10 years, the census data were 6 or 7 years out of date when used by Altman. During that time period, substantial changes in the labor force participation rates for all females and the participation rate for licensed RN’s were evident. Using the census data’s definition of labor force participation implies that the trained stock of RN’s, rather than the licensed stock of RN’s, must be used. However, good data on the trained populatian are very limited. In general, the model has some outstanding theoretical features. The principal deficiency is that it does not efficiently utilize the national data sources that are available. RTl’s Model Model Rationale The objective of this study, as stated in section II, is to develop a practical process which can be used periodically for making accu- rate national projections of registered nurse supply. To satisfy this objective, the model developed was made: (1) conceptually simple; (2) practical to implement; and (3) to use established national data 65 sources. Furthermore, the model included most variables that were found to be determinants of labor force participation. The model developed is a flow model similar to the Division of Nursing and Altman models. Two notable differences between these models and the RTI model are: (a) The RTI model is based on the stock of licensed nurses, while Altman’s model is based on the stock of trained nurses and the Division of N ursing’s model is based on employed nurses (supply). (b) The RTI model operates in two modes: an estimation and a projection mode. In the estimation mode “base year” data and subsequent licensure and graduation data are used to estimate nurse supply for the “origin year.” The projection mode uses the origin years as year zero and projects into the future, using projected values for the model parameters. The base year is the latest year for which good age-specific data are available (ANA inventory years) and the origin year is the most recent year for which licensure data are available. Neither of the other models permit adjustment of the base year data using licensure data. The licensed stock of nurses was used instead of the trained stock because of the availability of good age—specific licensure and em- ployment data from the ANA’s Inventory of Registered Nurses. Information on the trained stock from census data is not reliable for reasons stated in section III. The licensed stock can also be updated, using adjusted annual licensure data, whereas supply data used by the Division of Nursing and Altman’s trained stock cannot be similarly updated. Obviously, new entrants into the licensed stock must be included in the model. Since many of those being licensed in the United States for the first time were foreign trained, and because the factors affecting foreign trained nurses are different from those affecting U.S. graduates, it was decided to incorporate these en- trants explicitly in the model. It was hypothesized that the activity rate for those licensed for the first time (both foreign and US. graduates) might be higher than for the previously licensed nurses. Data availability limited the ability to check this hypothesis. However, the limited data that were available indicated that the age-specific activity rate was not significantly different from the activity rate for the newly licensed nurses. For this reason, the models use only one activity rate that applies to the complete licensed stock of RN’s. The annual numbers of reinstated nurses and nurses that allow all their licenses to expire are flow variables that must be estimated 66 in order to determine the size of the stock. Data limitations make reliable estimates of these variables difficult. The approaches used to make these estimates are discussed later. Husband’s income and number or presence of children have been shown to affect the participation of married nurses in the labor force. Though these factors are not explicitly included in the model, the age—specific activity rate reflects both. Since husband’s income and children are relevant variables, consideration was given to including marital status. Data exist that would permit the inclusion of marital status in the activity rate, but unless there is a significant shift in marital status of nurses, its inclusion would have little effect on the projection of nurse supply. Nurses’ wages relative to teachers was also found to be a determinant of admissions to US. schools of nursing and, hence, future licensed stock and RN supply. The model includes this ratio in the submodels that estimate admissions and licensed stock. These submodels are discussed later in this report. The basic flow logic of the Division of Nursing supply model is a suitable starting point for the development of a new model. A desirable feature of a new model would be to disaggregate the DN model into the components identified in the flow diagram in figure 1 (page 7). Some of the important components not explicitly consid- ered by the DN model include activity rates, RN expirations, reinstatements, and new foreign nurse supply. Some components, such as expiration due to death, could be readily established. Others can be obtained by using the licensure data as described below. Age was another important variable established in the previous section as being a determinant of labor force participation. This variable affects activity rates as well as mortality rates. The relevant age-specific RN stock and supply data are available for ANA inventory years; thereafter, it was assumed that the age- specific stock could be advanced through time without seriously disturbing the accuracy of the RN stock’s age distribution. This procedure is described below. The existence of a largely unused data source has been noted also. The licensure data discussed in Section 111 can be used to estimate current stock levels by converting licenses into the num- ber of RN’s holding active licenses. Another highly desirable fea- ture of using licensure data is that it is a national data source that is available annually. It is difficult to study the trends in the ANA data series because of the limited number of data points; whereas the licensure data can be subjected to analysis and projected into the future. 67 Model Form The general equation for estimating the number of employed RN’s is defined as: n S(t) = 2 ai(t)Li(t) (6) i=l where S(t) = the number of RN’s employed in the nursing profession at time t, ai(t) = the activity rate (the ratio of employed to licensed RN ’5) at time t for age category i. Li(t) = the stock of RN’s in age category i that hold licenses at time t. t = January lst of year T, and T = the calendar year beginning at time t. The equation for estimating the licensed stock of nurses is: Li(t) = piilLi—1(t_1)+ Gi'1(T—1) + Fi—1(T_1) (7) + Ni_1(T-1)— Oi_](T—1) where Pi = the 1-year female survival probability for the i"1 age category, Gi(T) = the number of graduates of U.S. schools of nursing in age category i licensed during year T, the number of RN’s trained in non-U.S. schools of nursing in age category i that have received a U.S. RN license during year T, Ni(T) = the number of nurses in age category i that are reinstated during year T, the number of living nurses in age category i that allow all their licenses to expire during year T. Fi(T) 01(T) || Equation 7 will be referred to as the licensed stock flow equation. Model Subcomponents ACTIVITY RATES: The age-specific activity rate is the ratio of the number of employed RN’s to the number of licensed RN’s for each age category. Data for calculating the activity rate are available from the ANA inventories. The age-specific activity rates for 1966 and 1972 are given in table 21. Projecting the future value of the activity rate is hampered by the lack of data. The limited number of years for which inventory data are available does not permit reliable projections of the 68 Table 21.