The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
11 CO NTEN TS The R e v i e w is published monthly by the Research Department of the Federal Reserve Rank of St. Louis. Single-copy subscriptions are available to the public free of charge. Mail requests for subscriptions, back issues, or address changes to: Research Department, Federal Reserve Rank of St. Louis, P.O. Rox 442, St. Louis, Missouri 63166. Articles herein may be reprinted provided the source is credited. Please provide the Rank’s Re search Department with a copy of reprinted material. The Productivity Problem JOHN A. TATOM T A HE American economy has been plagued by nu merous economic problems during the seventies. Of particular importance, of course, are the relatively high inflation and unemployment rates, the changing and growing influence of government in individual lives, and problems associated with the dramatic rise in energy prices. Such issues have been widely recog nized during this decade. One fundamental problem, however, has just re cently begun to attract increasing attention — the slow growth of U.S. labor productivity in the seven ties. The Joint Economic Committee in its 1979 Mid year Report refers to lagging productivity growth as the “Linchpin of the Eighties.”1 Likewise, a recent examination of the major economic issues in 1979 begins with five studies on the productivity problem.2 Despite the recent flurry of attention, however, there appears to be no consensus on the source of the slow down, its signficance, its impact on the economy or economic policy, or the possibility or desirability of policy reactions to the problem.3 Since productivity developments in the seventies appear to be a puzzle, it is useful to examine carefully the explanations that have been advanced in search ’ See Joint Economic Committee, Outlook 1980’s Midyear Re port and Staff Study, U.S. Congress, 96th Cong., (August 1979). 2See William Fellner, Project Director, Contemporary Eco nomic Problems 1979, American Enterprise Institute for Pub lic Policy Research, 1979. 3For example, the four factors cited most frequently at a re cent conference on productivity were a decline in capital investment, a slowing in the introduction of new technology, the changing composition of the labor force toward relatively more inexperienced groups, and an increase in service em ployment relative to manufacturing employment. Barry Bosworth, head of the President’s Council on Wage and Price Stability at the time, however, demonstrated that these ex planations are not significant in explaining the extent of the current productivity problem. See the discussion of the con ference by Bradley Graham, “Productivity Problem Outlined,” Washington Post, October 4, 1978. The article also points out of a clue.4 Such an examination shows why the pro ductivity problem has only recently been generally recognized and provides insight into the role of government policy in reversing the adverse produc tivity developments of the seventies. Prior to consid ering the various hypotheses about recent produc tivity developments, a brief review of what is meant by productivity, the sources of its growth, and recent productivity developments is necessary. PRODUCTIVITY AND ITS GROWTH RATE Productivity is a concept which relates the output of a production process to the inputs used in the process. There are several ways to view productivity since, in most processes, several resources are used. One may refer to labor productivity, which is the output obtained per worker employed or per hour of work. Similarly, the productivity of other resources, such as physical capital or energy, may also be of interest. A more general notion is that of total factor produc tivity which relates output to the entire set of physi cal resources used in the production process by as signing a weight to each resource employed based on its relative importance in the production process. that while Robert Strauss, then special assistant to the Presi dent on inflation and trade policy, noted at the conference that “Productivity is a terrible problem for this nation,” the government’s National Center for Productivity and the Quality of Working Life, an institution chartered at the beginning of this decade to study and promote productivity, was allowed to go out of existence in October 1978. 4Edward F. Denison, “Explanations of Declining Productivity Growth,” Survey of Current Business (August 1979), pp. 1-24, examines 17 factors which might account for the pro ductivity decline and finds them insufficient to explain the puzzle. Several of the factors he considers and rejects, in cluding, for example, the “people don’t want to work any more” hypothesis, are not discussed here. Denison’s views differ from those presented here in that he dismisses both energy developments and slow growth of the capital stock as factors responsible for the decline. Page 3 FEDERAL RESERVE BANK OF ST. LOUIS One may understand more clearly the different views of productivity by considering the production process itself. In any production process, output is obtained by organ izing the use of various re sources and utilizing a given technology. Output may be increased by using more re sources, by improving the organization of the produc ing unit, or by introduc ing an improved technology. If more output is obtained through improved organiza tion or technology without changing the level of re source use, total factor pro ductivity (the output per unit of resources employed) and the output per unit of any single resource used will increase. SEPTEMBER Chart l Real GN P Per Worker* ia 1972 D o llars R a tio S ea l* Thousands of D o lla rs P e r W orker 1948 49 50 51 52 53 54 R a tio Seal* Thousands of D o lla rs F o r W o rker 16 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 1979 Sources: U.S. Department of Commerce and U.S. Department of Labor 'R eal GNP divided by total civilian employment. Shaded areas represent periods of business recessions. Latest data plotted: 2nd quarter Output may be increased, however, by simply us ing more of a specific resource, such as labor, without altering the methods of organization, the technology, or the use of other resources. In this case, the output obtained per unit of other resources rises, so their productivity is higher. Whether output rises relatively more or less than the increase in labor employment determines how the productivity of labor changes. According to the “principle of diminishing returns,” the productivity of labor in such an example falls be cause the amount of other resources used per unit of labor declines. The analysis of productivity usually focuses on measures of the productivity of labor. For the firm, output per worker (or per hour) is important because labor costs represent a major share of the total cost of production. Consequently, changes in output per worker have an important bearing on employment and output decisions, as well as on wages and prices. Improvements in labor productivity tend to lower unit costs of output, and provide incentives for firms to expand production. This, in turn, tends to result in lower output prices (relative to those of other goods and services) and/or higher wages for employees. Labor productivity growth is also important at an aggregate level, primarily as a reflection of economic Page 4 1979 growth. Real income in an economy is often viewed as the product of the size of the working population and output per worker. Thus, growth in real income per capita is primarily determined by labor produc tivity growth. The only other source of change of real income per capita — a change in the percentage of the population that is working — varies over the business cycle because of fluctuations in employment, but does not show any systematic tendency to in crease.5 Moreover, gains in real income per capita from this source come at the expense of leisure and home production, unlike productivity improvements which permit increases in both nonmarket (for ex ample, leisure) and market income. According to the discussion above, labor produc5Since World War II, the ratio of civilian employment to the civilian noninstitutional population of age 16 and over has reached about 58 percent at the peak of periods of eco nomic expansion. In the recent expansion, this ratio rose slightly above 60 percent. In 1969 it was 58.0 percent, while in 1978 it was 59.4 percent. This difference given a constant level of labor productivity and population, ac counts for about .27 percent of the average annual real GNP growth rate of 2.93 percent over the period. Herbert Stein, “Why Did Consumption Not Reflect the Slackening of the Productivity Trend,” American