, BEA-SP 77-030 The Energy/Real Gross Domestic Product Ratio An Analysis of Changes During the 1966-1970 Period in Relation to Long-Run Trends Bureau of Economic Analysis Staff Paper No. 30 October 1977 >f °'% U.S. DEPARTMENT OF COMMERCE BUREAU OF ECONOMIC ANALYSIS BEA Staff Paper No. 30 BEA-SP 77-030 The Energy/Real Gross Domestic Product Ratio- An Analysis of Changes During the 1966-1970 Period in Relation to Long-Run Trends by Jack Alterman *«*0,c ; WE ; October 1977 **,.. of ^ This staff paper presents a reproduction of a report prepared by Jack Alterman (currently a consultant to Resources for the Future) under a contract with the Bureau of Economic Analysis. STANDARD TITLE PACE FOR TECHNICAL REPORTS 1. Report No. % Gmx, A«<5n ■jr. i -■■■■ 4. Title and Subtitle The Energy/Real Gross Domestic Product Ratio -- An Analysis of Changes During the 1966-1970 Period in Relation to Long-Run Trends 3. Recipient's Catalog No. 5. Report Date October 1977 6. Performing Organization Code 7. Auihor(s) Jack Alterman 8. Perfor Performing Organization Repi,. No. staff Paper No. 3D 9. Performing Organization Name and Address Resources for the Future, Inc. 1755 Massachusetts Avenue, NW Washington, D.C. 20036 10. Project/Task/Work Unit No. 11. Contract /Grant No. 12. Sponsoring Agency Name and Address Bureau of Economic Analysis U.S. Department of Commerce Washington, D.C. 20230 13. Type of Report & Period Covered 14. Sponsoring Agency Code 15. Supplementary Notes 16. Abstracts The initial objective of the study was to analyze the factors underlying the 1966-1970 reversal in the long-term decline in the BTU/GDP ratio. In attempting to place the 1966-1970 reversal in a longer-term perspective, it was found that there had been a slowdown in the secular rate of decline in the ratio since 1929, culminating in the most recent period, between 1966 and 1975, when there was a net increase in the ratio. As a result, the objective of the study was broadened to include an analysis of some of the f underlying the reduction in the long-term rate of decline in the BTU/GDP ratio and the connection, if any, between the 1966-1970 reversal and the slowdown in the secular rate. ctors 17. Key Words and Document Analysis. 17a. Descriptors Energy consumption BTU per dollar of gross domestic product Gross use of energy Net use of energy Effective use of energy Industry consumption of energy 17b. Identif iers/Open-bnded Terms Sector consumption of energy Input-output and energy use 17c COSATI Field/Group 18. Distribution Statement form CPSTI-3S (7 19. Security Class (This Report) UNCLASSIFIED 20. Security Class (This Pase UNCLASSIFIED 21. No. of Pages 22. Price USCOMM-OC eB00*-P70 Acknowledgments I should like to acknowledge the support and assistance provided in the course of the study by members of the staff of BEA, particularly Morris R.Goldman, formerly Deputy Director of BEA (retired), Martin L. Marimont, Associate Director for National Economic Accounts, who was contract officer of the study, and Philip M. Ritz, Chief of the Interindustry Economics Division. I should also like to acknowledge the encouragement and interest in the project by Hans Landsberg, Co-Director, Energy and Materials Division, Resources for the Future, Washington, D.C. Finally, the views expressed in this study are my own and should not be ascribed to the sponsoring organization or to the persons whose assistance and/or support is acknowledged. Jack Alterman The Energy/Real Gross Domestic Product Ratio -- An Analysis of Changes During the 1966-1970 Period in Relation to Long-Run Trends in the Ratio Introduction One of the critical questions involved in assessing the prospects for meeting the growing demand for energy, consistent with potential growth in the economy and reduced reliance on foreign sources of supply, is the extent to which increased energy requirements can be moderated by reductions in energy consumption per dollar of real output of the economy—the BTU/GDP ratio. ]_/ The long-run trend in this ratio, going back to 1929 and possibly even earlier, has been downward so that past growth in the economy has re- quired a less than proportionate increase in energy consumption. Continua- tion of this trend could make a substantial contribution to dampening increases in the demand for energy and, in fact, many of the analyses of energy demand over the next 10 to 25 years explicitly assume or project continued declines in the energy/output ratio. 2/ Given the critical role of the BTU/GDP ratio in evaluating the long- term outlook for energy supply-demand balances, it is important to determine what the trend in the ratio has been and to identify the factors affecting short- and long-term changes in the ratio. Although the trend in the BTU/GDP ratio has generally been downward, the changes have not been uniform and there have been subperiods with substantial reversals in the trend. The most recent reversal occurred between 1966 and 1970 and was sufficiently extensive that by 1970 the ratio was back up to where it had been in the yery early 1950' s. After 1970, the ratio started to decline again and continued to decline through 1975. 1. Energy consumption is conventionally measured in terms of the heat content or British thermal units (BTU's) of the individual energy sources. For reasons to be discussed later in the report, the measure of output used in this study is gross domestic product (GDP), in 1972 dollars, rather than the gross national product measure usually used in the past to derive the over- all BTU/output ratio. 2. See, for example, Walter G. Dupree, Jr. and John S. Corsentino, U.S. Energy Through the Year 2000 (Revised), Bureau of Mines, U.S. Department of Interior, December 1975. Also, the Federal Energy Administration report on National Energy Outlook, February 1976, and the report by David J. Behling, Jr., et. al., The Relationship of Energy Growth to Economic Growth Under Alternative Energy Policies, prepared for U.S. Energy Research and Development Administration by Brookhaven National Laboratory and Data Resources, Inc., March 1976. - 1 - At the time of the 1966-1970 reversal and subsequently, there were expressions of concern, reflected in at least one study as well as extended comments in other reports, as to whether the reversal represented a new trend or only a temporary aberration which would have no long-run effect on the trend. 3/ After 1970, with the return to a decline in the ratio, much of the previous concern was dissipated. Unfortunately, although the ratio has declined in every year since 1970, the declines have been modest and by 1975 the ratio was still above the level reached in 1966 at the beginning of the reversal period and about the same level as in 1953--22 years earlier. This lack of net reduction in the ratio over a period of more than two decades raises anew the question as to whether the 1966-1970 reversal and only partial recovery in the period since 1970 represents a break with the long-run trend or whether it reflects transitory factors with only limited implications, if any, for the underlying trend. The present study explores this and related questions and tries to place the changes since 1966 in historical perspective by analyzing longer- term changes in the BTU/GDP ratio, covering the period since 1929 at the aggregate level, and in more detail for the 1947-1975 period. 3. Energy Consumption and Gross National Product in the United States - An Examination of a Recent Change in the Relationship, (Bruce Netschert) , National Economic Research Associates, Inc., March 1971. Also, see discussion of the reversal by Joel Darmstadter in "Energy Consumption: Trends and Patterns," pp. 168-171, in Energy, Economic Growth, and the Environment , edited by Sam H. Schurr, Resources for the Future, Inc., 1972. The reversal is also discussed in the context of a broader paper by Harold J. Barnett on Energy, Resources and Growth, pp. 184-185, included in Resource Scarcity, Economic Growth and the Environment, U. S. Joint Economic Committee, 1974; John G. Myers, in an article on Energy Conservation and Economic Growth, the Conference Board Record, February 1975, p. 28, discusses the 1966-1970 trend reversal and con- cludes that the return to the long-term rate between 1970 and 1973 "leads one easily to suspect that the 1966 to 1970 period was a temporary aberration in the last half-century of energy use in the nation." - 2 - Section 1. The BTU/GDP Ratio and Underlying Measures -- A Description of Major Features Since the focus of the study is the analysis of changes in the BTU/GDP ratio, it is important to understand how changes in the overall ratio may be affected by the scope and characteristics of the underlying measures of energy consumption and real output of the economy. This section discusses the major features of these measures as well as the aggregate BTU/GDP ratio as they may affect changes in the ratio. Gross energy consumption The measure of energy consumption used in calculating the energy/GDP ratio is the series on apparent gross domestic energy consumption developed by the Bureau of Mines (BOM), U. S. Department of Interior. It is referred to as "apparent" domestic consumption because, for the most part, it is not derived directly from data on energy consumption but rather estimated as the sum of domestic energy production plus net imports (imports less exports) and net inventory depletions (inventory depletions less inventory increases). It should be noted that the inventory change estimates are incomplete, being based largely on inventory holdings of energy producers and only to a limited extent on the energy inventory change of users. Because the various types of energy are usually given in physical quantity terms, e.g., barrels of oil, tons of coal, which are not additive, BOM uses the heat or thermal content of energy, British thermal units (BTU'S), as the common unit of measure in deriving the aggregate energy consumption series. Separate conversion factors are used to convert the quantity measures of each energy type to the common BTU measure. In concept, the objective in deriving the measure is to count the primary energy content of domestic energy consumption. It includes the con- sumption of all fossil fuels (including coal, petroleum, natural gas and natural gas liquids but excluding fuelwood), hydropower and nuclear power. The BTU content of petroleum is measured at the petroleum refining stage rather than in terms of crude oil in order to take account of the different BTU conversion factors of the various petroleum products. Hydropower is converted to BTU's by imputing the same primary BTU content per kilowatt hour as for steam (fossil fueled) generating electric utility plants. The nuclear power conversion factor is based on the heat utilized in nuclear power plants per kilowatt hour generated. The BTU conversion factors can and do change over time reflecting changing quality of the primary energy sources, e.g., coal, as well as changing technology in the case of the impu- tation of the heat rate (BTU's per kilowatt hour) used in converting hydro- power to primary energy content. It should be noted that the BTU measure does not reflect the value added to the basic primary energy sources as they are converted to more convenient forms of energy, e.g., electricity, as well as the transportation and distribution services provided by electric and gas utilities, for example. Nor does the aggregate measure reflect the fact that different primary energy sources do not have the same ratios of energy content per dollar of output. For example, in 1972 based on BOM figures, the BTU content per dollar of crude petroleum output was about 1.7 million BTU's, whereas the figure for bituminous coal was 3.1 million BTU's per dollar, almost double. As a result, the BTU energy consumption series may move differently from either a constant dollar output series of primary energy consumption or a constant dollar value added series, including conversion, transportation, and distribution of energy. The distinction between the primary energy series, stated in BTU terms, and the series stated in constant dollars, particularly the one which reflects value added by conversion, etc., is commonly described as being analogous to the difference between an unweighted measure of man- hours and one which takes account of the different qualities such as education of labor. As noted earlier, the measure does not include all types of energy sources or energy uses and/or losses. The estimates of energy consumption developed by BOM exclude fuelwood and certain miscellaneous categories of fuels such as pulpwood bark and woodpulp residue which are used in the pulp and paper industry. These are relatively minor items and do not play a major role in explaining the significant changes in the BTU/GDP ratio during the post-World War II period, but they are not insignificant for specific industries. In addition to energy sources such as fuelwood which are ex- cluded, there are some uses or losses of energy sources which are not included in the gross BTU consumption estimates. A major instance of such explicit exclusion is the natural gas lost due to venting and flaring in the natural gas mining stage. There are several other aspects of the aggregate BTU consumption measure that need to be clarified before turning to the gross domestic product part of the ratio. The first relates to the growing use of energy as a material, rather than as a source of fuel and power. The present energy con- sumption measure includes such uses of energy, e.g., petroleum feedstocks used to make plastic products. Although the justification for such inclusion is that it is needed in order to take account of alternative uses of basic energy sources, a less ambiguous BTU/GDP ratio might be based on a measure of energy consumption excluding nonfuel uses of energy. The BOM estimates on energy consumption for individual sources of energy provide information on such uses and these have been used in the present report to analyze the effect of increasing nonfuel uses of energy on the BTU/GDP ratio. The other aspect of the BTU consumption measure that needs to be clarified is the distinction between gross, net, and "effective" consumption of BTU's. Gross consumption refers to the use of energy before deduction of conversion and transmission losses or own use of energy by the energy industries. Probably the most important instance of energy conversion loss is in electric power generation. The gross consumption measure includes the BTU content of fossil fuels, actual or imputed, used to generate electric power. However, about two-thirds of the potential BTU's in the fossil fuels used to produce electric power is dissipated in the form of waste heat and only about one-third of the initial BTU's is contained in the electricity generated. Specifically, - 4 - the current national average heat rate (BTU's consumed per net kilowatt hour produced) is 10,500 BTU's. A kilowatt hour, when converted to BTU's on the basis of the quantity of contained energy corresponding to 100 per- cent efficiency, is 3412 BTU's per kilowatt hour or about one-third of the initial 10,500 BTU's. The BTU consumption figure used in developing the BTU/GDP ratio is the gross consumption figure, before deducting conversion, transmission losses, or own use. A net consumption figure is also developed by the Department of Interior in which the conversion loss in electric utility generation is deducted. However, this is not the only category of conversion or trans- mission loss or own use. For example, energy is also lost in gas and electric power transmission but the BOM estimates make no deductions for such losses in arriving at the net energy consumption estimate. For the purpose of this report, a crude series on net energy consum- tion has been developed which is more comprehensive than the BOM net con- sumption measure. This has been done in order to clarify one aspect of changes in the BTU/GNP ratio, namely that a net BTU/GNP ratio, adjusted to include all losses and own use by energy industries, may move differently than the gross ratio and may help to explain part of the changes in the gross BTU/GDP ratio. The losses deducted from the gross measure to derive the net energy consumption measure must be distinguished from the loss in effective work. For example, gasoline used by the motorist to drive his car is counted in both the gross and net energy consumption measures by the same amount of BTU's, based on the BTU conversion factor for gasoline. However, a sub- stantial proportion of the BTU potential is dissipated in waste heat and only part is effectively used as motive power. It is, of course, quite difficult to derive estimates for each use of energy, of the effective work done by energy compared to the potential BTU content. Some estimates have been developed which indicate that of the total gross energy consumed, less than half is used for effective work, either because of conversion and trans- mission losses or loss in the application of energy. This latter type of loss could also affect the gross BTU/GNP ratio because in order to provide an effective amount of BTU's per dollar of GNP, the economy may have to provide a changing amount of gross BTU's, depending on the various categories of energy types, end uses, and associated losses. Finally, the energy consumption measure includes both energy used in producing, transporting, and distributing goods and services as well as energy used directly by consumers for household operations and personal automobile transportation and by government in providing public services. The distinction is important for two reasons: one, the overall BTU/GDP ratio is more inclusive and therefore not analogous to a conventional technical ratio in the production part of the economy such as the energy consumed to produce a ton of steel; and two, shifts between the various categories of uses of energy can affect the overall BTU/GDP ratio because of the widely different BTU/$ ratios for energy used directly by final de- mand categories and energy used by intermediate industries to produce non- energy final goods and services. - 5 - Gross domestic product The official estimates of gross domestic product (in constant 1972 prices), a component of the national income and product accounts developed and published by the Bureau of Economic Analysis, U. S. Department of Commerce, are used in deriving the energy/GDP ratio. 4/ The measure con- ventionally used in prior estimates of the BTU/output ratio has been GNP. The GDP measure, however, is more appropriate for this purpose since it re- lates to the output of the domestic economy, and is more consistent with the energy consumption measure, which also relates to the domestic economy. The difference between GNP and GDP is "net factor income from abroad", in- cluded in the former but excluded from GDP. It is not a large component of GNP and the change in the energy/output ratio would be about the same no matter which measure was used, except for an occasional year. Finally, one of the factors which can affect the change in constant dollar GDP and therefore the BTU/GDP ratio is the selection of the base year price level used to convert current dollar GDP to constant dollars. In general, it has been found that the use of a more recent year as the base for the price weights results in a lower growth rate in real output. Until recently, the real product estimates were based on 1958 prices and at the time of the 1966-1970 reversal in the BTU/GDP ratio and for the subsequent period, the ratio was based on the real product series in 1958 prices. Since then, the entire national income and product series have been bench- marked, including data revisions and some changes in concept. In addition, the constant dollar estimates have been shifted to a 1972 price base. The shifting of the base, however, has made little difference in the rate of change in the constant dollar GDP series. The BTU/GDP ratio - some general comments Because the energy/real output measure is so widely used and sometimes misinterpreted it may be useful to set down at this point some general observations on what the ratio attempts to measure and what it does not. Also, we will try to point up some of the factors which do affect the present measure, based in part on the earlier discussion in this section relating to the separate energy consumption and real output measures but going beyond the factors discussed there. In its simplest terms, the BTU/GDP ratio represents the thermal content of energy consumed domestically per dollar of real output of the economy. Note that it is not the only type of energy/output ratio that can be calculated, Other types would include measuring energy consumed in terms of constant prices, either limited to the primary energy sources at the extraction stage (perhaps including some imputed price for hydro and nuclear power) or by summing the constant dollar value added of all the energy industries, e.g., 4. For further information on the recently revised national income and product accounts, see the January 1976 issue of the Survey of Current Business , Bureau of Economic Analysis, U. S. Department of Commerce. - 6 - electric utilities, petroleum refining. (Value added is preferred in order to avoid double counting of energy implicit in the use of real gross output.) Unless the various energy sources had the same ratio of BTU constant dollar output, which they do not, it can be expected that either of the constant dollar energy consumption measures, particularly the one that is based on value added by the energy industries, would not show the same rate of in- crease as the basic BTU primary energy consumption measure. This is so because, other things being equal, users may be expected to shift to energy sources which are either cheaper in terms of BTU's per dollar or because of convenience in use or higher energy effective use rate. The point of this comment on possible differences in various measures of energy/output is that the BTU/GDP ratio is useful as an indicator of one major characteristic of energy, heat content, but it is not the same as the factor resources used to produce, transport, and distribute energy, measured in terms of the value added by the various energy industries. In addition, the gross BTU measure is not a comprehensive measure of all demands for energy in the economy which can then be compared to total supply since it does not include fuelwood, for example, a minor factor for the recent period. Perhaps more important, it excludes exports of energy, which in the case of coal, is a significant element in the market for coal. With these caveats as to what the present ratio does not include or purport to measure, we can turn back to a discussion of some of the factors that do affect the BTU/GDP ratio more directly. Perhaps the most important characteristic of the BTU/GDP ratio is that it is not based on a fixed set of output or final demand weights so that the ratio reflects both changes in the mix of output (or demand) and changes in energy intensity ratios for individual goods and services. The change in the ratio, therefore, is affected by much more than just changes in technology as they may be reflected in energy consumption per unit of output of individual sectors in the economy. A useful schema, therefore, for looking at the factors that can affect changes in the overall BTU/GDP ratio is the distinction between the demand mix effect and the energy intensity effect. The mix effect, in turn, can be further subdivided into the change in the relative importance of direct pur- chases of energy by consumers and government and the change in the patterns of demand of all the other nonenergy final goods and services. The energy intensity effect can also be further subdivided into two components, the change in the direct use of energy per unit of output for individual industries and all the other changes in the technology of production, transportation, and distribution which have an indirect effect on the amount of energy embodied in final products. For example, if the design of automobiles changes so as to use more aluminum and less steel, and if aluminum has a higher energy content than steel, then the effect of this change is to increase total energy embodied in the automobile, even though the energy used per unit of output remains un- changed in both the aluminum and steel industry. The energy intensity ratio, defined as the direct and indirect energy contained in a final product, may also be affected by the proportion of the final product that is made from - 7 - imported materials, and the change in that proportion. For example, with no change in technology in automobile manufacturing but an increasing use of imported steel to make automobiles, the total domestic energy needed to make an automobile would be reduced since the energy required abroad to make the imported steel is not included in the domestic energy consumption measure. Change in any one or all of the factors noted above can affect the BTU/ GDP ratio and it is obvious that trying to quantify the extent to which each of these has affected the change in the ratio is an extremely difficult and complex task. It is possible to develop such an analysis of the change in the aggregate ratio, partitioning the change into its component elements along the lines indicated above. This involves the use of input-output tables converted to BTU terms, but unfortunately such tables and analyses are limited to a few select years, with the latest year being 1967, Section 5 of this report summarizes the findings of the various input-output studies of the change in energy consumption, along the lines indicated above, supple- mented by references which can be drawn from a Bureau of Economic Analysis report on input-output tables for 1968-1970, which also includes an analysis in which the changes in industry output for the 1967-1970 years are partitioned into the mix vs. coefficient effect. Lacking the detailed input-output analysis for particular subperiods that may be most relevant for trying to understand the factors underlying the change in the aggregate ratio, particularly for the period after 1967, an alternative framework is also used for analyzing the changing structure of energy use, namely, the BOM consuming sector energy balance tables. These estimates, along with related data, are analyzed in later sections of the report, but first, the long-term historical record of changes in the aggregate BTU/GDP ratio needs to be presented and discussed, and this is the subject of the next section. 