—Age specific activity rates for 1966 and 1972 Activity rate Activity rate Age 1972 1966 Age 1972 1966 $21 ,,,,,,,,,,, 0.9266 0.9106 47 .- 0.7346 0.7163 22 ............ 0.9395 0.9266 48 - - 0.7346 0.7213 23 ............. 0.9148 0.8807 49 0.7395 0.7303 24 ............ 0.8811 0.8150 50 _- _ 0.7346 0.7343 25 W. . 0.8385 0.7492 51 ............ 0.7336 0.7383 26 ............ 0.7989 0.7034 52 _-...- 0.7356 0.7343 27 ....... .---. - 0.7534 0.6695 53 ............ 0.7326 0.7403 28 ............ 0.7227 0.6237 54 ............ 0.7326 0.7472 29 ............... 0.6940 0.6077 55 ......... - 0.7257 0.7363 30 ............ 0.6643 0.5898 56 --_-.--_..-- 0.7177 0.7393 31 ............. 0.6544 0.5759 57 - 0.7128 0.7432 32 ............. 0.6484 0.5679 58 ............. 0.7098 0.7383 33 ............. 0.6514 0.5679 59 -_-- ._ .- 0.6960 0.7403 34 ........... _ 0.6395 0.5769 60 --_..-.-.-._ 0.6989 0.7273 35 .......... - 0.6445 0.5798 61 ............ 0.6781 0.7084 36 ............ 0.6524 0.5918 62 ............. 0.6633 0.7163 37 ............ 0.6514 0.6077 63 ............ 0.6286 0.6655 38 .............. 0.6643 0.6097 64 ............ 0.6019 0.6536 39 ............ 0.6722 0.6187 65 ............ 0.5673 0.6366 40 ............ 0.6801 0.6307 66 .__.._,_- 0.4990 0.5769 41 ............ 0.6940 0.6476 67 ............. 0.4584 0.4493 42 ............. 0.7059 0.6615 68 ............ 0.4326 0.4493 43 ............. 0.7138 0.6745 69 ............ 0.4089 0.4493 44 ........... 0.7257 0.6855 70 ............. 0.3990 0.4493 45 ............ 0.7247 0.7024 71 ............... 0.3673 0.4493 46 ............. 0.7296 0.7074 72 ............. 0.3782 0.4493 273 ----..- 0.2724 Source: Computed from American Nurses' Association data, (2) activity rates. One approach investigated in this study for project- ing the activity rates is to link them to labor force participation (LFP) rates for all females. A technique to establish this linkage is to assume that the change in the female participation rate is equivalent to the change in activity rates recorded over the same period. Table 22 shows that between 1966 and 1972 a change of 4.7 percent in labor force participation for all females 20 years and older was accompanied by a change of 3.0 percent in the activity rate of licensed RN’s. The change in LFP rate for all females 20 and older was 1.3 percent from 1963 to 1966. The activity rate for licensed nurses for the same period changed 2.3 percent. Thus, according to the data for both time periods, the participation rates 69 Table 22.—-Labor force participation rates and activity rates for all females and licensed RN’s by year Labor force participation rate for women 20 or older ‘ Activity rate 2 Year Rate Change Rate Change 1963 _______________ 37.9 65.2 1.3 2.3 1966 ______________ 39.2 67.5 4.7 3.0 1972 ______________ 43.9 70.5 ‘ U.S. Department of labor data. (32) 1 American Nurses' Association data. (2) Table 23.—White female mortality rates (4) Mortality, rate, Mortality rate, deaths per year death per year Age per 1,000 persons Age per 1,000 persons 20 ____________ 0.65 45 ____________ 3.07 21 ____________ 0.65 46 ____________ 3.34 22 ____________ 0.66 47 ____________ 3.62 23 ____________ 0.66 48 ____________ 3.94 24 _____________ 0.66 49 ............. 4.28 25 ............ 0.67 50 ............. 4.65 26 ____________ 0.68 51 ,,,,,,,,,,,, 5.05 27 _____________ 0.70 52 ............. 5.47 28 _____________ 0.73 53 ______________ 5.92 29 ............. 0.78 54 ............. 6.10 30 ____________ 0.84 55 ,,,,,,,,,,,, 6.33 31 _____________ 0.90 56 ............. 7.51 32 .............. 0.98 57 _____________ 8.13 33 ............. 1.06 58 ............. 8.78 34 ............. 1.16 59 ............. 9.48 35 ____________ 1.26 60 ............ 10.26 36 _.-,__.--_.. 1.38 61 _____________ 11.13 37 ............. 1.51 62 ______________ 12.12 38 _____________ 1.66 63 ,_. ........... 13.25 39 __. .......... 1.82 64 ............ 14.51 40 _______________ 2.00 65 ............ 15.92 41 ,,,,,, . ______ 2.20 66 ____________ 17.45 42 _____________ 2.40 67 _____________ 19.23 43 ____________ 2.61 68 ____________ 21.31 44 ____________ 2.83 69 ............. 23.74 and the activity rate are growing at the rate of 1 to 2 percent per year. For the purpose of making projections, three assumptions are used for estimating the activity rate. The first assumption is that 70 the activity rates change by a constant amount based on the change observed in the activity rates for the 1966—1972 period, or .73 percent per year, i.e., ai(t + 1) = 1.0073a, (t) for all i. The second assumption is that the activity rates change in the same direction as the LFP rate for all females. Each percent change in the LFP rate will correspond to a change of 3.0/4.7 percent = .64 percent in the activity rate, i.e., ai(t + 1): ai(t) + .64ALFP(t). The third as- sumption is that the activity rate is constant, i.e., a,(t + 1) = ai(t). The effect of these different assumptions on supply projections is presented in section VII. MORTALITY RATES: Mortality rates were obtained from the 1973 Statistical Abstracts (4), which are based on NCHS Vital Statistics. These data show the expected number of deaths per 1,000 living by sex, race, and age for 1969. Since a vast majority of RN’s are white females, only these figures were used. These data are presented in table 23 for ages 20 through 69. For the projections made by the RTI model, it was assumed that these mortality rates remained constant. From this definition of mortality rates 1 Mi > = . _— 8 D(T) 2, L.> m5E=mm< wvwdn www.mv mmmdm Kiwi. 84. mod om; mfim ...... $4”me 9%.: momdv www.mm ; 3%.va mmé vmd vmé .oo.m ..... 3&me oQOn «3.3 mmmdm 35min 5; vmd mm; mad ‘‘‘‘‘‘ wwIEmF KQE. vmmdv 9.an v 3.9; R; NEN no; whw ...... Snow? «3.3 85.3. Soda ommdwr wmé EN 3; vmd ..... owlmnm— onmdn Kmdv www.mw mmfimvr 34‘ Few K; and ...... 31¢an omew wRdv mmndm 23.8? 004. ovd mu; n F .w ...... 3|me v3.3 $348 @5de «969 8.0 wNN mp; ems ..... Rumma— nmodw wmvdm mnmdm vmemr mod wtm $.— voN ..... 9.139 V 5mm mmvém «8.8 :5de N1” ZN mm; P: ..... 3.459 mmodm mvmdm wmmdm 9.de C‘m mod mm; is ...... 3:29 Elgar E ooé 8 mom... 3 m:_E:mw< $58 23.3. 9:920 .98. @9an EmEJS mEOEE _So.r 5m; 9m_oo$< .moomm 2m_00wm< -moomm mmumscma .mco_mw_Eum E EnEac 8629a Bozow :9; 99:3 3 :60th m mm m:o_mw_Eum 8629.”. 3.82 3:25 $.22 .5235 to 2:: 3 28:8 @595: 3:01.682: 2 m:o_mm_Eum nouofioildu 29a... 81 1983—84 this projected total has risen to 146,421 (for W) and 126,324 (for W*). In order to obtain the projected number of graduations, it was first necessary to estimate graduation completion rates. Recall that table 24 presents the actual completion rates observed until 1972— 73. Examination of these rates indicates that the BAC and AD rates have been increasing relatively rapidly since 1970, while the DIP rates have remained fairly constant since 1967. However, RTI did not feel justified in fitting models to the various rates and instead let the AD and BAC rates equal their 1972—73 value and set the DIP rate equal to .75. In the future RTI would suggest, for projection purposes, that these rates be set equal to their latest observed value. Table 30.