TN 295 No. 9082 . o » o ^ v ^^ °%W : a^ : ^ ^ ... "«k ""-' ,°V V » «o ^ \^W j &* „$ f^ . »5 ^ > -^. V *b>* • **o« • ^» -^ .veto-. ^ > ■ .: o > V . t ' » . G* V ^.7*' a A* * ^ -.1 \«? :. w v»> v»y vs*> *° \« , . /% * / V -.]»• «$*, IflP _/V -.ill; J^ • A v * i • ° I* < » • " « *$ ,0 V \3 •*T7T*" A •^^v^*.. o ^ '<$>, "o»«^ ^ o5°^ x°^ IC 9082 Bureau of Mines Information Circular/1986 Mineral Consumption Forecasting Standardizing and Comparing Forecasts By John B. Bennett UNITED STATES DEPARTMENT OF THE INTERIOR ^Ti^^i^H Information Circular 9082 Mineral Consumption Forecasting Standardizing and Comparing Forecasts By John B. Bennett UNITED STATES DEPARTMENT OF THE INTERIOR Donald Paul Hodel, Secretary BUREAU OF MINES Robert C. Horton, Director As the Nation's principal conservation agency, the Department of the Interior has responsibility for most of our nationally owned public lands and natural resources. This includes fostering the wisest use of our land and water resources, protecting our fish and wildlife, preserving the environment and cultural values of our national parks and historical places, and providing for the enjoyment of life through outdoor recreation. The Department assesses our energy and mineral resources and works to assure that their development is in the best interests of all our people. The Department also has a major responsibility for American Indian reservation communities and for people who live in island territories under ILS. administration. no** Library of Congress Cataloging in Publication Data Bennett, John B. Mineral consumption forecasting (Information circular; 9082) Bibliography: p. 11 Supt. of Docs, no.: I 28.27: 9082. 1. Mineral industries -Forecasting -Mathematical models. I. Title. II. Series: Informa- tion circular (United States. Bureau of Mines); 9082. TN295.U4 [HD9506.A2] 622 s [333.8'513] 86-600082 For sale by the Superintendent of Documents, U.S. Government Printing Office Washington, DC 20402 CONTENTS Paere Abstract 1 Introduction 2 Forecast standardization and comparison ,. 2 Problems in comparing mineral consumption forecasts 2 A standardization methodology and examples 4 GNP growth assumptions 10 Concluding statement 10 References 11 Appendix A. -Review of forecasting methodologies 12 Appendix B.- Implied consumption divided by actual consumption, 1980^83 17 ILLUSTRATIONS 1. Rebased consumption projections for 2000 divided by 1980 actual consumption: United States and world 7 2. Rebased 1980-83 implied consumption divided by 1980-83 actual consumption: United States and world 9 TABLES 1. Mineral consumption forecasts for the year 2000 3 2. Projected consumption in 2000 3 3. Standardization example 4 4. Mineral consumption 1970-83 4 5. Original base 5 6. Adjusted base 5 7. Original base divided by adjusted base 6 8. 1980 consumption and rebased projected consumption in 2000 6 9. Rebased projected consumption in 2000 divided by 1980 actual consumption 6 10. Comparison example: Trend versus actual 8 11. Growth rates used to compute implied values of consumption for 1980-83 8 12. 1980-83 implied consumption divided by actual consumption 9 UNIT OF MEASURE ABBREVIATIONS USED IN THIS REPORT kmt thousand metric tons lb pound MM lb million pounds Mmt million metric tons Mst thousand short tons mt metric ton Mineral Consumption Forecasting: Standardizing and Comparing Forecasts By John B. Bennett 1 ABSTRACT This Bureau of Mines report presents a method of standardizing forecasts of mineral consumption that attempts to resolve the problems caused by the use of different data bases, different definitions of minerals consumption, and different base years for making projections. Using this method, forecasts of U.S. and world mineral consumption, in the year 2000, for nine commodities (aluminum, chromium, cobalt, copper, manganese, nickel, tin, tungsten, and zinc) are standardized and compared. Also selected forecasts or associated growth rates are used to generate implied values of consumption for 1 §80-83, which are then compared to actual values of consumption for 1980-83. The economic growth predictions underlying these forecasts are examined, and some general conclu- sions are drawn. A description of each methodology used to generate these forecasts is given in an appendix. 'Economist, Branch of Economic Analysis, Bureau of Mines, Washington, DC. INTRODUCTION The Bureau of Mines develops forecasts of the future consumption of minerals and materials to aid in identifying and anticipating changes that might affect the national in- terest. Anticipating changes in the level and distribution of mineral demand is important for formulating long-run policies pertaining to the adequacy of mineral supply, for planning investment in U.S. mining facilities, for forming trade and development policies, and for planning the modernization and growth of rich and poor nations alike. The Bureau forecasts, made on both a national and a global basis, assist the Secretary of the Interior in carrying out his responsibilities under the Strategic and Critical Materials Stock Piling Act, which directs him to investigate the pro- duction and utilization of minerals and materials, and also assist him in helping to ensure the continued strength of the domestic mineral and material economy and the main- tenance of an adequate mineral and material base. Fore- casts of future consumption are published every 5 years in detail in Mineral Facts and Problems and, on an interim basis as necessary, in Mineral Commodity Profiles. The Bureau forecasts provide high, low, and probable levels of mineral and material consumption in the United States and the rest of the world as a whole. The Bureau also forecasts U.S. consumption by end use. The comparison of forecasts from different sources, ac- companied by a knowledge of how those forecasts were generated, offers the advantage of insight into how others view the world, and the possibility of learning from those views. The comparison of forecasts, however, frequently is hindered by differences in definitions, use of different base years, and other factors. In this study, a method of standardizing forecasts of mineral consumption is presented that attempts to resolve the problems caused by the use of different data bases, dif- ferent definitions of minerals consumption, and different base years for making projections. These problems have previously remained unresolved, or have been addressed in a patchwork fashion, in earlier works that presented forecasts of different forecasters (3). 2 Using the method described here, forecasts for nine minerals for the year 2000 are standardized and compared, for the United States and the world. The forecasts were for consumption of the follow- ing commodities: aluminum, chromium, cobalt, copper, manganese, nickel, tin, tungsten, and zinc. The forecasts ex- amined in some detail are from publications by Fischman (.4), Leontief (9), Malenbaum (11), Ridker (13), and the Bureau of Mines (14). Also selected forecasts or associated growth rates are used to generate implied values of con- sumption for 1980-83, which are then compared to actual values of consumption for 1980-83. The economic growth predictions underlying these forecasts are then examined, and some general conclusions are drawn. A description of each methodology used to generate these forecasts is given in appendix A. FORECAST STANDARDIZATION AND COMPARISON PROBLEMS IN COMPARING MINERAL CONSUMPTION FORECASTS In comparing forecasts of mineral consumption from various sources, several problems arise. First, the definition of consumption may vary. The consumption of a particular commodity may be defined, for example, as apparent con- sumption, primary consumption, primary plus secondary consumption, or industrial consumption. Sometimes the dis- tinctions between such definitions are made clear when, data are presented, sometimes not. Moreover, even when the same definition of consumption is used, the underlying data series used may not be the same. There are various sources of mineral consumption data, and over time, revisions, up- dates, and changes in definitions occur, not necessarily at the same time, for the different sources. Revisions of data series, in particular, occasionally go back a number of years. Also, the various data series may reflect different degrees of processing -for example, ore series are used occasionally by some forecasters. Third, when forecasts are made at dif- ferent points in time, the data base will cover a different time span, which might lead to different evaluations of like- ly trends. The data in table 1 illustrate these problems. These mineral consumption forecasts were gathered from the works reviewed in appendix A. The forecasts were made at approximately the same time; indeed one of the reasons they were picked was to avoid as much as possible the prob- lem of forecasts made when world conditions are quite dif- ferent. The Fischman, Ridker, and Bureau of Mines projec- tions were published in 1980, Malenbaum's in 1977, and Leontief s in 1983. The forecasts are of U.S. and world con- sumption in the year 2000, for the following commodities: aluminum, chromium, cobalt, copper, manganese, nickel, tin, tungsten, and zinc. They are given in their original units in table 1- short tons, metric tons, and pounds. In table 2, the forecasts are all converted to the same unit, thousand metric tons. The procedures used for converting the forecasts are given in the table notes. Of the five sets of forecasts, four different definitions of consumption are used: apparent, primary, gross domestic demand, and total consumption. Also, while both Fischman and Malenbaum use the apparent consumption concept, Fischman's