8 - Section 2. Changes in the BTU/GDP Ratio -- The Historical Record, 1929-1975 As the starting point for the analysis of the factors underlying changes in the BTU/GDP ratio, the historical record of the ratio for the years 1929-1975 is given in table 1 and chart 1. It is recognized that the beginning of the long-run decline in the BTU/GDP ratio is usually given as the period immediately after the World War I decade, but the analysis of the historical record should be affected very little by using 1929 as the starting point. The point is discussed more fully later in this section. The table provides estimates of gross domestic product, in constant 1972 dollars, energy consumption in terms of BTU's, and the derived BTU/GDP ratio for each year 1929-1975, as well as the percent change from the previous year for each of the above items. The general pattern which emerges from these estimates is that of a downward movement in the BTU/GDP ratio, with considerable fluctuations in the ratio, reflecting minor inter- ruptions and instances of major reversals, of which the 1966-1970 reversal is only the latest. * The wide swings in the ratio, going from valleys to peaks and then back again to new low points, are such a major feature of the historical record, as shown in chart 1, that it is difficult to discern a clear-cut pattern in the changes. If we define interruption as either no change in the ratio or a reversal involving less than a 5 percent increase in the ratio and a major reversal is defined as an increase in the ratio of more than 5 percent, then over the 46 years covered by the series, there have been 7 interruptions in the declining trend in the ratio, and 4 major reversals. In all, 19 years of the 46 years, or over 40 percent of the total, showed either no decline or an increase in the ratio. The BTU/GDP ratio declined between 1929 and 1941, just prior to World War II, with a major reversal in the early 1930' s and a minor interruption between 1938 and 1940. There were continued sharp reductions in the ratio between 1941 and 1944, due to a considerable extent to the wartime restric- tions on use of energy. There was a major reversal between 1944 and 1947 which could be explained largely by the dislocations during the reconversion period, and the return to less restrictive use of energy after the war. There was a sharp reduction in the ratio after 1947 until 1954, when a short reversal took place, from 1954 to 1956. The reversal was not yery large but the subsequent return to a declining trend was also rather modest. The ratio continued to decline after 1956, with minor interruptions, until 1966 when it reached the lowest level achieved during the entire 1929-1975 period, ex- cluding the 1942-1945 World War II years. The major reversal between 1966 and 1970, although not as large as the earlier 1944-1947 reversal, was sufficiently steep so that although the ratio started to decline again after 1970 and fell in each year after 1970; at the end of the period, it was still substantially above the low point reached in 1966. 9 - What does one make of this record in trying to measure the long-term trend rate of change in the BTU/GDP ratio and the relationship of the various reversals, particularly over the 1966-1970 subperiod, to the longer- term rates? One way of trying to put these changes into perspective is to calculate the average rate of change between terminal years selected to (a) avoid periods of major dislocations in the economy and/or (b) represent low points in the BTU/GDP series, prior to major reversals. The average rate of change for the subperiods selected and for the different phases within the subperiods can then be compared to see if there is any pattern to the changes over successive subperiods. Estimates along these lines have been developed and are shown in table 2. The subperiods selected cover the years: 1929-1941; 1941-1954; 1954- 1966; and 1966-1975. The first three subperiods cover about the same num- ber of years, 12 to 13; the last subperiod, 9 years. The 1929-1941 sub- period spans the years between the beginning of the Great Depression and U.S. major involvement in World War II. The 1941-1954 subperiod spans the years between the beginning of World War II and the year just prior to the first post-war reversal in the downward trend in the BTU/GDP ratio. The third subperiod, 1954-1966, covers the period between low points in the BTU/GDP series, and the last subperiod, 1966-1975, spans the latest reversal-decline period. Several points about these subperiods should be noted. First, 1929 was chosen as the beginning year of the first subperiod because it is the last year prior to the depression of the 1930' s. From the viewpoint of trying to derive trend rates between low years in the BTU/GDP time series, it may not be a good terminal year and this possibility is discussed later in the section. Second, 1941 has been chosen as a terminal year rather than 1944, which is a low year in the BTU/GDP series, in order to bridge the World War II and early post-war years which involved substantial distortions in the normal pattern of change in the BTU/GDP ratio. Finally, the 1966-1975 sub- period ends with the most recent period which may not represent the end of the recovery phase from the 1966-1970 reversal. It has already been noted that it covers a shorter timespan than the previous subperiods. Also, the last two years of this subperiod represent the special influences of a major recession, oil embargo, and \/ery large energy price increases. We shall have more to say about this later. Table 2 provides estimates of total percent change and ratio of change in GDP BTU's and BTU/GDP ratios for each subperiod and the decline and reversal phases within subperiods. In addition, long-term rates of change have been calculated over successively longer and longer periods by extending the time period covered to include each subperiod in turn. The first major finding that emerges from the estimates is the clear indication of a sharp falling off in the rate of decline in the BTU/GDP ratio over successive subperiods, with the average rate going from -1.3 percent in the 1929-1941 subperiod; to -.7 percent and -.2 percent over the next two subperiods; and culminating in an increase of .4 percent in the 1966-1975 subperiod. Long-term changes calculated to include longer and longer time periods show the same dampening in the rate of decline, - 10 - as follows: Average rate of change in BTU/GDP ratio 1929-1941 -1.3 1929-1954 -1.0 1929-1966 -.7 1929-1975 -.5 The figures suggest that the retardation since 1966 of the long-term downward trend is not an "aberration", unique to this period but may represent a continuation and perhaps intensification of long-term factors which have gradually dampened the rate of decline in the ratio. One aspect of this slowdown that needs to be explored further before trying to analyze these longer-run changes in more detail is the possibility, previously mentioned, that the first and last subperiods may give misleading indica- tions of longer-run changes because they may start or end in the middle of some phase in the fluctuations of the BTU/GDP ratio. The energy/output ratio for the year 1929, for example, may be unusually high because it is at the upper end of a reversal phase and the sharp decline in the ratio after 1929 may be no more than the normal catching up after a period of reversal . In order to explore this possibility, the BTU/GDP ratio has been ex- tended back to 1909, based on BTU estimates in the Census Bureau's Historical Statistics source cited in table 1 of the report, and the move- ment in the constant dollar GNP series in the recently revised national income accounts. The results are shown in table 3. Unfortunately the behavior of the ratios is so erratic, particularly during the World War I years and the immediate post-war years, that it is difficult to distinguish a clear-cut pattern to the changes in the ratio. What the figures suggest is that the 1910-1920 decade represented the highpoint in the long-term movement in the BTU/GNP ratio, with the 1920' s showing a decline from the earlier levels, particularly during the latter part of the 1920' s. The rate of change within the decade of the 1920' s, however, varies substantially from either -2.1 percent (1920-1929), or -1.9 percent (1923-1929), to 0.0 percent (1925-1929). The higher rate would suggest that the slowdown in the rate of decline started even earlier than 1929. The fact that there was little change in the ratios between 1925 and 1929, on the other hand, could be interpreted as a period when potential reductions in the ratio were post- poned until changed economic circumstances of the 1930' s provided the stimulus. If this view is taken and the first subperiod is dated from 1925, then the 1925-1941 rate of change in the BTU/GDP ratio would be -1.0 percent compared with the -1.3 percent rate for the 1929-1941 period. The general pattern of slowdown in the rate of decline in the energy/output ratio over time would be unchanged except that that fall off between the first and second subperiods would not be as abrupt. Turning to the most recent subperiod, 1966-1975, it has already been - 11 - noted that this subperiod is shorter than the first 3; 9 years as compared to the 12-13 years for the earlier subperiods. There is the possibility that 1975 does not mark the end of a declining phase in the fluctuation of the BTU/GDP ratio and that the reason the subperiod shows an average rate of increase of .4 percent rather than no change or possibly a decline is that 1975 may be in the middle of a declining phase, following the 1966-1970 reversal . In order to see what this possibility would imply for the average rate of change in the ratio for the subperiod if it were extended to cover a period of about the same duration as the earlier subperiods, i.e., about 12 years, an estimate has been made on the assumption that the rate of decline in the ratio for the remaining 3 years of the extended subperiod is the same as in the 1971-1973 recovery period after the 1970-1971 recession, -1.25 per- cent. This assumes no additional adjustment for the effect of the large increases in energy prices starting at the end of 1973, which may result in even larger reductions in the BTU/GDP ratio than occurred during the earlier 1971-1973 post-recession period. On this basis, the BTU/GDP ratio at the end of the hypothetical subperiod, extended to cover 12 years, would be 57,882 BTU's per dollar of GDP, almost exactly the same as in 1966 at the beginning of the subperiod. The rate of decline in the ratio between 1970 and hypothetical 1978 would be -1.0 percent, with no net change in the ratio for the period as a whole. These estimates are shown as an addendum to table 2. (The analysis of changes in the BTU/GDP ratio in this report covers the period through 1975. Preliminary data for 1976 indicate a decline of 1.1 percent in 1976. ) The previous pattern of a fall-off in the rate of change in the BTU/GDP ratio over successive subperiods, culminating in an actual increase in the rate over the last subperiod would be modified based on the hypothetical extension of the 1966-1975 subperiod, to indicate a continuation of the dampening in the rate of decline in successive subperiods, from -.2 percent in the 1954-1966 subperiod to little or no change in the period since 1966. At the yery least, it would suggest that over the period since about 1954, there have been no significant reductions in the BTU/GDP ratio, aside from short-term fluctuations. It modifies but does not substantially change the conclusion reached earlier that the retardation since 1966 of the long-term downward trend is not an "aberration", but may represent a continuation and perhaps intensification of long-term factors which have gradually dampened the rate of decline in the ratio. The next section analyzes the relationship of short-term changes, particularly the pattern of reversals and declines within subperiods, to longer-term changes. - 12 Rate Total percent increase Duration 3.0 6.1 2 8.3 27.0 3 2.7 5.6 2 2.1 8.5 4 Section 3. The Relationship of Short-Term Fluctuations to Longer-Run Changes in the BTU/GDP Ratio One possible explanation for the slowdown in the decline in the BTU/ GDP ratio from one subperiod to the next, is that the reversals have in- creased in severity and duration over time. To see if this may provide part of the explanation for the slowdown, the reversal phases in the sub- periods have been compared, in terms of rate of change, total percent change, and duration. Comparison of BTU/GDP Change During Reversal Periods 1931-1933 1944-1947 1954-1956 1966-1970 The rate of increase in the ratio during reversal phases of successive subperiods shows the 1966-1970 reversal to have the lowest rate of the four reversals, with the 1954-1956 reversal having the next lowest rate of in- crease. The largest rate of increase was in the 1944-1947 period. If this dampening of the subperiod rate of change in the BTU/GDP ratio is not ex- plained by increasing severity of increases in the rates during reversal phases, it may still be explained by the duration of the reversal phase so that the total increase in the ratio over the reversal phase may show in- creasing severity. Here, too, the figures do not indicate a continuing increase from one subperiod to the next in terms of duration of reversal phases or total percent increase, with the exception of the last subperiod, which does show a longer period: 4 years compared to 3 years for the 1944- 1947 reversal phase, and 2 years for the other two; and moderately higher total percent increases than the first and third reversals but much less than the sharp jump in the ratio between 1944 and 1947. Although the latest reversal, 1966-1970, does show a higher rate of in- crease than the first and third reversals, and therefore made some contribu- tion to the fact that the rates over the 1966-1975 period as a whole increased rather than declined, as in the previous three subperiods, this may not be the major factor in explaining the increase in the ratio. Rather, as indicated in the previous section, the increase in the ratio between 1966 and 1975 may be much more related to the fact that the last year of the subperiod, 1975, may in fact be in the middle of a declining phase of the BTU/GDP "cycle", and that when the estimates are extended to cover a period of about the same length as the previous subperiods, about 12 years, the ratio shows little or no change for the subperiod as a whole. 13 - In any event, it seems to be a fair conclusion from these figures that, at least through the first three subperiods, increasing severity in the reversals is not a major factor in explaining the slowdown in the rate of decline in the ratio, inasmuch as the third subperiod, 1954-1966, had the smallest of the three total percent increases in the ratio. The rate of the reversal in the latest subperiod is more conjectural, depending on what one makes of the fact that the elapsed time since the end of the reversal period is much shorter, only 5 years, as compared to the similar phases of 7 to 10 years in the earlier subperiods. This possibility is explored further at the end of this section as one aspect of the analysis of short-term fluctuations in the BTU/GDP ratio. Although the severity of the reversals does not seem to have been the primary explanation for the diminished rate of decline in the BTU/GDP ratio over successive subperiods, the pattern of decline-reversal -decline from one subperiod to the next may reflect underlying factors which do have implications for longer-term changes in the BTU/GDP ratio. One of these factors is the sensitivity of energy consumption to short-term changes in real output. In order to see if any pattern emerges in the relationship between the two for subperiods of declines and reversals in the BTU/GDP ratio, estimates for the subperiods, taken from table 2, were grouped into two categories—those with declines in the ratio and those with increases. It should be noted that the most recent subperiod was omitted from the comparison because, in contrast to the other subperiod which were characterized by declining real prices of energy or moderate changes, energy prices during the last subperiod increased sharply following the oil embargo at the end of 1973. Because the relationship between changes in GDP and the BTU/GDP ratio may be obscured by the substantially different behavior in energy prices compared to earlier subperiods, the most recent subperiod was omitted. The results of the comparison are shown in table 4. Of the ten sub- periods which cover the 1929-1973 historical record, six show declines in the BTU/GDP ratio and four have increases. All of the subperiods showing declines in the BTU/GDP ratio, with the exception of one subperiod, 1929- 1931, also show increases in real output of about 4 percent or more. The subperiods showing increases in the BTU/GDP ratio have a more mixed record. Of the four subperiods with reversals in the trend of the ratio, two show absolute declines in GDP, one has a low growth rate, 2.3 percent, and only one, 1954-1956, has a better than average growth rate, 4.4 percent. In general, one may conclude that during subperiods when the growth rate is about 4 percent or better, the BTU/GDP ratio declines; when real output declines or increases at substantially less than 4 percent, the BTU/GDP ratio increases. Eight out of the ten subperiods are consistent with this general conclusion. - 14 - The proposition that there is a fairly close connection between growth rates and the direction of change in BTU/GDP ratios may be examined further in two ways: first, by selecting subperiods based on "high" and "low" rates of growth in GDP, rather than on changes in the BTU/GDP ratio, and seeing whether the proposition holds; and second, by examining the relationship based on annual changes in the two variables. The relationship based on rates of change in GDP rather than the BTU/ GDP ratio is shown in table 5. "Low" growth is defined as an increase of less than 3.5 percent per year; "high" growth as 3.5 percent or more. Of the nine subperiods selected to cover the entire span of years between 1929 and 1973, five are the same as the subperiods chosen on the basis of changes in the BTU/GDP ratio and four differ. The earlier conclusion on the strong inverse relationship between growth in GDP and the direction of change in the BTU/GDP. ratio is reinforced by the comparisons shown in table 5. All of the subperiods showing "low" growth in GDP show either no declines in the BTU/GDP ratio or increases. All the subperiods with "high" rates of increase in real output show declines in the ratio of about 1 percent or more. The annual data on changes in GDP and the BTU/GDP ratio, shown in table 1, have also been analyzed to see if the finding of an inverse relation- ship between the size of increases in GDP and the direction of change in the BTU/GDP ratio also holds for annual changes. Setting the dividing line between "high" and "low" increases in GDP at 3.5 percent, the annual in- creases for the 44 years between 1929 and 1973 have been compared with the following results: GDP Change Direction of Change in BTU/GDP Ratio Increases Declines Total "High" growth (3.5 percent or more) "Low" growth (Less than 3.5 percent or declines) 12 18 25 19 Of the 44 years covered in the analysis, 30 years, or almost 70 percent of the total, showed an inverse relationship; that is, in 18 "high" growth years, the BTU/GDP ratio declined, in 12 "low" growth years, the ratio in- creased. The inverse relationship was somewhat better for years with "high" growth rates with 18 of the 25 years, or 72 percent, showing this inverse relationship; whereas 12 of the 19 years with "low" growth rates, or 63 per- cent, showed increases in the BTU/GDP ratio. The inverse relationship found in the changes, based on both the sub- period and annual data, suggests that a large part of the pattern of alterna- ting periods of declines and reversals found in the historical record of changes in the BTU/GDP ratio is related to alternating periods of "high" and "low" growth in the economy. This finding helps put the 1965-1970 reversal in 15 perspective by indicating that given the "low" growth rate for the sub- period as a whole, 2.3 percent, and with three of the four years of the period showing "low" growth rates, 2.7 percent or less, it is quite consistent with the historical record to find a decline in the ratio for this subperiod. This does not suggest that other forces, such as those discussed by Bruce Netschert in the NERA study (op. cit.), e.g., changes in heat rate in electric utility generation, may not have played a role in the 1966-1970 reversal. Rather, it does suggest that it is not necessary to rely primarily on factors, which are unique to this period, e.g., Vietnam War; or longer-run secular changes, e.g., increased use of electricity for air conditioning, to explain the 1966-1970 reversal. Further insight into the factors which explain short-term changes in the BTU/GDP ratio and implicitly the relationship between short-term and long-run changes in the ratio is provided by the equation developed by Joseph D. Parent and Henry R. Linden to explain annual changes in energy consumption for the 1947-1975 period. 5/ The explanatory variables in the equation are (a) civilian employment, labor force, (b) real price index for fossil fuels and power, and (c) population. Of the three variables, employment is the most important. The annual energy consumption estimates, calculated from the equation in logarithmic form, when related to the actual values of real GDP, provide a very close fit to the historical BTU/GDP data and particularly the reversals in the changes in the ratio, both for annual and subperiod changes. The equation 6/, fitted to data for the period 1947-1975, is: -0.1791 1.864 0.2156 E = 0.01414F M P Where 15 E = Annual primary energy consumption, 10 BTU s (quads) F = Relative price index for fossil fuels and power (1972 = 100) M = Employment, civilian labor force, millions P = Total population, millions 5. Joseph D. Parent and Henry R. Linden, U.S. Energy Consumption Through 1981, June 18, 1976, Institute of Gas Technology, Chicago, Illinois. Permission has been granted by the authors to use the material contained in their report. 6. The authors note that "the fit of this simple relationship to the historical data may be judged by the multiple correlation coefficient of 0.999 and Durbin-Watson coefficient of 1.85. It also passes, at the 5 percent level of significance, the Chow test of suitability of the coefficients found for the entire range of 1947 to 1975 and for the two intervals, 1947 to 1960 and 1961 to 1975." 