Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Electricity and Productivity Growth* I} . _ _Le_ . . 1 . . ........... .. . _ Technology ~ . By. . ~Price of I ncome _ ~ Electricity 51~£~1W Substitute Fuels Using Devices , ,,, i , Residential it' ~_ _ , _ .. ... . . .. ....... ,,,,, ........ . . . ~ . '! ~.Std ~! , ........................................................................ SU PP LY T Gross Nat or ~1: Product : | Consumption | DEMAN D The objective of this chapter is to analyze the role of electricity in the growth of productivity. The chapter touches on the shaded portions of the above reproduction of Figure 1-1. The concept of productivity figures prominently in analyzing economic growth. The relationship between electricity and *Much of the content of this chapter is based on Jorgenson (1984) and is incorporated here because of its special relevance to the co~runittee's task. 57
58 productivity began in the early decades of this century with the widespread electrification of many industrial processes, as described in the preceding chapter. Beginning in the 1970s, the decline in growth of the world's major economies, which followed the increase in prices of all forms of energy, demonstrated forcefully that the role of energy in economic growth should be more fully evaluated. Statistics for the United States show that decline in aggregate productivity growth contributed importantly to the decline in aggregate economic growth. Aggregate productivity rests on productivity of individual sectors of the economy, and these sectoral productivities are amenable to analysis. By adopting an econometric model of some generality, it has been possible to estimate empirically the quantitative dependence of sectoral productivity growth on technical change and the prices of the several inputs to production, including electrical and nonelectrical energy. For many industries, technical change is found to increase the shares, relative to those of other inputs to production, that given changes in electrical and nonelectrical energy input values contribute to change in output value. For such industries, technical change is said to be "electricity using" and "energy using," that is, it tends to increase the relative shares of electricity and nonelectrical energy in the value of output. For the same industries, lower prices of these inputs, in association with technical change, are found to enhance productivity growth. The significance of this analysis is that it provides an interpretation of the recent decline in economic growth in terms of higher energy prices associated with the Arab oil embargo of 1973. In addition, for the purposes of this study, the analysis provides increased insight into the interaction of electricity and economic growth and suggests areas for further research. The material of this chapter helps support one of the principal conclusions of the study, namely: o Productivity growth mav be ascribed partly to technical chant; in many industries technical chance also tends to increase the relet ive share of electric ity in the value of output, and in these industries productivity Growth is found to be the greater - the lower the real price of electric) ty/ and vice versa. THE CONCEPT OF PRODUCTIVITY Productivity means output per unit of input. In this sense, product) vity corresponds to the engineer' s concept of efficiency. Confusion may arise in characterizing productivity unless the measures of output and input are clearly specified. At the level of- individual industries in the economy, output is often expressed in physical units. For example, steel need not be measured in terms of monetary value added to iron--an economist's abstraction--but may be measured simply in tons. The output of the motor vehicle industry may be measured in numbers of vehicles produced. Similarly, the output of
59 the petroleum industry may be measured in barrels of petroleum, and so forth. It is also convenient to represent output by the monetary value (that is, the product of quantity and unit price) of the physical product. In fact, such a representation is needed if the outputs of diverse industries are to be compared. Inputs may be measured in physical quantities also, but to be compared they also have to be expressed in terms of their val ues. Since the output of production results f rom various input f actors such as capital, labor, and energy, it is possible to define a partial productivity with respect to any one input. For example, labor productivity is def ined as output per unit of labor input, the measures of both being specified, such as dollar value of goods produced per employee-hour worked. Energy productivity and electricity productivity may be defined similarly. In fact, electricity productivity, measured in constant dollar value of output per kilowatt hour of electricity input, is just the reciprocal of the quantity electricity intensity, introduced in Chapter 2. Total factor productivity is the ratio of some measure of output to some measure of all inputs--capital, labor, energy, materials, and so forth. Economists analyze the growth of output at the sectoral level in terms of the contributions of capital and labor inputs to a sector and the contributions of inputs to that sector produced by other sectors. Inputs produced by other sectors include both the raw materials and the energy that are produced by any given set of businesses and supplied to other sets. Growth of output also results from improvements in productivity. The idea of productivity growth at the sector al ~ evel is close to the engineering concept of an increase in eff iciency, and it is an easy idea to apprec late intuitively . Output, measured by its monetary value at producers' prices, say, may be considered a function of the various inputs, again measured in terms of their values. Fractional growth of output is allocable to contributions f ram the growth of each input, plus a contribution ascribed to productivity growth. Productivity growth may result from substitution of a cheap input for an expensive one to achieve the same measure of output for a smaller total measure of input. Productivity growth may also be achieved through technical change that of itself increases output or decreases input. Of course, substitution and technical change may occur simultaneously. Output at the level of total economic activity is given as gross national product (GNP), measured in dollars. Capital and labor inputs are the so-called primary factors of production that generate the whole of economic activity. To decompose economic growth (that is, percentage change in GNP) into its sources' we allocate growth among three components. The percentage growth of an economy is a combination of the percentage growth in productivity and the contributions of growth in capital input and labor input. Growth in capital input represents the increased stocks of capital equipment and structures that result from
60 investment. Growth in labor input represents an increase in the labor force, in hours worked per employee, in the education and experience of the labor force as ret lected in higher wage rates, or any combinat ion of these. As a matter of interest, in the United States the most important source of economic growth is the contribution of capital input. Growth in capital input accounts for about half of the growth that has taken place. The contribution of growth in labor input is the least important because of the stability of the labor force. The magnitude of productivity growth falls in between. Gains in the efficiency of production at the industry level will accumulate in the economy as a whole to provide greater growth of output than can be accounted for by the growth in both capital and labor inputs. Thus, productivity g rowth for the whole economy is def ined as the residual at ter accounting for the contributions of g rowth in capital and labor inputs to the g rowth of output. In engineering terms, productivity growth at the level of the entire economy may be thought of as the aggregate increase in the eff iciency with which economic resources are used at lower levels. THE: BACKGROUND The special signif icance of energy in economic growth was f irst established in the classic study, Energy and the American Economy, _850-1975: Its History and Prospects (Schurr et al., 1960) . Although this study covered only the United States, the experience of other industrialized countries is similar in many ways. In this study Schurr and his colleagues noted that, between 1920 and 1955, the energy intensity of production (defined as energy consumed per unit of GNP, and hence the reciprocal of energy productivity) fell in the United States, while both labor productivity and total factor productivity were rising.* The simultaneous decline in energy and labor intensities of production ruled out explaining the growth of productivity solely by the substitution of cheap energy for expensive labor. To explain the growth of output given declining energy and labor intensities required examining the character of productivity growth, engendered largely by technical change. Such an examination was further suggested by the fact that from 1920 to 1955 the use of electricity had grown more than 10-fold, while the consumption of all other forms of energy only doubled. The two most important features of technical change concerning electricity during this time were, first, that the thermal efficiency of conversion of fuels into electricity increased by a factor of three and, second, that "the unusual characteristics of electricity had made *This discussion is also based on Schurr (1983~. Berndt (1985) analyzed energy intensity and productivity growth in U.S. manufacturing for 1899 to 1939.