manganese and chromium projections are in terms of metal, while Malenbaum's are in terms of the respective ores. The ore figures were converted to refined metal using the same ratios used in the Global 2000 (3) study for this purpose. Mineral analysts could disagree with these ore conversion ratios, which serves to illustrate the prob- lem. Since the forecasts are for similar but not identical items, numerical comparisons of the forecasts with each other are strained. Comparisons of forecasts to currently available data are subject to the same kinds of problems. A series of forecasts from various sources, compared to current data, might offer useful information to decision makers or to those currently at work on their own projections. After all, many decisions are made on the basis of forecasts, and new forecasts are generated relatively often. One might argue that a forecast for a future time period, for example the year 2000, cannot be evaluated until that year arrives. At that point its ac- curacy could easily be determined. However, such forecasts italicized numbers in parentheses refer to items in the list of references preceding appendix A. do have implications for the intervening years. In 1999, to take an extreme example, one should be able to evaluate whether or not consumption in the following year will ap- proximate a given forecast for 2000. In a more general sense, these forecasts, in conjunction with the base period used to generate them, have implications for the general path consumption will travel on the way to 2000. If a forecaster, in say 1980, expects consumption of a given mineral to decline by 2000, and instead it increases every year in the 1981-85 period, one might properly doubt the ac- curacy of the forecast of decline. Of course, quite a different evaluation could result some years down the road. Table 1.— Mineral consumption forecasts for the year 2000 (Data in original units) Commodity and area Fischman'(^) Leontief 2 (9) Malenbaum 3 (11) Ridker 4 (73) BuMines*(74) Aluminum: U.S 11.1 Mmt 14.3 Mmt 13,073.0 kmt 16,493.0 Mst 17,200.0 Mst World 46.2 Mmt ND 36,516.0 kmt 60,965.0 Mst 63,700.0 Mst Chromium: U.S ND 1,150.0 kmt 1,601.0 kmt 1,107.0 kmt 1,240.0 Mst World 6.7 Mmt 8,220.0 kmt 16,018.0 kmt 2,016.0 Mst 7,950.0 Mst Cobalt: U.S 33.5 MMIb ND 16,608.0 mt 19.0 Mst 40 MMIb World 96.0 MMIb ND 57,532.0 mt 123.0 Mst 107 MMIb Copper: U.S 3.1 Mmt 4.3 Mmt 3,202.0 kmt 3,265.0 Mst 4,600.0 kmt World 16.3 Mmt ND 16,839.0 kmt 16,418.0 Mst 23,600.0 kmt Manganese: U.S ND 2,390.0 kmt 4,002.0 kmt 2,458.0 Mst 2,000.0 Mst World 17.1 Mmt 35,600.0 kmt 48,060.0 kmt 10,765.0 Mst 19,600.0 Mst N icKsl* U.S ND 355.8 kmt 280.1 kmt 525.0 Mst 600.0 Mst World ND 1,314.1 kmt 2,122.0 Mst 2,500.0 Mst Tin: U.S ND 113.0 kmt 67.0 kmt 114.0 Mst 65,000.0 kmt World ND 727.0 kmt 393.0 kmt 582.0 Mst 313,800.0 kmt Tungsten: U.S ND 17.4 kmt 12,006.0 mt 15.0 Mst 61 MMIb World ND 125.0 kmt 92,637.0 mt 69.0 Mst 219 MMIb Zinc: U.S 1.5 Mmt 3.5 Mmt 2,001.0 kmt 2,616.0 Mst 1,800.0 kmt World 10.7 Mmt ND 12,022.0 kmt 12,819.0 Mst 10,600.0 kmt ND No data. 1 Apparent consumption. 2 Primary consumption; technological change assumed. 3 Apparent consumption; Cr and Mn as ore, primary Al. 4 Gross domestic demand tor U.S. data; total consumption, primary and secondary, for world data; base scenario for U.S. data; standard case for world data. s Probable total consumption, primary and secondary, for both U.S. and world data. Table 2.— Projected consumption in 2000 (All data converted to thousand metric tons) 1 Commodity and area Fischman 2 (4) Leontief 3 (9) Malenbaum" (11) Ridker*(73) BuMines 6 (74) Aluminum: United States 11,100.0 14,340.0 13,073.0 14,962.4 15,603.8 World 46,200.0 ND 36,516.0 55,307.4 57,788.6 Chromium: United States ND 1,150.0 438.7 1,004.3 1,124.9 World 6,670.0 8,220.0 4,388.9 1,828.9 7,212.2 Cobalt: United States 15.2 ND 16.6 17.2 18.1 World 43.5 ND 57.5 111.6 48.5 Copper: United States 3,100.0 4,290.0 3,202.0 2,962.0 4,600.0 World 16,300.0 ND 16,839.0 14,894.4 23,600.0 Manganese: United States ND 2,390.0 1,921.0 2,229.9 1,814.4 World 17,100.0 35,600.0 23,068.8 9,766.0 17,781.1 Nickel: United States ND 355.8 280.1 476.3 544.3 World ND ND 1,314.1 1,925.1 2,268.0 Tin: United States ND 113.0 67.0 103.4 65.0 World ND 727.0 393.0 528.0 313.8 Tungsten: United States ND 17.4 12.0 13.6 27.7 World ND 125.0 92.6 62.6 99.3 Zinc: United States 1,470.0 3,520.0 2,001.0 2,373.2 1,800.0 World 10,730.0 ND 12,022.0 11,629.4 1 0,600.0 ND No data. ' Short tons multiplied by 0.9072, metric tons divided by 1,000, million metric tons multiplied by 1,000, pounds converted to short tons (divided by 2,000), manganese ore multiplied by 0.48, chrome ore multiplied by 0.274. 2 Apparent consumption. 3 Primary consumption; technological change assumed. 4 Apparent consumption; Cr and Mn as ore, primary Al. 5 Gross domestic demand for U.S. data; total consumption, primary and secondary, for world data; base scenario for U.S. data; standard case for world data. 6 Probable total consumption, primary and secondary, for both U.S. and world data. A STANDARDIZATION METHODOLOGY AND EXAMPLES The problem in mineral consumption forecast com- parison is identical to the familiar apples and oranges prob- lem. How do you compare values when they measure dif- ferent things? The problem would be eliminated if the forecasts could be recast into the same units or into a unit- free form. The following standardization procedure was developed to accomplish both these ends. The procedure is demonstrated first with a hypothetical example and then with data ^'•awn from the various works reviewed in appen- dix A. In the hypothetical example given in table 3, it is as- sumed that there are two forecasters, forecaster X and forecaster Y, using the base years 1975 and 1980, respec- tively. Forecaster X projects the level of consumption in the year 2000 to be 2,500 units, while forecaster Y projects that level to be 3,000 units. In other words, forecaster X expects consumption of this commodity to be 2V2 times its 1975 level in the year 2000, while forecaster Y expects consumption of the commodity, in the year 2000, to be twice its 1980 level. Suppose that another set of data is available, covering both the years 1975 and 1980. For the purpose of illustration, this latter set of data will be called the basic data. Suppose Table 3. —Standardization Example Base Year Projection (2000) Data Source 1975 1980 Original Rebased Forecaster X 1,000 NAp 2,500 3,000 Forecaster Y NAp 1,500 3,000 3,200 1,200 1,600 NAp NAp that both forecasters X and Y had used the basic data values for 1975 (1,200) and 1980 (1,600), respectively, instead of their original values, in making their projections. Then, assuming they still projected the same rate of increase be- tween their base years and the year 2000, the new or re- based projections would be 3,000 and 3,200 units, respec- tively. Now if the basic data (1,200 and 1,600) are in the same units, the new projections will also be in the same units, and comparisons can be readily made. This procedure was carried out on data collected from the works reviewed in appendix A. First, the base data from the various works were collected. None of the studies reviewed used the same base data set, even those published at the same time. The various bases used included a 1971-75 average, a 1975-77 average, 1978, 1972, 1970, and 1971. All of the base year data, with the exception of Ridker's, who used 1971, were published in the s works reviewed. Each forecast for the year 2000 was then divided by the base year figure. The resulting ratios showed the amount of increase between the base year and the year 2000. Next, a basic table of mineral consumption was con- structed, for the United States and the world, using the latest data available at the time of writing. This basje table included values for each of the nine commodities listed in table 1 over the period 1970-83; values of world chromium and manganese production are used in this table to repre- sent consumption data for these minerals, as no world con- sumption data were available. The period 1970-83 was used because it included every base year used by the various forecasters considered in this study. These data are presented in table 4. From this basic table, an adjusted ver- sion of the base data used by the various forecasters was compiled. A new value for each base year (or average) used Table 4.