Enterprise Institute Econo mist (June 1979), reprinted in Fellner, Contemporary Prob lems, pp. 13-15, has noted that much of the productivity slowdown of the seventies was compensated for by increased participation in the labor force so that growth of real GNP per capita did not slow as much. SEPTEMBER FEDERAL RESERVE BANK OF ST. LOUIS Output Per Hour, Private Business Sector R a tio S t a l l 1967=100 122 rr 118 - 112 106 - 100 - 94 - ■/ ' 1948 49 SO SI 52 S3 S4 55 56 57 i S8 59 l 60 61 62 l 63 l 64 l 65 l 66 l 67 68 69 70 71 72 73 74 75 76 77 78 1979 Source: U.S. Bureau of Labor Statistics Shaded areas represent periods of business recessions. Latest data plotted: 2nd quarter tivity is improved by increasing the employment of other resources used by workers, or by improving or ganization or the technology used in production. Exam ples of the sources of such changes are increases in the skill, experience, or health of workers; increases in the availability or quality of resources such as physical plant, equipment, and energy; successful research and development efforts; and the breaking down of artifi cial barriers to the use of resources such as labor in activities where productivity would be higher. LABOR PRODUCTIVITY SINCE 1948 Numerous measures of labor productivity in the U.S. economy can be obtained depending upon the choice of alternative measures of aggregate output and labor employment such as hours or the number of workers. Charts 1 and 2 show two of these mea sures. Chart 1 shows real GNP divided by the quar terly average of civilian employment, while Chart 2 shows output per hour in the private business sector, which excludes output originating in governmental units, households, nonprofit organizations, and the rest of the world. In both cases, the productivity measure is shown from 1948 to the present.8 6The charts show the logarithm of productivity to indicate trends in growth. Throughout this article, growth rates are 1979 Productivity growth in the seventies, particularly since 1973, has been extremely sluggish. From 1948 to early 1973, the trend rate of growth of real GNP per worker was 2.4 percent per year; since then, the average annual rate of growth has been essentially zero. The same pattern appears in Chart 2 where the trend rate of growth was 3.0 percent per year until the first quar ter of 1973, while the recent rate has been 0.5 percent. In both cases, productivity de clined sharply during 197374 and, despite the relatively rapid expansion of output and employment since 1975, the rate of productivity growth has remained un usually slow. The difference in growth rates in the two charts primarily reflects a systematic decline in hours of employment per worker. The average annual rate of decrease in hours per worker from 1947 to the present, which is 0.4 percent per year, accounts for almost all of the difference in growth rates.7 FACTORS INFLUENCING PRODUCTIVITY IN THE SEVENTIES Many explanations have been given for the disap pointing performance of productivity in the seven ties. While it is not possible to empirically assess the role of each factor, and while considerable disagree ment exists about their relative importance, a discus sion of them can provide some insight into the problem. The Cyclical View Until recently, the consensus was that the produc tivity problem was due to the stage of the business measured by changes in logarithms of variables to provide symmetry and ease of calculation. This is a standard approx imation of rates of growth. 7This trend is found by regressing quarterly hours per worker in the private business sector on a constant, the unemploy ment rate, and a time trend for the period I/1948-II/1979. The unemployment rate is included to account for a sig nificant negative relationship between hours per worker and unemployment. Page 5 FEDERAL RESERVE BANK OF ST. LOUIS cycle and was, therefore, temporary. Productivity growth usually slows during high-employment periods and recessions. During recoveries, however, produc tivity growth generally speeds up, restoring produc tivity to the previous trend growth path (see Charts 1 and 2). Unfortunately, such a recovery did not oc cur during the recent economic expansion. The rationale for cyclical productivity movements is that firms practice “labor hoarding.” During reces sion, firms are reluctant to lay off workers since they want to avoid the relatively large rehiring costs in the subsequent expansion. Similarly, output increases in the early stage of expansions are readily achieved, increasing output by increasing the utilization of labor at a given employment level. Thus, cyclical variation in output is larger than that in employment. It has also been observed that hours of employment per worker display a cyclical pattern. Output de clines during a downturn are achieved by reducing hours per worker rather than by simply reducing the number of employees. As a result, productivity meas ured relative to employment varies over the cycle more than output per hour. A related point often encountered in discussions of cyclical productivity is that some portion of labor employment is of the over head variety, such as management; normally, as out put changes over the cycle, little change in overhead labor employment occurs. As a result, productivity varies procyclically.8 Another explanation of cyclical labor productivity is simpler and, in most respects, compatible with the “labor hoarding” view. This explanation is that the use of capital relative to labor is procyclical — firms use relatively more capital-intensive methods as the economy expands and reduce capital usage relative to labor during economic contractions. Consequently, output changes relatively more than labor employ ment over the cycle, and labor productivity is procyclical.9 8A discussion of these arguments may be found in many text books. For example, see Rudiger Dornbusch and Stanley Fischer, Macroeconomics (New York: McGraw-Hill, Inc., 1978), pp. 348-9. They cite productivity developments in 1974-75 as an illustration of such an argument. 9This explanation of cyclical productivity and evidence sup porting it is presented more fully in John A. Tatom, “The ‘Problem’ of Procyclical Real Wages and Productivity,” Jour nal of Political Economy (February 1980), forthcoming. This explanation differs from the labor-hoarding argument pri marily because of its implications. Discussions of labor hoard ing often appear to imply that workers are allowed to idle about awaiting the recovery and that, consequently, output represents a departure from the normal production relation ship linking output to labor employment. Such an implica tion is unwarranted once the cyclical pattern of resource emPage 6 SEPTEMBER 1979 A recent example of cyclical changes in labor pro ductivity occurred prior to the 1970 recession. From the fourth quarter of 1964 to the first quarter of 1969, the economy operated under high-employment conditions, and real GNP per worker grew at an average rate of 2.2 percent per year (roughly the postwar average). During the next three quarters, productivity declined at a 1.5 percent rate, and dur ing the four quarters of the subsequent recession, this measure of productivity declined at a 0.5 per cent rate. In the ensuing expansion period, however, productivity initially grew faster than average. From the end of 1970 (the cyclical trough) to the first quarter of 1973, real GNP per worker expanded at a 3.6 percent rate and employment was restored to essentially a high-employment rate. Thus, an average annual rate of growth of 1.9 percent was achieved from the end of 1964 to early 1973, not much below the rate of expansion prior to the recession. Starting from high-employment conditions early in 1973, the cyclical explanation subsequently proved inadequate. Productivity declined at a 2.1 percent rate during the three quarters prior to the cyclical peak in the fourth quarter of 1973. During the reces sion (IV/1973-I/1975), real GNP per worker declined at a 3.3 percent rate. During the recovery, produc tivity rose faster than the prior trend rate, recover ing some of the cyclical loss, but the cyclical resur gence of labor productivity soon abated leaving the productivity growth path well below its prior trend. From the trough quarter (1/1975) to the second quarter of 1979, real GNP per worker rose at only a 1.4 percent rate, leaving the level of productivity below that of the first quarter of 1973. Capital Formation Another major hypothesis is that the productivity problem results from a slowing in the pace of capital formation and/or inefficiencies in the allocation of new capital expenditures. Growth of the net stock of nonresidential plant and equipment slowed after 1973, a decline that is espe cially pronounced when related to growth in the labor force over this period.10 Growth of the highployment is taken into account. Just as increasing the use of capital adds to output and labor productivity on a trend basis, cyclical variations in the mix of resources employed account for temporary movements in labor productivity over the cycle. There are no unusual departures of productivity from the basic empirical relationship between output and the use of labor and capital resources in the U.S. economy. 