16 - Of the statistical sources used by Parent-Linden in deriving their energy equation, two require some explanatory comment. The relative energy price index used in the equation is the fossil fuel and related products and power index, a component of the Wholesale Price Index (WPI), compiled by the Bureau of Labor Statistics, divided by the GNP implicit deflator. Because some of the fuels are used to provide secondary forms of energy, e.g., electric power, their values are duplicated in the index and as a result, the WPI energy price index also provides a duplicated measure of price change. This may be a major limitation of this particular measure of energy price change for some purposes, but apparently it "works" as one of the explanatory variables in the energy consumption equation. The other comment relates to the measure of employment used in the equation. This is the civilian employment component of the labor force series, published by the Bureau of Labor Statistics. It does not cover total employment since it excludes the Armed Forces. This presents a problem because it is used as an explanatory variable in deriving estimates of energy consumption, which in turn will be related to GDP, but GDP includes the "output" of the Armed Forces, but civilian employment, by definition, does not. In addition, the civilian employment series as published includes "breaks" in the series because of benchmark adjustments to the estimates which were not covered back to prior years. Although not very large, these noncomparabilities in the data may distort changes in employment for individual years and possibly some subperiods. Finally, it should be noted that the employment series, based on the labor force survey, does not agree with the movement in estimates of employment based on establishment reports. In comparisons of productivity change, these differences in the employment series may affect comparisons of the change in output per worker based on a variety of measures. With these qualifications in mind, we can turn back to the equation and see what it tells us about energy consumption and changes in the BTU/GDP ratio. The equation indicates that based on the historical experience of the 1947-1975 period and for moderate changes in the explanatory variables, for every 1 percent increase in the real price of energy, energy consumption (in BTU terms) would decline by .1791 percent; a 1 percent increase in civilian employment would be accompanied by an increase of 1.864 percent in energy consumption; and similarly, an increase in population of 1 percent would yield only a .2156 percent increase in energy consumption. The change in employment, along with its high elasticity of change in energy consumption, is the most important factor in explaining annual changes in energy consumption. Comparison of the actual and calculated annual BTU/GDP ratios (based on calculated BTU's and actual GDP) and changes in the ratios is provided in table 6 and chart 2. In general, the equation performs very well in tracking actual energy consumption and the direction of change in the BTU/GDP ratio. Of the 28 yearly changes over the 1947-1975 period, in only four years did the change based on the calculated ratio differ in direction from the actual. (In one other year, there was no change in the actual and a small decline of -.3 percent in the calculated.) Of the nine years with actual increases in the ratio, - 17 - in only two did the calculated ratio show a decline. In both these in- stances, the actual increase was slight, .1 percent and .2 percent. The direction of change in each year of the two reversal periods, 1954-1956 and 1966-1970, was correctly estimated by the equation. A similar comparison of subperiod changes is shown in table 7. Over the entire 1947-1975 period, the annual rate of change of the derived BTU/ GDP is wery close to that of its actual ratio, -.6 percent vs. -.5 percent. If the last two years are excluded from the calculations because the change in the BTU/GDP ratio in those years might have been outside the general experience of the post-1947 period, the rates remain the same. The rate of change in the derived BTU/GDP ratio for the 1953-1973 period, when there was no net change in the actual ratio, also showed little change, .1 percent. The rates of change for most of the subperiods, based on the derived ratio, were also quite close to the actual for the longer subperiods, somewhat less so for the shorter subperiods. Interestingly enough, the estimated change for the 1970-1975 subperiod was identical with the actual change, -.9 per- cent, but the two shorter subperiods, 1970-1973 and 1973-1975, deviated from actual by a fair amount but in opposite directions, offsetting each other. The direction of change, based on the derived ratio, was the same as the actual change in every subperiod, including the two reversal periods. In order to gain additional insight into the respective roles of the three variables in "explaining" the relationship between the factors deter- mining energy consumption and changes in the BTU/GDP ratio, the equation can be interpreted as follows: Increases in civilian employment lead to more than proportionate in- creases in energy consumption, i.e., the energy/employment elasticity is 1.864. However, the elasticity of real output (GDP) to changes in employ- ment may not be the same as the energy/employment elasticity and in fact, for the 1947-1973 period (which leaves out the last two years with a major recession and energy price increases), the average elasticity has been con- siderably higher--2.494. This means that increases in civilian employment have resulted in greater increases in GDP than in BTU's, and the BTU/GDP ratio has fallen, on the average, for the 1947-1975 period as a whole. More specifically, for this particular period, a 1 percent increase in civilian employment has been accompanied by a decline in the BTU/GDP ratio of -.630 percent. Since the average rate of increase in civilian employment for the period was 1.52 percent, the decline in the BTU/GDP ratio due to this factor alone would have been almost 1 percent per year. Declines in real energy prices and increases in population, along with their respective energy elasticities, offset part of this decline, however. Real energy prices fell during most of this period, declining at a rate of almost .8 percent per year. Given the energy/price elasticity of -.1791, the effect of the decline in energy prices was to increase the BTU/GDP ratio by .14 percent per year. Similarly, an average rate of growth in population of almost 1.5 percent per year with an energy/population elasticity of .2156 resulted in an additional increase in the BTU/GDP ratio of .32 percent per year. The net effect of the last two factors was to reduce the initial decline in the ratio from almost 1 percent to about one-half of 1 percent per year. - 18 This estimate of decline in the ratio for the 1947-1973 period, based on the equation, is about the same as the actual decline in the ratio, .6 per- cent, as shown in table 7. Although the elasticities for each explanatory variable is kept constant over the entire 1947-1975 period, the rates of change for the variables fluctuate from one subperiod to the next, resulting in estimated subperiod changes in the BTU/GDP ratio which, as already noted, track the actual changes quite well, including the reversal periods. The preceding interpretation of the contribution of each of the variables to the rate of change in the BTU/GDP for the 1947-1973 period as a whole can also be applied to each subperiod to identify those factors which might explain (a) the reversals in the decline of the ratio, and (b) the slowdown in the rate of decline in the ratio after the early 1950' s. Table 8 shows the rates of change for the various subperiods of the explanatory variables as well as the change in the real output/civilian em- ployment elasticity and ratio. The figures shown in the table have been rounded, although the actual calculations have been done in more detail. Table 9 uses the subperiod rates of change of the explanatory variables and the various elasticities to derive estimates of the contribution of each variable to the change in the BTU/GDP ratio, along the lines indicated above. In general, the BTU/GDP ratio will decline when (a) the GDP civilian employment elasticity is greater than the BTU/GDP elasticity of 1.864, or (b) real energy price index increases, or (c) population declines. During the entire 1947-1975 period, population increased, although at a declining rate; and for most of the period, until recently, relative energy prices declined so that both these factors tended to increase the BTU/GDP ratio rather than reduce it. It follows from the above that in order for the BTU/GDP to decline, the subperiod GDP/civilian employment elasticity had to be sufficiently higher than the long-term BTU/civilian employment elasticity of 1.864 to more than offset the two negative factors—real energy price and population growth. Table 8 indicates that for the first two subperiods which did show reductions in the BTU/GDP ratio, 1947-1954 and 1956-1966, this is what occurred. In both subperiods, GDP/civilian employment elasticities exceeded the long-term BTU/civilian employment elasticity by substantial margins. Conversely, during the two subperiods during which the BTU/GDP ratio increased, 1954-1956 and 1966-1970, the GDP/civilian employment elasticity fell substantially below 1.875. The anomaly of the increase in the BTU/GDP ratio in 1954-1956 at a time when growth in real output was quite high, 4.4 percent annually, noted in the earlier discussion of the relationship between high growth rates in GDP and declines in the BTU/GDP ratio, is now clarified. Although the growth rate during 1954-1956 was higher than average for the 1947-1973 period as a whole, the growth was due largely to the increase in civilian employment and not to a better than average increase in real output per civilian worker. The latter ratio increased substantially below the post World War II average rate of increase, through 1973, of 2.2 percent. It increased only 1.3 percent during the first reversal of 1954-1956 and even lower, .4 percent, during 1966-1970. As a result, the real output/employment elasticities increased substantially less than the long-term BTU/employment elasticity in both reversal periods. - 19 - The explanation for the change in the BTU/GDP ratio during the period since 1970 is more complex and requires some comment. In 1970-1973, relative energy prices, which had been declining, turned around and began to increase. The increase in prices helped dampen energy consumption and reduce the BTU/GDP ratio. Although the real output/employment elasticity, at 2.0 percent, was below the average for the post World War II period, it was still marginally above the energy/employment elasticity of almost 1.9 percent. This, plus the effect of higher energy prices, particularly the latter, was sufficient to more than offset the population effect, and the BTU/GDP ratio declined. Between 1973 and 1975, real output per worker and GDP actually de- clined, although there was a slight increase in civilian employment. As a result, the real output/employment elasticity became negative and energy consumption per dollar of real GDP would have increased if this had been the only factor involved. The increase in the ratio would have been con- sistent with past behavior during periods of less than average increases in GDP. In this instance, however, the large increase in real energy prices, following the 1953-1954 oil embargo and subsequent sharp increases in petroleum prices, more than offset the effect of the decline in real out- put per worker on the real output/employment elasticity so that the BTU/ GDP ratio continued to decline. Table 9 quantifies the contribution of the change in each variable, along with the related elasticities, to the change in the BTU/GDP ratio. As might be expected, the employment variable combined with the differential effect of employment on GDP and energy consumption was the most important factor, at least through the 1966-1970 subperiod. As we have already noted, the swings in this factor from successive periods of above average to below average increases in output per worker, as reflected in similar swings in the GDP/civilian employment elasticities, were the dominant influences in the total change in the BTU/GDP ratio. The population effect was to in- crease energy consumption and the BTU/GDP ratio throughout the long-term period, but at a diminishing rate, from .4 percent per year to about half that at the end of the period. The effect of energy prices was also quite small, with no appreciable effect in the beginning and rising to .4 percent in the 1966-1970 subperiod. The increase in energy prices during the period since 1970, however, did have a major influence in reducing the BTU/GDP ratio, and, as noted above, was the key factor in continuing the decline in the ratio after 1973, offsetting increases in the ratio due to the employment and population factors. What are the implications of this analysis of the factors underlying the swings in the BTU/GDP ratio over the 1947-1975 period for longer-term changes in the ratio. As noted in Section 2 of this report, the rate of decline in the ratio has been falling off over successive subperiods, from 1929-1941 on, with the most recent period either showing an increase in the ratio or no further decline, depending on whether the last period is considered to have effected a full recovery from the 1966-1970 reversal. The Parent-Linden equation, as modified and used in this report, covers only the period since 1947 and cannot be used directly to explain the longer- - 20 - term changes going back to 1929, but it may still be useful in identifying the factors underlying the dampening in the rate of decline in the BTU/GDP ratio for the post World War II period. We have already noted, that of the three explanatory variables, the change in employment and the disparity in the GDP and BTU elasticities associated with the employment change has been the most important factor, at least through 1970. Since the BTU/ employment elasticity, at 1.864, based on the 1947-1975 experience, is con- stant in the equation, the decline in the BTU/GDP ratio will be retarded if the GDP/employment elasticity is not constant, but begins to slacken and increases at a reduced rate. Since GDP is the product of employment and output per worker, it follows that the GDP/employment elasticity will slacken if the rate of increase in output per worker falls off or does not keep up with the rate of increase in employment. The record, shown in table 8, indicates that over the subperiods, 1947-1954, 1954-1966, and 1966- 1973, while the rate of increase in civilian employment was increasing from .8 percent per year to 1.7 percent and then 2.1 percent, the rate of increase in output per civilian worker was declining, from a 3.2 percent annual rate to 2.3 percent and down to 1.2 percent in the last subperiod. 7/ As a result, the GDP/civilian employment elasticity declined from 5.2 percent, to 2.5 percent and then 1.6 percent. To summarize, during most of the period 1947-1975, increases in popula- tion and reductions in real energy prices resulted in increasing energy consumption. During the early part of the post World War II period, the growth in real GDP associated with an increase in civilian employment was sufficiently greater than the comparable response of energy consumption to increased employment as to more than offset the population and real price effects. As a result, the BTU/GDP ratio declined. Between 1953 and 1973, however, the growth in GDP relative to increases in employment (excluding Armed Forces) was reduced from the earlier elasticity, and although still larger than the BTU/employment elasticity, was just enough along with the employment increase to offset the population and real price effects. The 1 • It should be noted that various measures of productivity yield different estimates of the slowdown in productivity gain for the period since 1947. These differences are due to a number of factors, including: (a) differences in concept and related measures, e.g., output per employed person, per man- hour, or per unit of total factor input; (b) the methods used to adjust for "quality" of factor input; (c) the level of aggregation used, e.g., total economy or nonresidential business; (d) the subperiods chosen for calculating trend rates; and (e) whether the measures refer to actual or potential productivity. The various measures yield somewhat different results on the question of whether there has or has not been a slowdown in productivity growth for the period through the mid-1960 1 s, but there is general agreement that since about 1966 there has been a fall-off in the growth in actual productivity. There is less agreement on whether there has also been a slow- down in the rate of increase in potential productivity during the most recent subperiod. - 21 - BTU/GDP ratio, therefore, in 1973 was about the same level as it had been about 1953 although in between these two years the ratio went through two periods of increases in the ratio followed by declines. Between 1973 and 1975, a major new element was added, namely, the very large increases in real energy prices. The increase in real energy prices reduced energy con- sumption substantially, enough to more than offset the increase in the BTU/ GDP ratio as a result of the sharp reduction in the GDP/employment elasticity relative to the BTU/employment elasticity. As a result, even though one might have expected an increase in the BTU/GDP ratio during a major recession, the ratio declined. What does this interpretation of changes in the aggregate BTU/GDP ratio over the 1947-1975 period imply for further changes over the next 10 to 15 years? Setting aside for the moment the possible effect of future real energy price changes on energy consumption and the BTU/GDP ratio, the post-recession period of economic recovery is usually accompanied by much better than average increases in output per worker. If the present recovery period is sustained and follows past patterns, we can expect that, as a result of the better than average increases in output per worker, the GDP/employment elasticity will be sufficiently larger than the BTU/employ- ment elasticity as to more than offset the population effect on energy consumption so that the BTU/GDP ratio might be expected to decline. In addition, the delayed effect of the large energy price increases since 1973, as well as any new relative energy price increase, should dampen energy consumption and reduce the BTU/GDP ratio even further. The course of the BTU/GDP ratio, after the short-term gains in productivity have been realized, is more problematical. A great deal depends on what the potential growth rate in output per worker is during this period and whether we are able to achieve gains in actual output per worker approximating the potential gains. In this regard, Edward F. Denison, senior fellow at Brookings Institution, is currently revising and updating his earlier work on accounting for U.S. economic growth to incorporate the recent revisions in the national income accounts. 8/ The revised version should provide new insights on the trend in potential growth and potential output per worker since 1969, the terminal date of his earlier study, and 8. Edward F. Denison, Accounting for United States Economic Growth, 1929-1969, Brookings Institution, 1974. - 22 whether a slowdown in potential output per worker has occurred or might develop over the medium term. Perhaps more basic to the outlook for changes in the BTU/GDP ratio is whether the functional relationships in the Parent-Linden equation, derived during a period of general decline in real energy prices, will hold under conditions of substantial pressure on energy prices. This could affect not only the elasticity of energy consumption relative to prices but also the energy consumption/employment elasticity which in the equation is much more important than the price variable. This can only be determined by seeing whether this equation holds up over the next few years when real energy prices may behave quite differently than in the past, accompanied by in- creasing concern with conservation of energy resources. - 23 Section 4. Gross, Net, and Effective Uses of Energy in Relation to the BTU/GDP Ratio In the previous section, the analysis of changes in the BTU/GDP ratio was largely limited to the movement in the overall ratio itself. In this section, we turn to an examination of changes in the composition of energy supply and demand to determine how such changes have contributed to the movement in the overall BTU/GDP ratio during the period since 1947. The analysis is limited to 1947-1975 because a major body of data used for the analysis of the components of energy consumption, namely, the BOM energy balance tables are not available for the period prior to 1947. It should be noted that starting the analysis with the year 1947 may give a somewhat misleading impression of a sharp slowdown in the rate of decline in the BTU/GDP ratio after the first subperiod, 1947-1954, because, as was noted earlier, this subperiod represented a strong recovery phase in the move- ment of the BTU/GDP ratio following the earlier 1944-1947 reversal in the ratio. Extending the analysis back to the pre-World War II period would still show a slowdown in the decline of the ratio but at a more moderate rate. The analysis in this section is based on an extension of the detail shown in the BOM energy balance tables to take account of certain categories and measures of energy use, mentioned in Section 1 of this report, which cut across the consuming sector detail shown in the energy balance tables. As background for the analysis which follows, it may be useful to briefly describe the measures of nonfuel , net, and effective uses of energy used in this report, and their relationship to the categories used in the energy balance tables. There are three categories of energy use whose movements may significantly affect the change in the overall gross BTU/GDP ratio, and may help explain the slowdown in the decline in the ratio over the 1947-1975 period. The first of these is the growing use of energy for nonfuel purposes, particularly as feedstocks for the chemical processing industries. If such use is sufficiently large as a component of total energy consumption and represents an increasing proportion of the total, this could contribute to an increase in the BTU/GDP ratio, or a slowdown :n the decline of the ratio, given the change in GDP. Excluding nonfuel uses from total gross energy consumption leaves energy consumption for fuel and power use. Part of total energy consumption for fuel and power is used or lost by the energy industries in the course of providing energy to the rest of the economy, the end-users of energy. End- use consists of both use by intermediate producers, excluding use by the energy sector itself, and final demand, e.g., personal consumption, government. Energy used or lost by the energy sector includes electric power conversion and distribution loss by the electric utility industry, energy used by the petroleum refining industry, and natural gas used by the natural gas industry 24 - as pipeline fuel, also in the extraction stage and gas lost in transmission. Increasing proportions of energy used or lost by the energy sector could also contribute to the increase in the BTU/GDP ratio or a dampening in the rate of decline. Deducting energy sector use and loss of energy from total energy con- sumed for fuel and power yields an estimate of net energy provided to end- users. Efficiency in the use of energy by end-users, i.e., household and commercial, industrial, and transportation consuming sectors, may offset part or all of the energy losses or uses by the energy sector because (a) the converted forms of net energy delivered to users may have higher thermal efficiency factors than the initial form and (b) there may be a gradual sub- stitution of energy with relatively high thermal efficiency factors for those with low ratios. An example of the former is electric power which has a high conversion loss, currently about two-thirds of primary energy input to electric utilities, but is generally given a 100 percent thermal efficiency factor for the kilowatt hours delivered. An example of the shift among energy types with varying efficiency factors is the shift from coal, which has a relatively low efficiency factor compared to other types. Deducting losses in the application of energy to end-users from total net energy yields a measure of effective use of energy. The extent which movement in effective energy differs from the change in gross energy used for fuel and power depends on how much the gains in effective use of energy offset the losses and uses of energy by the energy sector. The initial data base and framework for the analysis of energy con- sumption along the lines indicated above is the BOM energy balance table which, however, needs to be supplemented extensively to meet the needs of this type of analysis. The annual energy balance table provides information on energy supply, by detailed types of energy, cross-classified by each of five categories of domestic use. The types of energy included in the gross energy supply measure are coal (anthracite, bituminous, and lignite), dry natural gas, petroleum products (including still gas, liquified refinery gas, and natural gas liquids), hydropower and nuclear power. The five categories of energy use are household and commercial, industrial, trans- portation, electricity generation (utilities only), and miscellaneous and unaccounted. Since electricity generation is not a final use of energy but a conversion of energy from one form to another, information is also provided in the table on the use of utility electricity by the other major consuming sectors. As noted earlier, the electric power (KWH) consumed is converted to BTU's on the basis of 3,412 BTU's per kilowatt hour, the theoretical rate assuming 100 percent efficiency rather than on the BTU content of the fossil fuels, actual or imputed, consumed in generating utility electric power. Conversion losses, actual or imputed, are attributed to the electric utility industry in the energy balance table. Deducting the estimate for electricity conversion loss yields a -;timate of "net" energy consumption. As mentioned earlier, this is only a first approximation of a net energy consumption figure, since there are other categories of either conversion or transmission losses that need to be deducted in arriving at a more "net" energy figure. 