— Mineral consumption, 1970-83, thousand metric tons Year and area Al Cr' Co Cu Mn 1 Ni Sn W Zn 1970: United States 3,871.0 498.1 7.3 1,883.0 1,203.9 202.7 66.8 8.2 1,235.0 World 12,160.1 1,868.8 22.1 7,283.7 8,204.7 566.6 227.0 ND 5,055.9 1971: United States 4,347.3 366.5 6.2 1,880.0 1,061.4 180.2 62.1 6.7 1,199.0 World 12,936.9 1,981.3 18.4 7,349.7 9,070.2 516.2 235.2 31.8 5,172.3 1972: United States 5,025.9 508.0 8.8 2,185.0 1,239.2 213.5 61.2 7.1 1,377.0 World 14,156.8 1,981.3 26.5 7,984.7 9,082.9 573.7 235.2 35.5 5,709.4 1973: United States 5,941.0 545.2 10.0 2,208.0 1,409.8 239.4 65.1 9.9 1,474.0 World 16,359.9 1,999.5 29.9 8,445.9 9,707.0 652.3 253.3 39.2 6,283.0 1974: United States 5,625.0 567.0 10.9 2,210.0 1,353.5 256.8 72.6 10.7 1,311.0 World 16,705.9 2,207.2 29.2 8,390.8 9,253.4 703.8 243.9 37.4 5,998.1 1975: United States 4,079.0 372.0 6.4 1,467.0 1,027.9 198.9 53.7 6.3 1,331.0 World 13,891.5 2,540.2 21.8 7,457.5 9,797.8 576.2 218.0 32.7 5,092.4 1976: United States 5,196.0 479.9 9.1 1,946.0 1,237.4 220.5 57.9 7.8 1,298.0 World 16,857.8 2,669.0 24.5 8,538.7 9,979.2 670.3 239.1 36.1 5,764.4 1977: United States 5,649.0 517.1 8.2 2,045.0 1,381.7 231.3 58.3 8.5 1,154.0 World 17,563.3 2,883.1 24.5 9,056.4 8,709.1 642.2 231.2 42.2 5,808.4 1978: United States 6,111.0 535.2 9.1 2,369.0 1,236.5 247.4 58.9 10.1 1,154.0 World 18,569.0 2,820.5 25.9 9,530.4 8,618.4 697.4 231.5 48.6 6,209.3 1979: United States 6,030.0 553.4 8.6 2,432.0 1,134.0 204.8 53.5 10.8 932.0 World 19,452.8 2,935.7 24.5 9,829.8 9,797.8 750.2 236.8 51.2 6,323.8 1980: United States 5,223.0 532.5 7.7 2,175.0 933.5 186.7 46.7 9.9 951.0 World 18,773.0 2,972.9 22.7 9,361.3 9,707.0 715.3 234.6 49.1 6,124.3 1981: United States 5,209.0 462.7 5.9 2,278.0 931.7 186.9 52.5 10.3 1,146.0 World 18,216.5 2,786.0 18.6 9,508.0 8,437.0 663.0 225.6 47.0 5,994.4 1982: United States 4,811.0 289.4 5.0 1,761.0 609.6 163.7 30.3 6.1 869.0 World 17,823.5 2,489.4 16.3 9,067.6 8,709.1 651.8 215.4 40.0 5,916.9 1983: United States 5,442.0 298.5 7.3 2,020.0 606.0 185.4 38.1 6.5 1,005.0 World 19,358.6 2,490.3 20.9 9,113.2 7,983.4 683.0 215.2 41.1 6,132.0 ND No data. ' World production data used to represent world consumption data. Sources: U.S. Bureau of Mines, Mineral Facts and Problems, 1980 and 1985 Editions, for all data except as follows: Cobalt data from William Kirk, commodity specialist, Bureau of Mines; world data for Al, Cu, Ni, Sn, and Zn from World Metal Statistics; world data for W from Tungsten Statistics. by each forecaster was chosen (or computed). For example, if the forecaster used 1972 values as the base of his U.S. predictions, as did Leontief, new values for 1972 were taken from the basic table to form an adjusted base for Leontief. Thus, the values of the base year used by each forecaster were cast into the same unite, from the same table. The original and adjusted data bases are shown in tables 5 and 6, respectively. Table 7 shows the ratios of the values of table 5 to those of table 6. These ratios clearly show the differences in the underlying data bases and demonstrate the need for such an adjustment. Values from the new, "adjusted" data base were then used, along with the ratio of each forecast for 2000 to the original base, to recompute the 2000 predictions. This was done for each forecaster, except Ridker, since the tetter's base data were not known. Although the base years of the various forecasts in this recomputed form were still dif- ferent, all predictions were now based on data in the same unite coming from the same table. These recomputed or rebased values are given in table 8. The recomputed values were then divided by the actual values of 1980 consumption from table 4. The resulting ratios provide a unit-free method of comparing the expected change between 1980 and 2000 among the various forecasters. Since these ratios determine growth rates between 1980 and 2000, this method can also be used to standardize growth rates- i.e., the effective growth rates for various forecasters between a common year (in this case 1980) and the year 2000 can be deter- mined. The ratios are given in table 9 and are displayed graphically in figure 1. Table 5.— Original base, thousand metric tons Commodity and area Fischman ' (4) Leontief 2 (9) Aluminum: United States 5,480.0 4,800 World 16,050.0 ND Chromium: United States ND 460.5 World 3,680.0 2,100 Cobalt: United States 9.1 ND World 26.8 ND Copper: United States 1,890.0 1,760 World 8,370.0 ND Manganese: United States ND 1,236.8 World .' 9,000.0 8,240 Nickel United States ND 152 World ND ND Tin: United States ND 49.85 World ND 222 Tungsten: United States ND 6.45 World ND 33.6 Zinc: United States 1,080.0 1,310 World 5,750.0 ND ND No data. ' Average 1973-77. 2 1972 U.S. data, 1970 world data. 3 Average 1971-75. Malenbaum 3 (77) BuMines"(74) 4,388.1 6,111.8 12,248.7 18,270.1 314.8 498.96 1,901.8 3,538.1 7.282 9.18 23.434 24.66 1,886.1 2,380 7,922.7 10,100 926 1,236.5 9,762.7 8,692.8 160.5 217.4 618.4 881.8 52.6 53.9 232.5 281 6.51 10.21 40.18 47.29 1,159.8 1,229 5,506.26 6,779 1978. Table 6.— Adjusted base, thousand metric tons Commodity and area Fischman ' (4) Leontief 2 (9) Aluminum: United States 5,298.0 5,025.9 World 16,275.7 ND Chromium: United States ND 508.0 World 2,459.8 1,868.8 Cobalt: United States 8.9 ND World 26.0 ND Copper: United States 1,975.2 2,185.0 World 8,377.9 ND Manganese: United States ND 1,239.2 World 9,489.3 8,204.7 Nickel: United States , ND 213.5 World ND ND Tin: United States ND 61.2 World ND 227.0 Tungsten: United States ND 7.1 World ND ND Zinc: United States 1,313.6 1,377.0 World 5,789.3 ND ND No data. ' Average 1973-77. 2 1972 U.S. data, 1970 world data. 3 Average 1971-75. Malenbaum 3 (77) BuMines 4 (74) 5,003.6 14,810.2 6,111.8 18,569.5 471.7 2,141.9 535.2 2,820.5 8.4 25.2 9.1 25.9 1,990.0 7,925.7 2,369.0 9,530.4 1,218.4 9,382.3 1,236.5 8,618.4 217.8 604.5 247.4 697.4 62.9 235.8 58.9 231.5 8.2 35.3 10.1 48.6 1,338.4 5,651.0 1,154.0 6,209.3 1978. Table 7.— Original base divided by adjusted base Commodity and area Fischman (4) Leontief (9) Aluminum: United States 1.03 0.96 World .99 ND Chromium: United States ND .91 World 1.50 1.12 Cobalt: United States 1.02 ND World 1.03 ND Copper: United States .96 .81 World 1.00 ND Manganese: United States ND 1.00 World .95 1.00 Nickel: United States ND .71 World ND ND Tin: United States ND .81 World ND .98 Tungsten: Urifted States ND .91 World ND ND Zinc: U.S .82 .95 World .99 ND ND No data. Malenbaum (11) BuMines (14) 0.88 .83 1.00 .98 .67 .89 .93 1.25 .87 .93 1.01 .95 .95 1.00 1.00 1.06 .76 1.04 1.00 1.01 .74 1.02 .88 1.26 .84 .99 .92 1.21 .79 1.14 1.01 .97 .87 .97 1.06 1.09 Table 8.— 1980 consumption and rebased projected consumption In 2000, Commodity and area Actual 1980' Fischman (4) Leontief (9) Aluminum: United States 5,223.0 10,731.4 15,552.6 World 18,773.0 46,849.7 ND Chromium: United States 532.5 ND 1,268.6 World 2,972.9 4,458.4 7,315.0 Cobalt: United States 7.7 14.9 ND World 22.7 42.2 ND Copper: United States 2,175.0 3,239.7 5,325.9 World 9,361.3 16,315.4 ND Manganese: United States 933.5 ND 2,394.6 World 9,707.0 18,029.7 35,447.5 Nickel: United States 186.7 ND 499.8 World 715.3 ND ND Tin: United States 46.7 ND 138.7 World 234.6 ND 743.4 Tungsten: United States 9.9 ND 19.2 World 49.1 ND ND Zinc: United States 951.0 1,788.0 3,700.0 World 6,124.3 10,803.3 ND ND No data. ' Data from Mineral Facts and Problems, 1985. thousand metric tons Malenbaum (11) BuMines (14) 14,906.7 44,152.4 15,601.8 58,735.6 657.4 4,943.0 1,206.7 5,749.4 19.1 61.8 17.9 50.9 3,378.4 16,845.4 4,578.7 22,269.1 2,527.6 22,169.9 1,814.4 17,628.9 380.1 1,284.6 619.4 1,793.7 80.1 398.6 71.0 258.5 15.1 81.4 27.5 102.1 2,309.1 12,338.0 1,690.2 9,709.2 Table 9.— Rebased projected consumption In 2000 divided by 1980 actual consumption Commodity and area Fischman (4) Leontief (9) Malenbaum (11) Aluminum: United States 2.05 2.98 2.85 World 2.50 ND 2.35 Chromium: United States ND 2.38 1.23 World 1.50 2.46 1.66 Cobalt: United States 1.93 ND 2.48 World 1.86 ND 2.73 Copper: United States 1.49 2.45 1.55 World, 1.74 ND 1.80 Manganese: United States ND 2.57 2.71 World 1.86 3.65 2.28 Nickel: United States ND 2.68 2.04 World ND ND 1.80 Tin: United States ND 2.97 1.72 World ND 3.17 1.70 Tungsten: United States ND 1.93 1.52 World ND ND 1.66 Zinc: United States 1.88 3.89 2.43 World 176 ND 2.01 ND No data. BuMines (14) 2.99 3.13 2.27 1.93 2.32 2.24 2.11 2.38 1.94 1.82 3.32 2.51 1.52 1.10 2.77 2.08 1.78 1.59 5 2 CO UNITED STATES / / / / / / / & KEY Fischman (4) r~]Leontief (9) Malenbaum [11) BuMines [14) Al Cr Co Cu Mn Ni Sn Zn KEY Fischman (4) f~ni_eontief (9) Malenbaum [11) BuMines [14) Al Cr Co Cu Mn Ni Sn W Zn Figure 1.— Rebased consumption projections for 2000 divided by 1980 actual consumption: United States and world. The related problem of comparing forecasts to currently available data was approached in the following manner. If the base year and base data are given, along with the fore- cast, growth rates of consumption can be computed. Such growth rates are in fact commonly included in studies con- taining forecasts. With these growth rates, and the base year data, values of consumption can be computed for the interval of years between the base year and forecast year. These values would not be expected to be equal to actual values every year, or even any year, of the forecast period, since mineral consumption fluctuates from year to year, depending on, among other things, the business cycle. Still these values, when accumulated over a period of years, should give an idea of the cumulative consumption the fore- caster expected in that period. This sum would also give an idea, when compared to actual values, of how well that forecast seemed to fit the real world at some point in time. Of course, the accuracy of a given forecast, as judged by this method, would vary with the period chosen as the evalua- tion period. A hypothetical example of this approach is given in table 10. Forecaster X (of table 3) projected a 2V2 times increase in consumption between 1975 and 2000. This rate of in- crease implies a growth rate of 3.7%. Using the base year of forecaster X (1975) and the value of that year (1,200) from the basic data series of table 3, values of consumption for 1980, 1981, 1982, and 1983 were determined; these values are simply those values that result in the years 1980, 1981, 1982, and 1983 from a growth of 3.7% beginning in 1975 at a value of 1,200. Accompanying those values in table 10 are "actual" values, meant to represent actual mineral consump- Table 10.— Comparison example: Trend versus actual with forecaster X's growth rate (3.7%) and new base (1,200 units in 1975) Year Forecaster X trend values "Actual" values 1980 1981 1982 1983 1,440 1,488 1,548 1,608 1,200 1,400 1,200 1,800 Total . 