—Program graduations from professional nursing schools, by type of program, 1973—74 through 1984—85 Projected number of graduations ' Bacca- Associate Year Total Diploma lau reate degree Assuming w rises to 1.00 in 1980-81 1973—74 67,196 22,351 15,106 29,739 1974—75 ,, 76,145 22,386 20,244 33,515 1975—76 -M 82,673 22,724 22,554 37,395 1976—77 ,,_, 87,590 23,012 23,938 40,640 1977—78 ,,,,,,,,,,,,, 91,881 22,934 25,482 43,265 1978—79 95,124 22,634 26,957 45,533 1979—80 ,.,_ 98,233 22,300 28,446 47,487 1980—81 ,,,,,,,,,,,,,, 101,023 21 ,956 30,174 48,893 1981—82 ,,,,,,,,,,, 103,087 21,455 31,799 49,833 1982—83 _ ,, 104,356 20,807 33,085 50,464 1983—84 ,,_, , 104,373 19,967 34,258 50,148 1984—85 ,_, , _ 103,891 19,000 35,412 49,479 Assuming w constant after 1973-74 1973-74 ,,,,,,,,,,,,, 67,196 22,351 15,106 29,739 1974—75 ,1 76,145 22,386 20,244 33,515 1975—76 We , 82,673 22,724 22,554 37,395 1976—77 ,,,,,,,,,,,,,, 87,590 23,012 23,938 40,640 1977—78 W 91,220 22,934 25,482 42,812 1978—79 ,,,,,,,,,,,, 93,985 22,634 26,957 44,394 1979—80 2., ,, , 96,273 22,300 28,321 45,652 1980—81 _ 96,863 21,956 28,540 46,367 1981—82 , 96,739 21,455 28,758 46,526 1982—83 ,-_, _, _ 95,982 20,807 28,775 46,400 1983—84 ,,,,,,,,,,,, 94,076 19,967 28,590 45,519 1984—85 2,- , ,, 91,654 19,000 28,305 44,349 ‘Projected number of graduations , (Projected number of admlssions) x (Completion rate). Completion rates used were: AD , .67. BAC , .74, DIP '4 .75. 82 Figure 11.—Graduation from professional nursing schools by type of program (1.000) ACTUAL — - — - PROJECTION 1 _ xx PROJECTION 2 ,x" "‘ 100' ,/ I’x x x x I’x X ’ X 90- [,4 a ’1 ° / :53). 80‘ // a”; / ,_ 70' [I <1 I D I D I < 60- .1 CE (.9 ‘5 50— u: ‘33 2 40— D Z 30. I l l l I I 4_J 1960 64 68 72 76 80 84 YEAR Note: PYOjeCIIOH 1 assumes W increases to 1.00 In 1980—81. Projection 2 assumes W constant at .90 from 1973— 74 to 1980—81. Using the above graduation completion rates, the projected num- ber of admissions in table 29, and assuming the AD program takes 2 years to complete, the DIP program 3 years to complete, and the BAC program 4 years to complete, RTI projected the number of graduations from the three nursing programs and their sum. The results are given in table 30 and figure 11. Figure 11 shows that, while in 1972—73 the number of graduations from the three nursing programs was 59,427, this projected total in 1984—85 is 103,891 (for W) and 91,654 (for W*). Of course, it is very important to note here that the projections given in figures 10 and 11 and tables 28, 29 and 30 are based on many assumptions that may prove to be false in the future. For example, the growth function for the AD program may not rise as rapidly as predicted (table 28), or the graduation completion rates assumed by RTI may prove to be too high. Thus, it is very important that these and the other assumptions made in the projections of admissions and graduations be reexamined each year as additional data become available. In future years the effect of other factors that were not included in the present admissions and graduations model must be considered, e.g., men who enter nursing and the large number of students who do not enter the AD program directly after high school but do so at a much later age. LICENSURE' OF US. GRADUATES: The number of licenses issued to new US. nursing graduates was estimated using the data 83 Table '31.—Ex'amination and first time Iicensure data (3) Licenses issued to Percent passing US.1 registered examination nurses for the first a time by examina- First Subse- New US. Licensure Year tion time quently graduates ratio 2 1963—64 ,,,,,, 32,916 85.7 62.8 35,259 93.4 1964—65 ,,,,, - ,, 33,417 85.8 63.3 34,686 96.3 1965-66 ,,,,,, 33,968 85.6 58.5 35,125 96.7 1966—67 ,,,,,, 37,884 85.3 58.2 38,234 99.1 1967—68 ,._--,_ 41,155 84.6 58.3 41,555 99.0 1968—69 - 41,665 84.3 57.7 42,196 98.7 1969—70 ,,,,,, 42,866 83.2 53.8 43,639 98.2 1970—71 ______ 45,419 — — 47,001 96.6 1971—72 ,,,,,, 48,070 81.8 52.5 51,784 92.8 ’ Excludes RN's from foreign countries. 2 Licensure ratio f Licenses Issued in year T/new U.S. graduates In year T. in table 31. Note that the data in this table are not by type of nursing program, hence, it is not possible to compute Iicensure by type of training program. After examining the data in table 31, RTI used the following equation to estimate the number of licenses issued to US RN’s for the first time by examination in year T, G(T), for the years given in table 31: G(T) = PF'I‘ (T) g (T) + [l—PFI‘(T-1)] PR(T)g(T-1) (18) + [1—PFT(T—2)] [1—PR(T—1)] PR(T) g (T—2)N R where PFT(T) = Probability of passing the examination the first time during year T, Probability of failing the examination the first time during year T, PR(T) 2 Probability of passing retake of examination during year T, Probability of failing retake of examination during year T, gtT) = Number of new US. graduates, and 1—PFT(T) 1—PR(T) NR = Probability a student does not retake examination after failing twice. The results of using equation 18 to predict the number of licenses issued to new US. graduates are given in table 32. Table 32 indicates that equation 18 does a reasonable job in estimating G(T) for the years examined. Accordingly, RTI used equation 18 to project the number of licenses issued to new US. 84 Table 32.—Comparison of Iicensure ratio and estimated Iicensure rate Estimated ' Actual number number of of licenses Estimated Year licenses issued issued Iicensure rate 2 Licensure ratio 1963—64 .- .. -- . 33,944 32,916 96.3 93.4 "1968—69 -- -. .. .. ,- . 40,419 41,665 95.8 98.7 1969—70 -- - ._ . 41,410 42,866 94.9 98.2 1970-71 -- - 44,122 45,419 93.9 96.6 1971—72 __ .- _. .. .- - 48,413 48,070 93.5 92.8 ' In estimating LE(T), PFT(T) and PR(T) were allowed to vary; i.e., in estimating G (1968—69) PFT(T) and PR(T) given in table 331m 1968—69 were used 2The estimated Iicensure rate is the estimated number 01 licenses issued divided by the number 01 new U.S. graduates ’ Table 33.—Projected number of licenses issued to new U.S. nursing graduates, 1972—73 through 1984-85 Projected number of licenses ‘ Year Projection 1 2 Projection 23 1972—73 -.- .-...-- .. 55,250 55,250 1973—74 _ 62,510 62,510 1974—75 - - . ------ .-_- 70,846 70,846 1975—76 -_.-.._--..--_.-...-._.-__.-_---..-.- 77,314 77,314 1976—77 W_--._--.- - 82,269 82,269 1977—78 - - 86,476 85,934 1978—79 W-_ _ 89,710 88,713 1979—80 _--_---.--_.- 92,713 90,975 1980—81 -_ -...--_-- 85,406 91,770 1981—82 ------------------------------------- 97,470 91,801 1982—83 ----------------------------------- .- 98,801 91,189 1983—84 Wfl-- 99,005 89,549 1984—85 -------------------------------------- 98,654 87,356 ‘PrOJections for total over the three nurse training programs. ‘-' Projection 1 assumes W increases to 1.00 in 1980—81. ‘ Projection 2 assumes W constant at 0.90 from 1973-74 to 1980—81. nursing graduates. The results of these calculations are given in table 33. In calculating the projections in table 33, RTI used the projections given in table 30 and assumed PFT(T) = .82 and PR(T) = .53. FOREIGN NURSE SUPPLY: Table 34 presents the annual number of U.S. licensed RN’s trained in nursing schools outside of the United States. As this table indicates, the number fluctuates and has shown no discernible tendency. For this reason, and because foreign nurse supply is only a small portion of total first time licenses, a limited analysis was undertaken on this variable. 85 Table 34.