25 - It should be pointed out that these energy consuming categories do not correspond to the kind of classification system used in the national economic accounts, even in terms of broad aggregates. The major difference is that the BOM categories refer basically to type of use or function whereas the national economic accounts, and more particularly the input- output accounts, are based primarily on purchasing categories covering all functions. The difference can best be illustrated by an example: gasoline purchases by households are included in the transportation consuming sector in the BOM energy balance tables whereas they are included in personal con- sumption expenditures in the national economic accounts. Similarly, government purchases of fuel for all transportation purposes, including military use of petroleum products for aircraft and ships, are included in the BOM transportation sector. Use of gasoline, as shown in the input-output accounts, by intermediate producers such as agriculture, construction, mining, and industry for operating various types of vehicles, including off-highway vehicles, are also included in the BOM transportation sector. In general, the consuming sector system used by BOM is not only different from the "industry" system of the national economic accounts, but is not entirely consistently applied among the various types of fuels and from year to year. Natural gas used by government, for example, has been classified as part of "industrial" use, but more recently has been shifted over to "house- hold and commercial". With some exceptions, "household and commercial" use covers energy con- sumption for space heating by all sectors of the economy as well as use of energy for other household operations and commercial activities. It also in- cludes asphalt and tar used in highway construction. "Industrial" use covers energy consumption of mining, manufacturing, and gas utilities for operating purposes, including use of energy for chemical feedstocks and excludes trans- portation and space heating use of energy by these sectors. Use of diesel oil by off-highway vehicles in agriculture and construction is also included in industrial use. "Transportation" includes energy used for all transporta- tion activities in all sectors and is not limited to the transportation industries as defined in the Standard Industrial Classification (SIC) system. It also includes use of gasoline by off-highway vehicles and natural gas used as compressor fuel by pipelines to move fuel through the pipelines. The broad categories used in the BOM energy balance tables, the loose definition of these categories, and the fact that the allocation of specific uses of fuels to these categories has been changed from time to time, make it difficult to use these data for the analysis of changes in energy use by sector. Nevertheless, it represents the major comprehensive and continuing source of annual information on components of energy consumption, consistent with the aggregate BTU estimates, and will be used in this report for the initial disaggregation of overall energy use among consuming categories. This is supplemented and extended in various respects to provide estimates needed for the analysis discussed earlier. These are discussed below: - 26 - 1. Gross energy consumption, including allocation of energy used or lost by energy industries to end-use consuming sectors. In order to be able to relate the energy used or lost by the energy industries to the gains in thermal efficiency in end-use, the energy industries' use or loss of energy need to be allocated to the end-use con- suming sectors. The principal category of energy industry use or loss of energy is electric power conversion loss. This is distributed to the three major consuming sectors, excluding miscellaneous and unaccounted, on the basis of the use of electricity among the three sectors. A second category of energy use or loss is electric power distribution loss but this is already distributed implicitly to the consuming sector by the procedure used in developing the estimates of the BOM energy balance tables. Petroleum refining use of energy is deducted from the industrial sector consumption estimate and distributed to the three sectors as follows: The amount of energy used by petroleum refineries is allocated to the transporta- tion sector, based on that sector's share of total net consumption of petroleum products. The residual amount of energy used by petroleum re- fineries is added to the total amount of natural gas used or lost by the natural gas industry, and the total of the two is distributed to industrial, and household and commercial use, based on their respective shares of total net energy consumption. Since transportation does not use natural gas, no allocation was made to this sector. To avoid double counting, the natural gas used or lost by the natural gas industry was deducted from the BOM sectors in which they are initially included, prior to the redistribution, in the same way that energy used by petroleum refining was deducted. The reallocation of the various categories of energy used or lost by the energy industries to the three major end-use consuming sectors yields estimates of gross energy consumption, by using sector, which sums to the original total gross energy figure, except for energy used by the mis- cellaneous and unallocated category. 2. Nonfuel use Both the household and commercial and industrial sectors include non- fuel uses. These are identified in the BOM energy balance tables and can be deducted to derive the fuel and power use for each sector and in total. 3. Energy used or lost by energy industries These have already been discussed in item 1 above and are deducted from the fuel and power energy use figure to derive the net energy consumption estimates, separately by consuming sector and in total. 4. Energy lost in end-use application Each type of energy and end-use has its own thermal efficiency factor which is shown in table 10. The factors are largely based on those given in the major study by Sam Schurr et al , "Energy in the American Economy, 1850- 1975," Resources for the Future, Inc., 1960. Multiplying the efficiency - 27 - factors by the quantity of each type of energy used in each consuming sector, yields estimates of effective use of energy, by sector. The difference between net and effective use represents the energy lost in end-use applica- tions. It should be emphasized that the efficiency factors are engineering- type estimates, and should be considered as rough approximations useful primarily in providing some indication of relative differences in average thermal efficiencies among energy types and uses. In spite of the crudeness of such estimates, the need for including them in the analysis cannot be avoided since they represent an important element in the explanation of changes in the gross BTU/GDP ratio. Estimates of energy consumption based on the adjustments indicated above have been developed for selected years of the 1947-1975 period, including years which represent turning points in the movement of the gross BTU/GDP ratio. The estimates are given in table 11 and are shown as a percentage distribution of total gross energy consumption in table 12. The latter table is useful in showing the changing pattern of energy consumption over the period since 1947. Miscellaneous and unaccounted energy consumption, a small percentage of total consumption, increased between 1947 and 1956, reached a peak of 2.3 percent of the total in 1956, and declined to almost zero by 1975. Nonfuel use increased its share throughout the period, from about 4 percent in 1947 to about 6 percent by 1975. The increase between 1966 and 1970 is particularly large and is a significant factor in the increase of energy consumption by the industrial sector during that period. The respective shares of the miscellaneous and nonfuel categories moved in opposite direc- tions, thus, in combination, they accounted for about the same proportion of total energy consumption throughout the period—about 6 percent. As a result, 94 percent of total energy was used for fuel and power by the other sectors. The quantity of energy used or lost by the energy industries increased as a proportion of total energy consumed throughout the period, from 17 per- cent at the beginning to slightly over 27 percent by 1975. Therefore, the share of total energy available to end-use consuming sectors declined from about 77 percent in 1947 to about 67 percent in 1975. The relative loss of BTU's due to the increasing proportion of energy used or lost by the energy industries was more than offset by the increased thermal efficiency with which energy was applied by end-use sectors for the period between 1947 and 1966; thereafter, the gains in thermal efficiency were not enough to offset the energy sector's needs for energy. As a result, effective use of energy declined as a proportion of the total. In 1947, effective end-use of total energy consumption was about 36 percent; by 1966, it had increased to 40 percent. Thereafter, it declined and by 1975 only about 37.5 percent of total energy consumed in the domestic economy repre- sented effective end-use of energy for fuel and power, slightly above the proportion in 1947. The remaining 62.5 percent of gross energy consumption in 1975 was used for nonfuel purposes (5.9 percent), used or lost by the energy sector (27.2 percent), or lost in end-use applications (29.3 percent). - 28 These estimates for the various categories of energy use or loss are distributed among the three major end-use consuming sectors so that one can also see whether these changes were uniform among the consuming sectors or whether they exhibited divergent trends. Examination of the distribution of gross, net, and effective use of energy among the three major sectors for the 1947-1975 period (table 12) shows that for all three measures of energy use the household and commercial sector increased its share of the total, the industrial share declined substantially, and the transportation propor- tion varied within a narrow range. The relative importance of each sector is substantially different, depending on the particular measure and years selected for comparison. The sector distribution of gross (excluding miscellaneous and unaccounted), net, and effective use of energy for 1947 and 1975 is given in table 13. The proportions vary considerably over time and among the three measures of consumption. Probably the most important difference is the sharp reduction in the transportation share when based on effective use as compared to its proportion of gross consumption; and even more so when the comparison is with the net measure. For example, in 1975, transportation accounts for 27.5 percent of total gross consumption. Be- cause most of nonfuel use and energy used or lost by the energy industries occurs in or is allocated to the other two sectors, deducting these categories from sector gross consumption results in lowering the proportion of the house- hold and industrial sectors in net consumption and increasing transportation's share, to 37.1 percent. However, when losses in end-use are deducted from net to derive the effective consumption estimates, by far the largest propor- tion of end-use loss occurs in the transportation sector. In 1975, loss in the transportation sector because of low thermal efficiency accounted for 63 percent of total loss in end-use. As a result, transportation's share of effective use of energy was only 16.6 percent, as compared to the original 27.5 percent of gross consumption. How do these changes among categories of energy use affect the movement in the gross BTU/GDP ratios, at the aggregate level and among consuming sectors? One way of measuring the effect is to measure the change in energy consumption based on successively narrower coverage, starting with gross consumption and ending with effective consumption. The difference between the various measures would then represent the extent to which each of the categories excluded in turn contributed to the change in the overall gross BTU/GDP ratio. A complete accounting for the total change in the ratio would require one additional factor, namely, the change in effective use of energy per dollar of GDP, the effective use/GDP ratio. The relationship of the various measures to each other and to the gross BTU/GDP ratio follows: - 29 - Gross BTU GDP (1) Gross BTU X Gross BTU (excl. misc.) (4) (2) Gross BTU (excl. misc.) Gross fuel and power use (5) X (3) Gross fuel and power use Net use X Net use X Effective use Effective use GDP In index form, the various terms on the right hand side of the equation measure the extent to which each category of energy use, excluded in turn, contributes to the change in the gross BTU/GDP ratio. The categories covered are those discussed previously, that is: (1) miscellaneous and un- accounted use of energy; (2) nonfuel use,' (3) energy used or lost by energy industries; (4) energy lost in end-use application,* and finally (5) the effective use/GDP ratio. An increase in each ratio in the equation contri- butes to an increase in the gross BTU/GDP ratio and may be interpreted as follows for each ratio: (1) and (2) a decline in energy available for fuel and power use; (3) a decline in efficiency by the energy sector as a whole; (4) a decline in efficiency in end- use application; and (5) a decline in the productivity with which the economy uses effective energy to produce real gross domestic output. Conversely, a decrease in the various ratios indicates an increase in efficiency, the effect of which is to reduce the quantity of energy consumed per dollar of GDP. The estimates given in table 11 can be used to derive indexes and average annual rate of change for each term in the equation. The annual rates, except in the case of large numbers when the interaction effect may be signifi- cant, are additive and show how much each category of energy use or loss as well as the effective use/GDP ratio has contributed to the change in the gross BTU/GDP ratios. The results are shown in table 14. In this table, the annual rates for gross and effective use/GDP ratios for the major end-use consuming sectors are based on real domestic output for the economy. In the next section of the report, we will analyze how much the change in energy/GDP ratios for selected components reflect changes in energy intensity of the components and how much it reflects changes in activity levels of components relative to GDP. The estimates of annual rates of change in table 14 are arranged by time periods, with the estimates for the total 1947-1975 period first, followed by the three major subperiods (1947-1954, 1954-1966, and 1966-1975), and last, for periods based on turning points in the movement of the overall gross BTU/GDP ratio. 30 - Separate estimates are given for changes at the aggregate level as well as for each major end-use consuming sector. In addition, since changes in energy use or loss by the energy industries may be offset in part by gains in efficiency in end-use, the net effect of changes in the two categories of thermal efficiency is shown in the addendum column in the table. The estimates in table 14 will be used to see whether they shed any additional light on (a) the factors underlying the slowdown in the decline in the gross BTU/GDP ratio since 1947 and (b) the reversals in the decline in the ratio during 1954-1956 and 1966-1970. We will be particularly interested in those changes which may have longer-run implications. On the first point, namely, the longer-run deceleration in the reduction in the ratio, there are the two minor items of energy use which can be dis- cussed so that we can go on to the more important categories. As was noted earlier, miscellaneous and unaccounted use of energy is a small item which has been declining relatively over the period so that its movement has been in the direction of accelerating rather than dampening the rate of decline in the gross BTU/GDP ratio. Nonfuel use is also a small category and be- cause it has been increasing as a proportion of total energy consumption, has contributed to an increase rather than a reduction in the BTU/GDP ratio. However, the increase has been small, .1 percent, for the 1947-1975 period as a whole and for each major subperiod. Nonfuel use of energy is largely con- centrated in the industrial sector and for this sector it plays a more important role, particularly during the 1966-1970 reversal, and we shall come back to a discussion of this period later in the section. The analysis of the effect of changes in energy use or loss by the energy industries on the BTU/GDP ratio requires going somewhat beyond the figures given in table 14. The category "use or loss of energy by the energy industries" covers a number of such uses or losses. The largest single item in the category is electric utility conversion loss. The rest includes electric utility distribution loss, fuel used in petroleum refining, and various uses and losses of energy by the natural gas industry. If we combine conversion and distribution loss of the electric utility industry into one group, and all other identified and measured uses and losses of energy by energy industries into the second group, the contribution of losses and uses of energy by each group to the change in the BTU/GDP ratio is as follows: 1947-1975 1947-1954 1954-1966 1966-1975 Total Electric utilities .4 All other .5 .1 .9 .3 .6 .1 .3 -.1 .7 .7 .0 - 31 - Electric utility losses increased throughout the period, but during the earlier years the increase in these losses was held down to some extent by the gains in efficiency in electric power generation, i.e., the heat rate (BTU/KWH) declined. The heat rate stopped declining after the early 1960's and the losses in electric power generation and distribution in- creased in the most recent period relative to the earlier subperiods. As a result, this subcategory contributed about .3 percent in the rate of in- crease in the BTU/GDP ratio during the first two subperiods but more than double this, .7 percent, since 1966. The remaining uses and losses of energy by the energy sector increased rapidly during the 1947-1954 subperiod, declined relative to the growth in total gross fuel and power energy consumption in the middle subperiod and maintained about the same proportion in the period since 1966. As a result, the contribution of total energy use or loss by the energy sector was not uniform throughout the period, contributing to large increases in the overall BTU/GDP ratio during the first and third subperiods, .9 percent and .7 per- cent, respectively, but only .1 percent in the middle subperiod. The implications of changes in this category for longer-run trends in the BTU/GDP ratio depend on the changes for the two subgroups—electric utilities and all other uses and losses of energy by the energy sector. The losses in energy incurred by the electric utility industry represented about 66 percent of total energy used or lost by the energy sector in 1966. By 1975, this proportion had increased to 73 percent. If electric power continues to increase as a proportion of total energy consumed for fuel and power and there is no significant improvement in the heat rate with about two- thirds of primary energy lost in electric power generation, then this category may continue to contribute to significant increases rather than reduc- tions in the BTU/GDP ratio. Relative increases in energy used or lost by the energy sector, as noted earlier, have been offset by gains in efficiency in the conversion of energy by end-users to effective use. The increases in the gains, however, have been decelerating throughout the 1947-1975 period. In 1947-1954, these gains contributed 1.5 percent per year to a reduction in the BTU/GDP ratio. This dropped to .7 percent in 1954-1966. For the subperiod since 1966 the situation was reversed; there was a loss in overall efficiency in the con- version of energy to effective use by end-users. Consequently, this factor contributed .1 percent per year to the increase in the BTU/GDP ratio during the most recent subperiod. In the earlier subperiod, increases in end-use efficiency largely re- flected the displacement of coal by other forms of energy having relatively higher thermal efficiency factors (table 10). This process became less and less important over time, and in the case of the dieselization of railroads was essentially completed by the early 1960's. Toward the latter part of the post World War II period, the proportion of total net energy consumption accounted for by electric power increased. Because end-use of electricity is considered to be 100 percent efficient, this shift played a major role in continuing the gains in end-use thermal efficiency. The two major consumers - 32 of electric power, household and commercial, and industrial sectors showed continued gains in thermal efficiency, with the most recent sub- period contributing about as much to a decline in the BTU/GDP ratio as in the middle subperiod. However, the transportation sector uses very little electric power, and after the potential gains in efficiency from diesel ization of railroads had been realized, this sector showed no further gains. This alone would have had only a moderate effect in dampening the gains in thermal efficiency in end-use. One other factor which contributed substantially to the actual loss in overall thermal efficiency in end-use between 1966 and 1975--the growing proportion of total net energy consumed by the transportation sector. In 1966, its share was about 32 percent; by 1975, it was up to 37 percent. The average thermal efficiency in transportation, at 25 percent, is so much lower than the average efficiencies in household and commerical use, 73 percent, and in industrial use, 76 percent. Consequently, the increased weight of transportation in total net energy use resulted in a small decline in the overall thermal efficiency in end-use in 1975 relative to 1966. Part of the decline was related to the recession of 1974-1975 with the relative reduction in industrial activity and the corresponding increase in the transportation proportion. If the comparison is made between 1966 and 1973, we find no change in the overall end-use thermal efficiency ratio but, nonetheless, it does indicate a cessation of the prior gains in thermal efficiency. If we net out the changes in thermal efficiencies in the energy sector and in end-use and implicitly derive the change in the ratio of gross fuel and power to effective energy use, we find that there was a net gain in the combined efficiencies in the first two major subperiods which reduced the gross BTU/GDP ratio by .6 percent per year for the earlier subperiods. This situation was reversed in 1966-1975, and the net losses in efficiency since 1966 has contributed .8 percent per year to an in- crease in the overall gross BTU/GDP ratio. - 33 - Another way of analyzing the combined effect of the changes in thermal efficiency in the energy sector and in end-use on the BTU/GDP ratio is to decompose the net change into changes within components and shifts in the relative importance of components. For this purpose, the basic data have been regrouped so as to measure separately the effect of changes in the two categories that have already been noted as having \/ery high losses, either in the energy sector stage or in end-use. These are electric power losses and transportation end-use losses. Deducting these two categories from total gross fuel and power and effective use of energy yields a residual including all other uses which basically represents an aggregate of household and commerical and industrial use excluding electric power use. The estimates of gross fuel and power use, effective use, and the ratio of the two (which reflects the net effect of the thermal efficiency changes in the energy sector and in end-use) are shown in table 15. The table points up the fact that the effective/gross fuel and power ratios for electric power and transportation are both much lower than the average for "all other" uses and that an increase in the relative importance of either electric power or transportation would lower the net thermal efficiencies of energy consumption and increase the BTU/GDP ratio. At the same time, the table also shows substantial increases in the combined thermal efficiency ratios for all three component categories. In order to decompose the change in the overall effective use/ gross fuel and power ratio into component ratio changes and shift effects, the component ratio changes have first been combined with sub- period base weights (percent distribution of gross fuel and power consumption) and then subperiod ratios have been combined with changing weights. The results of this decomposition, admittedly at a broad level of aggregation, follow: - 34 - Decomposition