10See John A. Tatom, “Energy Prices and Capital Formation: 1972-77,” this Review (May 1979), pp. 2-9. Denison “Ex planation of Declining Productivity Growth,” p. 3, shows FEDERAL RESERVE BANK OF ST. LOUIS employment net capital stock per worker has practi cally halted when compared with the trend rate of growth of 2.9 percent per year from 1950 to mid-1972. From mid-1972 to mid-1979, capital per worker grew at a 0.6 percent rate, leaving the level of capital per worker by mid-1979 about 17 percent lower than that implied along the 1950-72 trend. Alternately, the net capital stock was about $200 billion (1972 prices) lower than would have existed if the previous trend rate of growth had been achieved. A 17 percent loss in capital per worker would reduce private business sector output per hour by approximately 4.8 percent. Distributing the 4.8 percent decline in output as sociated with these capital formation developments over the seven years, however, yields a 0.7 percent decline per year, while the growth rate of output per hour slowed from 2.8 percent from I/1950-II/1972 to 1.0 percent from mid-1972 to mid-1979. A sizeable estimate of the reduction in productivity growth due to the slowing in capital formation has also been made for the nonfarm business sector during the pe riod 1973-77.11 Capital formation has played a major role in the stagnation of productivity growth in the 1970s. Considered alone, however, it does not ap pear to account fully for the extent of adverse developments. There are reasons to believe that estimates of the role of slower capital formation may understate its ac tual impact. First, it is likely that the net stock of fixed nonresidential capital estimated by the Bureau of Eco nomic Analysis does not adequately capture important changes in the capital formation process in the last decade.12 In particular, the sharp increase in the rel virtually no slowing in the contribution of capital per per son employed during 1973-76. The reason for this discrep ancy apparently is due to his method of grouping factors which affect productivity and the cyclical differences be tween 1948 and 1973, on the one hand, and 1973 and 1976, on the other. As Denison points out, in each of the two pe riods capital per person employed grew at close to the same rate. Capital per hour of all persons employed slowed sharply, however, due to cyclical differences which show up elsewhere in Denison’s groupings. Corrected for cyclical effects, the growth of capital per worker and capital per hour show a sharp decline in capital accumulation. 11See J. R. Norsworthy and Michael J. Harper, “The Role of Capital Formation in the Recent Productivity Slowdown,” Bureau of Labor Statistics, U.S. Department of Labor Work ing Paper 87, (January 1979), Table IV. They also provide evidence refuting the claim that labor productivity growth slowed from 1965 to 1973 due to a slowing in capital stock growth. This claim has been made by Peter K. Clark, “Cap ital Formation and the Recent Productivity Slowdown,” The Journal of Finance (June 1978), pp. 965-75. Such evidence may also be seen in Chart IV in John A. Tatom, “Energy Prices and Capital Formation,” p. 7. 12This data is described by John C. Musgrave, “ Fixed Non residential Business and Residential Capital in the United States, 1925-75,” Survey of Current Business (April 1976), pp. 46-52, and Tatom, “ Energy Prices and Capital Forma tion,” p. 5. SEPTEMBER 1979 ative price of energy resources rendered some of the nation’s capital stock obsolete. To the extent that estimated series do not capture these losses in normal measures of “discards” and depreciation, the net capi tal stock measures lead to an overstatement of the growth of the net capital stock in the seventies.13 A second difficulty with existing measures is that since the late 1960s, some investment in business plant and equipment has been made for pollution abatement purposes. Since it is questionable to what extent, if any, such capital formation is “produc tive” of measured output (as opposed to a cleaner environment), the net capital stock measures may overstate the growth of plant and equipment used by workers to produce market output. Studies gen erally have indicated, however, that pollution abate ment investment has had little effect on the growth of the capital stock or on labor productivity.14 Ed ward F. Denison has estimated the impact of pollu tion abatement capital on productivity growth to be quite small; since he accounts for this reduction by reducing the trend growth of total factor produc tivity, rather than by reducing the size of the cap ital stock, his results are discussed below. There are many reasons for the slow pace of capi tal formation. Explanations usually emphasize the extent of slack in the economy’s use of existing re sources, the diversion of funds from investment in new plant and equipment to expenditures on health, safety, clean air and water, and large federal budget 13The theory of capacity and capacity losses due to a higher relative price for energy resources is discussed in Robert H. Rasche and John A. Tatom, “The Effects of the New Energy Regime ancl Economic Capacity, Production, and Prices,” this Review (May 1977), pp. 2-12. This article indicates that the bias that overstates capital stock or capacity meas ures is likely to understate capacity utilization rate measures. Thus, utilized capital measures that use a product of a utilization rate and a capital stock measure are unlikely to be biased. 14For example, Robert H. Rasche and John A. Tatom, “Energy Resources and Potential GNP,” this Review (June 1977), pp. 10-24, found no significant differences between produc tivity estimates using capital stock adjusted and not ad justed for pollution abatement capital, p. 15. Similarly, the growth rate of an adjusted capital stock series reported by Peter K. Clark, “A New Estimate of Potential GNP,” U.S. Congress, Joint Economic Committee, Hearings on the Eco nomic Report of the President, 95th Congress, 1st sess., January 19, 1977, pp. 39-54, is essentially the same from 1968-77, when the adjustments are present, as during the period 1954-68. A calculation of the capital stock in manu facturing, with and without adjustment for pollution abate ment capital, by Lawrence R. Forest, Jr., “Capacity Utiliza tion: A Discussion of Concepts and Selected Analytical Ap plications,” Measures of Capacity Utilization: Problems and Tasks, Staff Studies No. 105, Board of Governors of the Federal Reserve System, (July 1979) pp. 57-135, shows roughly a .3 percent decline in the annual rate of capital formation due to the adjustment from 1968 to 1977. Such a difference, one of the largest observed, has negligible effects on output per hour. Page 7 FEDERAL RESERVE BANK OF ST. LOUIS deficits which have drained funds otherwise avail able for new capital projects. Inflation is also an important source of reduced business capital formation. First, higher rates of in flation tend to reduce the purchasing power of fixed dollar depreciation expenses which results in lower real cash returns in future periods. Also, the U.S. tax system treats interest payments made by firms as in come to recipients and taxes it accordingly. When interest rates rise to compensate investors for the steady loss in purchasing power of original sums lent to firms, these receipts — which are necessary to main tain the real wealth of investors — are erroneously treated as income. Consequently, higher before-tax real rates of return are required to compensate for these taxes, further reducing incentives for firms to raise investment funds. Finally, higher inflation rates tend to increase uncertainty about the future inflation rate. Consequently, investors and firms view the cash flows expected from investment projects as riskier and are more reluctant to invest.15 This analysis reverses the traditional view of the relationship between productivity growth and infla tion. Until the seventies, the link was considered to run from productivity to inflation, instead of from inflation, via capital formation effects, to produc tivity