61 it possible to perform tasks in altogether dif ferent ways than if those fuels had to be used directly" (Schurr, 1983, p. 205~. Schurr emphasized the impact of the electrif ication of industrial processes, yielding much greater flexibility in applying energy to industrial production.* The importance of electrification in productivity growth was also documented by Rosenberg (1983~: Increasingly, the spreading use of electric power in the 20th century has been associated with the introduction of new techniques and new arrangements which reduce total costs through their saving of labor and capital. Perhaps the most distinctive features of these new techniques are (1) that they take so many forms as to defy easy categorization, and (2) that they occur in so many industries that they defy a simple summary. Rosenberg illustrates this point with examples drawn from the production of iron and steel, glass making, and the production and use of aluminum. Rosenberg, like Schurr and his associates, draws attention to the signif icance of electrif ication of industrial processes that was taking place during the first several decades of the century. Notably, electrical motors provided greater f legibility in supplying power to industrial processes and in organizing and physically arranging them. Rosenberg (1983, p. 295) reaches the following general conclusion concerning technical change that may rely more on electr ic ity and less on labor: It seems obvious that there has been a very wide range of labor saving innovations throughout industry which have taken an electricity using form. As a consequence, greater use of electricity is, from an historical point of view, the other side of the coin of a labor saving bias in the innovation process. Schurr (1982, 1984) recently extended the analysis of Energy and the American Economy, 1850-1975 through 1981. In this analysis, his assessment of the period 1953 through 1969 was as follows (1982, p. 61: Although the inverse relationship between total factor productivity and energy intensity virtually disappeared during the 1953 to 1969 period, it is still noteworthy that high rates of improvement in total factor productivity were essentially not associated with increases in energy intensity. *Recall f rom the discussion in Chapter 1 that electrification refers to the adoption of processes and activities based on the use of electricity. The term alone does not necessarily imply increased (or decreased) elect r ic ity consumpt ion .
62 In his later analysis Schurr also assessed the experience of the U. S. economy in the at termath of the oil embargo of 1973. He points out that the energy intensity of production has fallen steadily si nce 1973 and that the rate of decline accelerated sharply at ter the second oil price shock in 1979, following the Iranian revolution. He then makes the following point ~ 1982 , p. 10~: While energy productivity has been improving at a very high rate during the past decade, the overall productive efficiency side of the story has been highly unfavorable, and has become a matter of great concern. The post-1979 years that witnessed a new high in the rate of growth of national energy productivity also saw a decline in productive efficiency with a fall in total factor productivity of about 0.3 percent per year between 1979 and 1981. We can summarize this evidence about the relation between energy intensity and productivity growth by saying that energy intensity was falling while productivity was rising between 1920 and 1953. Between 1953 and 1969 energy intensity was relatively stable, while productivity continued to rise. After 1973 energy intensity resumed its downward trend, dropping faster af ter 1979, while productivity growth fell beginning in 1973, and turned into actual productivity decline at ter 1979. In exploring the determinants of trends in energy intensity and productivity growth, a useful framework is provided in the original study by Schurr et al. (1979), Energy in America's Future. This study emphasizes the role of change in the composition of national output, trends in energy intensity for industrial sectors, the significance of changes in the form of energy employed, and the role of energy pr ices . Chapter 2 of thi s report gave some attent ion to these points. Focusing on developments from 1975 through 1977, Schurr and his associates conclude that changes in the composition of national output offer "a useful but, at best, limited insight" (p. 88~. They also f ind that energy intensity has declined in some sectors and risen in others (pp. 89-90~. They find that the transformation of energy forms, especially in the direction of greater electrif ication and the use of fluid forms of energy such as petroleum and natural gas, has played an important role in economic activity: "ESuch] changes have made possible shifts in production techniques and locations within industry, agriculture, and transportation that greatly enhanced the growth of national output and productivity" (p. 92~. Finally, they argue that "quite apart from energy prices, technology developed its own momentum" ~ ibid. ~ . The framework suggested by Schurr and his associates and the historical evidence on trends in energy intensity and productivity growth suggest that explaining these trends must encompass a wide range of determinants. First, the gradual decline in real energy prices through the early 1970s and the sharp increases in energy prices that followed the oil shocks of 1973 and 1979 suggest an important role is played by the substitution that may occur between
63 energy and other productive inputs, especially labor input. While 'she real pr ice of labor input rose steadily during the early 1970s, this price has been declining since that time. These price trends would suggest that substitution of energy for labor occurred during the early 1970s and that substitution of labor for energy occurred thereafter. Second, productivity growth is an important element in explaining trends in energy intensity. In this regard, Schurr reviewed U.S. experience through 1969 as follows (1982, p. 91: The net result, then, was that strong improvements in both energy productivity and overall productive eff iciency were achieved without any special efforts being made to bring about this desirable combination of circumstances. Energy was abundantly available, and its price was low and, for the most part, falling during this period. Simple economic reasoning would tell us that the intensity of energy use should have risen because favorable energy pr ices would have encouraged energy consumption. But even though energy use rose relative to labor inputs, it fell in relationship to the f inal output of the economy. Did this decline in energy intensity take place in spite of low energy pr ices, or somehow becau se of them? The mechanisms of productivity growth, as described above by Schurr and by Rosenberg, indicate a specific role for electrification. Further understanding should result by analyzing the roles of the prices of both electricity and nonelectrical energy in determining productivity growth. THE RECENT DECLINE IN ECONOMIC GROWTH To assess the ef f ects of energy pr ices on economic g rowth, we beg in with a brief review of -several decades before the f irst oil crisis. World Economies* Rapid economic growth in the industrialized countries through 1973 has resulted in unprecedented world economic prosperity. An extreme example is provided by the Japanese economy, which between 1960 and 1973 grew at the astonishing rate of 10.9 percent per year. This g rowth quadrupled Japan ' s GNP, moving Japan f rom the ranks of the developing countries to its current status as a major industrial power. The largest industrialized economies of Europe participated fully in the great economic boom of the 1960s and early 1970s. The GNPs of France and Germany grew at 5.9 and 5.4 percent per year, respectively, *This section is based on Christensen et al. (1980, 1981), who compare patterns of economic growth in industrialized countries.
64 between 1960 and 1973. This rapid growth in Germany followed the "economic miracle" of 1952 through 1960, when Germany's GNP grew at 8.2 percent per year, exceeding even Japan's GNP growth at 8.1 percent per year during that time. From 1960 to 1973 Italy's GNP grew at 4.8 percent per year, while the United Kingdom's rate was a respectable 3.8 percent per year. The leading industrialized countries of Europe more than doubled their GNPs after World War II. By comparison, in North America the U.S. GNP grew at 4.3 percent per year from 1960 to 1973, and Canada's GNP grew at 5.1 percent per year during that time. In Europe the rapid economic growth took place with negligible growth in hours worked, while in North Amer ice hours increased at approximately 1.5 percent per year. The 1960s and 1970s also witnessed rapid growth among developing countries; GNP growth rates greater than 5 percent per year were not uncommon. Korea provided another extreme example, its GNP growing at 9.7 percent per year between 1960 and 1973, so that this country's economic expansion almost matched that of Japan. The impact of the first oil crisis on economic growth in industrialized countries was disastrous. GNP growth in the member countries of the Organisation for Economic Co-operation and Development (OECD) as a whole plummeted to 2.6 percent per year from 1973 to 1979. GNP growth in the United States dropped slightly less than the OECD average. GNP growth in Japan fell from its double digit rates of the 1960s and early 1970s to 3.9 percent per year--to almost the same rate as that of the United Kingdom, the slowest growing of industrialized countries f rom 1960 to 1973. The rate of GNP growth in Germany fell to 2.4 percent per year for the period 1973 to 1979, while GNP growth in France during this period was only 3.1 percent per year. The U. S . Economy To analyze in more detail the decline in U. S. economic growth following the f irst oil crisis, we can begin by decomposing the growth of output for the entire economy into the contributions of capital input, labor input, and productivity growth. * The results are given in Table 3-1 for the postwar period 1948 through 1979 and for a number of subperiods, each chosen because it represents a major business cycle. The first part of Table 3-1 provides data on growth in output and in capital and labor inputs. The second part of the table gives the contributions of capital and labor inputs to output growth. The third part of the table presents a decomposition of the rate of productivity growth for the U.S. economy as a whole. This growth rate is a weighted sum of productivity growth rates at the level of individual industrial sectors and the contributions of reallocations of value *An analysis of the slowdown in productivity growth in industrialized countries is given by Lindbeck (19831.