of Changes in Effective Use/Gross Fuel and Power Ratios Total Change in Change in Subperiod component ratios component shares 1947-1954 .6 .9 -.3 1954-1966 .6 .7 -.2 1966-1975 ' -.8 .0 -.8 For the first two subperiods, the dominant factor was the increase in net thermal efficiencies of the component categories, which was offset to a limited extent by the decline in thermal efficiencies resulting from the shift towards the categories with much lower levels of thermal efficiencies. In the most recent subperiod, however, the gains in thermal efficiency among component sectors ceased and the fall in overall efficiency due to the growing importance of electric power and transportation categories accelerated, The result of these changes was that overall thermal efficiency increased by .6 percent in the first two subperiods and fell by .8 percent in the period since 1966. (Note that increased efficiencies contribute to a decline in the gross BTU/GDP ratio and pluses in the above table should be interpreted as resulting in equivalent reductions in the gross BTU/GDP ratio, which is the way they are shown in table 14.) Although part of the shift effect may have been due to the recession induced decline in the relative importance of the "all other" category which includes energy consumed by the industrial sector, excluding electric power; nevertheless, the figures suggest a trend in overall thermal efficiency which may have longer-run implications. This is the growing share of total gross fuel and power use accounted for by electric power and transportation, both of which have relatively low net thermal efficiency ratios, and little or no increase in efficiency ratios within the categories. The effect of these trends can be offset by gains in the efficiency with which the economy uses "effective" energy to produce the real output of the economy. However, if the overall gross BTU/GDP ratio is to remain stable or decline, the latter type of gains, which may be broadly defined as "economic efficiency" as distinguished from "thermal efficiency," will have to accelerate to offset possible continued reductions in thermal efficiency. The record on economic efficiency in energy use is uneven. At the aggregate level, there was a sharp drop in the effective BTU/GDP ratio during the first subperiod of -2.1 percent per year but this was followed by an in- crease of .4 percent per year in the 1954-1966 subperiod (table 14). For the most recent subperiod, the ratio has declined, at the rate of -.3 percent per year. Among the component sectors, the industrial sector has accounted for most of the decline in the effective BTU/GDP ratio, with large reductions in - 35 - the component ratio in the first and third subperiods, but a slight increase in the middle subperiod. In contrast, the household and commercial sector has shown increases in the first two subperiods, particularly during 1954- 1966, and no change in the most recent subperiod. The transportation sector, after little or no change in the first subperiod, showed a decline in the second but this was followed by an average rate of increase of 1.4 percent in the period since 1966. The most encouraging aspect of these changes in effective BTU/GDP ratios among component sectors is their movement since 1970. Both the household and commercial and industrial sectors showed sharp reduc- tions in their component effective BTU/GDP ratios between 1970 and 1975, and this decline held for the 1970-1973 years prior to the 1974-1975 recession and large energy price increases as well as for the latter period. The trans- portation sector, on the other hand, showed no change between 1970-1973 and an increase since 1973. Because the two sectors that showed reductions in the ratio accounted for a much larger share of total effective energy use than the transportation sector, the overall effective BTU/GDP ratio declined at the rate of -1.7 percent during the 1970-1975 subperiod. In the next section we will examine the changes in component BTU/GDP ratios in more detail to see if there are trends among these components which, like those in thermal efficiency factors discussed in this section, may have longer-run implications for changes in the overall BTU/GDP ratio. Before doing so, however, we will want to analyze the changes during the two reversal periods, 1954-1956 and 1966-1970, to see if the factors explicitly identified and measured in the section, i.e., nonfuel use, thermal efficiencies in energy sector, and end-uses, help explain the short-term reversals in the downward trend in the gross BTU/GDP ratio. The figures given in table 14, page 2, indicate a substantially different pattern among the sources of change in the BTU/GDP ratio between the two reversal periods. The differences in the contributing factors are summarized in the following table. Annual rates of change 1954-1956 1966-1970 Total gross BTU/GDP ratio 2.7 2.1 Sources of change: Miscellaneous and unaccounted Nonfuel Energy sector efficiency End-use thermal efficiency Effective use/GDP Addendum: Net thermal efficiency -1.4 .1 -.2 .1 .2 -.2 .7 1.2 -.1 3.9 1.4 - 36 The major difference in the behavior of the enumerated factors between the two reversal periods is that in the first reversal, net thermal efficiency change did not contribute at all to the increase in the gross BTU/GDP ratio; in fact, it tended to reduce rather than increase the ratio. The opposite was true in the 1966-1970 reversal when it accounted for .6 per- cent of the 2.1 percent annual rate of increase in the ratio or almost 30 percent of the total change. In both reversal periods the effective BTU/GDP ratio increased, although at a substantially lower rate in the second reversal than in the first. The net effect of the changes in the miscellaneous and nonfuel categories was minor. Examination of the detail for the three major end-use consuming sectors indicates that the pattern of differences at the overall level between the two reversal periods also holds for the component sectors except for one item, the \/ery large increase in nonfuel use in the industrial sector during the 1966-1970 reversal. This increase accounted for .8 percent, or two-thirds of the 1.2 percent annual rate of increase in the BTU/GDP ratio for the sector, with most of the remainder of the increase attributable to the reductions in net thermal efficiency during the period. The large relative increase in nonfuel use by the industrial sector seems to have been largely confined to the 1966-1970 period, because this became a minor factor in the period after 1970. What is the explanation for the dramatic reversal in direction of changes in net thermal efficiency on the BTU/GDP ratios between the two reversal periods? Examination of the estimates for the two categories of thermal efficiency suggests that, in the case of end-use thermal efficiency, the difference between the two reversals is part of a long-term slowdown in the gains in efficiency since the early part of the post World War II period. This slowdown reflected the decline in the share of coal in total energy use, which was not significant in the latter part of the period. The fall off in the end-use thermal efficiency gains is evident in the figures, taken from table 14, for this category for the shorter subperiods. Contribution of end-use thermal efficiency to change in BTU/GDP ratio 1947-1954 -1.5 1954-1956 -1.2 1956-1966 -.6 1966-1970 -.1 1970-1973 .0 1973-1975 .5 The large difference between the changes in this factor for the two reversal periods is part of a trend and not primarily unique to the reversal periods themselves. On the other hand, this explanation does not seem to hold for the changes in energy sector thermal efficiency between the two reversal periods. Here the general trend seems to have been towards relative increases in uses or losses of energy by the energy sector throughout the period with 1954-1956 being the exception when it showed a relative decline. - 37 - i.e., increased efficiency. As noted earlier, the movement in two categories of energy used or lost by the energy sector has been different. The effect of changes in thermal efficiency of the two categories (electric utility conversion losses and all other energy sector use or loss of energy) on the change in the gross BTU/GDP ratio for the various shorter subperiods is shown below: Contribution of energy sector thermal efficiency to changes in gross BTU/GDP ratio Total Electric utility All other conversion loss .3 uses and losses 1947-1954 .9 .6 1954-1956 -.2 .1 -.3 1956-1966 .2 .3 -.1 1966-1970 .7 .6 ,1 1970-1973 .7 .7 .0 1973-1975 .7 .6 .1 The figures indicate that the main reason for the decline in the gross BTU/GDP ratio attributable to the change in thermal efficiency of the energy sector during the first reversal period was the reduction in the "all other energy uses or losses" category, not in the electric utility conversion loss which showed a small relative increase. During the later periods, however, electric utility conversion losses increased substantially relative to total energy used for fuel and power, whereas the "all other use or loss" category showed little relative change. As a result of these differences in the movement of the two components of thermal efficiency in the energy sector, the category as a whole contributed -.2 percent per year in the first reversal to the change in the overall gross BTU/GDP ratio, but a large .7 percent increase in the second reversal. - 38 - Section 5. Changes in Energy Intensities vs. Mix Effects As noted in Section 2 of the report, the change in the overall BTU/GDP ratio can be due to changes in energy intensities among individual categories of energy use or shifts in the mix of goods and services, each with its own level of energy intensity. In this section we shall examine the extent to which changes in the BTU/GDP ratios for two of the major con- suming sectors of the economy, transportation and industrial, are due to either changes in sector energy intensities or mix effects. Finally, to put this type of analysis in broader perspective, we shall review the input- output studies which partition the change in energy consumption between coefficient change and demand mix in order to determine whether these studies shed any additional light on the trend in the BTU/GDP ratio. Transportation In 1973, the transportation sector of the economy accounted for slightly more than a quarter of total gross consumption of energy, and some- what less than a quarter of the total when calculated on a net basis, that is, after deducting energy sector use or loss of energy allocated to the transportation sector. Deducting losses in the transportation sector itself in order to arrive at "effective" uses of energy reduces the proportion sub- stantially, to only about 6 percent of total gross energy consumption in the economy. Of the total gross energy consumed by the transportation sector, which amounted to 18.96 quadrillion BTU's in 1973, about 70 percent was used for passenger and intercity freight transportation for which estimates of energy consumption and transportation activity have been developed covering selected years over the 1950-1973 period. The analysis which follows is limited to the freight and passenger transportation segments of the transportation sector, and attempts to account for the relative contributions of changes in energy intensity of individual transportation modes vs. shifts among modes in explaining transportation BTU/output changes. In addition, the changing importance of freight and passenger transportation activity relative to GDP will be examined to see how this affects the transportation BTU/GDP ratio. Because many of the estimates used in the analysis are taken from the study by Eric Hirst, "Energy Intensiveness of Passenger and Freight Transport Modes, 1950-1970," Oak Ridge National Laboratory, April 1973, which covers only every fifth year starting in 1950, the years included will not coincide with those used in the earlier sections of this report. For the purpose of this report, a number of the Hirst estimates, particularly those relating to automobile passenger transportation have been modified and all the estimates have been extended to 1973. - 39 - Summary statistics for the transportation sector are given in table 16. Although the two major categories of transportation, intercity freight and passenger transportation, included in this analysis cover about 70 percent of total energy consumption by transportation during most of the years, their trends are quite different. The proportion of total transportation energy consumption accounted for by intercity freight declines continuously, from about 31 percent in 1950 to about 12 percent by 1973. Energy used for passenger transportation represents a much larger and increasing proportion of the total throughout the periods, going from 41 percent of the total in 1950 to almost 58 percent by 1973. The all other transportation category, which accounted for about 30 percent of transportation energy consumption during most of the period, includes urban and nonfreight truck transportation, military transportation, general aviation, international freight and passenger transportation, nonfuel uses of energy, use of off-highway con- struction and agricultural vehicles, pleasure boating, general aviation, etc. (The estimates for intercity freight, as used in this report, excludes air cargo transportation, because of the difficulties in the measurement of energy intensity for this mode. If included, it would result in dampening the decline in the ratio for intercity, particularly during the latter part of the period. ) Energy consumed for a particular category of transportation can change relative to the growth in real output of the domestic economy (GDP) because of changes in either energy/output (energy intensity) ratios or transporta- tion category output/GDP ratios. In index terms, the relationship among the ratios is as follows: Transportation category BTU = transportation category BTU GDP transportation category output transportation category output GDP When converted to annual rates, the changes are generally additive except for a small interaction term. The lower panel of table 16 shows the average annual rates of change in the BTU/GDP ratios for the two major transportation categories being analyzed. They show generally divergent patterns, with intercity freight BTU/GDP ratios declining through most of the period, due to declines in both energy intensities and ton-mile/GDP ratios for most subperiods. Conversely, energy intensities and passenger mile/GDP ratios have increased for most of the subperiod leading to an increase in the passenger BTU/GDP ratio. The form of transportation, passenger traffic, which has been growing in importance relative to GDP has also been becoming more energy intensive, whereas intercity freight which has been declining in importance relative to GDP has become less energy intensive. During the early part of the post - 40 - World War II period, the decline in the BTU/GDP ratio for intercity freight was sufficiently large as to more than offset the increase in the passenger BTU/GDP ratio, so that the BTU/GDP ratio for the two categories of transpor- tation combined was reduced. For most of the period after 1955, however, the opposite situation prevailed and the growing importance of passenger transportation and an increase in its energy intensiveness resulted in either an increase or no change in the BTU/GDP ratio for the combined categories, except for 1960-1965. The increase was especially large during the 1965-1970 subperiod when all component factors, both for intercity freight and passenger transportation, showed increases. During this sub- period, energy intensities of intercity freight and passenger transportation had the largest rates of increase experienced throughout the entire period. Similarly, the ton-mile/GDP ratio showed the only subperiod increase of the whole 1950-1973 period, and the passenger mile/GDP ratio showed the highest rate of increase for any subperiod. The coincidence of these negative factors helps to explain the wery high rate of increase in the BTU/GDP ratio for transportation during this subperiod, and its contribution to the reversal in the decline in the overall BTU/GDP ratio. The trends in the energy intensity ratios for the two major categories of transportation have moved in opposite directions, with the BTU/ton-mile ratios declining in almost all subperiods and the BTU/passenger mile ratio increasing throughout the 1950-1973 period. The changes in energy intensity of intercity freight and passenger transportation could be due to shifts among the various transportation modes, each with its own energy intensity level, or changes in energy intensity of the individual modes of transportation, or both. To determine the relative roles of the various factors requires more detailed analysis of each major category of transportation (tables 17 and 18). Table 17 provides data on the change in relative importance of the various intercity freight modes of transportation and also the change in energy intensity ratios for each mode. Table 18 shows similar information for passenger transportation. In order to partition changes in the average intercity freight and passenger energy intensity ratios into (a) the energy intensity change of individual transportation modes and (b) the effect of shifts in the relative importance of the various modes, estimates have been developed of the two factors in which the relevant weights have been held constant as of the first year in each succeeding subperiod. The sum of the two separate effects (energy intensity and shift effect) may not add to the average change because of the interaction effect. In those instances where there is an interaction effect, it has been distributed evenly between the two factors and the results are shown in the bottom panels in tables 17 and 18. The partitioning of the change in the BTU/intercity freight ton-mile ratio (table 17) indicates that, except for the first subperiod, the energy intensity and shift effect generally moved in opposite directions. The change in energy intensities among the various modes (trucks, railroads, waterway, oil pipeline), excluding effect of shifts, resulted in a decline - 41 - in the average BTU/TM ratio, except for the 1965-1970 period. In contrast, the shift among modes was generally in the direction of increasing the energy intensiveness of intercity freight transportation, except for the very first subperiod and to a lesser extent, 1965-1970. The upward drift in the shift effect was largely due to the increasing importance of inter- city trucking, which has the highest energy intensity ratio of all intercity freight modes, from 1955 on. As between the two effects, the change in average BTU/TM, excluding the effect of shifts, was the more important factor, particularly in the early part of the period when the dieselization of the railroads played such an important role in drastically reducing the BTU/TM ratio for that mode of transportation. The net effect of the two factors was a decline in the BTU/TM ratio for every subperiod except 1965-1970. In contrast to the pattern for intercity freight transportation, the separate energy intensity and shift effects for passenger transportation reinforced each other and were in the same direction, namely, an increase in the BTU/passenger mile ratios throughout the period. Here the growing importance of automobile and airline transportation, both of which have sub- stantially higher energy intensity ratios than the other passenger trans- portation modes (bus, school bus, electric mass transit, railroads), has increased the energy intensity of passenger transportation. The largest part of the increase, however, has been due to the continuing increase in energy intensity ratios for individual modes, particularly automobile and airline transportation. The increase in the BTU/passenger mile ratio for automobiles has been due to a number of factors, including the increasing proportion of automobile vehicle miles used for urban passenger transporta- tion relative to intercity automobile transportation. This has resulted in an increase in the average BTU/automobile passenger mile ratio for two reasons: (1) the energy efficiency of urban automobile transportation is considerably less than for intercity travel (Hirst assumes it to be about 40 percent less); and (2) the passenger occupancy ratio for urban automobile travel is also considerably less than intercity travel. (The estimates for 1972 are 1.9 passengers per vehicle for urban automobile travel vs. a 2.6 occupancy rate for intercity travel.) As a result of these and other factors, the average BTU/passenger mile ratio for automobiles, taxis, and motorcycles increased about .6 percent per year between 1950 and 1973. The increase in the BTU/PM ratios for airline travel was even larger, 2.2 percent per year over the same period, but because it still represented only a small part of total passenger transportation (only 5.3 percent of the total in 1973 as compared to the 90.7 percent for automobiles), the large increase in energy intensity did not have as much an effect on the overall increase in the BTU/PM ratio as automobiles. The other modes of public passenger transportation (railroad, bus, electric mass transit) all showed declines in traffic relative to total passenger transportation throughout the period. In 1950, they accounted for about 13 percent of total passenger transportation; by 1973, their combined share had fallen to only 2.6 percent of the total. During most of the period, from 1955 on, their individual energy intensity ratios were lower than that of automobiles or airlines, and the combination of lower BTU/PM ratios and the declining share of passenger transportation contributed to the increase in the average BTU/PM ratio for almost the entire period. - 42 School bus transportation represented a small part of total passenger transportation, ranging from 1.5 percent to 1.8 percent, but because of an increase in its BTU/PM ratio from 1950 to 1960 and some slight decline after that, also contributed to an increase in the average BTU/PM ratio between 1950 and 1973, but only to a modest extent. Industrial secto r As noted earlier in this section, the industrial sector in 1973 accounted for about 36 percent of total gross energy consumption in 1973 (including allocation of the energy sector's use or loss of energy) which is reduced to about 21 percent on a net basis and 16 percent in terms of effective use. In contrast to the other two major energy end-use consuming sectors, its share of energy consumption, whether measured as gross, or as effective use of energy, has declined during most of the post World War II period. In terms of the effective energy/GDP ratio, the industrial sector showed more of a decline in most subperiods than either of the other two sectors, and showed only a slight increase in the 1966-1970 reversal as compared to large increases in the ratio for the other two sectors during this subperiod. On the other hand, it showed a much larger rate of increase in the ratio during the 1954-1956 reversal than either the household or trans- portation sectors. As in the case of the transportation sector, we shall examine this sector in more detail in order to determine whether the changes in the energy/GDP for the sector are initially due to changes in the BTU/output ratio for the sector or the change in the importance of the sector relative to real domestic product for the economy. We shall then analyze the available data to gain some insight as to whether the change in energy intensity for the sector as a whole is due largely to changes in energy intensity of component industry groups or shifts among industry groups. A key element in the analysis is the measurement of output for the industrial sector and component industry groups in order to derive the appropriate BTU/output ratios for the sector as a whole and components. Here we encounter problems of comparability between energy input and sector output because, as noted earlier, energy consumption by the industrial sector includes manufacturing and part of other sectors such as construction, mining, and agriculture which are not included in the transportation and space heating portion of household and commercial energy consumption. In order to avoid the problem of trying to develop an output measure that is comparable to "industrial" energy consumption, much of the analysis is limited to the manufacturing sector and its component industry groups. Energy consumption in the manufacturing sector, defined on a "net" basis that is roughly comparable to the net measure developed in the previous section of the report for the three major energy end-use sectors, accounts for almost 90 percent of net energy consumption by the industrial sector. (It will be recalled that the net measure excludes both nonfuel use of energy, and energy used or lost by the energy industries.) The estimate of 90 percent coverage is based on the measure of energy purchased for fuel and power, developed by the Census Bureau for selected years between 1954 and - 43 - 1974. The Census Bureau's estimates, which are stated