—Estimated first time U.S. licensure of foreign trained nurses Estimated first time U.C. licensure of foreign Total first Percentage Year trained nurses 1 time licenses foreign 1959 ,,,,,,,,,,,,,,,,,, 2,106 _ _ 1960 _,_ 2,127 _ _ 1961 ,,,,,,,,,,,,,,,,,,,, 2,111 _ _ 1962 ____________________ 2,618 35,312 7.4 1963 ,,,,,,,,,,,,,,,,,,,,, 2,718 32,080 8.5 1964 ,,,,,,,,,,,,,,,,,,, 3,172 36,088 8.8 1965 ,,,,,,,,,,,,,,,,,,, 3,271 36,432 9.0 1966 ___________________ 3,803 37,466 10.2 1967 ,,,,,,,,,,,,,,,,,, 5,006 42,499 11.8 1968 ,,,,,,,,,,,,,,,,,, 5,772 46,456 12.4 1969 ,,,,,,,,,,,,,,,,,,, 4,043 45,708 8.8 1970 ,,,,,,,,,,,,,,,,,,,, 4,144 47,010 8.8 1971 ,,,,,,,,,,,,,,,,,, 4,749 50,171 9.5 1972 ,,,,,,,,,,,,,,,,,,, 6,608 54,682 12.1 1973 ___________________ 6,335 __ _ ‘Values are the sum of lirst time US. licensures of foreign trained nurses by endorsements and the number of foreign trained nurses receiving their license by examination. Some foreign trained nurses obtain their first license by endorsement and are requnred in another State to take the examination This results in some double counting and hence, the reported number may be high. On the basis of this analysis, and for the above stated reasons, a constant value of 6335 was used in the projections figure for 1974 through 1985. REINSTATED NURSES: The stock of licensed nurses can be increased, not only through first time licensure, but also by rein- statement of nurses who previously had let all their licenses expire. In estimating the number of reinstated nurses, the difference between reinstated nurses and reinstated licenses must be recog- nized. Some nurses who have an active license in one State may get a previous license reinstated in another State, hence, the possibility of a nurse having two or more licenses. Equation 19 indicates the relationship between reinstated nurses and reinstated licenses. N(T) = k2R(T) (19) where N(T) = Number of nurses reinstated during year T, R(T) Number of licenses reinstated during year T, and k2 = Ratio of nurses to licenses that applies to reinstate- ments. ’The number of reported reinstated licenses, R(T), has fluctuated over the years (see figure 7) without any discernible trend. The 86 average number of reinstated licenses between 1962 and 1972 was 12,400. R(T) was held constant at the 1972 value throughout the projection years. To estimate N(T) the value of k2 must be determined. This will be discussed in the next section on expirations. NURSES WHOSE LICENSES HAVE EXPIRED: The stock of licensed nurses is reduced when a nurse dies or all the licenses held by a nurse expires. It should be noted that when a nurse dies, the license stays active until its expiration date. Lack of data make estimating the reduction in the stock of nurses difficult. The approach used by RTI is as follows: Define G(T) = Q(T) + D(T) (20) H(T) = G(T) + F(T) (21) where G(T) = the number of RN expirations including death occurring during year T, 0(T) = the number of RN expirations due to reasons other than death occurring during year T, D(T) = the number of RN expirations due to deaths occur- ring during year T, and H(T) = the number of first time licenses issued during year T. The stock flow equation (equation 7) can be written as L(t) = L(t-l) + F(T—l) + G(T-l) + N(T—1) — 0(T—1) — D(T—1) (22) Substituting equations 19, 20, and 21 into equation 22 yields L(t) = L(t—1) + H(T-l) + k2R(T—1) — 0(T—1) (23) Also, the number of licensed nurses at time t equals all the licenses issued the previous year adjusted for multiple licensure and bien- nial renewals, or L(t) = H(T—l) + k1Q(T—1) + k2R(T—1) + k3E(T—1) (24) where Q(T) the number of licenses renewed during year t adjusted for biennial renewal, E(T) = the number of licenses issued by endorsement dur- ing year t, and k,,k.z,k3 = the ratio of nurses to licenses that apply to Q(T), R(T), and E(T) respectively. In equation 24 it is assumed that all nurses reported licensed for the first time in the United States do not have another license. 87 Setting equations 23 and 24 equal to each other and solving for Q(T) gives: Q(T) = L(t) — k1Q(T) — k3E(T) (25) Assuming k3 = O, i.e., all nurses who obtain a license by endorse- ment have an active license elsewhere, Equation 25 can be written as Q(T) = L(t) — lem. (26) The stock of licensed nurses, L(t), is known for ANA inventory years and Q(T) is available annually, however, k has yet to be estimated; Using the above two assumptions and an additional assumption that k = k1 = k2, equation 24 can be written as L(t) = H(T—1) + k[Q(T—1) + R(T—1)] (27) 01' k = L(t) — H(T—1) (28) Q(T—l) + R(T—l) The necessary data are available for solving equation 28, which permits solving equation 19 for N(T) and 26 for Q(T). Table 35 indicates the values for k, N(T) and Q(T) for the ANA inventory years. In summary, beginning with the ANA base year that provides a value of L(t) and given the licensure data for the number of renewed and reinstated licenses, an estimate of N(T) and Q(T) can be obtained. These values are then used to estimate the next year’s licensed stock. Consequently, for estimating the origin year, all that is required are the values of Q(T) and R(T) that exist for the base years prior to the origin year. In the projection mode it is necessary to project values of Q(T) and R(T). A discussion on the projection methodology is now presented. Table 35.—Estimates of k, N(T) and 6(T) for ANA inventory years Year L(t) H(T) Q(T) R(T) k N(T) Om 1972 ,,,,,,, 1,106,383 54,682 1,345,719 12,736 .795 10,100 36,500 1971 ______ — 50,171 1,317,394 10,956 — — — 1966 _______ 895,000 37,466 1,101,069 13,223 .797 10,500 17,400 1965 ,,,,,, — 36,432 1,064,458 12,595 — — — 1963 ______ 843,558 32,080 1,010,311 15,514 .810 12,600 25,200 1962 ,,,,,, — 35.312 985,409 12,146 — — — 88 Projecting Q(T), license renewals, is comparable to projecting the stock of licensed nurses. One factor contributing to this stock level during T is the value of Q(T-l), i.e., the number of renewals in the previous year. This follows from equation 24. A second determinant of Q(T) was hypothesized to be the expected wage rates adjusted for inflation of registered nurses employed in hospitals. However, analysis demonstrated that this was not a significant factor. A second determinant of Q(T) was the relative wages ratio of nurses to teachers. This occurs because this ratio is a determinant of new U.S graduates, i.e., first time licenses, and by equation 24 it follows that it is also a determinant of Q(T). Table 36 displays the statistical analysis of the linear model: Q(T + 1) = a0 + a,W(t) + 82Q(T) (29) a), a], a2 = Regression coefficients. Q(T) and W(t) are defined above. AGE: The age distribution for the licensed stock of RN’s is available for ANA inventory years. Equation 7 can be used to advance the stock by age from year t to t+ 1. However age—specific Table 36.—Regression coefficient estimate for linear model Coefficient T of H018 = 0' PROB > lTl an = —129920 ............................ . —1.57817 .1654 a1 = 436635 ,,,,,,,,,,,,,,,,,,,,,,,,,,,,,, 7 1.71718 .1368 32:.828391 A_.,,_--,2_.n._-.,__ . - N-.. . 6.73774 .0005 ‘Test of hypothesis that the estimated regressron coefftcnent equals zero. Table 37.—Sample of foreign trained nurses that passed nursing examination, by age 1 (10) Number passed Age examination Percent ‘1 20—24 “H,“ W v _ ,,_,, , 1,818 25.6 25—29 , 3,211 45.2 30—34 W, 1,329 18.7 35—39 ,,,,,,,,,,,,,,,,,,,,,,,,,,,,,, 440 6.2 40—44 ,,,,,,,,,,,,,,,,,,,,,,,,,,,, 191 2.7 45—49 _____________________________ 74 1.0 50—54 ,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,, 24 0.3 55 and over ,,,,,,,,,,,,,, , ,,,,,,,,,,, 18 0.3 Unknown ,,,,,,,,,,,,,,,,,,,,,,,,,,,, 128 2 Total Sample ,,,,,,,,,,,,,,,,,,,,,, 7,237 100.0 ‘ Number of foreign trained nurses that failed equaled 6.978. Data collected lrom eight States (California. Illinois, Louisiana, Massachusetts, Michigan. New Jersey. New York, and Texas) between July 1972 and February 1974. 2 Distribution excludes unknowns. 