growth. According to the earlier view, inflation is largely the result of labor costs growing faster than productivity. Consequently, inflation accelerates when productivity growth slows. In the long run, however, productivity growth determines the growth of wages relative to prices, but the rate of advance of prices including prices of labor services, is largely deter mined by money growth. Thus, monetary policy is an important influence not only on the rate of increase of prices (including wages) but also, through the effect higher inflation rates have on capital formation and trend productivity growth, on the relative rate of advance of wages compared with prices. SEPTEMBER 1979 intensive than the production of other products, a rise in energy prices raises the costs of capital goods relative to the future prices of the products that these capital goods eventually will produce. Taken together, these forces create incentives to reduce energy, plant, and equipment usage per unit of out put, by employing less energy per unit of capital and more labor-intensive methods of production. This ef fect has been shown to be quite substantial in tempo rarily reducing the growth of plant and equipment.16 The Rate of Technological Change Another important factor influencing productivity is the rate of technological progress, which conceptually represents improvements in total factor productivity not captured in measures of factors of production. For example, improvements in the quality of productive plant and equipment are not necessarily captured in standard measures of the capital stock. Also, improve ments in organization or technology (which repre sent additions to the nation’s capital stock) are not always embodied in plant and equipment measures. The contribution of these factors, as well as qualita tive changes in other physical resources, often are assessed by including a time trend in productivity studies to capture the secular rate of productivity change not measured by growth in the use of phys ical resources. Clark estimates that the trend rate of growth of total factor productivity in the private business sec tor was 1.9 percent per year from 1947-66, but slowed to a 1.4 percent rate after 1966. He attributes this reduction to a slowing in the rate at which resources have shifted from agriculture to nonfarm business.17 Rasche and Tatom find that the productivity trend rate was 1.6 percent in the private business sector from 1949-75; they provide some evidence supporting a smaller trend slowdown after 1966.18 Finally, the sharp rise in the relative price of energy since 1973 has been a major factor in the reduced pace of capital formation. When the price of energy rises relatively more than the price of business out put, firms find that the real net cash flows expected from plant and equipment are smaller because of higher operating costs. Moreover, to the extent that the production of capital goods is relatively more energy- An important factor that influences the trend rate of productivity growth is the increase in the stock of knowledge accumulated through research and de velopment outlays.19 While it is difficult to assess its 15Burton G. Malkiel, “Productivity — the Problem Behind the Headlines,” Harvard Business Review (May-June 1979) pp. 81-91, presents evidence of higher risk premiums built into stock and bond returns during the mid-seventies, as well as a discussion of some of the other factors inhibiting capital formation and productivity growth. 17See Clark, “A New Estimate of Potential GNP.” 8 Digitized for Page FRASER 16See Tatom, “Energy Prices and Capital Formation,” and Edward A. Hudson and Dale W. Jorgenson, “Energy Prices and the U.S. Economy, 1972-1976,” Natural Resources Jour nal (October 1978), pp. 877-97. 18See Rasche and Tatom, “ Energy Resources and Potential GNP,” pp. 16, 19. 19John W. Kendrick, “Productivity Trends and the Recent Slowdown: Historical Perspective, Causal Factors, and Pol- SEPTEMBER FEDERAL RESERVE BANK OF ST. LOUIS R ese a rch a n d D e v e lo p m e n t O u tla y s Ratio Scale B illio n s of 1972 Dollars lio Scale ions of 1972 Dollars 1955 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 1979 1979 increased business costs associated with unusually high rates of dishonesty and crime.20 The first two of these factors, under the general category of increased government regulation, have been cited by many analysts as a significant force in reducing American productivity. Denison contends that these factors divert real resources from the production of measured output. He estimates the ef fect of these forces, however, by assess ing their impact on output rather than on measured inputs. Furthermore, he finds that over the period 1969-75 these factors reduced productivity growth by an average of .26 percent per year, with pollution abatement accounting for .15 percentage points; health and safety, .07 points; and dishonesty and crime, .05 points. The effect of these factors is quite small when compared to the extent of recent productivity developments. S o u rce : U .S . N a tio n a l S cie n ce Foundation Latest d a ta p lo tted: 1978 quantitative impact on recent productivity develop ments, it is doubtless that a major decline in the con tribution of such outlays to productivity has occurred in the seventies. Chart 3 shows estimates of research and development outlays (1972 prices) from 1955-78. From 1955-68, the average annual rate of growth of such expenditures was 7.6 percent. From 1969 through 1976, such expenditures actually were lower in real terms than in 1968. While real research and develop ment outlays rose at a 4.2 percent rate from 1975-78, the record of the last 10 years is one of substantial reduction in research and development efforts, com pared with 1955-68. Denison has examined three other factors that have adversely affected the trend rate of growth of output per unit of input. The three factors are: increased costs of pollution abatement since 1967, increased costs for employee safety and health since 1968, and icy Options,” Contemporary Economic Problems 1979, pp. 17-69, attaches primary importance to this factor in account ing for the slowdown. One of his estimates is that the con tribution of advancing knowledge due to formal and informal research and development inventiveness, and diffusion of new technology and capital goods contributed 1.4 percent to annual productivity growth in 1948-66, 1.1 percent in 1966-73, and 0.8 percent in 1973-78. Denison and others have shown that these estimates are probably seriously biased upwards. See Denison, “Explanations of Declining Productivity Growth,” pp. 6-8. Labor Resources In the past decade, the age and sex composition of the labor force has changed. Many analysts have argued that the labor force has become less exper ienced on average, as women and young people have become a larger proportion of the nation’s labor force, and that this development should have a temporary influence on productivity.21 As Chart 4 shows, how ever, growth in the proportion of women in the labor force has not accelerated in the seventies and, since late 1973, the percentage of young people in the labor force has flattened out. If these trends have had any effect on productivity growth, they should have been positive. As the participation rate of women levels off and as younger workers mature, the composition of 20See Edward F. Denison, “Effects of Selected Changes in the Institutional and Human Environment Upon Output Per Unit of Input,” Survey of Current Business (January 1978), pp. 21-44, “The Puzzling Drop in Productivity,” Challenge ( May-June 1979), pp. 60-62, and “Where Has Productivity Gone,’ in William Fellner, Project Director, Contemporary Economic Problems 1979, pp. 71-77. 