6,084 5,600 tion, which does not grow along a smooth trend, but rather fluctuates. These "actual" values are also assumed to come from the basic data series and therefore are in the same units as the computed values. The sums of the two sets of 1980-83 values are shown; in this case, the sum of the "im- plied" values is fairly close to the sum of the "actual" values. This approach was carried out on the data gathered in this study in the following manner. First, growth rates that included the period 1980-83 were gathered from the various studies, if they were available. If they were unavailable, they were computed. These growth rates are given in table 11. These growth rates are not strictly comparable, since they cover various periods of time. Next, using these growth rates and the adjusted data base for each forecaster, "implied" values of consumption for 1980, 1981, 1982, and 1983 were calculated. These values were summed over each commodity, for each forecaster, and divided by the actual value of consumption over that period, taken from the basic table (table 4). The resulting ratios, given in table 12 and displayed graphically in figure 2, were computed from growth rates projected -or computed using projections -by Fischman, Leontief, Malenbaum, Ridker, and the Bureau. Appendix B contains similar ratios calculated on a year-by- year basis. Although Ridker did not publish base year data, he did publish growth rates for the consumption of various minerals in the United States for 1971-85. These growth rates, along with values for 1971 consumption from table 4, were used to calculate "implied" values of consumption for 1980-83. From the ratios presented in table 9 and figure 1, it is clear that quite a range exists for estimates of mineral con- sumption in 2000, at least for some minerals. Also, the ratios presented in table 12 and figure 2 show that the sum of the "implied" values of the forecasts or associated growth rates also followed the sum of the actual values of consump- tion for the period 1980-83 with varying degrees of ac- curacy, as one would expect given the range shown in table 9. In table 12 those ratios close to 1 indicate "implied" values quite close, in aggregate, to actual values over that period. As mentioned earlier, this kind of evaluation depends on the period of time chosen, and might show quite different results in a later time period. Table 11.— Growth rates used to compute implied values of consumption for 1980-83 Commodity and area Fischman'(4) Leontief 2 (9) Malenbaum 3 (11) Ridker 4 (73) BuMines 5 (74) Aluminum: United States 4.0 4.0 4.3 4.83 4.4 World 5.3 ND 4.2 ND 5.4 Chromium: United States ND 3.3 1.3 3.01 3.4 World 2.7 5.0 3.2 ND 3.3 Cobalt: United States 3.0 ND 3.2 3.41 2.9 World 2.3 ND 3.6 ND 3.0 Copper: United States 1.9 3.2 2.6 2.15 3.0 World 3.1 ND 2.9 ND 3.9 Manganese: United States ND 2.4 3.4 2.37 1.4 World 2.6 5.3 3.2 ND 2.7 NicKsl' United States ND 3.1 2.2 3.17 4.0 World ND ND 3.1 ND 4.3 Tin: United States ND 3.0 .9 2.16 .9 World ND 4.2 2.1 ND .9 Tungsten: United States ND 3.6 2.1 3.90 4.6 World ND 4.7 3.3 ND 3.5 Zinc: United States .7 3.6 2.6 2.36 1.7 World 2^5 ND 3J3 ND 2J ND No data. 'Computed; U.S. and world 1975-85. 'Computed; U.S. 1972-2000; world 1970-90. 3 Published; U.S. and world 1975-85. " Published; U.S. 1975-85. 5 Published; U.S. and world 1978-2000. Table 12. Commodity and area Aluminum: United States World Chromium: United States World Cobalt: United States World Copper: United States World Manganese: United States World Nickel: United States World Tin: United States World Tungsten: United States World Zinc: U.S World ND No data. -1980-83 Implied consumption divided by actual consumption (Adjusted data base of each forecaster) Fischman (4) Leontief (9) Malenbaum (77) 1.41 ND 1.27 1.04 1.75 1.22 1.29 .98 ND ND 1.60 1.62 1.44 ND 1.14 1.03 2.01 1.71 1.97 1.33 1.58 ND 1.39 1.09 1.93- 1.63 1.58 1.21 1.21 ND 1.15 .98 1.94 ND 1.59 1.15 Ridker (73) BuMines (14) 1.32 1.23 ND 1.09 1.67 1.53 1.08 1.10 ND 1.29 ND ND ND ND ND ND 1.38 1.13 1.38 ND 1.27 ND 1.37 ND 1.14 ND 1.76 ND 1.38 ND 1.85 ND 1.23 ND 1.54 ND 1.38 1.21 1.52 1.18 1.55 1.46 1.28 1.18 1.69 1.11 1.51 1.20 1.45 1.08 1.46 1.24 1.23 1.11 KEY ischman (4) □ F1 ^Leontief (9) Malenbaum (71) V^ Ridker i^S Watson (73) HI BuMines (74) KEY j J Fischman (4) Leontief (9) Malenbaum (17) BuMines (74) Figure Al Cr Co Cu Mn Ni Sn W Zn 2.— Rebased 1980-83 implied consumption divided by 1980-83 actual consumption: United States and world. 10 GNP GROWTH ASSUMPTIONS With the differences in forecasts clearly identified and measured, those interested in understanding the reasons for the differences can delve deeper into the various procedures and underlying assumptions. A description of the various methodologies used to generate the forecasts examined in the study is given in appendix A. These procedures are of varying degrees of complexity, with emphasis on different exogenous variables and relationships among variables. A thorough comparison of these procedures would involve assessing the assumptions behind each model, examining whether the procedures follow logically from the assump- tions, determining if the outputs could be replicated, and assessing the costs and benefits of each model (1, p. 307). The costs include the initial cost of development, the cost of maintaining the model, and operating costs; benefits include the improvement in forecast accuracy, the level of con- fidence provided by the model, and the ability to assess alternative scenarios of the future. An extensive com- parison of the procedures along these lines was beyond the scope of this study. One assumption, however, the assumed rate of growth of gross national product (GNP) or gross domestic product (GDP), was compared because every forecast examined was based on some assumption about GNP or GDP growth. Also, most of the forecasting studies reviewed did make ex- plicit their assumptions about growth in U.S. output, and some did so concerning growth in world output. Finally, the actual values of GNP and GDP growth in the period 1980-83 were readily available. In the period 1980-83 economic growth in both the United States and the world was quite low. Real U.S. GNP grew at the rate of -0.3%, 2.5%, -2.1%, and 3.7% for 1980, 1981, 1982, and 1983, respectively (5, p. 205). World output grew at 2.0%, 1.6%, 0.6%, and 2.6% over the same years (5, p. 205). These rates, on average, were much lower than those assumed by the various forecasters, whose assumptions, of course, covered longer periods of time. (As the period 1980-83 seems to have been a cyclical low for business, in the longer run, their assumptions will probably turn out to be much more accurate.) Malenbaum (11, p. 38), for example, assumed a growth in U.S. GDP of 3.3% and world GDP of 3.5% over the period 1975-85. Fischman (U, p. 142) assumed a growth rate of U.S. GNP of 3.0% over the period 1980-85. Leontief (9, p. 33) assumed a 3.1% increase in U.S. GNP over the period 1972-2000. Ridker (13, p. 141) assumed a 2.9% increase in U.S. GNP over the period 1971-85 and over 3.5% rates of growth in the rest of the world. The Bureau of Mines 3 in its 1980 statistical estimates of U.S. consumption in 2000, assumed a 3.0% growth rate of U.S. GNP between 1978 and 2000. Overall the various forecasters made quite similar assumptions about GNP growth rates. The methodology presented in the previous section resulted in ratios that when close to 1 indicate "implied" values quite close, in aggregate, to actual values over that period. Looking at table 12 in particular, what is striking is that almost all of the ratios are greater than 1. Thus most of the "implied" forecasts exceeded the actual value of con- sumption of these minerals during the period 1980-83. Given the actual low values of GNP and GDP growth in the 1980-83 period, it is not too surprising that so many of the values in table 12 exceed 1. What may be surprising is that a number of projections resulted in ratios fairly close to 1, say within 10%, under the assumption of growth rates in economic output that were much higher than were actually experienced during this period. Since mineral consumption generally varies directly with economic output, one can readily hypothesize that if these forecasters had assumed lower values of economic growth rates, their forecasts of mineral consumption would have been lower, everything else remaining the same, and the implied consumption derived in the previous section would have been smaller. Again their estimates of GNP were for longer periods of time, and the 1980-83 period seems to have been a cyclical low for business. If lower growth rates had been used, however, ratios of the type found in table 12 might then have been far less than 1. In other words, the forecasters whose ratios were close to 1 in table 12 might have under- estimated mineral consumption, if they had used in their procedures the growth rates that actually prevailed during the 1980-83 period! Were these forecasters assuming more substitution for minerals, for instance, than actually took place? Or were they assuming more technological change, or more of a shift in the composition of GNP toward the service sector than actually occurred? What do these questions, and the possible answers, imply about projections currently being for- mulated? These and other questions arise from an analysis of this kind and illustrate its usefulness. CONCLUDING STATEMENT The comparison of mineral consumption forecasts from various sources involves numerous difficulties. The stand- ardization procedure presented in this study provides a way, given adequate data, to overcome some of these difficulties. Standardized forecasts can be compared, in a precise fashion, to other standardized forecasts, and used for various purposes. For example, a set of standardized forecasts or standardized growth rates can be used to con- struct various scenarios of future events. Standardized forecasts can also be compared to available current data, us- ing the growth rates associated with the forecasts to venerate "implied" values for various years. The latter kind of comparison does not, however, provide a way of evaluating the underlying forecasting methodologies. In such an evaluation, assessments should be made of the pro- cedures or models and of the predictions of the important causal variables within the models. In the case of mineral consumption forecasts, GNP predictions should be ex- amined. The kind of examination of past forecasts for future time periods presented in this paper, along with a look at the underlying GNP predictions, was thought to be useful for decision makers and forecasters working with current data. Such an examination could help them adjust or confirm their own forecasts for future time periods. 