89 values for US. new graduates, foreign trained nurses licensed in the United States, reinstated nurses, and nurses whose licenses have expired need to be established. The previous sections dis- cussed how the values g(T), F(T), N(T) and 0(T) were obtained. This section discusses how to make these variables age specific. The age distribution of newly licensed nurses is given in table 4. Let hi = the percent of graduating nurses in age category i, then gi(T) = ng)hi (30) Since g(T) and h are both known, g. (T) can be computed. Similarly, the age distribution of new foreign nurse supply is defined in table 37. Let fi = the percent of foreign trained nurses newly licensed in the United States in age category i, then Fi(T) = F(T)fl. (31) Fi(T) can be computed since F(T) and fl are known. For the reinstated nurses it is assumed that N i(T) ”Si(t)ai(t) (32) where Si(t) = Li(t) ‘ Si(t)- Equation 32 assumes that the reinstatements in age category i are proportional to the inactive stock in that age category, and the age- specific activity rate. Thus, let U. : s. (t)a. / 2 Si(t)a;(t) (33> j=l where the subscriptj is used as an index for summing overall age categories. Then Ni(T) = N(T)Ui. (34) The age distribution of nurses whose license/s} have expired is obtained by assuming that expirations are proportional to the stock and inversely proportional to the activity rate; or: 0(T) N Li(t)/ai(t) (35) Thus let Vi 2 Li(t)/ai(t)/ Z Lj(t)/a_i(t) then Oi(T) = O(T)Vi. (36) where 0(T) = 0(T) — D(T) 90 is defined as expirations other than those resulting from death. The licensure data provides estimates of 0(T) total expirations. There— fore, it is necessary to calculate total deaths by using the mortality rates as described above. Model Validation Approach Both the estimation and projection modes of the RTI model were checked. To validate the estimation mode, 1966 ANA Inventory and 1966— 1971 licensure data were used to estimate the 1972 supply of RN’s. The 1972 estimates for different assumptions were compared with the 1972 ANA inventory data and the 1972 ICONS estimates. Table 38.—Licensed stock by age (1966) Age Stock Age Stock i L. i Li 21 ,,,,,,,,,,,,,,,,, 3313. 47 ,,,,,,,,,,,,,,,,,, 17660. 22 ,,,,,,,,,,,,,,,,, 16697. 48 .,--.._,--___,,_.. 18403. 23 .................. 27699. 49 ,,,,,,,,,,,,,,,,,,, 16400. 24 ,,,,,,,,,,,,,,,,,, 31367. 50 ,,,,,,,,,,,,,,,,,,, 15942. 25 ,,,,,,,,,,,,,, 1,- 28020. 51 ,,,,,,,,,,,,,,,,,, 15344. 26 ,,., __, ,.,__ .. 27114. 52 ,,,,,,,,,,,,,,,,,, 15764. 27 ,,,,,,,,,,,,,,,,,,, 25842. 53 ,,,,,,,,,,,,,,,,,, 16316. 28 ..................... 26803. 54 ........... . ..... .- 16574. 29 ................ .. _ _ - 26150. 55 .................. 15440. 30 .................. 25863. 56 ,,__..---_._-,-_,._ 16203. 31 ,__,. _____________ 25274. 57 _________________ 14333. 32 ................... 24623. 58 ................... 13722. 33 .................... 23050. 59 .................. 12278. 34 __ W--- .-,-_..,__ 24431. 60 ___________________ 10969. 35 ................... 23968. 61 ................. .- 10150. 36 ................... 24031. 62 ................... 8480. 37 .................. 20622, 63 ................... 7398. 38 .................. 19958. 64 .................. 7329. 39 ___________________ 23551 . 65 ................... 6053. 40 .................. 25614. 66 ................... 5919. 41 ................... 25636. 67 .................. 5314. 42 ,,,,,,,,,,,,,,,,,,, 24363. 68 ___________________ 4710. 43 ................... 22536. 69 ................. .- 4105. 44 __________________ 21271. 70 ,,,,,,,,,,,,,,,,,, 3485. 45 __________________ 21760. 71 .................. 3239. 46 ,,,,,,,,,,,,,,,,,,, 20729. 72 ........... . ,,,,,,,, 3186. Total ............... 895000. Source: Based on American Nurses‘ Association data. (2) 91 To validate the projection mode, projections were made to 1972, using the 1966 ANA Inventory data and other data available on or before 1966. The projections for different assumptions were com- pared with the 1972 inventory and ICON’S estimates. Estimation Mode BASE YEAR DATA: Table 38 illustrates the RN licensed stock distribution by age with the total number of RN’s as of January 1, 1966, estimated to be 895,000. This number was obtained by adjust- ing the 1966 ANA data to the January date. The ANA count of 909,131 apparently corresponds to a March lst date. LICENSURE DATA: Table 39 illustrates the licensure data used by the model. The licensure assumptions previously stated are used in making the computations. Also displayed in table 39 is the number of reinstatements and expirations implied by the model assumption for the years 1966 through 1971. THE 1.972 MODEL ESTIMATES: Table 40 displays model esti- mates for 1972 RN licensed stock and supply under the two activity rate assumptions. The 1972 ANA inventory estimates and two 1972 ICONS estimates are also presented. One of the ICONS estimates is made without the knowledge of 1972 inventory data. This estimate was developed using the Division of Nursing Nurse Supply Model described earlier. The other ICONS 1972 estimate adjusts the 1972 ANA inventory figure to correspond to January 1, 1972. THE DIFFERENTIAL OF STOCK AND SUPPLY ESTI- MATES BETWEEN 1966—1972: Table 40 also presents the growth estimates of licensed RN stock and supply between January 1, 1966, and January 1, 1972. Two sets of estimates are made using the RTI Model. These estimates correspond to the two following assump- tions: (1) the activity rates are a constant value based on 1966 ANA data; and (2) the activity rates grow according to the annual growth observed between the activity rates obtained from the 1963 and 1966 ANA inventory years. There is also an ICONS estimate that represents growth estimated by the Division of Nursing Model. The difference between the 1972 and 1966 census stock and supply values are also illustrated, and it is assumed that a good model is one that comes close to matching the “observed” ANA growth, which is also illustrated in table 40. MODEL COMPARISON—ESTIMATION MODE: From table 40, it can be seen that the licensed stock values for assumptions 1 and 2 are nearly equivalent. Both slightly underestimate the true stock growth. However, with respect to supply, the growth associated with assumption 1 and assumption 2 are markedly different. Be- cause of the dramatic increase in activity rates between 1963 and 36 Table 39.——Licensure data Year Q(T) H(T) R(T) L(t) N(T) 0(T) k 1972 ______________ 1,345,719 54,682 12,736 1,106,383 10,200 37,900 .7951 1éiimjifiiifiifiii """ {317.534 """"" é 6,171"m'm]bjééé """"""" I """Mmé'jbb """"""" A ,3ch """""" 397T 1970 ______________ 1,261,410 47,010 10,970 — 8,700 19,100 .7971 1969 ______________ 1,225,269 45,708 10,300 — 8,200 16,000 .7971 1968 ______________ 1,184,040 46,456 12,727 — 10,100 9,200 .7971 1967 ______________ 1,138,459 42,499 13,665 — 10,900 13,100 .7971 1966 ______________ 1,101,069 37,466 13,223 895,000 10,500 17,300 .7971 1ééé"fiiif:::: ::: 1156:4251 """""" é 6,1152‘"""”"?2’,ééé """"""" I """"""""" l ””””””””” I “““““““ — 1964 ______________ 1,042,532 36,088 11,439 — — — — 1963 ______________ 1,010,311 32,080 15,514 843,558 — — — 1962 ______________ 985,409 35,312 12,146 — — — 8102 Table 40.—Differential estimates of supply and stock RTI assumption 1 RTI assumption 2 ANA Estimated true growth ICONS Year Supply Stock Supply Stock Supply Supply Stock Supply Stock 1972 ________ 747,000 1,104,000 791,000 1,103,000 1 748,000 794,979 1,127,657 2 780,000 2 1,106,370 1966 ________ 3 604,000 3 895,000 3 604,000 3 895,000 621,000 613,188 909,131 3 604,000 3 895,000 Difference ___ 143,000 209,000 187,000 208,000 127,000 181,791 218,526 176,000 211,370 ' lCONS original estimate made without knowledge of the 1972 ANA Inventory of Registered Nurses. 2 ICONS adjusted estimate tor January 1, 1972, based on the 1972 ANA Inventory 01 Registered Nurses. 3 ANA Inventory adjusted to January 1, 1966. 