21For example, see George Perry, “ Potential Output and Pro ductivity," Brookings Papers on Economic Activity ( 1:1977), pp. 11-47, and William Fellner, “The Declining Growth of American Productivity: an Introductory Note,” Contempo rary Economic Problems 1979, pp. 3-12. Page 9 FEDERAL RESERVE BANK OF ST. LOUIS the labor force will be come relatively more ex perienced.22 Consequently, changes in the age structure toward older, more experi enced workers are expected to have a positive effect on productivity in the next decade. Another important factor influencing the quality of la bor resources and productiv ity is the health and educa tion of workers. In the past decade, there is no evidence that the trends of rising edu cational levels and improv ing health of workers have been reversed, or that the contribution of such influ ences to productivity have lessened.23 SEPTEMBER 1979 Chart 4 W omen and Youth as a Percent of the Labor Force 12 Ratio of the civilian labor force age 16-24 to total civilian labor force. Energy Resources Energy price developments have provided a major shock to the world economy in the seventies and have affected productivity, output, prices, and growth. The extent of the influence on productivity, however, re mains a matter of controversy. Rasche and Tatom have argued that the rise in the price of energy resources relative to that of business output reduced the economic capacity of the busi ness sector, raised prices of output, and reduced pro ductivity of existing capital and labor resources.24 These effects stem from efforts to economize on energy in production by substituting the use of other resources. For example, they estimate that the increase in energy costs in 1974 alone reduced labor and capi tal productivity by about 4 percent. For the 1973-78 22It is difficult to assess the importance of the composition of the labor force on productivity. For example, attempts to assess the role of the age structure of the labor force on output by including variables to control for it in a pro duction function framework are often unsuccessful in that such variables are not statistically significant. See Rasche and Tatom, “Energy Resources and Potential GNP,” p. 15. 23See, for example, Fellner’s comments on education in “The Declining Growth of American Productivity,” p. 8. 24See Rasche and Tatom, “The Effects of the New Energy Regime on Economic Capacity, Production, and Prices,” “Energy, Resources and Potential GNP,” and “Potential Out put and Its Growth Rate — The Dominance of Higher En ergy Costs in the 1970s,” U.S. Productive Capacity: Estimat ing the Utilization Gap (Washington University: Center for the Study of American Business, December 1977), pp. 67106. Page 10 period, higher energy prices reduced the growth rate of labor productivity by 1.3 percent per year, irrespec tive of the indirect effects of energy costs on produc tivity due to energy-induced reductions in capital formation. By lowering productivity, higher energy costs led to a decline in the demand for other re sources. This decline, in turn, was reflected in a fall in real wages and a reduction in firms’ desired capitallabor ratios. The observed decline in the capital-labor ratio since 1972 can be explained almost entirely by energy price developments alone.25 Other analysts have argued that the effect of in creased energy prices on the economy has been much smaller. For example, Fellner attributes only .3 per cent of the 1.9 percent per year decline in trend pro ductivity growth from the 1948-73 period to the 197377 period to the shift of production away from energyusing methods.26 Denison points out that the timing 25See Tatom, “Energy Prices and Capital Formation,” and Hudson and Jorgenson, “ Energy Prices and the U.S. Econ omy.” They have produced similar results for the impact of higher energy prices on capital formation and, consequently on productivity. Others have provided support for relatively large short-run effects of changing supplies of energy as reflected in relative prices of energy. See Clark, “A New Measure of Potential Output,” and Jacques R. Artus, “ Meas ures of Potential Output in Manufacturing for Eight Indus trial Countries, 1955-78,” International Monetary Funds Staff Papers (March 1977), pp. 1-35. 26See Fellner, “The Declining Growth of American Produc tivity,” p. 6. SEPTEMBER FEDERAL RESERVE BANK OF ST. LOUIS C h a rt 5 Industrial Sector Ratio of Energy to Output* S o u rc e : ♦ R a tio o f e n e r g y to r e a l in d u s t ria l G N P . a g r ic u lt u r e , a n d m in in g . L a te s t d a t a p lo tte d : 1 9 7 7 U .S . D e p a r tm e n t o f E n e r g y a n d U .S . D e p a rtm e n t o f C o m m e rc e used per unit of output would have had to occur.28 Likewise, both have noted that the reductions in en ergy use that followed the sharp rise in energy prices in 1973-74 were quite small, although both also have ac knowledged the difficulty of obtaining meaningful data on energy used by business. Perry claims that energy used per unit of output in the business sector declined 10.2 percent from 1973 to 1976, while the trend in such reductions from the past alone would have generated a 5.3 to 7.3 percent reduc tion in energy usage. Thus, he claims that energy price developments reduced en ergy use by only 2.9 to 4.9 percent over this period. Similarly, DeLeeuw, citing ratios of total and industrial energy consumption to total and industrial real GNP which show a downward trend from 1970-73, argues that this negative trend did not accelerate after 1973. Th e in d u s t ria l s e c t o r is co m p o s e d o f co n stru ctio n , m a n u fa c tu r in g , of energy price developments coincides with that of the adverse productivity developments. Nevertheless, he insists that a more reasonable estimate of the effect of higher energy prices on productivity is 0.2 percent for the period 1973-76, the upper bound of the size of the effect as discussed by Perry.27 Both Perry and DeLeeuw have argued that for productivity to have been adversely affected by energy price developments, a significant decline in energy 27See Denison, “Where Has Productivity Gone?” p. 76, “The Puzzling Drop in Productivity,” p. 62, and “Explanations of Declining Productivity Growth,” pp. 15-18 and George L. Perry, “Potential Output: Recent Issues and Present Trends,’ in U.S. Productive Capacity: Estimating the Util ization Gap, pp. 6-12. 1979 Chart 5 shows the ratio of industrial energy use ( E ) to “industrial” real GNP, where the latter is the real GNP originating in manufacturing, construction, agri culture, and mining sectors for the period 1960-77.29 This ratio declined after 1970, as DeLeeuw and Perry point out. There is no apparent trend in this ratio, however, over the longer period 1960-72. A simple regression of the logarithm (In) of this ratio over the period 1960-72 fails to reveal a significant negative time trend when the effects of changes in the rela28See Frank DeLeeuw, “ Comments on Rasche and Tatom,” in U.S. Productive Capacity: Estimating the Utilization Gap, pp. 137-45. 29The energy data is taken from U.S. Department of Energy, Federal Energy Data System (FEDS) Statistical Summary. This data series is available from 1960-77. Page 11 FEDERAL RESERVE BANK OF ST. LOUIS SEPTEMBER tive price of energy and eye-1 iical variations in the ratio are taken into account. The estimated equation, together with one that omits the time trend, is shown in Table 1. The price of energy, Pe, is measured using the producer price index for fuel, related products, and power deflated by the im plicit price deflator for pri vate business sector output. The cyclical adjustment var iable, UN, is the difference between the unemployment rate of the civilian labor force and a measure of the unemployment rate under high-employment conditions.30 Output Per Hour in Manufacturing 1948 49 SO 51 52 53 54 55 56 T a b le 1 The Industrial Energy Ratio In (E /R G N P ) = I n (E / R G N P ) = 1960- 72 ) 7 .6 3 4 1 .2 0 7 In P6 + ( 3 .4 0 ) ( - 2 .7 0 ) R2 = 0 .7 2 S .E . = 0 .0 2 6 5 .7 0 1 ( 6 .3 4 ) R2 — 0 .6 9 S .E . = 0 .0 2 5 D .W . = 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 1979 Source: U.S. Bureau of Labor Statistics significant positive coefficient on the excess unemploy ment rate suggests that the failure of the energy ratio to decline more substantially after 1973 is partly due to the unusually high levels of unemployment in 1975-77. Apparently, energy usage contains a signifi cant overhead component that is insensitive to cyclical variation in output. This contrasts sharply with Perry’s negative cyclical coefficient.31 Finally, since the time trend is insignificant, assessing energy reductions rela tive to a negative trend understates the extent of energy reductions after 1973.32 0 .0 4 3 U N — 0 .0 0 5 t ( 4 .2 3 ) ( - 0 .9 4 ) 1 .5 0 — 0 .8 2 5 In P e + (- 4 .4 1 ) 0 .0 3 9 UN ( 4 .2 2 ) D .W . = 1.31 30The full-employment unemployment rate is that prepared by Peter K. Clark, “ Potential Output in the United States 19481980,” U.S. Productive Capacity, Perry uses the ratio of real GNP to his estimate of potential real GNP which is highly correlated with such a measure of excessive unemployment. It should be noted that when the equations in Table 1 are estimated with the Cochrane-Orcutt technique, the auto correlation coefficient is not significant. Thus, autocorrelation can be rejected. Page 12 57 Latest data plotted: 2nd quarter The most notable features of these estimates are the size and significance of the relative price coeffi cient which indicate that the energy ratio is con siderably more sensitive to prices than either Perry or DeLeeuw believe. In addition, the statistically ( 1979 31The cyclical sensitivity of the energy ratio is essentially an empirical question since one could argue that over the cycle some firms, such as utilities and metal producers, are able to meet temporarily high demand using less efficient (espe cially with regard to energy use) plant and/or equipment, and this capital, in turn, is the first to be laid off during recessionary sales periods. If this effect dominated the use of overhead energy, the energy ratio would be reduced by slack in the economy. On the other hand, the energy pro ducing sector is the most energy intensive production sector. Given the rise in the relative price of their products due to reductions in the availability of other energy sources, some firms would temporarily have an incentive to expand the use of plant and equipment that would otherwise have been obsolete. This effect would raise the energy ratio during the recent slack period. 