65 u) c~ ~ o o o u, :D 41) 1 S . O tD V ~ tD c ~ ~ u: . - ·~4 v o · - ~ o ~ r~ s :, o a' :> .~d V: 0 a) · - =; 1 h' ~ ~ V E~ u: ~*a:,r ~*r ~c'3 a' ~kD ~r ~0 O* ~O ~oooo oOOOOOooOo ·eeeee·e·~ OOOOOOOOOO 1 11 kD 0\ kD r~ 0 ~D ~ a~ cr. cr. ~r ~ cn a' ~a - * ~O~ OOOO ··e·a OOOOO - 1'_L1~00 O*a'co0 ~O OOOO ···e OOOO O(Das0 ~ U ~u ~(D OOOCOOO ··*.·.~ OOOOOOO ~D KD OO ·. OO l~a~ O ~r~ OO ·~ OO - *kD 0\~ ~U. OO ·. OO OUt O~ ~O OO ·* OO t~ O~ OO ·~ OO Ot~ kD~ ~O OO ·* OO ~D o o . o O(D O OO ·* OO vC) ~O OO *~ OO Ir'~ O OO ·~ OO 0~LS-t FOO OOOO OOOO ··e* OOOO 11 O O~ OOOO oooO ···~ OOOo 11 a' o o o U~ , - o . o r~ o . o C~(D oo~ oOo ooo ··~ ooo 1 03a'0 o~ ooo ooo ··~ ooo 1 cr) ~r~r~u~ o ~u3,-o r~0 ~000 ~ oooooo _ ··**·. oo ~ooo 11~ ~* ~C:) O ~e 1 ~O ~cr' ~- cc C' =\ t~) ~,-4 0 r~t O O ~ ~O O O O O O O CJ~ ~ ~ ~O O O O O O O , -L1~ O OO OO ·~ OO 1 o U, C o V V w V ~4 ·~( tJ ~W .L ~W ·r~ ' ~>q C - C U) ::> 0 ~ ~O ~ ~0-4 V V ~4 ~ · - ~4 V ~ 0) ~:' · - ~ ~C ~ v v r ~ ~ ~ 3 ~ '- Q ~C ·' ~0 ~4 ~ ~O ~ O '~4 O 04 3 ( ~C_ ·_4 _ · ~·~4 ~ ~·~4 ~ ·~4 ·~4- ~ c ~ · ~V C ~ - ~ ·- O ' ~V ~ ~ ~V ·- · - ·~ ~·~4 ~ S · - ~ ~·~- ~ 0 S ~ ~ ~h Q, V L4 ~ 3 ~v ^ v Q ~ L4 V ~ ~` ~C 3 {--'4 t- Q 0 ~ ~= ·~-- ·4 £ ~Q~ ~° ~ 0 ·'4 0 V _' ~) Q4 =) S C ~ ~4 ~ - ~ O C r~, - ~ ~ ~·. ~ 3 ~3 ~3 ~ ~ ~· - ~ ~4'~ ~O V ~4 V O ~ ~O ~V Q4 ~' - ~ '- '- 3 ~4 ~4 ·4 ~V O O ' - ~C' ~C' O ~C ~- rJ ~ cn c:: =; c:: ~ `4 C a ~O C ~C: ~u~
66 added, capital input, and labor input among sectors to productivity g rowth for the economy as a whole. We have taken value added by capital and labor inputs to be the measure of output for the aggregate U.S. economy. Between 1948 and 1979, aggregate value added grew at 3.44 percent per year, while capital input grew at 4.04 percent per year, indicating that the ratio of capital to output rose during that time. By contrast, over the same time labor input grew at only 1.48 percent per year, and the productivity growth rate was 0.90 percent per year. The average growth rate of value added reached its high at 4.83 percent per year during the period 1960 to 1966; it grew at only 2.92 percent per year during the recession and recovery of 1973 to 1979. The growth of capital input was stabler, exceeding 5 percent per year from 1948 to 1953 and 1966 to 1969 and falling to 3.78 percent per year from 1973 to 1979. Growth of labor input reached a high of 1.99 percent per year in the period 1960 to 1966, falling only to 1.97 percent per year from 1973 to 1979, a value well above the postwar average growth rate. Finally, the productivity growth rate was at its high from 1960 to 1966, at 1.80 percent per year. In the following period, from 1966 to 1969, the productivity growth rate was almost negligible at 0.08 percent per year. This rate recovered during 1969 to 1973, rising to 0.78 percent per year; finally, the rate of productivity growth fell to 0.19 percent per year between 1973 and 1979. To provide additional perspective on the sources of U.S. economic growth, we next analyze the contributions of capital and labor inputs to the growth of value added. The contribution of each input is equal to the product of its growth rate and the average value share (or weight) of the input in value added.* Since the average subperiod value shares of capital and labor inputs remained fairly constant between 1948 and 1979, the changes of these contributions among subperiods largely parallel those in the growth rates of capital and labor inputs. For the entire period 1948 through 1979, the contribution of capital input, at 1.71 percent per year, is the most important source of growth in aggregate value added. The productivity growth rate is the next most important source, at 0.90 percent per year, while the contribution of labor input is the third most important source, at 0.84 percent per year. The contribution of capital input is the most important source of growth during six of the seven subperiods, all those but that from 1960 to 1966, during which time the productivity growth rate is the most important source of economic growth. The decline in the growth rate of aggregate value added between the two periods 1960 to 1966 and 1966 to 1969 appears to result primarily *Analytically, if In G = vK In K + vL In L + In P. where G is aggregate value added, K is capital input, L is labor input, P is productivity, and vK and vL are the value shares for capital and labor inputs respectively, then d In G = vK d In X + vL d In L + d In P.
67 from a dramatic fall in the rate of aggregate productivity growth between these periods. The growth of capital input actually increased, while the growth of labor input declined only slightly. The revival of productivity growth during 1969 through 1973 was offset by declines in the growth of capital and labor inputs, leaving the growth rate of value added almost unchanged. The productivity growth rate declined gain between 1973 and 1979. Thus the deal ine in g rowth of value added s ince 19 66 has been associated with productivity growth rates that are the lowest of the postwar per iod. U. S. Productivity Growth As noted above, the productivity growth rate for the U. S. economy as a whole can be decomposed into four components--a weighted sum of the rates of sectoral productivity g rowth and reallocations of value added, capital input, and labor input. The weights ref. lect the contr ibution of productivity g rowth in each sector to the 9 rowth of output in that sector . The weights also ref lect the cant r ibut ion of productivity growth to the growth of inputs to each sector that are produced by the other sectors. The contribution of the reallocation of value added to aggregate productivity growth involves the redistribution of value added among sectors from low value to high value components of output. Similarly, reallocations of capital and labor inputs involve the redistributions of these inputs among sectors from low remuneration to high remuneration uses. For the entire period from 1948 to 1979, sectoral rates of productivity growth account for almost all of the rate of aggregate productivity growth. The reallocation of value added is 0.21 percent per year, while reallocations of capital and labor inputs are -0.05 and -0.09 percent per year, respectively. The collapse in the rate of aggregate productivity growth after 1966 resulted from a drop in the weighted sum of sectoral rates of productivity growth from 1.62 to 0.13 percent per year from the period 1960 to 1966 to the period 1966 to 1969. Between 1969 and 1973 sectoral rates of productivity growth recovered to 0.44 percent per year; the most important contribution to reviving the aggregate productivity growth rate between those two periods was the increase the reallocations of value added from 0.11 percent per year in the period 1966 to 1969 to 0.48 percent per year in the period 1969 to 1973. Between 1973 and 1979 the weighted sum of sectoral rates of productivity growth declined to -0.72 percent per year. To summarize these findings about the decl ine in U. S . economic growth during the past decade, we can see that this decline took place in two steps. First, productivity growth at the sectoral level essentially disappeared as a source of economic growth after 1966. A very sizable decline in sectoral productivity growth rates began in