11 REFERENCES 1. Armstrong, J.S. Long-Range Forecasting. Wiley, 1985, 689 pp. 2. Cammarota, V. A. Jr., W. Y. Mo, and B. W. Klein. Projections and Forecasts of U.S. Mineral Demand by the U.S. Bureau of Mines. Pres. at "109th Annual Meeting, AIME (Feb. 24-29, 1980, Las Vegas, NV), 4 pp.; available upon request from V.A. Cam- marota, Jr., BuMines, Washington, D.C. 3. Council on Environmental Quality and Department of State. The Global 2000 Report to the President. Volume 2: The Technical Report. 1980, p. 582 (table 22-1). 4. Fischman, L. L. World Mineral Trends and U.S. Supply Prob- lems. Resources For The Future, Inc., Res. Paper R-20, Oct. 1980, 535 pp. 5. International Monetary Fund. World Economic Outlook. Washington, DC, Apr. 1985, 283 pp. 6. Lansberg^JI. H. Natural Resources for U.S. Growth. Johns Hopkins Press, 1964, 260 pp. 7. Lansberg, H. H., L. L. Fischman, and J. L. Fisher. Resources in America's Future. Johns Hopkins Press, 1963, 1040 pp. 8. Leontief, W., A. P. Carter, and P. A. Petri. The Future of the World Economy. Oxford Press, 1977, 110 pp. 9. Leontief, W., J. CM. Koo, S. Nasar, and I. Sohn. The Future of Nonfuel Minerals in the U.S. and World Economy. NY Univ., 1983, 512 pp. 10. Malenbaum, W. Materials Requirements in the United States and Abroad in the Year 2000. Natl. Comm. on Mater. Policy, Washington, DC, 1973, 40 pp. 11. Malenbaum, W. World Demand for Raw Materials in 1985 and 2000. Natl. Sci. Found., NS/RA-770421, Oct. 1977, 153 pp. 12. Meadows, D., D. Meadows, J. Randers, and W. W. Behrens. The Limits To Growth. Universe Books, 1974, 205 pp. 13. Ridker, R. G., and W. D. Watson. To Choose a Future. John Hopkins Press, 1980, 463 pp. 14. U.S. Bureau of Mines. Mineral Facts and Problems, 1980 Edi- tion. BuMines B 671, 1981, 1060 pp. 12 APPENDIX A.- REVIEW OF FORECASTING METHODOLOGIES RESOURCE PROJECTIONS Long-term projections of the U.S. demand for nonfuel minerals have been made for over 20 years. In the pioneer 1963 book Resources in America's Future (7), Landsberg projected for 1980 and 2000 both fuel and nonfuel minerals requirements, along with those for land, lumber, water, chemicals, and labor. The study compared the projected resource requirements with the 1960 U.S. resource base. These results were also presented in a condensed version of the original study entitled Natural Resources for U.S. Growth (6). The starting point for these demand projections was a series of projections of population, labor force, and gross na- tional product. Requirements for food, clothing, shelter, heat and power, transportation, durable goods, military equipment, outdoor recreation, etc. were then calculated. Next, these goods and services were translated into re- quirements for resource products such as agricultural raw materials, steel, lumber, and textile fibers. From these, in turn, the various demands on land, water, fuels, and other resources were estimated. Nearly all projections were made at three levels -low, medium, and high -with the middle levels considered most likely. (The low projections for 1980, in most cases, were closest to the actual 1980 levels of con- sumption.) Long-term projections of resource needs on a global basis have been made for over 10 years. With the publica- tion of Limits to Growth;(12) in 1974, long-term global pro- jections, already a subject of some debate, became con- troversial. In this book it was argued that the limits to growth on the planet would be reached sometime within the next century, if present trends continued. It was also argued that, given present resource consumption rates and the projected increase in these rates, most important nonrenewable resources would be extremely costly in the future (12, p. 66). These conclusions were reached with the use of a simplistic world model that featured one general population, one geographic unit, one composite industrial output, one nonrenewable resource, and one class of pollutants. Within this highly aggregated model, the basic behavior mode of the world system was asserted to be exponential growth of populatirn and capital, followed by collapse. Technological change had no impact on the essential problem, exponential growth in a finite and complex system (12, p. 145). Since the publication of Limits to Growth, arguments have been ad- vanced both for and against the notion that man's progress would be limited in the future by resource scarcity. STATISTICAL AND CONTINGENCY FORECASTING In 1970 the Bureau of Mines began making mineral con- sumption projections for the United States and the rest of the world. These forecasts are presented in editions of Mineral Facts and Problems, which is published every 5 years. For the United States, two kinds of consumption forecasts are developed: statistical projections and con- tingency forecasts (2). The statistical projections are calculated by the Division of Minerals Policy and Analysis, using linear regression analyses. Simple linear regression equations are used to ap- proximate the end-use consumption of mineral com- modities, where macroeconomic variables are selected as possible explanatory variables. For each end use of a par- ticular mineral, a set of likely explanatory variables is chosen from among 75 possibilities (gross national product (GNP), gross private domestic investment (GPDI), U.S. population, new construction activity, and 71 different Federal Reserve Board industrial production indexes). Each variable within the chosen set is then used as the ex- planatory variable in a series of simple linear regressions over some historical period, usually starting in 1960, with the dependent variable being the end use of the material. From the equations so generated, the one with the highest R 2 (smallest sum of squared predicted errors) is chosen as the basis for deriving the statistical projection. The coeffi- cients of this equation, plus a value of the explanatory variable projected to the year 2000 by a selected economic forecaster outside the Bureau, is then used to generate a projection for the end use. The contingency forecasts are formulated by commodi- ty specialists. These forecasts are judgmental and are based both on historical trends and on the specialists' knowledge of all developments, current and anticipated, that might have an impact on the future use of a commodity. In making these forecasts, the commodity specialists identify those problems or opportunities that might cause the consumption of a particular commodity to deviate from its historical trend. Using the statistical projections as a guide, the specialist arrives at estimates of low, high, and most prob- able consumption growth. Other specialists then critique these estimates, using, among other things, their knowledge of how the growth in consumption of the commodity in ques- tion might be affected by the growth of consumption of other commodities. All of the contingency forecasts utilize the statistical projections as their point of departure and assume the same economic conditions that underlie the statistical projections. Individual commodity specialists also make the world consumption forecasts. For these forecasts, reliance is placed on trends in world economic growth, population growth, and the growth in consumption of selected minerals and materials. Also commodity specialists have access to forecasts made by knowledgeable consultants, mining com- panies, international commodity associations, and other economic groups. INTENSITY OF USE In a March 1973 study (10) for the U.S. Commission on Materials Policy, Professor Wilfred Malenbaum of the University of Pennsylvania's Wharton School of Finance and Commerce prepared projections of mineral demand by a methodology known as intensity of use. In this study, trends in overall economic growth, population growth, and growth in primary consumption of 11 minerals and materials were projected for the world, and for 10 world regions. The U.S. National Commission on Materials Policy was responsible for examining the feasibility of striking a balance between the national need to produce goods on the one hand and to protect the environment on the other. In October 1977 (11), following the publication of Limits to Growth and the oil price shocks brought on by the Organiza- tion of Petroleum Exporting Countries (OPEC), Malenbaum 13 updated his 1973 study at the request of the National Science Foundation. The 1977 report dropped some of his original commodities, added others, and revised downward the projections for consumption of aluminum, copper, iron, steel, and zinc. Underlying the Malenbaum reports is the premise that long-term growth is not governed by supply limitations of any specific input materials. He argued that economic ex- pansion is based on the human resources of the world's societies (11, p. 