32The White House, Fact Sheet on the President’s Program, reprinted in Daily Report for Executives, (April 5, 1979), supplement p. 4, points out that industrial energy use has dropped by 6 percent while output increased 12 percent since 1973. This represents a drop in the ratio of about 18 percent. FEDERAL RESERVE BANK OF ST. LOUIS Estimates such as those in Table 1 should be viewed skeptically, however. First, as Perry and DeLeeuw point out, the data are problematical.33 Second, the data cover only direct energy purchases and not the energy component of intermediate products purchased from other industries including transportation. Third, the energy measure is in Btu equivalents rather than an index of the quantity of energy which is compati ble with the producer price index for energy. SEPTEMBER 1979 well. In fact, the evidence from the manufacturing sector supports this explanation. Productivity growth in the manufacturing sector has slowed relative to its past trend (see Chart 6). From 1948 to 1973, manufacturing sector output grew at an average annual rate of 3.8 percent, while hours grew at a 1.0 percent rate. From 1973 to 1978, manufac turing output growth slowed to a 1.7 percent rate and hours growth declined at a 0.1 percent rate. Con sequently, output per hour in manufacturing slowed from a 2.8 percent rate of growth to a 1.6 percent rate. Comparable figures for the private business sec tor show a decline in growth of output per hour from a 2.9 percent to a 1.0 percent rate. While the decline in the growth rate in manufacturing is smaller, this Most important, however, the size of the price ef fects on the energy-output ratio is unimportant to the hypothesis that relates energy price developments to the sharp decline in productivity in 1974 and its subsequent slow growth. The Rasche-Tatom estimates of the effect of the relative price of energy resources on productivity apparently are not biased by implicit T a b le 2 assumptions about energy demand. If the assumptions Production Functions for M anufacturing introduced such biases, they and The Private Business Sector would be detected in other ( 1948 - 1 9 7 6 ) coefficients of the model esti - 0 .1 2 8 I n P«m + 0 .0 2 1 1 in x„ 1 .4 2 6 + 0 .5 2 4 I n h „ + 0 .4 7 6 I n k „ mated. In this regard, it is (4 .4 5 ) ( 4 .1 0 ) ( 2 .4 8 ) (2 .2 5 ) (- 2 .1 7 ) curious to note that, if en ergy use is less price sensi D .W . = 1 .5 9 R - 0 .9 3 S .E . = 0.021 p = 0 .8 3 tive than the approximation used by Rasche and Tatom + 0 .0 1 8t I n X = 1 .5 4 5 + 0 .7 1 5 I n h + 0 .2 8 5 I n k — 0 .1 0 9 I n implies, their estimate of the ( 1 2 .3 4 ) ( 1 3 .9 5 ) ( 5 .5 5 ) ( - 5 .1 6 ) ( 1 0 .4 7 ) energy and energy price elasticity of output would be D .W . = 1 .8 6 R = 0 .9 8 S .E . = 0 .0 0 9 p = 0 .5 9 too small, not too large.3* Thus, the claim of a low does not imply that the energy price effect observed price elasticity of demand for energy made by Perry, for the private business sector was smaller or absent DeLeeuw, and Denison would reinforce the Raschein manufacturing. Tatom conclusions if it were correct. In a related fashion, some have argued that produc tivity developments in the manufacturing sector have not been unusual.35 If increased energy prices are an important causal factor in the productivity decline, they should produce a slowing in manufacturing as 33For example, the regressions end in 1972 because the indus trial sector energy use data in the U.S. Department of Energy, Monthly Energy Review, reported since 1973, show a lower level of energy use in 1973. The more recent data also show that energy use declines much more rapidly from 1973-78 than in the Federal Energy Data System estimates. 34See Rasche and Tatom, “Energy Resources and Potential GNP,” Appendix II, p. 23. Denison, “Explanations of De clining Productivity Growth,” pp. 16-17, criticizes the price sensitivity assumption while arguing that the energy price elasticity of output must be smaller than Rasche and Tatom estimate. 35See, for example, “Behind the Slump in Productivity Growth,” Morgan Guaranty Survey (November 1978), pp. 7-11. An estimate of the Rasche-Tatom production func tion for manufacturing shows that the energy effect on manufacturing is essentially the same as that for the private business sector.36 Table 2 shows estimates of PBS output (X ) and manufacturing output (Xm) for the period 1948-76. In each case, output is a func tion of hours employed (h), the product of the sec tor’s lagged net capital stock and the Federal Reserve Board capacity utilization rate (k), and the relative price of energy in each sector (Pe), which is found by deflating the producer price index for fuel and related products and power by the implicit price de flator for output in the respective sectors. There is little difference between the size of the effect of the relative price of energy on the private business sector 3eThe theory and deviation of the equations estimated are discussed more fully by Rasche and Tatom, “Energy Re sources and Potential GNP.” Page 13 FEDERAL RESERVE BANK OF ST. LOUIS or on manufacturing output. Furthermore, the effect is statistically significant at conventional levels in each case.37 The smaller decline in productivity growth in manufacturing apparently results from the fact that, while the growth of the net capital stock in manu facturing slowed after 1973, it grew faster than hours employed. Growth in the capital stock per hour of all persons employed slowed for the private business sector during 1973-78, but accelerated in manufacturing.38 The rise in the relative price of energy was largely due to OPEC actions since 1972; the relative price of energy rose by a similar amount on a worldwide scale, except to the extent that various nations may have offset the increases by lowering their specific taxes on energy products. Consequently, if the energy priceproductivity hypothesis is correct, one would expect to observe adverse productivity developments and slower capital accumulation in other nations. Table 3 shows the trend growth rate for 11 indus trial countries from 1960-73 and the average annual growth rate for each country from 1973-78, except for the Netherlands and Belgium where growth rates are computed through 1977. In each country, produc tivity growth declined in the later period. In half the foreign countries, the reduction in the five-year aver age annual rate was more than the standard error of the prior trend-adjusted annual growth rate ( Canada, Sweden, United Kingdom, Italy, and Japan). In Italy and Sweden, the reduction was more than twice the standard error of the prior annual grow th rate. The pattern of the reduction also conforms to the U.S. experience. In every country but West Germany, pro- SEPTEMBER 1979 T a b le 3 Productivity G ro w th Rates for 1 1 Industrial N ations (O u tp u t p e r h o u r) A n n u a l G ro w th Tren d 1 9 6 0 -73 * A v e ra g e A n n u a l G ro w th Rate 1 97 3 -78 D ifference U n ite d S tate s 2 .8 % ( 2.8 ) % 1 .7 % 1 .1 % Belgium 7 .0 ( 3 .5 ) 6 .4 * * 0 .6 2 .5 1 .9 4 .6 2 .6 4 .8 * * 2 .8 1 .5 5 .2 0 .2 3 .8 Canada 4 .4 ( 1 .2 ) D enm ark 7 .2 ( 2.8 ) N e th e rla n d s Sw e d e n 7 .6 ( 3 .2 ) 6 .7 ( 1. 2 ) U n ited Kingdom 4 .0 ( 2.0 ) Fran ce 5 .9 ( 1 .5 ) 4 .7 1 .2 W e st G e rm a n y 5 .4 ( 1 .3 ) 5 .0 0 .4 It a ly 6 .6 ( 1 .3 ) 2 .6 4 .0 Jap an 9 .9 ( 4 .0 ) 3 .5 6 .4 * ( Figures in parentheses are standard error o f annual growth rate after adjustment for trend.) 