93 1966, assumption 2 overestimates supply by 11,000 (1.4 percent) and assumption 1 underestimates supply by 33,000 (4.2 percent). A comparison of the RTI model estimates with the original ICONS estimate reveals the value of using the licensure data. The growth of the supply estimate from the Division of Nursing Model, which is 127,000, is less than the actual supply growth attributed to the increase in licensed stock alone (see table 41). The justification for using the licensure data in the RTI model was to enable the model to accurately estimate the increase in supply caused by an increase in stock, i.e., true stock growth times the 1966 activity rates. This has been accomplished. In addition, RTI’s model has estimated the effects of the change in the age distribution on supply, i.e., because of the increase in number of graduates, the average age is decreas- ing during the 1966—72 time period. Both the Division of Nursing and RTI model assumption 1 have failed to estimate the growth in supply caused by changes in the activity rates. The magnitude of the effects of these factors on the 1966—72 supply differential is examined in table 41. Projection Mode For validating the projection mode of the model, 1966 was used as both the base year and the estimation year. Values for the 1966 age- specific stock and activity rates of RN’s are given in tables 38 and 21. For projecting the activity rate, two alternative assumptions were made: (a) there would be no change in the activity rate over the projection period; and (b) the activity rate would continue to change at the same rate as it did between 1963 and 1966, i.e., 1.16 percent per year. Table 41.—Distribution of supply growth by cause Cause of supply change Size Percent Size Percent Total supply differential _- ,, ,. _ -— — 176,000 100.0 Supply differential due to change in size of stock “flesh”, _ 142,675 81.1 — — Supply differential due to change in age , ,,, 1,067 .6 — — Growth differential due to change in age and size of stock ,, ,_ — — 143,742 817 Growth in supply due to change in activity rates , _ ,. ,, , — — 32,258 18.3 94 Table 42.—Estimated number of renewals Year Renewals 1966 _, V , WWW- 1,096,000 1967 W ,_W, 1,131,000 1968 W_ , W_ 1,165,000 1969 __ ,WWWWW , WWW_ W,” W , _WW_WW_WW 1,205,000 1970 ,WWW , _ WW_ _WWW WWWWWWW __,W_WW 1,248,000 1971 W___ _W,_W,,_WW_ 1,295,000 Table 43.—Frequency and distribution of age at graduation AD DIP BAC Age 62 65 62 65 62 65 Total Percent 19 128 351 0 0 0 0 478 .0178 20 W W W 128 351 692 661 O 0 1,832 .0682 21 W W 521 1,366 693 661 431 505 4,177 .1555 22 V W W 520 1,366 1,723 1,904 432 1,719 6,450 .2402 23 V W W 49 117 1,723 1,904 1,175 1,719 6,687 .2491 24 ,W . 48 115 49 41 1,174 45 3,146 .1172 25 , W , 48 115 40 33 38 38 319 .0119 26 ,W , 48 115 40 33 33 38 307 .0114 27 , W , 48 115 40 33 33 38 317 .0114 28 ,,,,, 48 115 40 33 33 38 317 .0114 29 , W , 48 115 40 33 33 38 317 .0114 30 , ,_ 48 115 40 33 33 38 317 .0114 31 W ,_ 48 115 40 33 33 38 317 .0114 32 WW , 48 115 40 33 33 38 317 .0114 33 W W 26 56 40 33 33 38 226 .0084 34 ,,,,, , 26 50 1O 1O 33 38 167 .0062 35 . 26 50 3 4 5 8 96 .0036 36 . W , 26 50 3 4 3 3 89 .0033 37 W __ , 26 50 3 4 3 3 89 .0033 38 . W , 26 50 3 4 3 3 89 .0033 39 , 26 50 3 4 3 3 89 .0033 40 , W , 26 50 3 4 3 3 89 .0033 41 , W , 26 50 3 4 3 3 89 .0033 42 , W , 26 50 3 4 3 3 89 .0033 43 W W _ 158 286 3 4 3 3 457 .0170 44 W W 0 0 10 7 3 3 23 .0009 45 , ,_ . 0 O O O 9 8 17 .0006 Several models utilizing 1962—1966 data were postulated for esti- mating the number of renewals. The model judged most appropri- ate was: Q(T+ 1) = —53062 + 1.07993Q(T) (37) 95 Using the 1965 value for renewals (see table 39) Q(T) was com- puted using equation 37. The values are given in table 42. The number of reinstated nurses and the number of foreign- trained nurses receiving their first US. liscense were assumed constant at the 1965 value, 12,595 and 3,271 respectively. The values were obtained from tables 34 and 39. A value of 92.8 percent was used as the licensure rate estimate. The age-specific distribution of the new graduates was derived from the 1962 and 1965 Career-Pattern figures. This distribution was obtained by assuming an average length of stay of 2, 3, and 4 years for each of the associate, diploma, and baccalaureate degree pro- grams, and then assuming an average of 1 year for obtaining a license. This distribution is shown in table 43. Using the procedure developed previously in the RTI report, the data given in tables 44 and 45 were used to project admissions and graduations from the three nursing programs (as well as their total) from 1965—66 to 1970—71. The first step in the projection was to predict AD(T), BAC(T) and DIP(T) where AD(T) is the percent of female high school graduates admitted into the AD program in year T [similarly for BAC(T) and DIP(T)]. Examination of the data in tables 44 and 45 lead to the following assumptions about AD(T), BAC(T) and DIP(T): (a) AD(T) could adequately be described for the years being considered (i.e., 1959—60 through 1970—71) by a logistic growth function - __q_ AD(T) _ “beg“, (38) Table 44.—Admissions to professional nursing schools by type of program, 1959—60 through 1964—65 Admissions as a percent of female high school graduates ‘ Number Of admissnons Associ- Associ- Bacca- ate Bacca- ate Year Total Diploma laureate degree Total Diploma laureate degree 1959—60 ._... 5.79 4.71 .89 .19 49,166 40,013 7,555 1,598 1960—61 5.12 4.01 .90 .22 49,487 38,702 8,700 2,085 1961-62 2.. 4.92 3.78 .89 .25 49,805 38,257 9,044 2,504 1962—63 ., _ - 5.03 3.70 .98 .35 49,521 36,434 9,597 3,490 1963—64 ., ., . 5.31 3.83 1.04 .45 52,667 37,936 10,270 4,461 1964—65 - 4.94 3.39 1.01 .53 57,604 39,609 11,835 6,160 ' Number of admlsSlOnS from American Nurses‘ Assomauon data 13, 1966, p. 83) 96 Table 45.—Relative earnings of registered nurses relative to public school teachers (W), 1956-57 through 1964—65 and number of female high school graduates, 1965—66 through 1970-71 Number of female high school graduates2 Year W ' Year (in thousands) 1956—57 Mn”. .785 1965—66 , ,4 _ 1.346 1957—58 . .776 1966—67 _ - 1.348 1958—59 _,-_._,_ .768 1967—68 , , , 1.360 1959—60 .759 1968—69 - _ .._, 1.427 1960—61 .750 1969—70 1.403 1961—62 .750 1970—71 1.487 1962-63 ,W _ .750 1963-64 vb, .750 1964—65 .763 ' Earnings ol registered nurses In 1956—57 are from ANA data. (3, 1958. p. 121). Note that a small adjustment of the figure was made to make the figure correspond WIth wages in later years. The 1960—61 and 1963—64 earnings for registered nurses are from table 26 in this report. Earnings at public school teachers are from Digest of Educational Statistics. (33. p. 48) 2 Female graduates figures are from Projections of Educational Statistics. (34, p. 47) where q, b, and parameters to be estimated, (b) BAC(T) could be described for the years being considered by BAC(T) = a, + B1(T) (39) where al and B, are parameters to be estimated, and (c) DIP(T) could be described for the years being considered by DIP(T) = a2 + BZ(T). (40) Note that W was not considered in predicting AD(T), BAC(T), and DIP(T) for the current projections. The reason for this was that for the time period being considered, W was relatively constant (see table 45). Using the assumptions given above, the following equations were used to project AD(T), BAC(T), and DIP(T). (The data given in table 44 for 1959—60 were used to estimate the unknown parameters in equations 38, 39, and 40.) 2.0 = -___ 4 AD(T) 1+2.77e‘-2““’ ( 1) BAC(T) = 1.0310 + (.0317)t (42) DIP(T) = 3.3876 — (.2063)t (43) Equation 41 was obtained as described on pages 84-87, and equa- tions 42 and 43 were obtained by linear regression. 97 Table 46.—Projected admissions as a percent of female high school graduates to professional nursing schools by type of program, 1956-66 through 1970-71 Year Total DIP BAC AD 1965—66 -... 4.88 3.18 1.06 .64 1966—67 - .-,-.--.. 4.83 2.98 1.09 .76 1967—68 .. 4.78 2.77 1.13 .88 1968—69 ,,,,,,,, 4.74 2.57 1.16 1.01 1969—70 4.69 2.36 1.19 1.14 1970—71-....._ .