68 the period 1966 through 1969 and persisted through 1973. Second, between 1973 and 1979 sectoral productivity growth rates plummeted. Whatever the causes of the decline, they are to be found in the collapse of productivity growth at the sectoral level rather than in decline in the growth of capital and labor inputs at the aggregate level or in the reallocations of value added, capital input, or labor input among sectors. However, our measure of sectoral productivity growth is simply the unexplained residual between growth in sectoral output and the contributions of sectoral capital, labor, energy, and materials inputs. The problem remains of explaining the fall in productivity at the sectoral level. THE: ECONOMETRIC MODEL Our general conclusion from reviewing postwar U.S. economic history is that understanding the role of energy in productivity growth or decline requires analyzing productivity growth by individual industrial sectors. As Rosenberg and Schurr have indicated, this analysis should encompass the substitution of electricity for other forms of energy. Schurr has suggested that the relatively lower electricity prices (compared to those of other forms of energy) that have accompanied the dramatic increases in thermal efficiency of electricity generation have been an important force in electrif ication. The relative price reduction and ensuing electrification have accelerated the productivity growth rate through innovations in many industrial activities. A Model of Sectoral Productivity Growth The Form of the Model To assess the role of energy in stimulating productivity growth, it is necessary to do more than merely describe the trends in energy use and in productivity. For this purpose we employ an econometric model of sectoral productivity growth. Details are given in Appendix C. In assessing the signif icance of changes in the form of energy employed, we identify the inputs for each sector as capital, labor, electricity, nonelectrical energy, and materials. Our econometric model treats the substitution among productive inputs in response to changes in relative prices. The model also determines sectoral productivity growth rates as a function of relative prices. For each industry our model of production is based on a sectoral price function (Appendix C, Equation 1) that encompasses possibilities for substitution among inputs as well as patterns of technical change. Each price function gives the price of output for an industrial sector as a function of the prices of capital, labor, electricity, nonelectrical energy, and materials inputs and, in addition, time; in this formulation time represents the level of
69 technology in the sector (Samuelson, 1953~. Thus the passage of time represents technical change. Obviously, when there is an increase in the price of one input, and the pr ices of the other inputs and the level of technology remain unchanged, there must be an increase in the pr ice of output. Similarly, if the productivity of a sector improves, say, through a change in the level of technology, and the pr ices of all sectoral inputs remain the same, tare pr ice of output must fall . Price functions summarize these and other relationships among the prices of output, capital, labor, electricity, nonelectrical energy, and materials inputs, and in addition the level of technology. Sectoral price functions provide a complete model of production patterns for each sector. We find it useful to express the model in two parts. First, we can express the value shares of changes in each of the five inputs--capital, labor, electricity, nonelectrical energy, and materials--in the change of output value as functions of the prices of these inputs and time, again representing the level of technology (Appendix C, Equations 3~.* Second, supplementing these five equations for the value shares, we can obtain an equation that expresses productivity growth rate as a function of the prices of all five inputs and time (Appendix C, Equation 4~. This equation is our econometric model of sectoral productivity growth (Jorgenson, 1985; a useful survey of studies relating energy prices and productivity growth is given by Berndt, 1982~. Parameters of the Model As in any econometric model, the relationships determining the value shares of capital, labor, electricity, nonelectrical energy, and materials inputs and the productivity growth rate involve unknown parameters (Appendix C, Equations 2) which must be estimated from data for the individual industries. Among these unknown parameters are the so-called biases of productivity growth (Appendix C, Equations 6), which indicate the effects of change in the level of technology (that is, technical change) on the value shares of each of the five inputs. (Hicks, 1932, introduced the concept of bias of productivity growth; an alternative definition was introduced by Binswanger, 1974a, 1974b. The definition employed in our econometric model is discussed by Jorgenson, 1985. ~ The biases of productivity growth for each of the f ive inputs appear as the coefficients of time, again, representing the level of technology, in the f ive equations for the value shares of each k ind of input. For example, the bias of productivity growth for capital input *Our sectoral price functions are based on the transcendental logarithmic (or translog) price function introduced by Christensen et al. (1971, 1973~. The translog price function was f irst employed for sectoral analysis by Berndt and Jorgenson (1973) and Berndt and Wood (1975~. Berndt and Wood (1979) provide references to sectoral production studies encompassing energy and materials inputs.
70 gives the change in the value share of capital input in response to change in the level of technology, represented by time.* We say that technical change, or productivity growth, which is proportional to technical change, is capital using if the bias of productivity growth for capital input is positive. That is, technical change would tend to increase the relative share of capital in the value of output. Similarly, we say that productivity growth is capital saving if the bias of productivity growth for capital input is negative. It is important to observe that the sum of the biases of all five inputs must be zero (Appendix C, Equations 2), since the changes in all five value shares with a change only in technology must sum to zero (from Appendix C, Equation 8~. That is, if productivity growth is electricity using, then productivity growth must be input saving in some other input. For example, productivity growth could be labor saving and electricity using, as suggested by the quotation from Rosenberg cited earlier. This last example would be represented by a positive bias of productivity growth for electricity and a negative bias of productivity growth for labor. Our econometric model, then, classifies each of the 35 industries of our study among the 30 logically possible patterns of productivity growth, that is, all the possible combinations of positive or negative values for each of the five biases. We can rule out only the possibility that all five biases are negative and that all five are positive, on the basis of purely analytical considerations. We have pointed out that our econometric model yields an equation (Appendix C, Equation 4), for each industrial sector of the U.S. economy, giving the sectoral productivity growth rate as a function of the prices of the five inputs and time (or level of technology). The same biases of productivity growth that appear in the value share equations also appear as coefficients of the prices in the equation for the sectoral productivity growth rate. This feature of our econometric model makes it possible to use information about changes in the sectoral value shares with time and about changes in the sectoral productivity growth rate with prices in estimating the biases of productivity growth. The model also contains alternative interpretations of the biases of productivity growth. The first interpretation is as the rate of change of the value share of each input with respect to time. This interpretation focuses on the biases of productivity growth as the coefficients of time in the equations for the value shares, that is, each bias is the weight of the level of technology in that value share (Appendix C, Equations 3~. The second interpretation is as the rate of change of the negative of sectoral productivity growth rate with respect to proportional input prices. This interpretation focuses on the biases of productivity growth as the coefficients of prices in the equation for the negative of productivity growth rate, that is, each *In other words, the bias of productivity growth for capital is the weight of the level of technology as one contribution to the value share of capital, and similarly for other inputs.