3"The estimate of the energy elasticity of output from the equa tions in Table 2 is 11.4 percent in the manufacturing sector and 9.9 percent for the private business sector. Denison’s claim that this elasticity must be larger in manufacturing appears to be correct. The difference is not statistically significant, however, since the standard error for the manu facturing elasticity is 5.2 percent, while that for the private business sector is 1.9 percent. See Denison, “Explanations of Declining Productivity Growth,” pp. 16-17. The coeffi cient on hours in the manufacturing sector is low, given that the share of labor cost for the manufacturing sector during the sample period averaged 69.9 percent. The implied hours elasticity of output in Table 3 for manufacturing is only 46.5 percent, but given its standard error of 18.7 per cent, the hypothesis that the hours elasticity and labor share are equal cannot be rejected at the 95 percent confi dence level. When the equations in Table 3 are estimated with the output elasticity of hours constrained to equal the average shares of labor cost during the sample period, the estimate of the energy elasticity of output is 9.9 percent in both sectors and significant. Moreover, an F-test of this con straint fails to reject it at the 5 percent significance level. 38When a second time trend is included in the equations in Table 2 to allow for the slowing in the rate of technological change after 1966 discussed earlier, it is not statistically significant at the 20 percent significance level in either case. Digitized forPage FRASER 14 <=*Growth rate 1973-77. Data for 1978 are unavailable at this time. SOURCE: U.S. Department of Labor. ductivity growth was more than one standard error below the prior trend during 1974 or 1975. In 8 of the 11 countries, productivity growth in 1974 or 1975 declined by more than twice the standard error of the 1960-73 trend-adjusted growth rate (the exceptions are West Germany, the Netherlands, and Denmark).39 39The worldwide decline in productivity growth has also been noted recently in “ Whatever Happened to Productivity?” The Economist (August 25-31, 1979), pp. 60-61. As in the case of manufacturing above, it is possible that the labor productivity decline has been partially avoided in countries such as West Germany and Belgium by institutional arrange ments such as “indexing” which keep real wages from fall ing and lead to faster capital growth relative to employment. Such growth is manifested primarily by negative or slow em ployment growth. FEDERAL RESERVE BANK OF ST. LOUIS Finally, the rise in the relative price of energy led to incentives to reduce the desired capital-labor ratio in production through a reduced pace of capital for mation. Again, this would be expected to occur world wide. A recent report by the Bank for International Settlements shows that for the United States and five of the foreign countries in Table 3, real nonresidential fixed investment slowed dramatically after 1973 com pared with a 1955-73 trend. In the five foreign countries, such investment was virtually flat from 1973 to 1978, compared with annual trend rates prior to 1973 of 15.5 percent in Japan, 5.3 percent in Italy, 6.5 percent in Germany, 7.9 percent in France, and 5.8 percent in the United Kingdom.40 Capital formation patterns are similar in the other five foreign nations.41 In Bel gium, real nonresidential capital formation slowed from a 5.2 percent average annual rate of growth in 1965-73 to a 0.3 percent rate in 1973-76. Comparable slowings occurred in Denmark, (4.7 percent to -1.0 percent) and the Netherlands (4.5 percent to -2.3 percent). Data available through 1978 indicate that real nonresidential gross capital formation slowed in Sweden from a 3.6 percent rate in 1965-73 to a -1.3 percent rate in 1973-78. In Canada the slowing was smaller: from a 4.6 percent rate in 1965-73 to a 3.1 percent rate in 1973-78. Thus, two of the major implications of the energy price rise in the seventies appear to be borne out in other nations as well. SUMMARY AND CONCLUSION Productivity growth declined substantially during the seventies. Many analysts, regarding the decline as a cyclical phenomenon which reflected unusually high levels of unemployment, did not recognize the seriousness of the problem. In recent months, how ever, it has become more widely recognized that the productivity problem of the seventies is not a tempo rary cyclical aberration. Nonetheless, the source of the productivity problem remains controversial. Economic theory suggests several sources of the problem. With the possible exception of energy price developments during the decade, however, 40See Bank for International Settlements, Forty-Ninth Annual Report (Basle, June 1979), p. 28. The United States shows the most favorable growth of real nonresidential fixed in vestment relative to the prior trend when compared with the five other nations. I am indebted to Allan H. Meltzer for calling my attention to this dramatic evidence. 41The data used for these countries is fixed capital formation less residential construction, both in 1970 prices, from the OECD, National Accounts of OECD Countries, Vol. II, 1976. For the 1973-78 period, the data (in 1975 prices) for Sweden and Canada is from the OECD, Quarterly National Accounts Bulletin, 1978/IV and 1979/1 issues. SEPTEMBER 1979 no single factor appears to account for the entire problem. While the changing composition of the labor force has been erroneously tagged as the main factor adversely affecting productivity in the seventies, the relatively slow pace of capital formation, including research and development capital, has quietly played an important role in productivity’s decline. In addi tion, the diversion of capital formation from marketoriented production to government-regulated goals (such as clean air, occupational health and safety) has been a contributing factor. The reasons for the slowing of capital formation are as diverse as those that account for productivity developments. Both the increased relative cost of pur chasing new plant and equipment and the higher relative cost of operating capital account for a large part of the slowing. These, in turn, are largely due to energy price developments. Inflation and uncertainty about future inflation have also provided strong disincentives for business investment. Much of the disincentive effect of inflation arises through the tax system, but high inflation rates also increase the variability of inflation, which adds to the riskiness of business ventures. Coupled with increased risk aris ing from growing government influence in the market place, risk .premiums built into the costs of funds have escalated, further reducing capital formation incentives. The influence of some of the factors affecting pro ductivity in the seventies may be reversed in the eighties. For example, the sharp acceleration of govern ment regulation in the seventies, to the extent it rep resented substantial “start-up” requirements of new programs, may slow in the coming decade. The labor force will tend to become more experienced in the eighties as the proportion of young workers declines and participation rates of women level off. Barring further disruptions, large increases in the relative price of energy are less likely since pricing decisions of OPEC appear to be based upon major changes in market structure such as the formation of the car tel or production policy changes in member countries. In addition, such increases will become increasingly limited by the competitive cost of alternative sources of energy. In spite of such favorable possibilities, however, other adverse factors influencing both capi tal formation and productivity growth remain. More over, the substantial losses of productivity in the seventies are not likely to be recouped without con siderable effort. The list of possible policy options to restore produc tivity growth is quite long. Essentially it includes any Page 15 FEDERAL RESERVE BANK OF ST. LOUIS efforts that increase incentives to improve labor skills, increase the quantity and quality of plant, equipment, research and development, and increase incentives to produce and utilize energy resources efficiently. Most of the proposals involve cutting various taxes or re moving regulatory restraints to stimulate the supply of resources and output in the economy. Monetary policy can also be useful in the promotion of productivity growth. Inflation and its variability re Digitized forPage FRASER 16 SEPTEMBER 1979 tard the expected returns from investment and increase risk. Since inflation depends principally upon the rate of growth of the money supply, a policy which obtains lower and more stable growth rates of the money stock would enhance productivity growth. The most practical means for removing variability of monetary growth and achieving a lower targeted rate of growth is to