-- 4.64 2.16 1.22 1.26 Table 47.—Projected admissions of professional nursing schools by type of program Year DIP BAC AD Total 1965—66 ............. 42,962 14,321 8,046 65,929 1966—67 --.---.. .. .- 40,111 14,671 10,230 65,012 1967—68 ...-.-.-_._ 37,340 15,232 11,862 64,434 1968—69 -_ .. ._ . 34,952 15,776 13,736 64,464 1969—70 -----. 33,677 16,981 16,268 66,926 1970—71 .- .-... _ .- 31,601 17,849 18,434 67,884 Table 46 gives the projected values of AD(T), BAC(T), DIP(T) and their sum. Using the percents given in table 46 and the number of female high school graduates given in table 45, the number of admissions of the three types of nursing programs were projected from 1965—66 through 1970—71. The results are given in table 47. Finally, the projected number of graduations from the three types of nursing programs were projected through 1970—71, using the admissions given in table 47 and the following graduation comple- tion rates: Program Completion Rate AD .56 BAC .59 DIP .74 The above graduation completion rates were those observed in 1964—65. Table 48 presents the projected graduations. Using the data displayed in table 43, the model was exercised using two activity rate assumptions. The assumptions are: (a) no change from the year estimates, and (b) a constant change of 1.16 percent per year in each age category. 98 Table 48.—Projected graduations from professional nursing schools by type of program Year DIP BAC AD Total 1965-66 ,,,,,,,,,,,,, 28,073 5,662 3,450 37,185 1966—67 ,, ._ , 29,311 6,059 4,842 40,212 1967—68 .. ,, ,1 31,792 6,983 5,729 44,504 1968—69 _, ,, 29,682 8,449 6,643 44,774 1969—70 , , ,, we 27,632 8,656 7,692 43,980 1970—71 H ,1, 25,864 9,041 9,110 44,015 Table 49.—RN supply projections by year Assumption 1 ‘ Assumption 2 2 Year Supply Stock Supply Stock 1966 __,, _ , ,, we 604,290 895.000 604,290 895,000 1967 ,, W, 630,088 922,040 630,088 922,040 1968 n.1,, - “,1. 649,906 952,813 657,314 952,813 1969 _, ,, i-” , , 671,959 986,747 687,355 986,747 1970 _, ,, ,_ , ,, , , 692,802 1,018,992 716,736 1,018,992 1971 _,,, , , 714,310 1,052,892 747,384 1,052,892 1972 ,1 .. ,u. ,, 737,789 1,090,369 780,717 1,090,369 ‘ No change In actwnty rates ‘1 Constant change In actlvny rates. The resulting RN stock and supply projections by year are pre- sented in table 49. Using the preceding data, projections of RN stock and supply were made for the time period 1966 to 1972. The values are shown in table 49. The 1972 projected values of RN stock and supply for assump- tions 1 and 2 are compared with adjusted ANA Inventory data in table 50. The comparison shows the stock to be off for both assumptions by 16,000 or 1.4 percent. The supply for assumption 1 is off by 42,000 or 5.4 percent. For assumption 2 the projected supply is essentially the same as the adjusted inventory value. 99 Table 50.—Comparison of projected 1972 values of RN stock and supply with adjusted ANA 1972 Inventory data Stock Supply Assumption 1 ,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,, 1,090,000 738,000 Assumption 2 ,,,,,,,,,,,,,,,,,,,,,,,,,,,,,, 1,090,000 781,000 Adjusted inventory ”cw-” ,_,v._-r,,_ t ,rr. 1”. 1,106,000 780,000 Conclusion In both the estimation and projection, the RTI model computed value of the RN stock is within 2 percent of the best known value of the actual stock. Using assumption 1, the estimated supply was about 4.2 percent, and the projected supply about 5.4 percent different than the actual supply. With assumption 2 the estimated supply was about 1.4 percent and the projected supply about .1 percent of actual supply. The RTI model appears to be valid, particularly in estimating and projecting the stock of RN’s. The sensitivity of the estimates and projections to activity rates results in the computed values of the supply being less certain. VII. REGISTERED NURSE SUPPLY PROJECTIONS 101 This section presents the set of projections made by RTI’s RN supply model. Each projection set provides estimates of RN stock and supply for the origin year, 1973, and for the projection years 1974, 1975, 1980, and 1985. Each projection set given in table 51 corresponds to a specific set of assumptions. These assumptions are: (1) The age distribution of new US. nursing school graduates will remain constant, (2) The age distribution of new foreign nurse supply will remain constant, (3) The licensure rate of new US. nursing school graduates will remain constant, (4) The percentage of nurses holding multiple licenses will remain constant, (5) The number of reinstated nurses will remain constant, (6) The number of new foreign nurses will remain constant, Table 51.—Stock and supply projections for selected years and assumptions Assumptions Year Stock Supply Alternative 1: W constant at .9 after 1973 1,132,000 809,000 1972; activity rates constant at 1972 1974 1,164,000 832,000 value 1975 1,192,000 852,000 1980 1,278,000 928,000 1985 1,300,000 950,000 Alternative 2: W has no effect; activity 1973 1,132,000 809,000 rates constant at 1972 value 1974 1,184,000 845,000 1975 1,237,000 884,000 1980 1,524,000 1,100,000 1985 1,841,000 1,322,000 Alternative 3: W constant at .9 after 1973 1,132,000 809,000 1973; activity rates increase at 1974 1,164,000 838,000 1966—72 rate 1975 1,192,000 865,000 1980 1,278,000 981,000 1985 1,300,000 1,041,000 Alternative 4: W has no effect; activity 1973 1,132,000 809,000 rates increase at 1966—72 rates 1974 1,184,000 852,000 1975 1,237,000 897,000 1980 1,524,000 1,158,000 1985 1,841,000 1,444,000 102 (7) Mortality rates will remain constant, and (8) The rate of increase of new US. graduates will begin to decline in 1974 and the total number of new US. graduates will peak in 1984. In addition, each projection set corresponds to a specific combina- tion of nurse-teacher earnings ratio (W) and activity rate (a). Three values for each were considered; however, only those values that resulted in a high or low projected value of RN stock and supply are presented in table 51. The value of nurse-teacher earnings ratio considered are: (1) W held constant at .9 after 1972, (2) W had no effect on projections of renewed licenses but was the same as (3) below for entrants into nursing schools, Figure 12.—Projected trends in RN stock and supply ALTERNATIVE 1.9 _ 2 & 4 1.8 — I 1.7 - / 1.6 — ,’ 1.5 — / 1.4 — / 1.3 _ / 12 _ RN STOCK I 1.1 - x’ 1.0 - I Number of RNs (millions) \ RN SUPPLY 1962 66 70 74 78 82 86 YEAR 103 (3) W increased from .899 in 1972 to 1.00 in 1981 and held constant thereafter. Only numbers 1 and 2 are included in table 51. The values of activity rates that were considered were: (1) Activity rate held constant at the 1972 value, (2) Activity rate increased at the 1966—72 rate, and (3) Activity rate increased in proportion to the increase in the female labor force participation rate. Only 1 and 2 above are included in table 51. The projections that correspond to these four assumptions are presented in table 51 and figure 12. As this table and figure indicate, the model is extremely sensitive (in absolute terms) to changes in the activity rates (a) and earnings ratio (W) variables. 