in terms of kilowatt hour equivalent of purchased fuel and power, have been converted to BTU's based on 3412 BTU's per kilowatt hour. They have also been modified to ex- clude energy consumed by petroleum refining and include captive energy con- sumption by the steel industry, thus making the Census Bureau's manufacturing energy consumption estimate comparable to the definition of net consumption by the industrial sector (table 11). Estimates of net energy consumption, output (gross product originating, in billions of 1972 dollars), and BTU/output ratios for the total manufac- turing sector, excluding petroleum refining, are given in table 19 for each of the years covered by the Census Bureau's estimates of energy consumption: 1954, 1958, 1962, 1967, 1971, and 1974. The annual rates of change in the energy intensity ratio for manufacturing are also calculated for the total period and subperiods and compared to similar rates of change for the energy intensity ratio, based on gross energy consumption, for the total economy. The gross consumption measure is used in calculating the energy intensity ratio for the total economy because estimates of net energy con- sumption have been developed only for the years selected as turning points in the trend of the BTU/GDP ratio, and these do not match the years covered by the Census Bureau's energy estimates, except for 1954. However, in order to provide some comparison of changes in energy/output ratios between manu- facturing and that for the economy as a whole based on more comparable measures of energy consumption, annual rates of change have also been com- puted for the total period and longer subperiods which approximate the same time periods and are based on the net measure. These are shown at the bottom of table 19. Rates of change in the manufacturing (excluding petroleum refining) net BTU/GDP ratio have also been computed for the total period and subperiods and these have been partitioned into the change in the energy intensity ratio for manufacturing and the change in the importance of manufacturing gross product (excluding petroleum refining) relative to total GDP, both in con- stant 1972 dollars. The comparison of the change in the manufacturing energy/output with that for the economy as a whole, the latter based on gross consumption, indicates that in every subperiod when the latter ratios declined, the manu- facturing energy intensity ratio declined more than the overall ratio. In the two subperiods when the ratios increased, the manufacturing ratio showed a higher rate of increase in 1954-1958 but did not show as much of an increase in 1966-1971 as the overall ratio. For the 1954-1974 period as a whole, whereas the overall ratio showed a slight increase of .1 percent per year, the manufacturing ratio declined at an annual rate of -.8 percent. If we examine the comparison based on net energy consumption for both ratios, we find about the same results--a substantially greater rate of decline in the manufacturing BTU/output ratio than for the overall ratio. For the period as a whole, the manufacturing BTU/output ratio showed the same rate of decline as previously indicated, -.8 percent; but the overall ratio (1954-1973) has a small decline, -.2 percent, rather than the increase - 44 of .1 percent based on gross consumption (1954-1974). Also, both the manu- facturing and total ratios show the same fall-off or cessation in the rate of decline previously discussed; in the case of manufacturing, from -.9 per- cent to -.6 percent. For the total ratio, there is a cessation of the rate of decl ine--from -.3 percent in the earlier subperiod to .0 percent in the period after 1966. Turning to the partitioning of the change in manufacturing net energy consumption relative to growth in the domestic economy as a whole, the ratio shows an interesting pattern in which the change in the two components, i.e., manufacturing energy intensity and manufacturing share of total domestic output, are in opposite directions. In every subperiod that the energy intensity ratio shows a decline, the manufacturing share increases and the same reverse relationship holds during periods when the manufacturing intensity ratio increases. The explanation seems to be that during periods of relatively high growth, the manufacturing share of total output increases and the energy intensity ratio declines. Conversely, when growth slows down, the manufacturing share slows down more than average and the BTU/output ratio goes up. This is quite consistent with our earlier discussion of the relationship between rates of growth and changes in BTU/output ratios. The net effect of the changes in opposite direction of the two components was to reduce the manufacturing BTU/GDP ratio in every subperiod. For the period as a whole, 1954-1974, the manufacturing share of total output did not change significantly so that the entire decline in the manufacturing BTU/total GDP ratio was due to the decline in the energy intensity ratio of the manufacturing sector. As in the case of the analysis of the energy intensity ratio for trans- portation, we shall want to partition the change in the energy/output ratio for the manufacturing sector into the average change in energy intensity of the industry components of manufacturing, with fixed weights, and the shifts in relative importance among industry groups. Table 20 provides estimates of net energy consumption, output (gross product originating, in 1972 dollars, developed by the Bureau of Economic Analysis, U. S. Department of Commerce) and BTU/output ratios for the five industry groups that are particularly energy intensive and collectively account for about 80 percent of total net energy consumed by the manufacturing sector, excluding petroleum refining. The five energy intensive industry groups are food and kindred products, paper and allied products, chemicals and allied products, stone, clay and glass products, and primary metals. These energy intensive industries account for a much smaller share of real gross product of the manufacturing sector, about 33 to 35 percent of the total, again excluding petroleum refining. This disparity in the share of energy consumption and real output, accounted for by the five energy intensive sectors, reflects the much higher energy intensity ratios for these industries than for the remaining industry groups. Estimates of energy consumption, output, and BTU/output ratios for the remaining manu- facturing industries combined are also shown in table 20. The BTU/output ratios of the energy intensive industries are about 8 times higher than that of all the other manufacturing industries combined. In 1974, for example, the average BTU/output ratio for the energy intensive industry groups was 113.0 thousand BTU's per dollar of real gross product originating (1972 dollars) 45 - whereas the average for the remaining industries was only 13.7. The average for total manufacturing, excluding petroleum refining, was 46.7. Average annual rates of change in the BTU/output ratios for the individual energy intensive industry groups, all other manufacturing industries, and total manufacturing have been calculated and are shown in table 21. These estimates indicate that for the 1954-1974 period as a whole, the averages for all energy intensive industries combined and for all other manufacturing industries declined, with the latter showing more of a reduc- tion, -.7 percent per year vs. -.5 percent. The fact that the decline for total manufacturing was -.8 percent indicates that there was a significant shift effect, with the less energy intensive industries increasing in relative importance. In general, the industry groups with the largest reductions in energy used per unit of output are chemicals (Industry 28) and food (Industry 20). Stone, clay and glass (Industry 32) was in the middle and the remaining industry groups--paper (Industry 26) and primary metals (Industry 33)-- showed relatively little change in their energy intensity ratios between the terminal years of the 1954-1974 period. Although there is no uniformity in the average rates of change from one group to the next or from one subperiod to the next, there is a general pattern which emerges which is consistent with our previous analysis; namely, that periods of relatively high rates of growth are more likely to have reductions in the BTU/output ratios and con- versely, periods of below average growth may show either dampening in the rate of decline or increases. This is reflected in the fact that 1954-1958 and 1967-1971, the subperiods with the lowest growth rates, are also the sub- periods with increases in the average BTU/output ratios for both the energy intensive industries as a whole and for the remaining manufacturing industries. We have noted earlier that part of the decline in the BTU/output ratio for total manufacturing, excluding petroleum refining, was due to the decline in the relative importance of the five energy intensive sectors and relative increase in the manufacturing industries with lower energy intensity ratios. This is seen in the greater rate of decline over the 1954-1974 period, in the BTU/$ ratio for total manufacturing (-.8 percent) than in the rate for the energy intensive industries (-.5 percent) or all other industries combined (-.7 percent). In order to partition the change in the BTU/output ratio for total manufacturing into (a) the energy intensity effect, excluding effect of shifts, and (b) the shift effect, estimates have been developed of these two factors for the two longer subperiods which come closest to matching the longer subperiods used in earlier sections of the report and for the comparison with the change in the net BTU/GDP ratios shown in table 19. The longer sub- periods are 1954-1967 and 1967-1974. The method used for partitioning the change in the BTU/output ratios for the sector as a whole is similar to the one used in the analysis of the transportation BTU/output ratios. Estimates are initially developed for changes in the energy intensity ratios of component industries (five energy intensive industry groups plus "all other" manufac- turing) with fixed output weights as of the first year of each subperiod. The shift effect is calculated by keeping the first year BTU/output ratios fixed and varying the relative importance of industry output based on the - 46 - terminal years of each subperiod. Any interaction effect (the residual difference between the sum of the two component effects) and the actual change is then distributed equally between the energy intensity and shift effects. The results are shown in table 22. They indicate that the main reason for the slowdown in the rate of decline in the manufacturing BTU/ output ratio between 1954-1967 and 1967-1974 is not in the change in energy intensity of the individual industry groups but is largely due to the shift in relative importance of the various industry groups. In the first sub- period, the shift effect reduced the BTU/output ratio by -.4 percent per year whereas the reverse took place in the more recent subperiod when shifts among industry groups had the effect of increasing the BTU/output ratio by .3 percent per year. The net change between the two subperiods due to shifts was .7 percent. Energy intensity, on the other hand, calculated with fixed output weights, declined more rapidly in the more recent period, -.9 percent vs. -.5 percent for the earlier period, which suggests that the individual manufacturing industry groups have, on the average, been able to continue to improve on their efforts to economize in the use of energy relative to output. The improvements were largely concentrated in four of the major energy intensive industry groups: paper, chemicals, stone, clay and glass, and primary metals. The food products industry continued to reduce its BTU/output ratio but at a slower rate than in the earlier period. The "all other" manufacturing group showed an increase in the BTU/ output ratio during the 1967-1974 period compared to a decline in the earlier subperiod. The gains in reducing the energy intensity ratios were especially marked during the shorter subperiod, 1971-1974, when all the industry groups, with one exception, achieved their best performance of the entire 1954-1974 period in reducing energy consumed per dollar of real product. The exception was the chemicals industry which had about the same 2,5 percent reduction in the BTU/output ratio in both 1967-1971 and 1971- 1974 (table 21). The figures on annual rates of change in the energy intensity ratios for the two longer subperiods are given below. Food and kindred products Paper and allied products Chemicals and allied products Stone, clay and glass Primary metals "All other" manufacturing 1954-1967 1967-1974 -1.93 -.96 +1.06 -2.12 -1.32 -2.54 -.49 -.76 + .33 -.21 -1.22 + .26 - 47 - Input-Output Analysis of Effect of Changes in Coefficients and Demand Mix on BTU/GDP Ratio A conventional procedure in input-output analysis of output changes in the economy is to partition the change in output of industries in the economy between changes in (a) final demand and (b) input-output coefficients. This technique can be used to partition the change in energy consumption into the same two factors, based on a series of input- output tables. To relate input-output analysis of energy consumption to the change in the overall BTU/GDP ratio requires that the partitioning exercise be based on input-output tables in which all transactions in- volving energy consumption be converted to BTU terms. Several studies along these lines have been developed but the most recent is that by William A. Reardon, Battelle Pacific Northwest Laboratories. 9/ In order to be able to compare the Reardon analysis with the BOM estimates of energy consumption used in this study, a comparison is made of the two sets of domestic energy consumption estimates and the BTU/GDP ratios based on the two sets of energy estimates. The comparison is shown in table 23. The levels of the two estimates of domestic energy consump- tion, although fairly close, are sufficiently different in that changes in the BTU/GDP ratios between the particular years for which 1-0 tables have been converted to BTU terms by Reardon show somewhat different rates of change between the two sets of estimates. The changes for the BTU/GDP ratios for the first subperiod, 1947-1958, are almost identical, but the ratio based on the Reardon energy estimate shows much more of a decline in the second subperiod, 1958-1963, than the ratios based on the BOM estimate and conversely is lower than the BOM based ratio in the last subperiod, 1963-1967. In describing Reardon 's results, therefore, the difference should be kept in mind in relating the 1-0 partitioning analysis to the change in the BTU/GDP ratio discussed earlier. Although the rate of change in the BTU/GDP ratio do vary somewhat between the two energy estimates, they both show the same pattern previously described, namely, a slowdown in the rate of decline in the BTU/GDP ratio after the sharp fall in the ratio during the first subperiod. 9. Reardon, William A., An Input-Output Analysis of Energy Use Change from 1947 to 1958, 1958 to 1963, and 1963 to 1967, Battelle Pacific Northwest Laboratories, Richland, Washington, 1976; prepared for Electric Power Research Institute, Palo Alto, California. - 48 - Reardon's study partitions the change in total domestic energy con- sumption between final demand change and input-output coefficient change for three subperiods: 1947-1958, 1958-1963, and 1963-1967. The estimates are based on 1-0 tables for 1947, 1958, 1963, and 1967, using an input- output 38 sector table with 35 intermediate industries and 3 final demand categories. All energy transactions in the 1-0 tables have been converted from 1958 constant dollars (the valuation basis used for all four 1-0 tables) to BTU's. The procedure used in partitioning the subperiod change in total domestic energy consumption is to calculate the change in domestic energy consumption that would have occurred given the actual change in 1-0 coefficients with the final demand of the first year of each subperiod held constant (the coefficient effect) and alternatively, the change in energy consumption given the actual changes in final demand over the sub- period with the 1-0 coefficients of the last year of the subperiod held constant (the final demand effect). The Reardon estimates of actual domestic energy consumption distributed between direct energy consumption by final demand categories (personal con- sumption expenditures and government) and intermediate use are summarized in table 24. The partitioning of the total change in domestic energy con- sumption between the effect of final demand changes and that of 1-0 coefficients is shown in the second panel of the table. The final demand effect on energy consumption combines two factors: (a) growth in aggregate final demand and (b) changes in the mix of final demand goods and services, each with its own energy intensity ratio. In order to determine how much of the total change in the BTU/GDP ratio is due to the final demand mix effect alone, the change (in index terms) in energy consumption due to the final demand change is divided by the GDP index, thus excluding the first factor, i.e., the effect of aggregate growth in final demand. The 1-0 coefficient effect is, by definition, calculated as the change in energy consumption per unit of output due to changes in all 1-0 coefficients and therefore already excludes the effect of aggregate growth in the economy. The partitioning of the change in the BTU/GDP ratio between the two factors-- final demand mix and 1-0 coefficient change--is shown in table 25, with the difference between the sum of the two factors and the actual change in the BTU/GDP ratio shown as a residual. The residual may include an interaction effect, which has not been distributed. The partitioning analysis indicates that the decline in the overall BTU/ GDP ratio for the three subperiods did not follow a uniform pattern in terms of the contributions of the two factors. In the first subperiod, 1947-1958, both factors contributed to the reduction in the overall BTU/GDP ratio, with the change in 1-0 coefficients providing the major part of the explanation for the reduction in the ratio. In the second subperiod, 1958-1963, the entire decline in the overall BTU/GDP ratio was due to the final demand mix effect, with little or no change in the ratio due to 1-0 coefficient changes. In the last subperiod, 1963-1967, the effect of final demand mix was to in- crease the BTU/GDP ratio but this was more than offset by the sharp reduc- tion in the overall ratio due to the effect of the 1-0 coefficient changes. From the viewpoint of any particular trend in these changes, the large 49 - increase in the BTU/GDP ratio due to the final demand mix effect during the last subperiod, following declines in the earlier subperiods, is potentially the most significant factor which emerges from the partitioning analysis. If on net balance, the change in final demand mix continues to be in the direction of increasing the relative importance of more energy intensive goods and services, as occurred during the 1963-1967 period, then it will require a continuation or acceleration in 1-0 coefficient changes which are energy saving in direction in order to offset the final demand mix effect on the BTU/GDP ratio. Unfortunately, we do not have direct information on the relative contribution of the two factors to the change in the overall BTU/GDP ratio for the more recent period, since the Reardon analysis carries the story only through 1967. Some indirect evidence is available, however, from several sources which do throw some light on this question for the period after 1967. For the 1967-1970 period, there are estimates in BEA Staff Paper No. 27, Summary Input-Output Tables of the U. S. Economy: 1968, 1969, 1970, 10/ of both the change in industry constant dollar output, as well as the derived change in output based on the actual change in final demand, with 1967 1-0 coefficients held constant. These estimates have been used to derive a rough partitioning of the change in industry output/GDP ratios between the final demand mix effect and coefficient change for the three energy sectors that are relevant to our analysis. The results are shown in table 26. The estimates in the table can only be used as rough indicators of the relative roles of final demand mix and 1-0 coefficient change in accounting for the change in the BTU/GDP ratio for the 1967-1970 period, when the ratio increased by about 2.6 percent per year. The reason the estimates can only be used as rough indicators is that the output of the energy industries is in constant dollar terms, not BTU's, and the BTU/dollar ratios are not the same across energy industries. In addition, from the viewpoint of analyzing the ultimate effect of final mix or coefficient changes on energy consumption, the energy industries in the table represent some over- lap since coal and petroleum products are used as inputs to one of the major components of the utility industries, namely electric utilities. Nevertheless, keeping in mind that the coal industry accounts for only about one-fifth of total energy consumption and therefore the results for this industry should be given less weight than the other two energy indus- 10. Paula C. Young and Philip M. Ritz, Summary Input-Output Tables of the U.S. Economy: 1968, 1969, 1970, BEA Staff Paper No. 27, U.S. Department of Commerce, September 1975. These estimates have since been extended to 1971 in BEA Staff Paper No. 28, March 1977. - 50 - tries, some tentative conclusions can be drawn. First, the figures suggest that both final demand mix and coefficient change contributed to the in- crease in the BTU/GDP ratio over the 1967-1970 period, and that, in this case, it was not due to the increase in one factor more than offsetting the decline in the other. All of the estimates of the contribution of each factor to the change in output of the energy industries are positive, with the exception of coal. In this instance, the coefficient change effect indicates a decline in the ratio but because of coal's modest proportion of total energy consumption, noted above, it is believed to be more than offset by increases in the coefficient change effect of the two other energy industries. Second, the estimates also suggest, that as between the two factors, the change in final demand mix was probably more important than 1-0 coefficient changes in accounting for the increase in the overall BTU/GDP ratio during the 1967-1970 period. This reading of the results is based on the fact that in two of the three energy industries, the final demand mix effect contributed substantially more to the increase in the ratio than the coefficient effect. In the third industry, petroleum refining, the effects of the two factors, are about even. For the period after 1970, any partitioning analysis is even more rough and tentative than for the 1967-1970 period. Estimates of direct energy consumption (in BTU terms) by households can be derived for household opera- tions (residential use) from Hirst J]/ and for automobile transportation based on U.S. Department of Transportation data. The change in direct energy use by households per dollar of GDP is a major component of the contribution of final demand mix to the change in the overall BTU/GDP ratio and may serve to indicate at least the direction of the effect of the component for the 1970- 1973 period. (In 1970, direct energy purchases by households in BTU terms accounted for almost 30 percent of total domestic energy consumption.) Un- fortunately, there is no similar short-cut indicator for the effect of the change in mix of the nonenergy components of personal consumption expendi- tures or the other components of final demand. As to the effect of changes in 1-0 coefficients for the post-1970 period, there are three indicators discussed in earlier sections of this study: the change in (a) BTU per dollar of real product in manufacturing (excluding petroleum refining); (b) BTU/ton mile of freight transportation; and (c) energy sector efficiency. All three indicators reflect a mixture of mix and coefficient change effects but the limited analysis of this in earlier sections indicates that the coefficient effect is probably the more important factor. 11. Eric Hirst, William Lin, Jane Cope, "An Engineering-Economic Model of Residential Energy Use," Oak Ridge National Laboratory, ORNL/TM-5470, July 1976. 51 The average rates of change for the mix and coefficient change effects, based on the indicators discussed above, are shown in table 27. The estimates indicate that during the post-1970 period (1970-1973 or 1971-1974), the decline in the overall BTU/GDP ratio was attributable to both changes in the household direct energy purchase component of final demand mix and in 1-0 coefficients. They also suggest that the more important factor in reducing the overall BTU/GDP ratio was the change in 1-0 coefficients. 