48), and that the aspirations and commit- ment to economic expansion on the part of public and private leaders mattered more than resource endowment and technology (11, p. 26). Thus, the direction of determina- tion would run from gross domestic product (GDP) to material use. This assumption made it possible to project na- tional and world economic growth in one part of the research effort without regard to the material needs ap- praised in the other part of the study. Malenbaum thought that the record of growth in various parts of the world in the early 1970's lent support to the decisive role of quality inputs, both in rich lands where growth was impressive and in poor lands where the economic performance was more uncertain (11, p. 26). His appraisals of this situation, especially within the third world, led to his lower estimates of average GDP growth rates in the 1977 study than in the earlier study. In the 1977 study, he forecast the world in 2000 as having per capita GDP some 50% above 1971-75 levels in real terms, and as using two to three times the volume of raw materials in a year compared with average annual usage over 5 recent years. These results contrasted markedly with the conclu- sions of the 1973 study, where comparable ratios were reported between three and four. Thus, his expectation in 1977 was for a relative weakening of demand, and lower relative mineral prices. It was not a picture of a world con- fronted with limited resources (11, p. 122). In brief, intensity of use (IOU) analysis is a procedure for translating GDP and population projections into mineral and material consumption projections, using tables that estimate the intensity with which minerals or materials will be consumed within a given country or region relative to per capita GDP levels. As used by Malenbaum, this analysis con- siders only primary use and disregards subsequent shipments of processed or manufactured minerals or materials to other regions. Hence, Japan, for example, is represented as having exceptionally high consumption levels, since Japanese exports are disregarded. Total world consumption of a given mineral or material is calculated as the sum of the consumption levels for the commodity pro- jected for each region. The following country groupings were used by Malen- baum: 1. Western Europe -Organization for Economic Cooperation and Development (OECD) countries in Europe. 2. Japan. 3. Other developed lands -Australia, Canada, Israel, New Zealand, Republic of South Africa. 4. U.S.S.R. 5. Eastern European Countries -Soviet bloc countries plus Albania and Yugoslavia. 6. Africa minus South Africa. 7. Asia minus Israel, Japan, China, Mongolia, North Korea and North Vietnam. 8. Latin America. 9. China plus Mongolia, North Korea and North Vietnam. 10. United States plus Puerto Rico and overseas islands. Each of the 10 groups was considered to have a high enough degree of homogeneity to justify common assump- tions with respect to changes in IOU of raw materials and in rates of total economic growth. Calculations regarding future mineral consumption levels are independent of similar calculations involving that region's consumption of other minerals and materials, and independent of any other region's consumption levels of any commodity. They are also independent of any explicit considerations regarding potential changes in supply levels, prices, or strategic or balance of payment positions. Consumption of a given mineral or material within a given year is calculated on the basis of just three com- ponents: an exogenous projection of the level of overall economic activity (GDP) within a given region in a given year, an exogenous projection of the total population within the same region in the same year, and an "IOU table" show- ing the quantity of a given mineral or material per unit of that region's total GDP (a ratio known as the commodity's intensity of use) likely to be consumed within that region at various levels of regional per capita GDP. From the IOU table, the appropriate IOU value is obtained (expressed in terms of commodity units per unit of total GDP) for the regional per capita GDP level in question, interpolating or extrapolating as necessary. This IOU value is then multiplied by the exogenously estimated total regional GDP for that year. On analysis of the historical record, Malenbaum found what seemed to be patterns strong enough to allow IOU projections. He thought that these patterns had a technological dimension, reflecting changes in use and effi- ciency of inputs to outputs, both taking account of changes in techniques (for input or output) and changes in market relationships associated with supply, demand, and public policy. Also he thought the analytic and descriptive literature on material use provided some guides for projec- ting the appropriate intensity levels. Primarily, however, it was the apparently systematic behavior of the measure that underlay his conviction of its usefulness in demand analysis for raw materials (11, p. 22). The primary pattern found was formed by the IOU in a region and its per capita GDP. The IOU statistic increased as a function of increasing per capita GDP for less developed countries, and decreased for industrialized coun- tries moving toward postindustrial service economies. Thus, mineral and metal consumption levels within a region whose economy is moving from industrialization to postin- dustrialization are projected using IOU statistics at various levels of per capita GDP that form an inverted U-shaped curve. For most of the materials Malenbaum considered, world intensity of use seemed to have already reached historical peak levels, mostly a decade or more back (11, p. 49). This conclusion supported his view that for the entire world, a large proportion of total world use will long con- tinue to occur in the wealthier lands (11, p. 121). The materials analyzed in the 1977 study were alumi- num, chrome, cobalt, copper, iron ore, manganese, nickel, platinum, crude steel, tin, tungsten, and zinc. IOU data for these minerals were assembled for the same time intervals, 1951-75, with occasional data for 1934-38, on the same regional bases as were GDP and population. Most of the na- tional product data and all of the population data were taken from United Nations sources. The output was converted to U.S. dollars in 1971 prices on the basis of exchange rate data. For the most part, the historical record was examined in the form of 5-year averages with recourse to individual years only to trace patterns of marked change within a 14 5-year period, a problem that arose especially for regions composed of countries not usually analyzed as a single unit (11, p. 25). The material use data were from the Bureau of Mines, the United Kingdom's Summary of the Mineral Industry, the U.N. Department of Economic and Social Affairs, and occasional specialized private groups associated with pro- ducing and processing interests (e.g., the publishers of Ger- many's Metal Statistics).The consumption concept was "ap- parent consumption" or production minus exports plus im- ports plus changes in stocks (11, p. 29). INTENSITYOF-USE VARIATION A variation on the IOU methodology was utilized in World Minerals Trends and U.S. Supply Problems (4) by Leonard L. Fischman of Resources For The Future. In this book the historical patterns of consumption for seven non- fuel minerals -aluminum, chromium, cobalt, copper, manganese, lead, and zinc -are examined. Projections of future patterns of consumption are offered, based for the most part on demographic and macroeconomic projec- tions -that is, growth in GDP, or GDP per capita. The prin- cipal conclusion of his study was that the U.S. faces only one important type of mineral supply problem, based on its dependence on imports of certain minerals. The imports of a few of these minerals, chromium in particular, may be sub- ject to disruptions that could cause a sharp upward move- ment of prices, with serious economic impacts (.4, p. 3). The general procedure for making the mineral com- modity projections in this study was, first, to relate con- sumption of the refined metal to the appropriate macro in- dicator, usually GNP or GDP, and then to derive the ore or semirefined input into the refined metal; steel production was used as a macro indicator in several cases, yet it in turn was related to GNP or GDP (.4, p. 144). Five-year moving averages of the ratio of consumption to GNP or GDP (averages of IOU ratios) were calculated for historical data periods of selected countries, in order to smooth out varia- tions due to business cycles. Projections of the moving averages of these ratios were made, with the intention of achieving a similar smoothing pattern. The precise pro- cedure used to make the projections was not specified, but it seems that the projections were judgmental. Elaborate statistical "filtering" methods were rejected as infeasible, and straight-line trends were rejected as inappropriate. Fischman's world totals were based on the selected in- dividual countries that account for the bulk of each com- modity's utilization. By and large, these are the leading in- dustrial countries for the refined metals, and a combination of industrial countries and other mineral-rich countries for the cruder forms of metals. It takes projections of the macroeconomies of only about a dozen countries, all told, ac- cording to Fischman, to provide the base for projecting the bulk of the consumption of all 18 commodity forms covered, plus crude steel. The aggregate consumption of the leading consumers of each form, though gradually declining in rela- tion to world totals as new consumers enter the picture, may, it