bias is the weight of the price of the corresponding input in the negative productivity growth rate (Appendix C, Equation 4~ . The two interpretations are equivalent, since the value shares and the productivity growth rate are both generated from an underlying price function for each sector. In other words, the biases of productivity growth express the dependence of the value shares of the five inputs on the level of technology and also express the dependence of the productivity growth rate on the prices. Thus, capital-using productivity growth, represented by a positive bias of productivity growth for capital input, means that an increase in the price of capital input decreases the productivity growth rate. Analogous relationships hold for the biases of labor, electricity, nonelectrical energy, and materials inputs. Modeling the Role of Electricity in Productivity Growth In assessing the role of electricity in productivity growth, the critical parameter in our econometric model is the bias of productivity growth for electricity (Appendix C, CITY. This bias gives the change in the value share of electricity in response to changes in the level of technology (that is, technical change). We say that productivity growth (or technical change) is electricity using if the bias of productivity growth for electricity is positive. Similarly, we say that productivity growth is electricity saving if the bias of productivity growth for electricity input is negative. Thus, to test the hypothesis that productivity growth is electricity using for a particular industrial sector, we evaluate the bias of productivity growth for electricity along with other parameters describing substitution and technical change in that sector. We then test the hypothesis that the bias of productivity growth is positive. The dual role of the bias of productivity growth--expressing the effect of technical change on the value share of an input and the effect of change in price of that input on the productivity growth rate--is central to our assessment . Hi stor ical evidence, as summer ized by Rosenberg, suggests that much of the innovation in the twentieth century is electricity using. That is, innovation increases the share of electricity in the value of output for a given set of input prices, including that of electricity. Entirely different evidence, analyzed by Schurr and his assoc fates , has linked the: reduction in the cost of electricity, f rom increasing thermal efficiency in electricity generation, to enhanced productivity growth. Within our econometr ic model these two instances of historical evidence are consistent with the hypothesis that the bias of productivity growth for electricity is positive. According to the model, electricity-using productivity growth (or technical change), associated with ~ positive bias of productivity growth for electricity, means that electricity use tends to increase as technology changes. This is what occurred during the first several
72 decades of this century, according to the evidence reviewed by Rosenberg. Electricity-using productivity growth also implies that the rate of productivity growth increases as the price of electricity declines, which is consistent with the historical evidence on the growth of output, productivity, and energy consumption analyzed by Schurr . This historical evidence suggests the hypothesis that technical change at the level of individual sectors of the U. S . economy is electricity using. This relationship implies that electricity plays a central role in productivity growth. It remains to test the hypothesis using our model. Results from the Model Applying the Model To apply our econometric model of production we assembled a data base for 35 industries in the U.S. economy (Jorgenson and Fraumeni, 1981~. These industries encompass all sectors of the U.S. economy. Manufacturing is subdivided among 21 industries at the two-digit level of the Standard Industrial Classification. These industries differ greatly in their relative importance to the economy and in their energy intensities of production. The complete set of industries also encompasses the primary production sectors of agriculture and mining; the energy-intensive transportation and public utilities industries; and the construction, communications, trade, and service industries. For capital and labor inputs we first compiled data by sector on the basis of the classification of economic activities employed in the U.S. National Income and Product Accounts. We then transformed these data into a format appropriate for the classification of activities employed in the U.S. Interindustry Transactions Accounts. Regarding electricity, nonelectrical energy, and materials inputs, we compiled data by sector on interindustry transactions among the 35 industrial sectors. For this purpose we used the classification of economic activities employed in the U.S. Interindustry Transactions Accounts.* The data on capital inputs employed in this study are based on estimates of the stock of capital goods for each of the 35 industrial sectors. For each sector we prepared estimates for corporate and noncorporate business, separated into four asset types--producers' durable equipment, nonresidential structures, inventories, and land. Stocks of capital goods, broken down by legal form of organization and type of asset are aggregated by property compensation for each type of capital. Estimates of property compensation incorporate data on rates . *Data on energy and materials are based on annual interindustry transaction tables for the United States, 1958 through 1974, compiled by Jack Faucett Associates (1977~. Data on capital and labor inputs are based on estimates by Fraumeni and Jorgenson (19841. These data have been updated through 1979 by Jorgenson (1984~.
73 of return for each sector, rates of revaluation of each type of asset, and variables that describe the tax structure for each type of capital input. Measures of capital input are obtained by aggregating over both legal forms of organization and all four types of asset for each sector. Our measures of labor input are based on data on hours worked and labor compensation for each industrial sector, cross-classified by sex, age, education, employment status, and occupation of workers. Control totals for each industry are based on surveys of business establishments. To disaggregate labor input by demographic characteristics we exploited the detail on employment, hours worked, weeks paid, and compensation available from household surveys. For each year the complete distributions of hours worked and labor compensation are broken down by 81,600 categories. Measures of labor input are obtained by aggregating over both sexes, eight age groups, five levels of educational attainment, two employment classes, and ten occupational categories within each industry. For each sector we compiled annual data on the value shares of capital, labor, electricity, nonelectrical energy, and materials inputs, for 1958 through 1979. We also compiled price indexes for sectoral outputs and all f ive sectoral inputs for the same period. Finally, we compiled transcendental logarithmic indexes of sectoral rates of productivity growth. There are 21 observations for each equation since unweighted two-period averages of all data are employed. The sample period contains changes in energy prices that were momentous by any standard. The data set includes 7 years, 1973 through 1979, out of a total of 22 years, 1958 through 1979, that can be characterized as a period of high energy prices. The results, as reported in detail by Jorgenson (1984), show very low standard errors on the key parameters relating energy prices to productivity growth. One reason for this precision is the sharp increase in energy prices and the equally sharp decline in productivity growth after 1973. Estimating the Biases of Productivity Growth We first consider the bias of productivity growth with respect to the price of capital input, using the dual interpretations explained previously. Technical change is capital using for 20 of the 35 industries i ncluded in our study; it is capital saving for the other 15. We conclude that the productivity growth rate decreases with the (increasing) price of capital input for 20 industries and increases with this (increasing) price for 15 industries. Interpreting the biases of productivity growth with respect to the prices of labor, electricity, nonelectrical energy, and materials inputs is analogous to interpreting the bias with respect to the pr ice of capital input. Considering the bias of productivity growth with respect to the pr ice of labor input, we f ind that technical change is labor using for 26 of the 35 industries and labor saving for 9 of
74 these industries. Hence the productivity growth rate decreases with the price of labor input for the 26 industries and increases with this price for the other 9. Considering the bias of productivity growth with respect to the price of electricity input, we find that technical change is electricity using for 23 of the 35 industries included in our study and electr icity saving for 12. Hence the productivity growth rate decreases as the price of electricity rises for the same 23 industries and increases as this price rises for the same 12. Turning to the bias of productivity growth with respect to the price of nonelectrical energy input, we find that technical change is nonelectrical energy using for 28 of the 35 industries and nonelectrical energy saving for only 7. We conclude that the productivity growth rate decreases with the price of nonelectrical energy for 28 industries and increases with this price for the remaining 7. Finally, technical change is materials using for 8 of the 35 industries and materials saving for the other 27. Thus, the rate of productivity growth increases with the increasing price of materials for 27 industries and decreases with this price for the- remaining 8. Patterns of the Biases of Productivity Growth Table 3-2 classifies industries by their patterns of biases of productivity growth. The most frequent pattern of the biases corresponds to technical change that is capital using, labor using, electricity using, nonelectrical energy using, and materials saving. This pattern occurs for 8 of the 35 industries. In this pattern the productive ty growth rate decreases with increasing prices of capital, labor, electricity, and nonelectrical energy inputs and increases with increasing price of mater ials input. The second most f requent pattern corresponds to technical change that is capital saving, labor using, electricity using, nonelectrical energy using, and materials saving. This pattern occurs for 5 industries. In this pattern the productivity growth rate decreases with increasing prices of labor, electricity, and nonelectrical energy inputs and increases with increasing prices of capital and materials inputs. These two patterns of the biases of productivity growth differ only in the role of the pr ice of capital input. The Hypothesis Concerning Electricity and Productivity Growth We have estimated biases of productivity growth with respect to prices of capital input, labor input, electricity input, nonelectrical energy input, and materials input. These biases are undetermined parameters of the econometric models for the 35 industrial sectors included in this study. To test the hypothesis advanced by Schurr and Rosenberg about the importance of electrification in productivity growth, we