12/ The figures also suggest that, given the generally greater reductions in energy/output ratios for the several indicators used (with the exception of the energy sector) than the decline in the overall BTU/GDP ratio, the effect of the changing mix of the other components of final demand, excluding household direct purchase of energy, may have been neutral or perhaps tended to increase the aggregate BTU/GDP ratio. This possibility is consistent with the fact that the "services" proportion of final demand, which accounted for about 45 percent of total GDP in 1970 (in constant 1972 dollars), and which is, in general, less energy intensive than the goods and structures com- ponents declined between 1970 and 1973 relative to GDP, the decline amounting to about 1.0 percent per year. Finally, if subperiod estimates are combined to approximate the longer periods used in the other sections of the report (and allocate the "residual" to the two factors for the periods included in the Reardon analysis), the figures given below indicate that within periods coefficient changes contributed more than demand mix to the reductions in the BTU/GDP ratio during periods when the ratio declined and less to the increase in the ratio during the most recent sub- period when the ratio increased. However, analysis of the extent to which the two factors contributed to the slowdown in the rate of decline in the ratio from one subperiod to the next reveals that both factors contributed to the slowdown but demand mix also played the major role in the fall-off in the reduction in the ratio from the first to the second subperiod but coefficient changes contributed more to the reversal in trend which occurred between 1958-1967 and 1967-1973. Annual rates 1947-1958 1958-1967 1967-1973 -.3 .1 .6 -.9 -.7 .2 -1.2 -.6 .8 Demand mix Coef f i ci ent change Total BTU/GDP The estimates for 1967-1973 are, of course, only approximate since they are based on a variety of indicators rather than a comprehensive set of calcu- lations along the lines of the Reardon analysis but they are believed to provide a reasonable distribution of the change in the BTU/GDP ratio between the two factors. As the annual updating of the 1-0 tables become available, it should be possible to develop a better set of estimates for the most recent period. 12. BEA Staff Report No. 28, op. cit., shows a mixed picture regarding effect of changes in 1-0 coefficients on energy output between 1970 and 1971. Coal shows an acceleration in the decline attributable to coefficient change; petroleum products continues to show an increase due to coefficient change but at a sub- stantially reduced rate and utilities indicate no significant effect, due to coefficient change. - 52 Section 6. Conclusions The initial objective of the study was to analyze the factors under- lying the 1966-1970 reversal in the long-term decline in the BTU/GDP ratio. In attempting to place the 1966-1970 reversal in a longer-term perspective, it was found that there had been a slowdown in the secular rate of decline in the ratio since 1929, culminating in the most recent period, between 1966 and 1975, when there was a net increase in the ratio. As a result, the objective of the study was broadened to include an analysis of some of the factors underlying the reduction in the long-term rate of decline in the BTU/GDP ratio and the connection, if any, between the 1966-1970 reversal and the slowdown in the secular rate. The relationship between growth and BTU/GDP ratio Analysis of the historical record since 1929 reveals that the 1966- 1970 period is not unique, that there have been three other reversals (1931-1933, 1944-1947, 1954-1956) and these do not display a pattern of in- creasing severity or duration. The 1966-1970 reversal lasted four years, the longest of the four reversals but the annual rate of increase in the BTU/GDP ratio over the period was 2.1 percent, the lowest rate of the four. The total increase in the ratio was larger than in two of the other re- versals but less than the 1944-1947 reversal. Further analysis of the long-term record indicated a fairly close in- verse relationship between growth rates in GDP and the direction of change in the BTU/GDP ratio leading to the conclusion that during periods of reduc- tions in GDP, or low rates of increase (2.5 percent or less), there was either an interruption to the downward trend in the ratio or an actual in- crease. The 1954-1956 reversal was an exception to this general finding, showing an increase in the ratio in spite of the fact that GDP was increasing at a better than average growth rate of 4.4 percent per year. The inverse relationship between changes in GDP, defined in terms of "high" or "low" growth rates and the BTU/GDP ratio, suggests that a large part of the pattern of alternating periods of declines and increases found in the historical record of changes in the BTU/GDP ratio is related to alternating periods of high and low growth in the economy. This finding helps put the 1966-1970 reversal in perspective by indicating that given the "low" growth rate for the subperiod as a whole, 2.3 percent, it is quite consistent with the historical record to find an increase in the ratio for the subperiod. This does not mean that specific developments during the period may not have played a role, e.g., Vietnam War, increased use of electricity for air conditioning and home heating, etc., but rather it suggests that primary emphasis on these factors is not necessary to explain the 1966-1970 reversal. Further probing of the relationship between growth and changes in the BTU/GDP ratio, at least for the 1947-1975 period, revealed that a major factor in "explaining" short- and long-term changes in the BTU/GDP ratio is the differential elasticities of real domestic output (GDP) and energy (BTU's) - 53 - to changes in civilian employment. For the period as a whole, the higher response of GDP, as compared to that of energy consumption, to changes in civilian employment resulted in declines in the BTU/GDP ratio. This factor more than offset the effect of two other variables—reductions in relative energy prices and increases in population—which tended to in- crease the BTU/GDP ratio during most of the period. However, the elasticity of GDP to changes in civilian employment was not uniform throughout the period and the pattern of the departures from the average GDP/civilian employment elasticity during the 1947-1975 period provides part of the explanation for the short-term reversals in the trend of the BTU/GDP ratio, and the longer-term slowdown in the decline in the ratio. In particular, during both the 1954-1956 and 1966-1970 reversal subperiods, the GDP/civilian employment elasticity fell sharply below the average elasticity for the period as a whole. This helps explain the anomaly of the reversal in the ratio in 1954-1956 when the GDP growth rate was above average, 4.4 percent, based largely on a much higher than average increase in employment and a less than average GDP/employment elasticity (1.4) so that given a constant BTU/civilian employment elasticity (almost 1.9), energy consumption rose more rapidly than GDP. The same thing happened in the 1966- 1970 reversal, although the increase in employment and GDP was not as large as in 1954-1966. In both instances, the fall-off in the GDP/civilian employ- ment elasticity reflected a slowdown in the rate of increase in GDP per civilian worker. The explanation given above for the short-term reversals in the decline of the BTU/GDP ratio since 1947 also applies to the longer-run deceleration in that decline, culminating in an actual increase in the ratio between 1966 and 1973. The estimates indicate that over the 1947-1973 period (which leaves out the major recession and very high energy price increases of 1974- 1975), the average annual increase in output per civilian worker was about 2.2 percent. This rate of increase was not uniform over the period, however, declining from an annual rate of 3.2 percent in 1947-1954 at the beginning of the period, to 2.3 percent in 1954-1966, and down to 1.2 percent in the last subperiod, 1966-1973. At the same time, the rate of expansion in civilian employment was increasing from .8 percent per year, to 1.7 percent and then 2.1 percent. As a result, the GDP/civilian employment elasticity fell from 5.2 to 2.5, and then 1.6 in the last subperiod. Given the constant energy/ civilian employment elasticity of almost 1.9, the BTU/GDP ratio declined during the first two subperiods but at a diminished rate, and then actually increased between 1966 and 1973. The further deterioration in output per civilian worker during the 1974-1975 recession would have resulted in a further decline in the GDP/employment elasticity and an increase in the BTU/ GDP ratio except for the major factor working in the opposite direction, namely, the sharp increases in energy prices since the oil embargo at the end of 1973. These findings suggest that the outlook for long-term changes in the BTU/GDP ratio depends to a considerable extent on whether the recent (1966- 1973) relatively low rate of increase in output per worker, and its dampening effect on GDP/worker elasticity, represents a new secular trend as compared to the much higher rates of earlier periods, and the extent to which this may - 54 be offset by continued increases in the relative price of energy. In this regard, the pending revision of Denison's study on accounting for U.S. economic growth should provide new insights on the trend in potential growth and potential output per worker and whether the slowdown in actual output per worker for the more recent period is also found in the growth rate of potential output per worker. Changes in thermal efficiency and effective BTU/GDP ratTb~ The slowdown in the rate of decline in the BTU/GDP ratio could have occurred because the thermal efficiency of energy use for the economy as a whole has been declining (or perhaps declining at an increasing rate re- flecting the growing proportion of electric power use in total energy con- sumption) or because the rate of decline in effectively used BTU's per dollar of GDP has been slowing down. To provide some tentative answers to this question, estimates were developed in this study of changes in thermal efficiency and effective BTU/GDP ratios for the three major subperiods (1947-1954, 1954-1966, and 1966-1975). Thermal efficiency has been broadly defined to reflect the proportion of initial primary energy which is effectively used for heat or work after deducting (a) all energy used or lost in conversion by the energy sector itself (which is broader in scope than just electric power conversion losses) and (b) energy lost in non- energy sector use, i.e., end-use. Deducting these losses, as well as non- fuel and miscellaneous uses of energy, yields a measure of effective use of energy which when related to GDP may be considered as an indicator of the economic efficiency in energy use as distinguished from thermal efficiency. The estimates, indicate that for the first two subperiods thermal efficiency, broadly defined to include both categories of thermal efficiency, increased significantly thus reducing the gross BTU/GDP ratio but this was reversed in the latest subperiod and thermal efficiency declined. The analysis of these changes on thermal efficiency indicates that the decline in thermal efficiency in the energy sector between the second and third sub- period was due largely to the growth in the relative importance of electric power with its very high conversion losses, which were no longer offset by gains in efficiency in electric power generation. The gains in thermal efficiency in end-use sectors, which were very high in the first subperiod, and were based to a large extent on the substitution of more efficient types of energy for coal, fell off rapidly as this source of increased efficiency diminished. During the last subperiod, there were practically no further gains in thermal efficiency in end-use, taken as a whole, although not necessarily for each end-use sector. The gains in thermal efficiency in the first subperiod were accompanied by increases in effective energy/GDP efficiency resulting in the very large rate of decline in the gross BTU/GDP ratio. In the second subperiod, the gains in thermal efficiency were maintained but the effective BTU/GDP, instead of continuing to decline, increased resulting in a substantial slowdown in the rate of reduction in the gross BTU/GDP ratio. Finally, in 55 - the last subperiod, the gains in overall thermal efficiency were reversed and, as noted above, actually fell at a substantial rate, which would have resulted in an increase in the gross BTU/GDP ratio of .8 percent per year. This was partially offset by increases in efficiency in effective energy use per dollar of GDP, .4 percent per year, resulting in a net increase in the gross BTU/GDP ratio of .4 percent per year. At least for the more recent subperiod, gains in effective use of energy have not been sufficient to offset the losses due to the decline in thermal efficiency. The recent low rate of gain in the effective BTU/GNP ratio may well be directly related to the earlier finding that there has been a slowdown in labor productivity over the post World War II period, particularly since the mid-1960's. The implica- tion for the long-run trend in the BTU/GDP ratio is the same, namely, given the assumption that electric power will continue to grow as a proportion of total energy consumption and therefore tend to reduce thermal efficiency of energy, both effective use of energy and, by inference, labor productivity will have to increase at a faster pace than the average for the period since 1966 if the gross BTU/GDP ratio is to resume its long-term downward trend. Sector energy/output changes The study analyzed in some detail energy/output changes for three energy using sectors of the economy, first examining the extent to which energy consumption in the sector increased or declined relative to growth in the economy (the sector BTU/total GDP ratio) and whether these changes were due to changes in the sector BTU/output ratio or the growth in sector output relative to GDP. The sector BTU/output change was then further factored into mix vs. intensity effects, i.e., the change in the mix of component industries or modes of transportation vs. changes in energy/output ratios for the individual industries or types of transportation. The sectors included in the analysis were passenger transportation, intercity freight transportation, and manufacturing, excluding petroleum refining. In total these three sectors accounted for 52 percent of total net energy consumption in 1973, with passenger transportation accounting for 21.6 percent of the total, intercity freight, 4.6 percent, and manufacturing, 26.0 percent. The major category of energy use not included in the analysis was "household and commercial" because time series are not available on energy consumption by major com- ponents of the group, i.e., households, commercial, government--nor are there readily available measures of "output" which could be used to derive esti- mates of energy/output ratios comparable to those developed for the other sectors. 1. Passenger transportation Energy consumption per passenger mile increased during the 1950-1973 period, at successively higher rates from one subperiod to the next. (The subperiods are selected to approximate those used in other parts of the study and cover 1950-1955, 1955-1965, and 1965-1973.) Underlying this trend has been the continued shift in passenger transportation towards automobiles, which are more energy intensive than bus and mass transportation, the decline in vehicle miles per gallon, and the increasing proportion of automobile miles used for urban transportation. Because urban automobile transportation - 56 - is less energy efficient than interurban and also has fewer occupants on the average, this has contributed to the continuing increase in the BTU/PM ratio between 1950-1973. The increase in the relative importance of air- line transportation, which is the most energy intensive of the various modes of passenger transportation, has also contributed to the continued increase in the BTU/PM ratio. As a result of all of these factors, the BTU/PM ratio increased at the rate of .4 percent per year during 1950- 1955; .7 percent in 1955-1965; and 1.1 percent in 1965-1973. Passenger transportation miles also increased throughout the period relative to the growth in GDP but the relative increase was not uniform, being close to 1 percent per year during the beginning and end of the first period but dropping to only .1 percent in the middle period. The combined effect of both factors, i.e., changes in the BTU/PM ratio and the passenger BTU/GDP ratio, was to increase energy consumption for passenger transportation relative to GDP by 1.2 percent per year in 1950- 1955, which then declined to .8 percent in the middle period, but jumped to an annual rate of 2.1 percent between 1965-1973. These increases contributed to the slowdown and subsequent reversal of the overall BTU/GDP ratio, and a major part of the outlook for changes in the overall ratio depends on whether all or most of the underlying trends indicated above can be turned around. 2. Intercity freight transportation The change in energy consumed per ton mile (BTU/TM ratio) in inter- city freight transportation also contributed to the pattern of slowdown in the BTU/output ratio for the first two subperiods followed by an in- crease for the most recent subperiod. The chief factor in this pattern for intercity freight was the diminished rate of decline in the BTU/TM ratio for railroad freight operations as the transition from the use of coal to diesel power was completed in the earlier years of the period. Freight ton miles fell relative to GDP for the period as a whole and for each of the subperiods but here again the rate of decline in the last subperiod was substantially below that of the earlier subperiods. The combined result of both factors was a fall in energy consumption for inter- city freight transportation relative to GDP for the first two subperiods but at a declining rate: -12.5 percent in 1950-1955; -3.5 percent in 1955- 1965; followed by an increase in the ratio during 1965-1973 at the rate of .3 percent per year. 3. Manufacturing, excluding petroleum refining The long-term trend in the manufacturing BTU/output ratio has been downward, with rates of decline considerably in excess of that for the comparable economy wide ratio. For the entire 1954-1974 period, the manu- facturing net BTU/output ratio declined by about -.8 percent per year compared to -.2 percent for the total economy. Both ratios declined at a diminishing rate, from -.9 percent (1954-1967) to -.6 percent (1967-1974) 57 - for manufacturing and for the total economy, from -.3 percent (1954-1966) to no change (1966-1973). Thus, the manufacturing sector has also played a role in helping to dampen the decline in the overall BTU/GDP ratio, but to a much lesser extent than some of the factors described earlier. The mix effect in the manufacturing sector has not been uniform, helping to reduce the BTU/output ratio in the first subperiod but increasing it in the second. In fact, the change in energy intensity for the sector, excluding the effect of shifts in industry composition, shows a substantial acceleration in the rate of decline in the ratio, from -.5 percent in the first subperiod to almost double, -.9 percent in the second. In the first period, from 1954 to 1967, the less energy intensive industries increased their relative share of total manufacturing real pro- duct (excluding petroleum refining) from 64.3 percent in 1954 to 67.6 per- cent in 1967. Because these industries, on the average, have a much lower BTU/output ratio than the energy intensive industries (the ratio for the former is only about one-eighth that of the latter), this resulted in a substantial reduction in the BTU/output ratio for the sector as a whole. The reverse happened between 1967 and 1974, leading to an increase in the ratio which partially offset the larger gains in energy efficiency of the energy intensive group of industries. The outlook for further improvements in reducing energy consumption per dollar of output for the sector depends in part on whether this recent reversal of the earlier change in the relative growth of energy intensive industries continues, dampening the decline in the BTU/output ratio. Preliminary indications, based on changes in employment for the two groups of industries, is that the energy intensive industries (food, paper, chemicals, stone, clay and glass, and primary metal products) have again begun to decline in relative importance, which should help to further reduce the BTU/output ratio for the manufacturing sector as a whole. Input-output partitioning of change in overall BTU/GDP ratio The recent report by William A. Reardon, using input-output analysis to partition the change in energy consumption between coefficient change and demand mix, for subperiods between 1947 and 1967, provides the basis for partitioning of the change in the BTU/GDP ratio. In addition, the analysis of coefficient changes contained in BEA Staff Paper No. 27, used in conjunction with supplementary estimates, makes it possible to develop crude estimates for the more recent subperiods. If the subperiod estimates are combined to approximate the longer periods used in the other sections of the report and the "residual" is allocated to the mix and 1-0 coefficient change, we find that both factors contributed to the dampening in the rate of decline in the BTU/GDP ratio and subsequent reversal. Demand mix contributed proportionately more than changes in 1-0 coefficients in accounting for the slowdown in the ratio between the first and second subperiods, but coefficient changes accounted for most of the reversal in the rate of change between the second and third subperiods. 58 To summarize, the analysis of the factors underlying the slowdown in the long-term rate of decline in the BTU/GDP ratio, culminating in no net change in the ratio over the past two decades, suggests that this development reflects long-term trends in the economy and that, other things being equal, the ratio might be expected to continue to fluctuate during successive subperiods, but with no discernible trend over the long-term. However, there are two factors that might change this prognosis. The first is the possibility that the slowdown may reflect the deterioration in productivity gains since the mid-1960 ' s and that sustained increases in productivity more nearly in line with those achieved during the earlier years could result in a return to the long- term decline in the BTU/GDP ratio. The other factor is the new element in the whole energy picture—namely, the recent sharp increases in relative energy prices after a long history of either little change or declines in relative energy prices, accompanied by a growing emphasis on conservation as a means of curbing energy demand without seriously curbing growth. It is obviously too early to tell what the long-term effect of higher energy prices and the conservation drive will have on reductions in the BTU/GDP ratio but the limited analysis of the 1974-1975 experience, when increased relative energy prices more than offset the effect of the fall off in productivity gains, indicates that these could have a significant effect in reducing the BTU/GDP ratio. - 59 - Table 1 Gross Domestic Product (1972 Dollars), Energy Consumption (BTU's), and BTU/GDP Ratio, 1929-1975 Percent change GDP BTU's BTU/GDP (Billion 19 * (BTU's per GDP BTU BTU/GDP 1929. 1930. 1931. 1932. 1933. 1934. 1935. 1936. 1937. 1938. 1939. 1940. 1941. 1942. 1943. 1944. 1945. 1946. 1947. 1948. 1949. 1950. 1951. 1952. 1953. 1954. 1955. 1956. 1957. 1958. 1959. 1960. 1961. 1962. 1963. 1964. 1965. 1966. 1967. 1968. 1969. 1970. 1971. 1972. 1973. 1974. 1975. 