was hypothesized, be "blown up" fairly accurately to a world total for any given future year by extrapolating their declining share (4, p. 136). Fischman argued that the difference between his macro assumptions and those of others should be taken as the ex- pression of how he judged the future would unfold. He main- tained that "whenever long-term commodity projections are based directly or indirectly on long-term projections of gross economic output, the commodity-projection accuracy tends to depend more upon the accuracy of the gross na- tional or domestic product (GNP or GDP) projections than upon the parameters tha,t join GNP/GDP to individual com- modities" (4, p. 3). He concluded that "prior projections of world mineral consumption have been almost uniformly too high -principally because of overestimation of long-term growth rates in gross economic output of the principal con- suming countries" (4, p. 137). INPUT-OUTPUT A quite different approach to the projecting of mineral requirements was used by Leontief in two studies. In The Future of the World Economy (8), published in 1977 with U.N. funding, a large and complex input-output model was constructed and used to, among other things, project the production and consumption of six nonfuel minerals. Overall, the model addressed the question of global resource requirements and the availability of food, mineral, and energy resources to meet these requirements to the year 2000. Macroeconomic variables, such as gross domestic product, consumption, investment, and the balance of payments, were projected for 15 regions into which the world's countries were aggregated. Also projected were a large number of sectoral outputs, including those of 30 manufacturing and service sectors, 4 agricultural sectors, 3 energy resource outputs, and 6 nonfuel mineral outputs: aluminum, copper, iron, lead, nickel, and zinc. In addition to output levels, the model tracked imports and exports in minerals, as well as all other traded goods. For nonrenewable fuel and nonfuel minerals, the model traced the cumulative resources produced in each region after each decade. In a second book, The Future of Nonfuel Minerals in the U.S. and World Economy (9), the earlier study was updated and expanded, and the outlook for nonfuel minerals was stressed. Twenty-six nonfuel minerals were included in the second study: iron and ferrous metals (chromium, manganese, molybdenum, nickel, silicon, tungsten, vanadium), nonferrous metals (aluminum, copper, gold, lead, magnesium, mercury, platinum, silver, tin, titanium, and zinc), fertilizer-related minerals (phosphate rock, potash), and , miscellaneous chemicals (boron, chlorine, fluorine, soda ash, and sulfur). An input-output table shows the amounts of goods and services individual industries buy from and sell to each other in a particular year. Input coefficients are derived from such a table by dividing the column entries by the respective sec- toral outputs. The coefficients show the amount of each in- put required to produce one dollar's worth of a sector's out- put. Each column describes the structure of a particular in- dustry and, by giving a detailed, quantitative description of the inputs used by the industry, serves as an implicit representation of that industry's technology. Because each industry has its own column, the matrix is a structural description of the entire economy for a particular year. Similarly, a separate set of capital coefficients describes the stocks of buildings and equipment, as well as all kinds of working inventories, that each producing sector has to maintain to transform the proper combination of its inputs into its final output of goods and services (9, p. 20). The in- puts of primary natural resources, such as agricultural land, water, and minerals, required by all producing sectors of the economy as well as households can also be depicted and 15 analyzed. Input-output tables are prepared by the Bureau of Economic Analysis (BEA) every 5 years (9, p. 23). The world economy (in the U.N. World Input-Output Model) is subdivided into 15 regions that fall into 3 main groups: the developed regions, characterized by relatively high per capita income (North America, Europe, the Soviet Union, Oceania, South Africa, and Japan); the less- developed regions rich in natural resources (the Middle East, some of the South American countries, and some countries in tropical Africa); and the less-developed coun- tries with few resources (9, p. 210). The model describes each region in terms of 45 sectors of economic activity, in- cluding various types of agriculture, mining, manufactur- ing, utilities, construction, services, transportation, com- munication, and pollution abatement. Though each region is initially treated separately, the model contains linkages that permit its users to trace the complex interconnections of trade, foreign investment, loans, interest payments, and foreign aid. The rates of regional or world economic growth, as determined by the regional rates of population growth, technological change, and savings, will to a great extent determine the global long-term requirements for nonfuel minerals. Certain assumptions and projections underlie the Leon- tief projections. Assumptions about technological change and changes in recycling rates were both incorporated through changes in the mineral input coefficients. Changes in import requirements were handled in a similar fashion. Bureau of Mines projections of ratios of high and low levels of expected imports to primary demand were reconstituted into import coefficients, the ratio of imports to domestic output, so that they would be compatible with the structure of the model. By surveying special studies of future trends in material use, making extrapolations based on past trends, and then qualifying these crude estimates by discussions with experts on material use, the interindustry coefficients prepared for the World Model study were updated to reflect expected technological change. Labor and energy coeffi- cients were similarly updated. The final U.S. demand projections were based on pro- jections made with the Bureau of Labor Statistics (BLS) macroeconomic model. The BLS model takes into considera- tion detailed projections of demographic trends and cor- responding changes in the pattern of consumption, invest- ment, exports, imports, and labor productivity. The size of the U.S. economy, measured by gross domestic product (GDP), was projected by the BLS to grow in real terms from 1972 to 2000 at an average annual rate of 3.1% (9, p. 23). INPUT-OUTPUT AND INTENSITY OF USE COMBINED A study published in 1980 used a combination of input- output and intensity of use methodologies in making projec- tions. This study, To Choose a Future (13), by Ronald G. Ridker and William D. Watson, used a system of models, in- cluding a core input-outpift model, for its U.S. projections. Projections for the rest of the world were based on intensity-of-use calculations and projections of the growth of per capita income. The purpose of this study was to analyze the resource and environmental impacts on the United States of alternate rates of population and economic growth, with attention to international and technological developments and energy prices. The system of models consisted of a number of special- purpose models linked to INFORUM, the University of Maryland's 185-sector, dynamic, macroeconomic-cum-input- output model of the U.S. economy, developed over a series of years by Clopper Almon and his students at the Universi- ty of Maryland (13, p. 6). This system is known as SEAS/RFF (Strategic Environmental Assessment System/Resources For The Future). The SEAS/RFF system develops national U.S. economic forecasts through 2025 based on an exogenously specified set of demographic, macroeconomic, energy price, environmental policy, and resource policy assumptions. In turn, these forecasts form the basic economic inputs used by other models in the system to develop their more specialized forecasts. Forecasts are made at both national and regional levels. The model computes both dollar and physical estimates for all major fuels, some 20 nonfuel minerals, and 42 pollutants. INFORUM is a dynamic forecasting model that joins ag- gregate GNP forecasts to the markets in which products are sold. The model determines industry outputs year by year based on forecasts for all product markets, for capacity, and for the availability of labor. Most of the final demand com- ponents are based on econometric equations derived from regressions performed by Clopper Almon and his associates (13, p. 418) on historical time series. Past levels of personal consumption expenditures were regressed against levels of disposable income, changes in disposable income, relative prices, time trends, and levels of consumption from previous years. The six special-purpose models integrated with IN- FORUM into a common model were PRICE, which uses relative energy prices and price elasticities to alter energy demands, capital requirements, and GNP growth; TECHNOLOGY, which uses technological change assump- tions to alter current and capital account flows; INSIDE, which provides greater detail on industrial output; ABATE, which calculates costs for abating pollution and sector pur- chases for abatement; ENSUPPLY, which uses assump- tions about fuel availability and energy supply technology to determine energy supply and demand mixes; and MINERALS, which allows exogenous specification of stockpiles and import levels for selected minerals (13, p. 412). Together, these seven models form the national economic forecasting model (FORECAST) for the SEAS/RFF system. Linkage among all of the models allows population growth, energy price effects, technology change, abatement, energy supply constraints, and stockpiling and import constraints for minerals to be reflected in SEAS/RFF economic forecasts. A number of assumptions and scenarios were used in the study. For instance, the study used the Census Bureau projection series D, E, and F for population projections. The E series, which was adopted as the baseline projection, assumes that the groups of women just entering the reproductive years will have a completed fertility rate of 2.1 births per woman. The D series assumes an average of 2.5 births per woman, reflecting a continuation of the trends of the last 50 years. Series F assumes 1.8 births per woman, reflecting a continuation of the trends of the last 5 years (13, p. 20). The study also assumed that the unemployment rate, which stood at 8.5% at the end of 1975, returned to between 4% and 4.5% by 1980 and remained within that range thereafter (13, p. 25). Two productivity assumptions were used: (1) worker-hour productivity returned by 1980 to the trend line of the 1948-68 period and (2) after 1968 the long- term growth rate in labor productivity shifted downward by 0.3% per year and only half of the shortfall from this new trend was made up by 1980 (13, p. 27). 