75 TABLE 3-2 Classif ication of Industries by Their -Patterns of Biases of Produc t ivity Growth Pattern of Biases Capital using Labor us ing Electr ic ity using Nonelectrical energy using Materials saving Capital using Labor saving Electricity using Nonelectrical energy using Materials using Capital using Labor using Electricity using Nonelectrical energy saving Materials saving Capital using Labor using Electricity saving Nonelectrical energy using Materials saving Capital using Labor saving Electricity using Nonelectrical energy saving Materials using Capital using Labor using Electricity saving Nonelectrical energy saving Mater ials saving Capital using Labor saving Electricity saving Nonelectrical energy using Mater ials saving Industries Tobacco, textiles, apparel, lumber and wood, printing and publishing, fabricated metal, motor vehicles, transportation Electrical machinery Metal mining, services Nonmetallic mining, miscellaneous manuf acturing, government enterprises Construction Coal mining, trade Agriculture, crude petroleum and natural gas, petroleum refining
76 TABLE 3-2 Classif ication of Industries by Their Patterns of Biases of Productivity Growth (Concluded) Pattern of Biases Industr ies Capital saving Labor u s ing Electricity using Nonelectr ical energy using Materials using Capital saving Labor using Electricity using Nonelectrical energy using Mater ials saving Capital saving Labor using Electr ic ity saving Nonelectr ical energy using Mater ials us ing Capital saving Labor saving Electricity using Nonelectrical energy using Mater ials us ing Capital saving Labor using Electricity saving Nonelectrical energy using Materials saving Cap ital savi ng Labor using Electricity us ing Nonelectr ical energy saving Materials saving Capital saving Labor saving Electricity using Nonelectrical energy using Materials saving Capital saving Labor saving Electr ic ity saving Nonelectrical energy saving Materials using Food, paper Rubber; leather; instruments; gas utilities; finance, insurance, and real estate Chemicals Transportation equipment and ordnance, communications Stone, clay, and glass; machinery Primary metals Electric utilities Furniture
77 focus on the bias of productivity growth with respect to electricity input. Recall that if this bias is positive, then technical change is electricity using; if the bias is negative, technical change is electricity saving. If technical change is electricity using, the value share of electricity input in the value of output increases with technical change, while the productivity growth rate increases with a decrease in the price of electricity. We have found that technical change is electricity using for 23 of the 35 industries included in our study. Our first and most important conclusion is that electricity plays a very important role in productivity growth. A decline in the price of electricity stimulates productivity growth in 23 of the 35 industries and dampens productivity growth in only 12. Alternatively and equivalently, we can say that technical change results in an increase in the share of electricity input in the value of output, holding the relative prices of all inputs constant, in 23 of the 35 industries. Technical change results in a decrease in the share of electricity input again in only 12. Our empirical results provide strong confirmation of this hypothesis about the relationship of electrification and productivity growth in a wide range of industries. Schurr et al. (1979) have shown that the price of electricity fell in real terms through 1971. This decline in real electricity prices promoted electricity use through the substitution of electricity for other forms of energy and through the substitution of energy for other inputs, especially for labor. In addition, the decline in the real price of electricity stimulated the growth of productivity in a wide range of industries. The spread of electrification and the rapid growth of productivity through the early 1970s are both associated with a decline in real electricity prices. This decline was made possible in part by advances in the thermal efficiency of electricity generation. Beginning in the early 1970s the downward trend in the real price of electricity reversed. This reversal has been associated with a marked decline in advancing the thermal efficiency of electricity generation, a decline which began in the late 1960s. However, the diminishing rate of technical change in the electricity-generating industry only partly explains the reversing trend in real electricity prices. In addition, the prices of primary energy sources employed in electricity generation rose sharply following the oil price shocks of 1973 and 1979. Rising electricity prices, then, have slowed productivity growth in U. S . industr ies throughout the 1970s . These price increases play an important role in explaining the decline of U. S . productivity growth s ince 1973. Subs id iary Hypotheses In linking electrif ication and productivity growth, Schurr advanced an important subsidiary hypothesis, namely, that electrif ication is especially significant in stimulating productivity growth in the
78 manufacturing industries. Schurr's hypothesis is supported by the fact that technical change is electricity using in 15 of the 21 manufacturing industries included in our study, while technical change is electricity using in only 8 of the 14 nonmanufacturing industries. Schurr's explanation for this phenomenon is that the electrification of industr ial processes led to much greater f legibility in the application of energy. Rosenberg's examples of the importance of electrification--in iron and steel, glass, and aluminum production--are also drawn from manufacturing. Rosenberg advanced another subsidiary hypothesis in analyzing the link between electrification and productivity growth. This hypothesis is that electricity-using technical change is the "other side of the coin" of labor-saving technical change. We have been unable to find support for this hypothesis in our empirical results. In fact, technical change is labor saving for only 9 of the 35 industries and labor using for the remaining 26. However, we have pointed out that the sum of biases of productivity growth for all f ive inputs must equal zero. The predominance of electricity-using technical change therefore must be balanced by technical change that saves other inputs. We have found that technical change is materials saving for 27 of the 35 industries and materials using for only the remaining 8. For all other inputs, including labor and electricity, technical change is predominantly input using. We conclude that technical change that uses electricity input and inputs of capital, labor, and nonelectrical energy is balanced by technical change that saves materials. Nonelectrical Energy and Productivity Growth We have found that electricity plays an important role in productivity growth, and we have also examined the use of nonelectrical energy. Our findings are that technical change is nonelectrical energy using for 28 of the 35 industries included in our study and nonelectrical energy saving for 7 of these industries. A decline in the price of nonelectrical energy stimulates productivity growth in 28 of the 35 industries and dampens productivity growth in only 7. Correspondingly, we can say that technical change results in an - increase in the share of nonelectrical energy input in the value of output in 28 of the 35 industries and results in a decrease in nonelectrical energy input share for only 7. Again considering the evidence on energy price developments presented by Schurr and his associates, we find that the price of nonelectrical energy fell in real terms through the early 1970s, reaching a minimum for natural gas and fuel oil in 1970 and for gasoline in 1972. This decline in real nonelectrical energy prices promoted greater use of nonelectrical energy through the substitution of these forms of energy for capital, labor, and materials inputs. In addition, the decline in the real price of nonelectrical energy, like the decline in electricity prices we examined earlier, stimulated the
79 growth of productivity in a wide range of industries. We conclude that the greater use of nonelectrical energy in relation to other inputs such as labor and the rapid growth of productivity through early 1970s are associated with the decline in the real price of nonelectrical energy. Beginning in the early 1970s the downward trend in the real price of nonelectrical energy reversed, and the increase in use of nonelectrical energy relative to other inputs in U.S. industries slowed dramatically. This reversal in the trend of nonelectrical energy prices, as well as an important part of the reversal in the trend of electricity prices examined above, was associated with the oil price shocks of 1973 and 1979. Rising prices of nonelectrical energy have reinforced the negative effects of rising electricity prices on productivity growth throughout the 1970s. Increases in the real prices of both electricity and nonelectrical energy help explain the decline in U. S. . productivity growth s ince 1973 . In linking greater use of nonelectrical energy with productivity growth, Schurr et al. (1979) advanced another important subsidiary hypothesis: that greater use of fluid forms of energy has enhanced productivity in agriculture, transportation, and manufacturing. We find that technical change is nonelectrical energy using in agriculture and transportation, as suggested by Schurr and his associates. We also find that technical change is nonelectrical energy using for 19 of the 21 manufacturing industries included in our study. Technical change is nonelectrical energy using for only 7 of the 12 industries other than agriculture, manufacturing, and transportation. We conclude that greater use of nonelectrical energy has a significant role in productivity growth for an even wider range of industries than has the use of electrical energy. Summary We have now completed our analysis of the role of electrical and nonelectrical energy in productivity growth employing an econometric model of production. Given the f ramework of our model we can offer a tentative explanation of the disparate trends in energy intensity and productivity growth. These trends first drew the attention of Schurr and his associates to the special role of electrification. Between 1920 and 1953 energy intensity of production was falling while productivity was rising. While the fall in real prices of electricity and nonelectrical energy resulted in substitution of energy inputs for other inputs, especially for labor, these price trends also generated suff icient growth in output per unit of energy input that the energy intensity of production fell. This explanation is consistent with that advanced by Schurr and his associates. Between 1953 and 1973 energy intensity was stable, while productivity continued to grow. During this period real energy prices continued to fall, but at slower rates than between 1920 and 1953. As