1972 $) 313.5 284.1 262.8 227.3 222.6 240.3 261.8 298.1 312.1 298.7 321.0 346.2 410.3 486.5 544.2 574.3 564.1 474.6 466.7 485.9 488.8 531.5 574.7 596.7 619.9 611.4 652.2 666.1 678.0 676.5 717.3 733.6 751.2 794.3 825.8 868.7 919.9 975.6 1,001.9 1,045.7 1,073.1 1,069.8 1,100.3 1,164.1 1,227.4 1,206.9 1,186.8 10 12 dollar) 23,756 75,777 22,288 78,451 18,799 71,533 16,392 72,116 16,900 75,921 17,937 74,644 19,107 72,983 21,398 71,781 22,751 72,897 19,880 66,555 21,589 67,255 23,908 69,058 26,625 64,892 27,897 57,342 30,442 55,939 31,821 55,408 31,541 55,914 30,494 64,252 33,035 70,784 33,880 69,726 31,488 64,419 33,992 63,955 36,775 63,990 36,458 61,099 37,586 60,632 36,263 59,311 39,703 60,876 41,700 62,603 41,706 61,513 41,696 61,635 43,140 60,142 44,569 60,754 45,319 60,329 47,422 59,703 49,308 59,709 51,240 58,985 53,343 57,988 56,412 57,823 58,265 58,155 61,763 59,064 64,979 60,553 67,143 62,762 68,698 62,436 71,946 61,804 74,754 60,904 73,176 60,631 71,370 60,137 -9.4 -6.2 -7.5 -15.7 13.5 -12.8 -2.1 3.1 8.0 6.1 8.9 6.5 13.9 12.0 4.7 6.3 -4.3 -12.6 7.5 8.6 7.9 10.7 18.5 11.4 18.6 4.8 11.9 9.1 5.5 4.5 -1.8 -.8 ■15.9 -3.3 -1.7 8.3 4.1 2.6 .6 -7.1 8.7 8.0 8.1 8.2 3.8 -.9 3.9 3.1 -1.4 -3.5 6.7 9.5 2.1 5.0 1.8 0.0 -.2 0.0 6.0 3.5 2.3 3.3 2.4 1.7 5.7 4.6 4.0 4.0 5.2 3.9 5.9 4.1 6.1 5.8 2.7 3.3 4.4 6.0 2.6 5.2 -.3 3.3 2.9 2.3 5.8 4.7 5.4 3.9 -1.7 -2.1 -1.7 -2.5 3.5 5.3 -1.7 -2.2 -1.6 1.6 -8.7 1.1 2.7 -6.0 -11.6 -2.4 -.9 .9 14.9 10.2 -1.5 -7.6 -.7 .1 -4.5 -.8 -2.2 2.6 2.8 -1.7 .2 -2.4 1.0 -.7 -1.0 0.0 -1.2 -1.7 -.3 .6 1.6 2.5 3.6 -.6 -1.0 -1.5 * Trillions. Note: For further information on the newly revised and benchmarked national income and product accounts, see the January 1976 Survey of Current Business , BEA, U.S. Department of Commerce. The estimates of GDP in table 1, for the years 1929-1946 inclusive, have since been superseded by those given in the BEA Supplement to the Survey of Current Business , The National Income and Product Accounts of the United States, 1929-1974, Statistical Tables, 1977. The difference be- tween the preliminary and final estimates of GDP for these years are not sufficiently large to affect the analysis in the report on changes in the BTU/GDP ratio. Sources: Column 1 - 1929-1945 - Unpublished BEA table 1.8; 1946-1971 - 1976 Economic Report of the President, table B-7, p. 179; 1972-1975 - BEA, Department of Commerce, Survey of Current Busin ess. July 1976, table 1.7, p. 26. Column 2 - 1929-1946 - Historical Statistics of U.S. Part Series M83 and M90, p. 588, Bureau of the Census, U.S. Department of Commerce, 1975; I U.S. Energy Through the Year 2000, Walter G. Dupree, Jr. and James A. West, Appendix table 1, J.S. Department of Interior, 1972; 1970-1973 - 1974 Minerals Yearbook (to be published), table 17, U.S. Department of Interior, U.S. Bureau of Mines; 1974-1975 - U.S. Department of Interior, News Release, Annual U.S. Energy Use Drops Again, April 5, 1976, adjusted to include natural gas transmission loss and unaccounted, to be consistent with previous measures of natural gas consumption. 60 Table 2 Real Gross Domestic Product (GDP), BTU's, and BTU/GDP Ratio Selected Subperiods, 1929-1975 Total percent change Annual rates of change Subperiods 1929-1941 GDP 30.9 -29.0 -16.2 -15.3 84.3 49.0 13.7 40.0 -18.7 31.0 59.6 9.0 46.5 21.7 9.7 10.9 14.7 -3.3 BTU's 11.9 -28.9 -20.9 -10.1 57.3 36.2 24.1 19.6 3.2 9.8 55.6 15.0 35.3 26.5 19.0 6.3 11.3 -4.5 BTU/GDP -14.5 0.0 -5.6 6.1 -14.6 -8.6 9.1 -14.6 27.0 -16.2 -2.5 5.6 -7.6 4.0 8.5 -4.2 -3.0 -1.3 GDP 2.3 -8.2 -8.5 -8.0 8.0 3.1 2.2 11.9 -6.7 3.9 4.0 4.4 3.9 2.2 2.3 2.1 4.7 -1.7 BTU's .9 -8.2 -11.1 -5.2 5.8 2.4 3.7 6.1 1.1 1.3 3.8 7.2 3.1 2.7 4.5 1.2 3.6 -2.3 BTU/GDP -1.3 1929-1933 0.0 1929-1931 -2.8 1931-1933 3.0 1933-1941 -2.0 1941-1954 -.7 1941-1947 1.5 1941-1944 -5.1 1944-1947 8.3 1947-1954 -2.5 1954-1966 -.2 1954-1956 2.7 1956-1966 -.8 1966-1975 .4 1966-1970 2 1 1970-1975 - 9 1970-1973 -1 1973-1975 - 6 Long-term changes 1929-1941 30.9 1929-1954 95.0 1929-1966 211.2 1929-1975 278.6 11.9 -14.5 2.3 .9 -1.3 52.7 -21.7 2.7 1.7 -1.0 137.5 -23.7 3.1 2.4 -.7 200.4 -20.6 2.9 2.4 -.5 Addendum Hypothetical extension of 1966-1975, subperiod for 3 years: 1966-1978 0.0 1966-1970 2.1 1970-1978 -1.0 1970-1975 -.9 1975-1978 -1.25 61 Table 3 Gross Domestic Product (1972 Dollars), Energy Consumption (BTU's), and BTU/GDP Ratio, 1909-1929 1909. 1910. 1911. 1912. 1913. 1914. 1915. 1916. 1917. 1918. 1919. 1920. 1921. 1922. 1923. 1924. 1925. 1926. 1927. 1928. 1929. GDP BTU BTU/GDP (Billion 12 1972 $) 10 179.9 13,531 75,214 185.0 14,800 80,000 189.7 14,624 77,090 200.5 15,708 78,344 202.4 16,719 82,604 193.4 15,503 80,160 191.7 16,076 83,860 206.8 17,781 85,982 208.2 19,597 94,126 233.8 20,436 87,408 225.5 17,558 77,863 215.6 19,782 91,753 196.8 16,410 83,384 227.9 17,215 75,538 255.5 21,685 84,873 254.9 20,453 80,239 276.3 20,899 75,639 292.6 22,495 76,880 292.3 21,828 74,677 294.0 22,381 76,126 313.5 23,756 75,777 - 62 - Table 4 Relationship between Changes in BTU/GDP Ratio and Changes in GDP (Subperiods Selected on the Basis of BTU/GDP Changes) Rates of Change BTU/GDP GDP Declines in BTU/GDP 1929-1931 -2.3 -8.5 1933-1941 -2.0 8.0 1941-1944 -5.1 11.9 1947-1954 -2.5 3.9 1956-1966 -0.8 3.9 1970-1973 -1.0 4.7 Increases in BTU/GDP 3.0 1931-1933 -8.0 1944-1947 8.3 -6.7 1954-1956 2.7 4.4 1966-1970 2.1 2.3 - 63 - Table 5 GD?(72$) and BTU/GDP Growth Rates (Subperiods Selected on the Basis of GDP Growth Rates) Subperiods "Low" growth or actual declines in GDP 1929-1933 1944-1947 1953-1960 1966-1970 Rates of Chanqe GDP BTU/GDP -8.2 -6.7 2.4 2.3 0.0 8.3 0.0 2.1 "High" growth in GDP 1933-1941 1941-1944 1947-1953 1960-1966 1970-1973 7.9 -2.0 11.9 -5.1 4.9 -2.6 4.9 -0.8 4.6 -1.0 - 64 - cn c fO a. _c Q o CD -«■* -4-> ZD C I— CU CO u S- Ol -a CU o n3 CJ «^-LOkDc\jcoo«X3 Q c_> ro CD CU =D +-> h- u CQ 1 — 3 ra -o-o 3 O CD 4-J S- +-> CJ D_ (0 d r— CJ 3 •r- u ■M >— i_n C£> to rO CD o CD E i — i — o -o i -O Q C r>. rO rO ■=d- (— CO CX> "O CO i — i — cu O rO 4-> S- 3 rD CD -M *- — ^ i — ~^ U C\J 3 ^--=a: 1 — o ID o 1 — h- CQ ZZ- ro o Or-ONCvj^uD^ciicoowor-in^-LnrooiDMiMMi-i-o^aicn cm i — wnwconloi — cnj^-^i — oaic\j-=3-cor^.O"=j"cr>coi — <3- o ^J- ■ — co wa^c\irv>3-OLOCTi( v )c\nDrv'* o a~> cm i — cororoir)(\i^tcTicoCTiLncoroir)^i-fO<\ik£)"*<*i — r^ co cm i — LncriCTiooi — i — o i — co^LocMOOoooocMir>coi-ncQcooococo SN^tCTiCTiomrooocoirnDi — i~~. co r^- r^. a~> cr> oo ■ — ouii — >3- co a~> co i — O CXi ■=* CO CO i— OCTiOCMi — r-OOOmCTiCONNCOOiOMWr-OOO NkDco^DiDiKiOLOix>co^o^iOiDniLf)inir)ir)ir)iriir)kDcoiDU3vDcoco CO un^j-r-^CDi — CTic\j«3oowwifi^tf»i<^cTiNir)m«tixcam'd-o^coMcD ocx>i^c\j(Ti , *unLn>^LOcTiCTir~.'^ _ 'd-c\jc\i , 5t-o~icTiLncrii — c\i co Oi — o ai fT*C0a->d-- ro Z3 U inooowLnaDcoMcooiflcooaiCJiwcoofocviinoiaicooocD'a-coo nCOCOCTiNLDCOCOOOOCTlitlDi — CM O i — ■ — i — m^lon cru — co co co ■ — ^n co > — >vj- co i — c^-- o <=o- O- •i — Q i — CM CD i — r-^ •j— CTv CO r— I — CTiCO^NNCri'^Mi — OLnmcOMCOCONCTiCDWNi — CO CO i — >cj- CTi CO coLnoOi — "^-coctip — McococoNroi-^i-LncocMni — Lr)noiO"*NcoiD CO 00 00 CO I — Oli — i — Ifl IDNISi — CO LT> CX> C\J CO i — r-.O*M0O^tMO0D - NCOCTiOi — CMOO^J-LncOr^COCTiOi — CMOOC^CTiCTiCT>O^CTiC71CJ1CTiCTiCTiCTvO^CTiCT»CTiC3^CT^CT^ Table 7 Changes in BTU/GDP Ratios Actual and Calculated 1947 - 1975 Annual rates Actual Calculated Long-term 1947 - 1975 f - 5 1947 - 1973 -.6 -.5 1953 - 1973 .0 .1 Subperiods 1947 - 1954 - -2.5 -2.0 1954 - 1966 -.2 -.3 1954 - 1956 2.7 1.9 1956 - 1966 -.8 -.7 1966 - 1975 .4 .3 1966 - 1970 2.1 1.9 1970 - 1975 -.9 -.9 1970 - 1973 -1.0 -.5 1973 - 1975 -.6 -1.5 1966 - 1973 .7 .9 - 66 - c o •i — +j (O ZS CL O «3" LO LO CO (O q: Q- Q CD >> CO co >— cx> • • • s- 1 1— a> 1 c CD 00 ai a> > u •r— *r- +-> S- (D Q- i^ a; a: co ro co cn rv fO (13 -C ce: OLO IV i — •4-JCTl (13 u>— 3 C1J 1 C co M-rv C M-^l- <=C cn i — r— CD -Q +J CD (T3 (0 (ti 1— .c: i- \— CD > 00 <£ CD r— _Q n3 •i — s- fd > -o CD -l-> O CD r— CD OO +-> c: c (O CD ■r- E — >J •i- o > I— ■r- Q. O E CD CT> CXI CO CXI CO CO LO LD "3" <0 CM O CM i— CM i — oo CXI r— CXI CM CM ,— CM CO LD ^j" CM O CO i — CM CM CM IV 4-> C C a. (d CD ^O •r- E > *«— ' •r- O o > i— •r- Q. O E J * -r- O > I— •i- Q. O E CD LO -51- CM i — r— i— i— CM CM co o oo cn «d- cm OO I — I— CM Q_ Q CD <3" 00 LO CO oo oo cn oo O *3" CT> ■=d- <* oo CM CXI CM "=d" i — ^- i— CD I cn c o oo oo rv rv cn CT> (V oo «3" LO cn cn oo •o o 5- OJ CL -O I— r— I— 00 OO LO 00 rv IV -a to to O LO CT> cn o LO tO IV IV i — i — •i — "vf CO CT> CTi LO CT) CT> i i C" S- LT LO r— r— rv i — i — O oo IV CD a- CTi 1 1 cn i i i — IV o- Q. r— «* tO ■— to o cn CT. 1 1 1 LD U") 1 tO IV i — , — 1 +-> |v *d" CT> CT> to cn cn iX S- ■5* LT) i — i — lOi-r- X o CT cn CT> a- .c i — 00 ^1" CO tO CD 0O LO LD Ln cd rv rv i^ cn cn cn cn cn cn i i i i i i N^aiDOro ■^t- lo lo to rv rv cn cn cn cn cn cn Table 9 Contribution of Explanatory Variables to Changes in BTU/GDP Ratio 1947-1975 Average Annual Rates (2) (3) [4) (6) (7) (8) Contribution of Variables to Long-term 1947-1975 1947-1973 1953-1973 Sub periods 1947-1954.... 1954-1966.... 1954-1956.. 1956-1966.. 1966-1975 1966-1970.. 1970-1975.. 1970-1973 1973-1975 1966-1973.... Short-periods 1947-1954.... 1954-1956 1956-1966 1966-1970.... 1970-1973.... 1973-1975.... A GDP 1.864 less col. 2 -.5 -.6 -.3 Civilian employment BTU/GDP Ratio Change ACivi lian employment elasticity Civilian employment Relative energy price -.1 .1 .2 Po pulation Total Calculated 2.4 2.5 2.1 1.4 1.5 1.6 -.7 -1.0 -.5 .3 .3 .3 -.5 -.5 .1 5.2 2.5 1.4 2.9 1.3 1.2 1.4 2.0 -7.6 1.6 5.2 1.4 2.9 1.2 2.0 ■7.6 -3.4 -.6 .4 -1.0 .6 .6 .5 -.1 9.5 .3 -3.4 .4 -1.0 .6 -.1 9.5 1.6 3.0 1.3 1.7 1.9 1.5 2.4 .2 2.1 3.0 1.3 1.9 2.4 .2 -2.5 -2.5 1.3 -1.4 1.2 -.2 2.1 .0 1.0 .3 1.3 .2 1.4 .3 1.0 -.9 1.2 .4 .7 -1.9 -.2 -.5 2.1 -3.8 .0 .0 .2 .3 .4 -.5 -3.8 •2.0 -.3 1.9 -.7 .3 1.9 -.9 -.5 -1.5 -2.0 1.9 -.7 1.9 -.5 -1.5 Note: Calculations based on unrounded numbers. Figures shown in table have been rounded. Sources: Column 2 - table 8, column 4; column 3 - BTU/ Employment 1947-1975 elasticity less GOP/Employment subperiod elasticity; column 4 - table 8, column 3; column 5 - column 3 x column 4; column 6 - BTU/Real price elasticity of -0.1791 x table 8, column 6; column 7 - BTU/Population elasticity of .2156 x table 8, column 7; column 8 - sum of columns 5, 6, and 7. 68 Table 10 Thermal Efficiency Factors in End-Use Coal Petroleum. . . Natural gas. Electricity, Household and Industrial Trans po rtation commercial 50 60 5 60 65 25 70 80 -- 100 100 100 Source: Based primarily on Sam H. Schurr et al , Energy in the American Economy, 1850-1975* Resources for the Future, Inc., 1960, pp. 172, 177-178. 69 - CO"=J-CnCM"3-CMLOCO iooinwi — on- cd >d- CO CO o Ln i— «* ^3- oo >— C\J CO CD LO [-- LO ld I — i — ro ■3- Ln r-^ en «3- . — co in i — lo cm r-^ won oo* ID LD Cn «d- i — CM C\J CvJ i— Cn Ln col CM CVJ Ln un cm Ln i — r-. co en Ln kDCOVOi- NOi-O Cn >* Cn co *$■ r-~ [~~ Ln Ln in co co O I~~- 1 — *3" "3" CI LD CO in CO LO LD >* "* CM Cn i — Ln i — CM CM i— . — CO CD r-. LO CO LT> Cn CM CM r— i— o r-. t— o i— Ln ld Ln co i— =3- co co co i— CM i — CM «* •=J- evil CM m en ^a- Ln 53- co 1— ^d- co en 00 Ln co 1 LD CD "3" Ln CD LD CO LD Ln 1 — cn id ^3- ld ld co 1 — Ol LD cn CD CO r— LD LOtvCOCvJ I— 1— ■* 1 — CM CTv LD LD LD Ln >3- 1-- co CM CM 1 — 1— CM CO LD CM CvJ r— r— CO LD 1— LD uivtino sTOi — LD r— CM 1 — 1— CO LD LD 0-1 ■^1- ol LD CO r^ >=*• >* CO CM CM CO CTi Cn 1 Cn mvtON *J- co cm en co Ln co Ln o^tno Ln en 1— cm CO CO CO "3" ld ld 1 — co O r-~ ld ld 0-1 >* CM 1— CM LO LD Ln Ln Or-3- en ^t- 'd- co en ld co 5- O CO LD O 1— <~ 00 CO CM LD O CO co >3- 1 — en CM en r— en *d- co co r^ Ln LD 1-^ CM Ln en LO CM col <* LD CM CM 1— 1— O LD CTl LO lo ld ld *3- 1 en LD t-~- 1— CO •=3- ir> en 00 LO 1 — CM LD co ld CD ld en 1— co Ln 00 <=r r^ ld co co r--. lo CM CM CM CM CM 1— LO CO LO CO cm co en 1 — ■- CO CO r— cm en r-^ CO CO 1 — LD CM 00 en co cm CM CM LO CO LD LD CM O O r^ ^^ <* en ■": en s- <=> CO r— col LD 10 lo en ld cm r-^ Ln co CO CO co lo 1 CM co cm en en en r^ Ln ld en co r-- ld 1 — 1^ r-~ Ln NOiCOLO ld 1— en 00 lo cm en «* «d- 1— en CM lo en 00 co CVJ CO p^. 1 — CO LO <— o r-. "£ o JD+J •"- (DC/)' I— CD "~ ° i 03 5- +-> CO S 3 o S- enl p o CD O LO (J "O fO +J 4- — •!- S- C O S- O O C 4-> D_^ CD 00 00 \— LO 1— rO CD ■O (O+J 3 S- Ll_ O D. c 10 ra i— s- a> in L0 3 3 -O O C o t_ .c ■<-> CD 00 L0 3 3 -O O C 3: 1— 1 o Ol •r- +J 4-> fO (O o •M o S- 1— O 1— Q.CC LO o s- o ^ +-> D_ 0) L0 L0 •• LO 3 C LO 3 T3 fO 1— O C I. (5 X l-H I— 3 C7 >,TD s- s- Ol E E QJo D- O T3 "o ^ ^ +-> CD LO 00 3 3 -O O C -(-> t- (0 +J 4-> S- 10 o o Q-_J on cu > o +-> •r- (J +J CD ro 4- +-> 4- S_ LU O C 00 (O >— S- (O 1— 3 CT (0 1— nj ■OOP .— -i- i- O S- o r: +-> CD- CD 00 00 00 3 c: 3 -a re o c s- IH|- CD 00 3 03 (J CD .a T3 -a >> CD Q o ■— 1 CTi ^-i — *a- cn loo i — no WOWCV1 ■— r^ «S- 00 N CO COO CO CTi LO lo CTi «3- CM o o CTi CTi r-~ lo i~- ld|i — «a- «3- fOfOM CTi in i — r~ i — OO CO CM CM ^f CD CXI LO ,— i— LO r— o «* cn CM CM CSJ CXI lo <3- co r^ r- co LO LO LO r-. cn o col r^ i— CTi r~~ o It — i — cm r — <*coirii- ■3" ^3" OO LD CD ^3" CM OO O CM i — CO O CM O o o CTi cn r^ lo ld lo |i — ^t- co CO CO CM CTi LO i — LD LD CO CO CM CM CO O CM t-» i— ,— lo CM i — ^" CTi CM CM CM CM LO LT> CO CO i— CO 1X3 LO LD o en o col r~- «d- <* 00 CTi IlO CM 00 CO CM CO CO ID 00 "3" CD CM CO LO CTi IslOlOi — LD r-- CTi o o cr> CTi r-~ lo ld un|i — «a- CO CO CM i co CTi LD CM LO ^3" CO CO CM CM CM CTi CM CTi ■— LO CO CM CO CTi CM CM CM CM CD LO r^ CTi i — co ID CD UO ID CTi o r-l CTi i — r~- CM O |r-~ *d" 0~i <*■ CO CM CTi oo r-. lo o r-~ co r~- CTi CM CD O i — lo o r-~ o o - CO CTi ld r-~ LO LO |i — CO CO CO CM 1 CO CTi 5t "S-LOCJ CO CO CM CM CD CTi CM i — ■ — r-^ 0O rd- CM O CM CM CM CO i— ^r ID CO LO o col r^ cn lo co loIlo o CM CO LO CO CO lo i— r-. oo 00 ^" LD CD CD i — CTv CO MMIX o o cm! CTi i — cn — CO CO CM CM CO CD CM •— 1 — 1 — ■ — 3- CO CO CM i— CM 1 CO CTv i— *3- r^. r— CO CO CM CM CTi CTi CM i — CM CM CM CO ISNCON i— CO >3- r-» lo c o CJ c >v O cn •!- s- +j Ol 3 I C -O CT. -O. 4-> 113 CO I— CU E o Q 4- O CU <- 3 4-1 O 3 o col r^ OlDOr-|Oi- LO CD LO O CM LD CO CO «3" *WNi — lill — CO CO CTi LO CTi o o "■' CO CT C0 >— CTi ^J-|i — CO CM ■* CM 1 >* CTi 1 — 00 CTi r-. CM CO CM i — LO CO CM r»» O CD LD i— M to CM q- CO CD i— LD i— CM CO ■— CTi ^d- +-> CD 03 3 +-> 4- •— -r- S- c o o T3 a 03 "O 03 o t- -C 4-1 CU 00 VI 3 3 "D o C nr i — i TJ ID P 3 i — -r— S_ L0_ O D. o s_ -C 4-> CU L0 L0 3 3 -n o c 00 •• c oo ID i — S- ID (— 3 CT 03 CU 3 o _ cME o o T3 O S- .c +-> CU 00 00 3 3 T3 O C rC •— i 4-> ID 03 (J 4-> O i- r- O I— a. s_ i— ID HI "O ID +-> C i— •!- i- LU o s- o SZ 4-> Q. CU oo oo • • 00 3 C 00 3 "O l"0 03 CL C 03 O •i- C 4-> •!- (D 4-> +-> i— t- i. 00 O O D I X) O S- -C 4-> CU 00 00 3 3 T3 O C n: i—i s- 00 o a T3 C ID "O ID O C ^ 4-> CU 00 00 3 C 00 O 4-> •i- O 4-) CU ID 4- 4-) 4- 3 -O O C 03 i— S- 03 03 • 4-> i — 03 •a in 4-) o Z. o -C +-> Q. CU 00 00 00 3 C 3-D m O C i_ re — i h- u a> 03 4-J O +J o 4^ -o X3 03 o? ID +J O) a cu +J o CD > ■M CD CD =3 CXI i— to <*■ <*• r— O o 4-> CD to O i- C9 CXI i — co co <* co CO CXI O O O o 0J > J ( — +-> O CD <4- i/i 4- S- LU O 4-> ■o o C CD CO O0 •> en ■M C CD -t- 2: E 3 * to to c CO O o c_> S- C3 en c CO 4- O 1 — o E sz i — o c -O •r- o ra +-> •,- I— 3 +-> -Q Q. •r- E S- 3 -t-> to to c •I- o Q O CD >> en en (O S- +-> CD C E CD LU o S_ a> Q- cu > •1 — +-> cu o CO CD 4- 4- LlJ 3 CM CO CO CO tO I— o o CXI r-. 0) cn c X5 1 — to CD T7 3 +J =5 O C CD rs O c o X ITS o CD i — o i — CO s- o a. to CO -M O r— O i— f— *$• LO i ro i Q -M Z3 u u CD u 1— CD c CD ?*> to CD =1 CO H 'to >, C_> ca en LO cn-,- C o S. <+- to •i- to S- CD M- to E CD cn se CD o =3 +-> LxJ i- CO rO c: >* tS c ce + O C C_J> I— 01 CD 13 en 0) i- 3 C to -D to O C to 3 ru , 1 1 1 i — CD t_> 1 -C C +-> CD CD U to 'i— rs 4- 00 <3- o ro i locono t— O C\J o r- r— i— CM O O f0 A3 ,f0 c c c T5 13 fO e c c 13 fD T3 C C c c c c; LO Lf) C\J CO tr» c\j «=d- ro >T3 E C O ro C i— ro O CJ LO CD .C +-> Q_ Q. 1 CTi i — CD CO CO .Q ro co rs c: ^ +J DD f0 0O'4-J is o o c S- o <^i- 1— X ■— i I— I— CX> ^j- LO CO O lo lo rs jc CTt CT> CT\ CU i — CO i rs is <3- lo <3- lo lo : CT\ CTi CTi <^J- LO LO S_ LO LO is +-> CT> CTi CT\ LO ■ill -a i is -^j- lo : s* lo LD i o> CT> CO CTi CT» CTt I I I ro I I I : is <^- lo S- rs ^t- lo ^ LO LO I— ^f LO LO OS CT1 0""> O^ CTi cn + KO ■— CO CO CO CT> ro N ■* n CO IT ■* ■* l£3 ^3" C\J .— Csl i— II i — CO I I I I I c I/) i- il- in t- t/i o CU 4- in E OJ <- C CU O 3 +■> cr UJ S- LO fO o ca c o + c: o.— •r- fO O) i CM CO I ■ — i — i— i — C\J c\j r-^ CM P-- CM o CM CM i — CM i — CO CM O CM <3" CTt CM ^f U0 i— CM i — CO . — I CM LD CM MD i — CM O LD "^ tD 1£HD l£) N "^ CTt LT> LT> (£) CO LO i — ■ — I I I I I t I I I I I I co r^ co o o o o (Ti c\J cm k rv rx ix en en lo «3- *3- lo ro cmcmi — en <— on «d- i— i — cm i — cm « — i i i i • •— O CM O i— i— CM CM O i— O »— CM O i— CM CO O r— r- O O O O CM CO O <— i— i— CM O O r— LnrscOr-o>O^D --co^OOinro CM CM I CM I •— I I CO l i — I CT» <3" CO CM UD LO r-» CO LO r- i— i— i— r— rocniMN^ CO r— CM CO LD 03 en en ~o ^ *£> l£> O LO I I O LDLfl^NNOCOr en en en cr en r^. ix OM^CA CTi (Ti N N (/) i — i — i — i— i — cn en 3 .— r- -O <^- LnmiON •— i CT» cn CT» CTi cn > CO LO ■ — CM CM "^t" i— I— CTl i— OCON'sf O CTi CTi O r-« i — i — co O I— CM 1^ co co co cm co co «d- co CM I— co co *a- co 00 CO CO o O CTl CM O o r-^ *d- 0O CM ^J" 00 CO CO CM CM ■* o en Or-^- co co **■ ^}- cm i— en i — CO CO CM O CO cn r-- CM CO CM CO n ion co co r~- cn CO r— i— CM co co co i~~. CM i— i— i— CM CO ■* fOCMLf) m n id CM CM 'd- >> cn v CD c LXJ <4- CO o c o CD •!- CO -M => 3 -Q CD t- > S- LO •1- +-> 1 — +-> CO O «r- C1J CD O 1 — M- JD <+- -a CD fC 3 cn o a. -o CD -o +-> C fO (C r— CD i— cn CD rs -o Ll_ C fO CO CO O <~ CD I — MOID CTlOCMN CO CO O 00 CO CO CM 00 CM CM ^ CD CO r— i— CM CD 00 00 O 00 O CTi CO CM CO CO LD CM i— CO >=d- CO <3" cn o co cn tONOO CO i— CM ■=3- >=a- 1— cn CTi CTi CTi O CO CM CM i — CO O CM «Cf- LO ^J" i — O CO CO CTi CO CO r— CO CD CTi O CM CM CO <* CO i— CM O CTl r^ i — r— CTi CO CD CO .— -=3" CTi 1-^. CO i— i— r— CTi CM CM <=3- CD r- i — co co CM CTl CM CTl "^f CM CM LO Nr-CDCO O CO CO «* i— 00 <— CO >* CM CM CO cn n cti o CTl CO CO CO CTi co i — cn CO CM CM CO r^ CM CTi CD co >=J- r-~ co CO O CD *d- CO CM i — CO i_ CD o Cl|S- CD 2 o CL s- o Q- co i- C CD I— o CD O 3: 1- o +-> Q. ns +-> O i- ■r- o i_ Q. p w s. (J C CD CD rt3 -C i— s- +-> UJ I— O S- o CL T3 C CD O O +-> +-> o s_ •p- o CTi r— O CO CO CO CM CO O CM CM CO CM CD CO CM CM CO 00 00 CO r— CM CO CD O O CO CM CO r--. CTi co i— CM CO CO CTi CO CD CD CD co o co i — CO CO CD +-> I— o Q. +-> CO s- U C CD CD rO -C S_ +J I— O a; o CL c: 03 •13 +J s- o Q. CO i_ C CD fO -C S- -t-J I— o Table 16 Transportation Summary Statistics 1950 1955 1960 1965 1970 1973 1 p Energy consumption (BTU 10 ) Intercity freight 2,698 Passenger transportation.. 3,497 Subtotal 6,195 All other 2,445 Total 8,640 Percentage distribution Intercity freight 31.2 Passenger transportation 40.5 Subtotal 71.7 All other 28.3 Total 100.0 Intercity freight Ton miles (10 y ) 1 , 350 BTU's (10 12 ) 2,698 BTU's/Ton mile 1,999 Passenger transportation Passenger miles (10 y ). 946 BTU's (10 12 ) 3,497 BTU' s/Passenger mile 3,697 GDP (Billion 1972 dollars) 531.5 652.2 733.6 919.9 1,069.8 1,227.4 1,701 1,628 1,678 2,165 2,296 4,549 5,571 6,922 9,330 10,904 6,250 7,199 8,600 11,495 13,200 3,595 3,637 4,132 4,994 5,758 9,845 10,836 12,732 16,489 18,958 17.3 15.0 13.2 13.1 12.1 46.2 51.4 54.4 56.6 57.5 63.5 66.4 67.6 69.7 69.6 36.5 33.6 32.4 30.3 30.4 100.0 100.0 100.0 100.0 100.0 1,560 1,600 1,880 2,210 2,460 1,701 1,628 1,678 2,165 2,296 1,090 1,018 893 980 933 1,207 1,423 1,717 2,181 2,486 4,549 5,571 6,922 9,330 10,904 3,769 3,915 4,031 4,218 4,386 1950- 1950- 1955- 1960- 1965- 1970- 1973 1955 1960 (Annual 1965 rates) 1970 1973 -3.3 -11.4 -1.4 -2.6 1.9 -1.6 -1.0 -1.2 -1.8 -1.3 .2 -1.0 -4.2 -12.5 -3.2 -3.8 2.1 -2.6 .7 .4 .8 .6 1.2 .8 .6 .8 .9 -.8 1.8 -.2 1.3 1.2 1.7 -.2 3.0 .6 Intercity freight BTU's/Ton miles + Ton miles/GDP = BTU/GDP..... Passenger transportation BTU' s/Passenger mile + Passenger mile/GDP = BTU/GDP Total intercity freight and passenger transportation BTU/GDP -.3 -3.8 .5 -1.0 2.8 - 76 - Table 17 Intercity Freight Transportation 1950 1955 1960 1965 1970 1973 Intercity ton miles (10 9 ) Truck 1 70 Rai 1 roads 630 Waterway 420 Oil pipelines 130 Total 1,350 Intercity ton-miles, percent distribution Truck 12.6 Railroads 46.7 Waterway 31.1 Oil pipelines.., 9.6 Total 100.0 BTU/Ton-mile ratios Truck 2,218 Railroads 3,100 Waterway 730 Oil pipelines 450 Total 1,999 220 660 480 200 1,560 14.1 42.3 30.7 12.8 100.0 2,218 1,200 690 450 1,090 290 600 480 230 1,600 18.1 37.5 30.0 14.4 100.0 2,600 790 620 450 1,018 360 720 490 310 U! 19.1 38.3 26.0 16.5 100.0 2,240 720 450 450 893 410 770 600 430 2,210 18.6 34.8 27.1 19.5 100.0 2,540 670 680 450 980 500 860 590 510 2,460 20.3 35.0 24.0 20.7 100.0 2,190 655 680 450 933 1950- 1973 1950- 1955 1955- 1960 1960- 1965 1965- 1970 1970- 1973 'Annual rates' BTU/Ton mile Change in model BTU/Ton mile Shift among transportation modes.. Total •10.9 -.5 ■11.4 -2.4 1.0 -1.4 -3.0 .4 -2.6 2.2 -.3 1.9 -2.5 .9 -1.6 77 Table 18 Passenger Transportation 1950 1955 1960 1965 1970 1973 Passenger miles (10 ) Automobile, taxi, motorcycle... 799 Airline 9 Railroad 33 Bus - intercity 26 Bus - urban 28 School bus 14 Electric mass transit 37 Total 946 Percent distribution Automobile, taxi, motorcycle... 84.5 Airline 1.0 Ra i 1 road 3.5 Bus - intercity 2.7 Bus - urban 3.0 School bus 1.5 Urban mass transit 3.9 Total 100.0 BTU/Passenger mile Automobile, taxi, motorcycle... 3,802 Airline 4,500 Railroad 7,400 Bus - intercity 615 Bus - urban 2,340 School bus 714 Electric mass transit 2,180 Total 3,697 1950- 1973 BTU/Passenger mile Change in model BTU/Passenger mile Shift among transportation modes Total 1,071 1,291 1,559 1,970 2,256 21 32 54 no 132 29 22 18 11 9 26 19 24 25 26 21 19 18 15 14 19 25 31 38 38 20 15 13 12 11 1,207 1,423 1,717 2,181 2,486 88.7 90.7 90.8 90.3 90.7 1.7 2.2 3.1 5.0 5.3 2.4 1.5 1.0 .5 .4 2.2 1.3 1.4 1.1 1.2 1.7 1.3 1.0 .7 .6 1.6 1.8 1.8 1.7 1.5 1.7 1.1 .8 .6 .4 100.0 100.0 100.0 100.0 100.0 3,916 3,986 4,031 4,174 4,322 4,800 6,900 8,200 8,400 7,439 3,700 2,900 2,700 2,900 2,900 1,038 1,421 1,458 1,480 1,480 2,720 2,760 2,800 3,010 3,110 894 1,120 1,065 1,026 1,026 2,240 2,300 2,250 2,460 2,490 3,769 3,915 4,031 4,278 4,386 1950- 1955- 1960- 1965- 1970- 1955 1960 (Annual 1965 rates) 1970 1973 .2 .6 .6 .6 1.2 .5 .3 .8 78 COOID IDCOr- CO ID cn • cd I-. . CO r~^ co <* en t-» i— >X> IX) co • -co - "00 - -v»- mCMU3 CO i — o r— CSJ r— «cr r^ -ix> i — I— <*■ Cn CT) CD o CO i— CD IT) IX) CO i — i — co ix) r~- en en vf- >x> r-» CO CO IX) en csj i— ■ o en • co i — CO i r— >— o «* en • r-~ i — - IX) - - O - -co ID 1 — CO CM i — 00 l— C\J r— CSJ en en i — IX) IX) - o <* iX) r^. • o IX) • LT> IX) 1^. 1 *■ — s en o r*- r~- CSJ r— r— o • CSJ c~- (/) i — -*3- - - o - »>* IX) C\J CSJ CO 00 O 00 CSJ CSJ en en ■4-> r— vt" IX) -IX) r ~ 1 ' to ^i- r-~- oo csi i — i 1 IX) Lf) lO en en • — > — i — i— fO CJ> CSJ *d- co >— csj co co «* IX) ix> • i— CSJ . o vi- r»- i en en o r— vt «a- r». IX) • CO CSJ i — -oo - - en - -csi IX) IX) en i — ld r-. i — en csj csj en CTi IX) «3- IX) i — i — i — IX) IX) en en o a> ■r- cc +-> (O E Qi 3 CJJ >> 4->i— fc- 3 O o CL S_ c +-> -4-> o 3 0) u oa. UJ :d en i — en 1— c - +-> ■O o CJJ T3 3 1— >— Ci— XI ID U o <0 X +-> h CLJ O c •r- — O +J C7> •(— O-C +-> E-- m 3 S- i — l/) 3 CI) C-(-> Csi O U C_) (O c M- •1— >)3 CJ) c S- (0 cus: IX) en lOOr- CTi • en ix) en -<* - co r— r-» IX) CSJ 00 (X) oo • **■ cti ix) r^ -c3- - r-- i— «d- IX) IX) IX) IX) en . co IX) IX) IX) - r-~ - I— IX) r— "* IX) IX) • i— <*■ on IX) IX) en en 1 — i — CO o I-- i — CSJ I 1 vf ro IX) r^ en cn 1 — i — LT> IX) IX) CO O • -CO co csj "3- <3- IX) [-» en en vt- vr ix) r-- CT) CT) c/) o S- +J l/l (3)3 - ^^ o. =1 +J 1— -l-> 3 CQ 3 O +-> 4-> ZD 0)31- Z O CO Q CD CJJ o s- o o. cn CD ■ CD O O- S- CD CD CD (/) Q_ - Q ID CD cn C ■a CQ Q- CD +-> CD CJJ -*. s +-> 4-> U O 1 s: £ S- U cn ^ CO o c cn en c c S- S- U <_) rO to 14- 14- +J o_ at Q c C^ en^) cr t— r* ca S~ ~1 +j +J cu (O o Table 20 SIC Ind. No. Manufacturing Sector and Industry Groups Net Energy Consumption, Output, and BTU/Output Ratios Title 1954 1958 1962 1967 1971 1974 20 Food and kindred products BTU's (1012).... 805 766 802 899 1,025 958 Output (gross product, billion 1972 dollars) 16.8 BTU/Output 47,928 2(5 Paper and allied products BTU's 648 Output 5.4 BTU/Output 120,056 28 Chemicals and allied products BTU's 1,276 Output 6.9 BTU/Output 184,986 32 Stone, clay, and glass BTU's 905 Output 5.8 BTU/Output 156,034 33 Primary metals BTU's 2,873 Output 17.2 BTU/Output 167,035 Subtotal - Energy intensive industries BTU's 6,507 Output 52.1 BTU/Output 124,894 All other industries (ex- cluding petroleum refining) BTU's 1,475 Output 93.7 BTU/Output 15,741 Total manufacturing (ex- cluding petroleum refining ) BTU's 7,982 Output 145.8 BTU/Output 54,746 Addendum 29 Petroleum refining and coal products BTU's 1,761 2,013 2,304 2,538 2,816 2,730 Output 3.8 4.3 5.4 6.2 7.4 7.8 BTU/Output 463,421 468,140 426,667 409,355 380,541 350,000 18.7 40,963 20.1 39,925 24.2 37,178 26.2 39,145 27.6 34,739 806 5.8 139,017 928 7.0 132,671 1,156 8.4 137,655 1,315 9.8 134,183 1,327 11.2 118,509 1,541 8.5 181,400 1,876 11.1 169,090 2,460 15.8 155,703 2,778 19.8 140,308 2,936 22.6 129,942 945 6.2 152,419 1,056 7.2 146,667 1,229 8.4 146,309 1,304 8.9 146,517 1,332 9.6 138,750 3,049 15.9 191,761 3,544 17.6 201,364 4,079 23.4 174,316 4,026 19.9 202,311 4,310 25.1 171,713 7,107 55.1 128,984 8,206 63.0 130,254 9,823 80.2 122,481 10,448 84.6 123,499 10,863 96.1 113,039 1,489 93.9 15,857 1,758 117.8 14,924 2,250 167.7 13,417 2,666 172.0 15,500 2,635 192.9 13,660 8,596 149.0 57,691 9,964 180.8 55,111 12,073 247.9 48,701 13,114 256.6 51,107 13,498 289.0 46,706 80 r-^ r--~ cr> cr> CTi ■ — LO CO co oo ^t cxi i— ir> i i i i i o r~^ i — co r-~ en cr> miDiOOCO i— I CXJ O CO en c CXI I>> MD CD cn en 1 1 ^MOOCO o cxi I CXI I CXI I "vl- CXI it- CD en CD ^— i — CO woo) o s- +-> •r- +-> C 3 3 -r- c CL"0 E +-> rs 0) CD > CO CXJ LO CO en CTi 1 — 1 — 1 CO U~> LD CT. CT) 1 — ■ — CD CXJ 1^. O CXI I I I coixunoLn OO CO I I CO CXI LT> I CO CXJ CXI I CO ld r^ en en CD >— CO CD I co I -M o rt3 C re o CO CO "O 4J +-> O O O S- 3 3 aw "O "O to O OH BJ S_ S_ CU r— CL O-T- CD -O -O r— CU CU CO s- •!- -a i— -a c r— c •i— 03 (O "a oo -o c i— C (0 n3 (O CJ S- -r- XI (U E o am our u_ o_ o X) c f0 to >,4-> (13 CU ■— E a CU "3 C E o t- 4-> S- CD S- CD to I cu i— si n3 +-> 4-> tO O =5 +-> -o .a c =5 -r- c/i T3 rs CJ CU CD CD to CD •i- s_ c CO ■a o +J 4- (_) CU (t3 S- 4- rs E c rs f0 CU E i— o i— S- ro 4-J +-> CD O Q- O -o • »— i C O IOWZ O -o o a. _o +■» w 0) E m o iv 10 .J o <* 5 2 (U o _l Q Q ■ CD i_ C i — 0) Q_ 4- 1- CQ O s. CM < = a> O CO < 3 i ^ .— co i_ _^ a. ^ i | o Q S-