16 Four different GNP projections were made for various world regions, using the following scenarios: high popula- tion and high economic growth, low population and high economic growth, low population and low economic growth, and two intermediate cases. These projections were based on a division of countries into four groups, each occupying a somewhat unique position in the world economy. The first group consists of the relatively developed, industrialized countries (OECD members, South Africa, Eastern Europe, and U.S.S.R.), where economic growth is largely a matter of maintaining growth in labor productivity and full employ- ment of the labor force. Projections for countries other than the United States in this category begin by applying assumptions about growth rates in GNP per capita (an ap- proximation for labor productivity) to the population projec- tions (an approximation for the labor force). The projections are then modified to take into account the effects of higher petroleum and other import prices. Projections for the United States are derived from runs of the SEAS/RFF model. The second group of countries consists of OPEC members. The third group comprises countries whose growth rates are strongly dependent on foreign trade earn- ings in volatile international markets; for the most part, these countries are the principal nonfuel mineral exporters. The last group of countries consists of the resource-poor LDC's (13, pp. 42-44). To solve the models, a trial value of disposable income was run through the system in order to determine personal consumption expenditures. Personal consumption expen- ditures were then used with estimates of public abatement expenditure, number of households, and interest rates to determine residential and public construction and other variables. This process continued until a GNP figure was ob- tained. The calculated GNP was then compared with a target GNP. If they differed, the level of disposable income was changed and the calculations began anew. Mineral de- mand projections for the United States were obtained by first estimating a value of gross domestic demand (13, p. 104) -that is, demand excluding exports that can be satisfied from primary production, secondary production, or imports. In those cases in which the SEAS/RFF model in- cluded a specific sector that could be associated with a specific mineral- aluminum, copper, iron, lead, or zinc -the demand projections were derived directly from the model. In all other cases, information from the Bureau of Mines (1975) on unit requirements (and changes in unit re- quirements over time) for each major mineral-using sector was combined with projections of the output of these sec- tors. With the exception of the fertilizer sector (a major user of phosphate rock, potash, and sulfur), for which projections were developed by other methods, sector projections were derived from the model. Secondary production, estimated by assuming that the percentage of demand satisfied by recycling remained the same as in the base period (roughly, the 1971-74 average), was then subtracted from these gross demand projections. The mineral demand projections for the rest of the world were based on intensity-of-use relationships. It was assumed that mineral consumption per capita was a func- tion of GNP per capita, and that this function changed over time as GNP per capita increased. The assumption was made that the relationship for other regions approaches that for the United States, though it never reaches it until the GNP per capita of the subject country catches up to that of the United States. The projections were made from four points in time, starting with historic data for 1971 for the region in question. A straight line was drawn from this first point through a point representing the projected relation- ship between consumption per capita and GNP per capita for the United States in 1985. The relationship for the region in 1985 was then found by locating its 1985 GNP per capita on that straight-line segment. The same procedure was repeated using the U.S. relationship in the year 2000 and then again in 2025. These estimates were then multiplied by population (13, p. 107). 17 APPENDIX B.— IMPLIED CONSUMPTION DIVIDED BY ACTUAL CONSUMPTION, 1980-83 Commodity Implied/actual consumption, relative value and area 1980 1981 1982 1983 Average ADJUSTED BU MINES DATA (14) Aluminum: United States 1.28 1.34 1.51 1.39 1.38 World 1.10 1.19 1.29 1.25 1.21 Chromium: United States 1.08 1.27 2.11 2.12 1.52 World 1.02 1.11 1.29 1.34 1.18 Cobalt: United States 1.25 1.68 2.04 1.45 1.55 World 1.21 1.52 1.79 1.44 1.46 Copper: UnitedStates 1.15 1.13 1.52 1.36 1.28 World 1.10 1.12 1.23 1.27 1.18 Manganese: UnitedStates 1.36 1.38 2.15 2.18 1.69 World 95 1.12 1.13 1.27 1.11 N i cKsl ' UnitedStates 1.38 1.43 1.71 1.56 1.51 World 1.06 1.20 1.27 1.27 1.20 Tin: UnitedStates 1.29 1.16 2.02 1.62 1.45 World 1.01 1.06 1.12 1.13 1.08 Tungsten: UnitedStates 1.12 1.13 2.01 1.97 1.46 World 1.06 1.15 1.40 1.42 1.24 Zinc: UnitedStates 1.25 1.06 1.42 1.25 1.23 World 1.05 1.11 1.14 1.12 1.11 ADJUSTED FISCHMAN DATA (4) Aluminum: UnitedStates 1.23 1.28 1.45 1.33 1.32 World 1.12 1.22 1.31 1.27 1.23 Chromium: UnitedStates ND ND ND ND ND World 94 1.03 1.19 1.22 1.09 Cobalt: UnitedStates 1.34 1.80 2.19 1.56 1.67 World 1.28 1.59 1.86 1.50 1.53 Copper: UnitedStates 1.00 .97 1.28 1.13 1.08 World 1.04 1.06 1.15 1.18 1.10 Manganese: UnitedStates ND ND ND ND ND World 1.10 1.30 1.31 1.46 1.29 N icksl* United States ND ND ND ND ND World ND ND ND ND ND Tin: UnitedStates ND ND ND ND ND World ND ND ND ND ND Tungsten: UnitedStates ND ND ND ND ND World ND ND ND ND ND Zinc: United States 1.42 1.19 1.58 1.38 1.38 World 1.07 1.12 1.16 1.15 1.13 ND No data. 1.41 ND 1.75 1.22 ND ND 1.44 ND 2.01 1.71 1.58 ND 1.93 1.63 1.21 ND 1.94 ND Commodity Implied/actual consumption, relative value and area 1980 1981 1982 1983 Average ADJUSTED LEONTIEF DATA (9) Aluminum: UnitedStates 1.31 1.37 1.55 1.42 World ND ND ND ND Chromium: United States 1.24 1.47 2.44 2.43 World 11.02 1.15 1.35 1.42 Cobalt: UnitedStates ND ND ND ND World ND ND ND ND Copper: United States 1.30 1.28 1.70 1.54 World ND ND ND ND Manganese: United States 1.61 1.64 2.56 2.64 World 1L41 1.71 1.75 2.01 Nickel: UnitedStates 1.46 1.51 1.77 1.61 World ND ND ND ND Tin: United States 1166 1.52 2.71 2.22 World 1,46 1.58 1.72 1.79 Tungsten: UnitedStates 95 .94 1.67 1.61 World ND ND ND ND Zinc: UnitedStates 1.93 1.65 2.25 2.01 World ND ND ND ND ADJUSTED MALENBAUM DATA (77) Aluminum: UnitedStates 1.18 1.24 1.39 1.29 1.27 World 97 1.04 1.11 1.06 1.04 Chromium: UnitedStates 94 1.10 1.78 1.75 1.29 World 84 .93 1.08 1.11 .98 Cobalt: UnitedStates 1.27 1.72 2.10 1.49 1.60 World 1.32 1.67 1.98 1.61 1.62 Copper: UnitedStates 1.03 1.01 1.36 1.21 1.14 World 97 .99 1.07 1.10 1.03 Manganese: UnitedStates 1.54 1.60 2.52 2.63 1.97 World 1.13 1.35 1.35 1.52 1.33 Nickel: UnitedStates 1.29 1.33 1.54 1.40 1.39 World 98 1.09 1.15 1.13 1.09 Tin: UnitedStates 1.40 1.26 2.20 1.77 1.58 World 1.12 1.18 1.27 1.29 1.21 Tungsten: UnitedStates 92 .90 1.57 1.49 1.15 World 84 .91 1.10 1.12 .98 Zinc: United States 1.59 1.35 1.85 1.64 1.59 World 1.08 1.14 1.19 1.20 1.15 ADJUSTED RIDKER DATA (73) Commodity Implied/actual consumption, relative value (United States data only) 1980 1981 1982 1983 Average Aluminum 1.27 1.34 1.52 1.41 1.38 Chromium 90 1.07 1.76 1.76 1.27 Cobalt 1.09 1.47 1.80 1.28 1.37 Copper 1.05 1.02 1.35 1.20 1.14 Manganese 1.40 1.44 2.25 2.31 1.76 Nickel 1.27 1.31 1.55 1.41 1.38 Tin 1.61 1.47 2.58 2.10 1.85 Tungsten 95 .95 1.69 1.64 1.23 Zinc 1.55 1.32 1.78 1.57 1.54 C 67 ^*5 «r, •♦***** '$%$: /% ^.\ A fw-VV-W" / \ ' >* J. ^ • * -i o_ ♦ "*bv* > ^ - ^o< •■ ^ ^0^ ? N' ^ ^ .."•« ^°- *bV w %'^^-^ ^-:^-^ ^'^^Z V T ^^\^ \^ f '/ V : ^' > A ,G V V */J7T» A V ♦* ^ > ^ "•■ v-o^ v ^> r- ^ ,G^ \s *7^T* A • cS^V?v.«.* . O . ^. ^°\^ft^,^ .v^^iV-.V ^,*i^:%*> o . » - g v \» *'T7T* A . o » o r ^0 x A^*V v ^> -o.^ .G v \5, *'VVT* A '* A° °w *'^7* A y ^:: v ,0^ ,G^" ^ %V.V A G • c«^w' o ^0* g ^ ^:* % v .. MM ^p <> ♦*..« 4 ,G • A V «x. **..;£&.% y , .-^fa.% c«*.^j,.*«i V^'V v^V V^V V-~-V ^ % <^ ^ • >n t 4* .Wr *,. A* * •w . '•% y#fe' % /■«-% /'XfeS /slmkS ^^A » o 4 o^ a ^°* ^ X y ..... °*. ^o 1 •^o? 5»^ V %^\f V^'/ \^?\^ V O^ »o . » * .A ^ ^»- ^.^ c 'gsm**. ^~ .V ^.^^V„ Tft. « c sj&^>* v - ^"^. ^ "oV° ^ V c"" J. BOOKBINDING II** 0^ o' P"9j. <> *'TVT« ,c^ ^ *- CONGRESS ■ I 0k ■