80 before, the fall in real prices of electricity and nonelectrical energy resulted in the substitution of energy inputs for other inputs. Yet these increases were almost completely matched by the g rowth in output per unit of energy input, leaving the energy intensity of production unchanged. Finally, real energy prices began to rise in the early 1970s, increasing dramatically after the first oil shock of 1973 and again after the second oil shock of 1979. These price trends resulted in substitution of capital, labor, and materials inputs for inputs of electricity and nonelectrical energy, thereby reducing energy intensity of production. At the same time, energy price trends contributed to a marked decline in productivity growth. Although much research is still required to understand the role of energy use in productivity growth, our analysis has made progress toward that goal. We have analyzed the character of productivity growth in industries representing the whole U.S. economy. We have tested hypotheses advanced in earlier research, by Schurr and his associates and by Rosenberg, with empirical evidence and found support for the hypothesis that electrification and productivity growth are related. We have found that the use of nonelectrical energy and productivity growth are even more strongly related. Pursuit of this inquiry should provide a deeper understanding of the relationship between energy use and productivity change. Given support for the hypothesis that technical change is electricity using and nonelectrical energy using, we can assess the potential for electrif ication and greater use of nonelectrical energy in reviving productivity growth at the level of individual industries in the United States. Schurr has summarized this potential as follows (1982, p. 7~: If this line of theorizing is correct, one of the keys to reconciling the future growth of energy productivity and labor and total factor productivity would be (a) through the vigorous pursuit of these energy supply technologies which assure the renewed future availability, on favorable terms, of those energy forms which possess the highly desirable flexibility features that have characterized liquid fuels and electricity, and (b) through the search for counterpart energy consumption technologies that can put these characteristics to efficient use in industrial, commercial, and hou Behold appl icat ions . INTERPRETATION OF THE RECENT DECLINE IN GROWTH The sharp decline in economic growth in industrialized countries presents a problem comparable in scientific interest and social importance to the problem of mass unemployment in the Great Depression of the 1930s. Conventional methods of economic analysis that look only at aggregate changes in productivity have been tried and found inadequate. Clearly a new framework based on understanding changes at the sectoral level will be required. The findings we have presented
81 contain some of the elements for analyzing the prospects for the world economy in the last half of the 1980s. At first sight the finding that higher energy prices are an important determinant of the decline in economic growth seems paradoxical. In studies of sources of aggregate economic growth, energy appear s as both an output and an input for ind ividual industries, but cancels out for the economy as a whole.* It is necessary to disaggregate the sources of economic growth by sector to define the correct role of energy in economic growth. Within such a framework for analyzing economic growth, it is still not sufficient to decompose the growth of sectoral output among the contributions of inputs and of productivity growth. It is essential to explain the growth of sectoral productivity. In the absence of such an explanation the growth of sectoral productivity is simply an unexplained residual between the growth of output and the contributions of growth of capital, labor, electricity, nonelectrical energy, and materials inputs. Finally, the significance of energy prices for sectoral productivity growth must be determined empirically. From a conceptual point of view, energy prices can have positive, negative, or no effects on sectoral productivity growth. From an empirical point of view, the influence of higher energy prices has been negative and highly significant. This empirical finding can be substantiated only through an econometric model of productivity growth. The steps we have outlined--disaggregating the sources of economic growth by sector; decomposing the growth of sectoral output into productivity growth and the contr ibutions of capital, labor, electricity, nonelectrical energy, and materials inputs; and modeling the growth of productivity--have been taken only recently. Although much additional research is required to explain the decline of economic growth in industrialized countries, we find it useful to employ this framework in assessing future growth prospects for industrialized countries. We begin by comparing our methods with alternative approaches to energy demand forecasting. A gradual decline in real energy prices through 1973 provided a mild stimulus to the growth of energy demand. However, rapid economic expansion in the industrialized world and in the less developed countries provided by far the main source of energy growth. Forecasts of energy demand were based on projections of economic growth with little or no attention to energy prices. This method of energy demand forecasting prevailed up to the time of the first energy crisis in 1973. The Arab oil embargo of late 1973 and early 1974 resulted in a dramatic increase in world oil prices. Between 1973 and 1975 crude oil import prices increased by two and one-half times in real terms for the seven major OECD countries--Canada, France, Germany, Italy, *A leading proponent of this view is Denison (1984~. Comparisons of studies of energy and productivity by Berndt (1985) , Jorgenson (1984), and Schurr (1984) are given by Sonenblum (1985) and Wood (19851.
82 Japan, the United Kingdom, and the United States.* Japan was the country most affected by the oil price increases, experiencing a tripling of real crude oil import prices. Of European countries France was not far behind Japan in experiencing increases in the real price of crude oil imports. Real energy prices to final users increased considerably less than did real oil prices in all major OECD countries. The average increase for these seven countries from 1973 to 1975 was 23.9 percent. Japan and Italy were at the high end of the range, with increases in excess of 50 percent. Meanwhile, Canada experienced only a 3.9 percent increase under a regime of price controls on domestic petroleum and natural gas. Similar controls in the United States d id not prevent an increase of 23 percent in real energy prices to final users. The Iranian revolution beginning in late 1978 sent a second wave of oil pr ice increases through world markets . Between 1978 and 1980 crude oil import prices almost doubled in real terms for the seven ma jor OECD countries. Real energy prices to final users climbed by 33.5 percent for these countries. Again, Japan was hard hit with an 80.3 percent increase, while Canada experienced an increase of only 8.7 percent. In the United States, the real price increase was 34 percent, while major European countries had increases below the average. Energy analysts generally agreed that the great discovery emanating from the first oil crisis was the price elasticity of demand for energy. One of the f irst post-embargo pro jections of energy demand in the United States was provided by the Energy Policy Project. This projection featured two low energy growth scenarios--a Technical fix" scenario, exploiting available technologies to achieve energy conservation, and a "zero per capita energy growths scenario, featuring energy growth at the same rate as population growth. Although the low energy growth scenarios presented by the Energy Policy Project met with considerable skepticism in 1974, the zero per capita energy growth scenario actually overestimated U. S. energy consumption in 1980, only six years later, by 15 percent. By 1982 U. S. energy demand had fallen below the level that prevailed in 1972 before the f irst oil crisis. U. S. oil consumption fell even more dramatically, declining to 1971 levels by 1982. The world price of petroleum was taken to be exogenous by the Energy Pol icy Proj ect t Hudson and Jorgenson, 1974a, 1974b). Although the price increases associated with the first oil crisis were taken into account, the additional increases after the Iranian revolution were not included in the analysis. As a consequence, energy demands were overestimated. Methods for energy demand forecasting that take account of the price elasticity of energy demand became co~runonplace by the end of the 1970s. However, the new orthodoxy was itself overtaken by events. l *Fuj ime (1983) compares energy prices and energy demand patterns in industrialized countries. Hogan (1984) provides projections of U.S. energy demand.
83 The oil price increases that accompanied the Arab oil embargo presented ~ unique challenge to economic policymakers. Some policy analysts interpreted price increases from whatever source as inf let ionary . Careful students of open economy mac roeconomics po inted out the deflationary impact of an increase in the price of an imported commodity--oil. As the debate among policy analysts continued, policymakers were hesitant to take precipitate action. With inflation at double digit levels in 1973 anti-inflationists held ground in the United States well into the first oil crisis. As a consequence, the deflationary impact of oil price increases was reinforced by tight monetary and f iscal policy, leading to the most severe economic decline since the Great Depression. As unemployment rose, orthodox Keynesianism experienced a brief revival, only to be banished with the resulting "stagflation"--combined economic stagnation and inflation. By 1978, after the f irst oil crisis, economists began to analyze the role of energy prices in economic change. The central theme that emerged was the substitution between energy and other productive inputs--especially labor and capital inputs (Hudson and Jorgenson, (1978a, 1978b; Jorgenson, 1983b). Economists recalled that energy and capital are complements if an increase in the price of energy reduces the demand for both energy and capital, while energy and labor are substitutes if an increase in energy prices leads to an increase in the demand for labor. It transpired that energy and capital are on the borderline between substitution and complementarily, so that the increase in energy prices left the demand for capital largely unaffected. Of course, the short-run effect of higher prices for imported petroleum was to reduce the return to capital, which is fixed in supply. However, energy and labor proved to be highly substitutable, so that the demand for labor rose with increases in energy prices. In Europe, this effect resulted in an increase in real wages, since labor supply was inelastic with respect to price. In the United States, the inc rease in labor demand led to unprecedented increases in employment. By 1981 it was clear that the concept of substitution between energy and other productive inputs, combined with the analysis we have presented of energy prices and productivity growth, could explain the decline in economic g rowth in industr ialized count r ies (Jorgenson, 1983a, 19841. The process of substitution requires five to seven years, since the accumulation or decumulation of capital stock takes time. The process of adjustment to the second world oil crisis has now been completed in most industrialized countries. However, the effect of higher energy prices on productivity is permanent and has led to widespread declines in productivity growth. As a consequence, the growth prospects for industrialized countries have been permanently reduced to levels below those that prevailed from 1973 to 1979.