105 VIII. CONCLUSION AND RECOMMENDATIONS The objective of the study was to develop a practical model for accurately projecting U.S. nurse supply. The model developed by the Research Triangle Institute (RTI) provides a relatively practi- cal model for making projections of nurse supply that, in a retro- spective validity check, is more accurate than any existing model. Because of the increased accuracy, it is recommended that the model be used annually to update national projections of nurse supply. The RTI model is similar to the models used by the Division of Nursing and that developed by Altman in that it is a flow model. Many of the factors that have been shown to affect the supply of nurses have been included in the RTI model. This has increased the complexity of the model; however, national data sources exist that permit annual adjustment of model parameters. This ability to annually adjust the parameter values greatly enhances the near future projections of nurse supply. The RTI model is based on the licensed stock of nurses and the flow into and out of this resource pool. The supply of RN’s is estimated by multiplying the stock by the activity rate (the propor- tion of licensed nurses that are active). The licensed stock was used for two primary reasons: (1) The ANA inventory of registered nurses provides reliable data for determining the base year age-specific stock and for computing the age—specific activity rate, and (2) The availability of annual licensure data which permits ad— justing the size of the stock of licensed nurses. If the ANA inventory is discontinued or if census data relating to RN’s becomes more reliable, then alternative models may become more appropriate. In reviewing the literature and performing analyses, some of the primary factors that affect the supply of nurses are: 0 Husband’s income 0 Presence and number of children 0 Nurse’s age 0 Relative wages. Though several analyses were performed by RTI, the study was primarily to be based on existing knowledge; hence, there are some areas where additional investigation should be conducted. These are as follows: 0 Present estimates of the number of admissions to schools of nursing are based on the number of female high school gradu- 106 ates. It is known that many of the admissions to the associate degree program do not enter directly out of high school. Since the associate degree program is the dominant program, consid- eration should be given to determining if there is a better means for estimating the number of admissions. 0 Characteristics related to race apparently affect the labor force participation rate. Consideration should be given to including race in the projection model. Model projection can be improved if reliable national annual data can be obtained on: (1) the number of nurses who have allowed all licenses to expire; (2) the actual number of nurses who have been reinstated; (3) the number of foreign trained nurses who are licensed for the first time in the United States; and (4) nurses’ and teachers’ salaries. 10. 11. 12. 13. 14. 107 IX. REFERENCES . Altman, Stuart H., Present and Future Supply of Registered Nurses. DHEW Publication No. (NIH) 72—134, Division of Nursing, National Institutes of Health, Public Health Service, November 1971. Washington, DC: US. Government Printing Office. . American Nurses’ Association. The 1962, 1966, and 1972 In- ventories of Registered Nurses, 1965, 1969, 1974. New York: American Nurses’ Association. . American Nurses’ Association. Facts About Nursing, 1958— 1973 editions. Kansas City, Missouri: American Nurses’ Asso- ciation. . US. Department of Health, Education, and Welfare. Statisti- cal Abstracts, 1973. Vital Statistics of the United States, Washington, DC: US. Government Printing Office. . Knopf, Lucille, From Student to RN. A Report of the Nurse Career-Pattern Study. DHEW Publication No. (NIH) 72—130, Division of Nursing, National Institutes of Health, Public Health Service, 1972. Washington, DC: US. Government Printing Office. . Johnson, Walter L. “Admission of Men and Ethnic Minorities to Schools of Nursing 1971—1972.” Nursing Outlook, January 1974. . National League for Nursing. “Educational Preparation for Nursing: 1969.” Nursing Outlook, September 1970. . Bauer, David. “The Foreign—Born and US. Manpower Sup- plies.” The Conference Board Record, February 1971. . Wogan, Lynn. “No Simple Prescription to End Nursing Short- age.” The Raleigh Times, July 25, 1974. Unpublished data made available through the courtesy of the Bureau of Health Resources Development and the National League for Nursing. Informal communication from the Division of Nursing, Bu- reau of Health Resources Development, July 18, 1974. Bognanno, M.F., J.S. Hixson, J.R. Jeffers. “The Short-Run Supply of Nurses’ Time.” The Journal of Human Resources, IX, 1972. Burke, Sandra. Inactive Nurses—A Missouri Study. Missouri Division of Health, June 1968. Benham, Lee. “The Labor Market for Registered Nurses: A Three-Eq uation Model.” The Review of Economics and Statis- 108 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 23. 27. 28. 29. tics, Vol. 53, No. 3, August 2, 1971, Massachusetts: Harvard University Press. American Nurses’ Association, unpublished data from the 1972 Inventory of Registered Nurses. Owen, John D. The Price of leisure: An Economic Analysis of the Demand for Leisure Time, 1969, Rotterdam: Rotterdam University Press. Yett, Donald E. “The Supply of Nurses: An Economist’s View.” Hospital Progress, February 1965. Archibald, K.A. The Supply of Professional Nurses and Their Recruitment and Retention by Hospitals, July 1971, The New York City Rand Institute. Bognanno, Mario F. An Economic Study of the Hours ofLabor Offered by Registered Nurses, 1969, Ph.D. dissertation, Uni- versity of Iowa. Cohen, Malcolm 8., Samuel A. Rea, Jr., and Robert L. Ler- man. A Micro Model ofLabor Supply. BLS Staff Paper 4, 1970, Washington, D.C.; U.S. Government Printing Office. Kosters, Marvin. Income and Substitution Effect in a Family Labor Supply Model, 1966, Santa Monica, California: Rand Corporation. Benham, Lee. An Economic Analysis of the Labor Market for Registered Nurses, 1970, Ph.D. dissertation, Stanford Univer- sity. U.S. Bureau of the Census. Public Use Samples of Basic Records from the 1970 Census: Description and Technical Documentation, 1972, Washington, D.C.; U.S. Government Printing Office. U.S. Bureau of the Census. 1970 Census Users’ Guide, 1970, Washington, D.C.; U.S. Government Printing Office. U.S. Bureau of the Census. Census of Population: 1970, Detailed Characteristics, 1970, PC (1)—D Series. Washington, D.C.; U.S. Government Printing Office. Johnston, J. Econometric Models, 1963, New York: McGraW- Hill Book Company. Cain, Glen G. Married Women in the Labor Force: An Eco- nomic Analysis, 1966, Chicago: University of Chicago Press. Goldfield, SM. and RE. Quandt. Nonlinear Methods in Econo- metrics, 1972, Amsterdam: North—Holland Publishing Com- pany. Ashenfelter, Orley and James Heckman. “The Estimation of Income and Substitution Effects in a Model of Family Labor Supply.” Econometrics, XLII, N0. 1, 1974, pp. 73—85. 30. 31. 32. 33. 34. 109 U.S. Department of Health, Education, and Welfare. Health Manpower Source Book, Section 2, Nursing Personnel, PHS Publication No. 263, Division of Nursing, National Institutes of Health, Public Health Service, 1969, Washington, DC: US. Government Printing Office. Meyer, Burton. “Development of a Method for Determining Estimates of Professional Nurse Needs.” Nursing Research, June 1957, pp. 24—28. US. Department of Labor, Bureau of Labor Statistics, Special Labor Force Reports No. 40, 50, 80, and 153, reprinted from the Monthly Labor Review. National Center for Educational Statistics. Digest of Educa- tional Statistics, 1972, DHEW Publication No. (OE) 73—11103. National Center for Educational Statistics. Projections of Educational Statistics, 1972, DHEW Publication No. (OE) 73— 11105. L) S GOVERNMENT PRINTING OFFICE 1976 0 5977536 a.‘ , .3?» L53. DEPARTMENT OF HEALTH. EDUCATION. AND WELFARE PUBLIC HEALTH SERVICE PosTAG: AND was run U.S. DEPARTMENT OF H.E.W. HEALTH RESOURCES ADMINISTRATION "EV/.390 . BETHESDA. MARYLAND 200M OFFICIAL BUSINESS PtNAITv foa PIIVAYE use 3300 DHEW Publication No. (HRA) 76-15 U C BERKELEY LIBRARIES III 'IIIIIIIIIIIIII“ CUEHBBHUB?