84 REFERENCES Berndt, E. R. 1982. Energy price increases and the productivity slowdown in United States manufacturing. Pp. 60-89 in The Decline in Productivity Growth, Proceedings of a Conference held in Edgartown, Mass., June. Boston: Federal Reserve Bank of Boston. Berndt, E. R. 1985. Electrif ication, embodied technical progress, and labor productivity growth in U. S. manufacturing, 1889-1939. In Electricity Use, Productive Efficiency, and Economic Growth, S. H. Schurr and S. Sonenblum, eds. Palo Alto, Calif.: Electric Power Research Institute. Forthcoming. Berndt, E. R., and D. W. Jorgenson. 1973. Production structures. Chapter 3 in U.S. Energy Resources and Economic Growth, D. W. Jorgenson and H. S. Houthakker, eds. Washington: Energy Policy Project. Berndt, E. R., and D. O. Wood. 1975. Technology, prices, and the derived demand for energy. Review of Economics and Statistics 56~3~:259-268. August. Berndt, E. R., and D. O. Wood. 1979. Engineering and econometric interpretations of energy-capital complementarily. American Economic Review 69~3~:342-354. September. Binswanger, H. P. 1974a. The measurement of technical change biases with many factors of production. American Economic Review 64~5~:964-976. December. Binswanger, H. P. 1974b. A macroeconomic approach to induced innovation. Economic Journal 84~336) :940-958. December. Christensen, L. R., D. Cummings, and D. W. Jorgenson. 1980. Economic growth, 1947-1973: an international comparison. Pp. 595-698 in New Developments in Productivity Measurement and Analysis' Studies in Income and Wealth, vol. 41, J. W. Kendrick and B. Vaccara, eds. Chicago: University of Chicago Press. Christensen, L. R., D. Cummings, and D. W. Jorgenson. 1981. Relative productivity levels, 1947-1973. European Economic Review 16 (11: 61-94. May. Christensen, L. R., D. W. Jorgenson, and L. J. Lau. 1971. Conjugate duality and the transcendental logar ithmic production function. Econometrica 39~3) :255-256. July.
85 Christensen, L. R., D. W. -Jorgenson, and L. J. Lau. 1973. Transcendental logar ithmic production f rontiers . Review of Economics and Statistics 55~1~:28-45. February. Denison, E. F. 1984. Accounting for slower economic growth: an update. Pp. 1-45 in International Comparisons of Productivity and Causes of the Slowdown, J. W. Kendrick, ed. Cambridge: Ballinger. Fraumeni, B. M., and D. W. Jorgenson. 1984. The role of capital in U.S. economic growth, 1948-1979. In Capital and Productivity Growth, A. Dogramaci, ed. Boston: Kluwer Nijhoff. Fujime, K. 1983. Structural Changes in Energy Demand in the OECD Nations, with Emphasis on the U. S . and Japan, and a Cyclical Behavior in Oil Markets. Energy Laboratory Working Paper. Cambridge: Massachusetts Institute of Technology. MIT-EL 83-015WP. March. Hicks, J. R. 1932. The Theory of Wages. (2nd edition, 1963.) London: Macmillan. Hogan, W. W. 1984. Patterns of Energy Use. Energy and Envi ronmental Policy Center. Cambridge: Harvard University. April. Hudson, E. A., and D. W. Jorgenson. 1974a. U.S. energy policy and economic growth, 1975-2000. Bell Journal of Economics and Management Science 5~2~:461-514. Autumn. Hudson, E. A., and D. W. Jorgenson. 1974b. Economic analysis of alternative energy growth patterns. Pp. 493-511 in A Time to Choose, D. Freeman et al. eds. Cambridge: Ballinger. Hudson, E. A., and D. W. Jorgenson. 1978a. Energy policy and U.S. economic growth. American Economic Review 68~2~:118-123. May. Hudson, E. A., and D. W. Jorgenson. 1978b. Energy prices and the U.S. economy, 1972-1976. Natural Resources Journal 18~4) :877-897. Octobe r . Jack Faucett Associates. 1977. Development of 35 Order Input-Output Tables, 19S8-1974, Final Report. Washington: Federal Preparedness Agency. October. Jorgenson, D. W. 1983a. Energy prices and productivity growth. Pp. 133-153 in Energy Productivity and Economic Growth, S. Schurr, S. Sonenblum, and O. O. Wood, eds . Cambr idge: Oelgeschlager, Gunn, and Hain.
86 Jorgenson, D. W. 1983b. Econometric and process analysis models for the analysis of energy policy. Pp. 9-62 in Perspectives on Resource Policy Modeling: Energy and Minerals, R. Amit and M. Avriel, eds. Cambridge: Ballinger. Jorgenson, D. W. 1984. The role of energy in productivity growth. American Economic Review 74~2) :26-30. May. (Expanded version reprinted in The Energy Journal 5~3) :11-25. July. Further expanded version reprinted as pp. 270-323 in International Comparisons of Productivity and Causes of the Slowdown, J. W. Kendrick, ed. Cambridge: Ballinger. ~ Jorgenson, D. W. 1985. Econometric modeling of producer behavior. In Handbook of Econometrics, vol. 3, H. Z. Griliches and M. D. Intriligator, eds. Amsterdam: North-Holland. Forthcoming. Jorgenson, D. W., and B. M. Fraumeni. 1981. Relative prices and technical change. Pp. 17-47 in Modeling and Measuring Natural Resource Substitution, E. R. Berndt and B. Field, eds. Cambridge: Massachusetts Institute of Technology Press. Lindbeck, A. 1983. The recent slowdown of productivity growth. Economic Journal 93~369~:13-34. March. Rosenberg, N. 1983. The effects of energy supply characteristics on technology and economic growth. Pp. 279-305 in Energy, Productivity, and Economic Growth, S. Schurr, S. Sonenblum, and O. O. Wood, eds. Cambr idge: Oelgeschlager, Gunn, and Ha in . Samuelson, P. A. 1953. Prices of factors and goods in general equilibrium. Review of Economic Studies 21~1) :1-20. October. Schurr, S. 1982. Energy efficiency and productive efficiency: some thoughts based on American experience. Energy Journal 3~3) :3-14. Schurr, S. 1983. Energy efficiency and economic efficiency: an historical perspective. Pp. 203-214 in Energy, Productivity, and Economic Growth, S. Schurr, S. Sonenblum, and D. O. Wood, eds. Cambridge: Oelgeschlager, Gunn, and Hain. Schurr, S. 1984. Energy use, technological change, and productive efficiency: an economic-historical interpretation. Annual Review of Energy 9:409-425. Schurr, S., J. Darmstadter, H. Perry, W. Ramsay, and M. Russell. 1979. Energy in America's Future. Baltimore: Johns Hopkins University Press.
87 Schurr, S., and B. C. Netschert, with V. E. Eliasberg, J. Lerner, and H. H. Landsberg. 1960. Energy in the American Economy, 18SO-1975: Its History and Prospects. Baltimore: Johns Hopkins University Press for Resources for the Future. Sonenblum, S. 1985. Electricity and technological progress: implications for productivity growth. In Electricity Use, Productive Efficiency, and Economic Growth, S. H. Schur r and S. Sonenblum, eds. Palo Alto, Calif.: Electric Power Research Institute. Forthcoming. wood, D. O. 1985. Discussion. In Electricity Use, Productive Ef f iciency, and Economic Growth, S. H. Schurr and S. Sonenblum, eds. Palo Alto, Calif.: Electric Power Research Institute. For thcoming .