conduct policy to achieve targets for the monetary base rather than targets for interest rates. Repurchase Agreements NORMAN N. BOWSHER I N recent years, an increasing number of major fi nancial developments have occurred, including such innovations as savings accounts subject to automatic transfer, NOW accounts, and money market mutual funds. Another significant financial development has been the rapid growth in repurchase agreements (RPs). RPs, or “repos” as they are frequently called, have existed for decades, but have risen substantially only in the last few years. In view of the rapidly expanding role of RPs and the general lack of understanding of them by those outside the financial community, this article discusses RPs and their principal uses and provides a perspec tive on their recent growth. Further, this article ex plores the implications and potential importance of RPs for both financial markets and stabilization policy. and lending short-term funds. RPs frequently are made for one business day (overnight), although longer maturities are not uncommon. An illustration of a “typical” RP transaction is help ful in understanding this financial instrument. Sup pose that’ the treasurer of a large corporation calcu lates the firm’s cash position for the day and determines that the firm has funds that are not re quired immediately, but will likely be needed to meet expected expenditures in a day or two. The treasurer, wishing to earn interest on these “excess” funds for a day, arranges to purchase a government security from a commercial bank with an accompanying agreement that the bank will repurchase the security on the following day. This type of transaction is illustrated in the following accounting entries. Bank C o rp o ra te Custom er B e fo re RP $1 M illio n D eposit $1 M illio n D eposit A repurchase agreement, as the term is used in the financial markets, is an acquisition of funds through the sale of securities, with a simultaneous agreement by the seller to repurchase them at a later date.1 Basically they are a secured means of borrowing C re a tio n of RP —$1 M illio n D epo sit - $ 1 M illio n D epo sit - b $ l M illio n RP B o rro w in g + $ 1 M illio n C o lla te r a l ize d Loan (R P ) JFor a detailed discussion of RPs, see Charles Lucas, Marcos Jones, and Thom Thurston, “ Federal Funds and Repurchase Agreements,” Quarterly Review, Federal Reserve Bank of New York (Summer 1977), pp. 33-48, and Thomas D. Simp son, “The Market for Federal Funds and Repurchase Agree ments,” Staff Studies 106, Board of Governors of the Federal Reserve System (July 1979). C o m p letio n o f RP A g re em en t + $1 M illio n D epo sit -|-$1 M illio n D epo sit —$1 M illio n RP B o rro w in g —$1 M illio n Lo an (R P ) Description of the RP Market Page 17 SEPTEMBER FEDERAL RESERVE BANK OF ST. LOUIS At times the corporate treasurer may decide that excess funds will be available for a longer period. For example, if the corporation has just sold long-term bonds but will not make the capital expenditures for, say, 30 days, an RP can be arranged for this specific period. This would be beneficial because the transac tion costs of a term contract are less than those for a series of daily contracts. The advantages of RP transactions are numerous. Through the RP mechanism, corporations and other holders of large cash balances can earn a secured market rate of return on these balances until they actually are used for payments. At the same time, banks find these RPs a useful source of short-term funds. Interest rate ceilings do not apply to typical RP transactions, and member banks are not required to hold reserves against funds obtained through RPs as long as the securities involved are obligations of the United States government or federal agencies.2 In addition, the parties to the transaction have greater flexibility in setting maturities than they do with cer tificates of deposit (CDs), which must have at least a 30-day maturity. 1979 on or changing market prices of the government se curities bought and sold. The role of government se curities is only to provide collateral for the lender. The interest rate on RPs usually approximates the federal funds rate, but frequently is slightly lower because RPs are collateralized borrowings whereas federal funds are not. Most RP transactions are in amounts of $1 million or more, but a few are smaller than $100,000. De spite the large sums involved, there is little financial risk in RP transactions since securities issued or guar anteed by the federal government are used for collateral, since most transactions occur between in stitutions with high credit standings, and since repur chase is usually scheduled for only a short period after sale. There is no central physical marketplace in which RPs are arranged. Transactions are negotiated by tel ephone, either on a direct basis between parties sup plying and acquiring funds or through a small group of market specialists (U.S. government securities dealers). Most large banks and business firms employ traders who maintain telephone contact with potential suppliers (or borrowers) of funds, making offers to borrow (or lend) at specific interest rates. RP transactions must be settled in immediately available funds such as deposits in Federal Reserve Banks or collected liabilities of commercial banks that may be withdrawn in cash the same business day as the transaction occurs. Most new deposits in a commercial bank are not immediately available, since they result from the deposit of checks which may take several days to collect. During this time, the credit to the depositor’s account is only provisional. A cus tomer can make funds immediately available to a bank other than the bank where funds are deposited, how ever, by transferring them through the Federal Re serve electronic communications network. The term “reverse repurchase agreement,” fre quently used by participants in the RP market, signi fies the same transaction viewed from the perspective of the lender. In an RP, the borrower sells a security in order to receive funds and repurchases it at matu rity. In a reverse RP, the lender buys a security and resells it at maturity. Large commercial banks and government securities dealers frequently arrange re verse RPs in order to obtain government securities with which to engage in an RP. In an RP contract, the borrower pays interest on the funds acquired at a rate negotiated with the lender. This interest rate is not determined by yields Growth of the RP Market 2Effective August 1, 1979, interest rate ceilings were imposed on repurchase agreements of less than $100,000 with maturi ties of 90 days or more. To prevent undue hardship, a threeyear phase-out period was provided. Effective in the statement week beginning October 11, 1979, a marginal reserve requirement was placed on “man aged liabilities” of member banks, Edge corporations, and U.S. agencies and branches of foreign banks. These liabilities include repurchase agreements against U.S. government and federal agency securities as well as large time deposits ($100,000 and over with maturities of less than a year), Eurodollar borrowings, and federal funds borrowings from a nonmember institution. The base for the marginal reserve is $100 million or the average amount of managed liabilities held as of the two statement weeks ending September 26, 1979 whichever is larger. Any increase in managed liabilities above this base is subject to an 8 percent reserve requirement. Page 18 The market for RPs has increased dramatically in the past few years, and it has become one of the major financial markets in the nation (Table 1). Since 1969, RPs of commercial banks with the nonbank public have grown at an average 26 percent annual rate, although the pace has been uneven. The sharp est gain occurred in 1973 when outstanding RPs more than doubled. In the following two years, RPs in creased at an average annual rate of 8 percent. Dur ing 1976, there was another jump (80 percent) in RPs, and since then the average annual growth rate has been 21 percent. SEPTEMBER FEDERAL RESERVE BANK OF ST. LOUIS T a b le 1 O utstan d in g RPs* (B illio n s o f D o lla rs ) Y e a r-e n d A m oun t 1969 $ 4 .9 1970 2 .8 1971 4 .9 1972 6 .0 1973 1 3 .3 1974 1 4 .8 1975 1 5 .5 1976 2 7 .9 1977 3 6 .3 1978 4 3 .8 1979 4 5 .0 (J u n e ) ♦Estimates o f RPs o f all commercial banks with the nonbank public by staff o f Federal Reserve Board o f Governors. The growth in RPs over this period reflects several factors. The most important motivations for lending in the RP market are the prohibition of interest pay ments on demand deposits and the higher market interest rates since the mid-1960s. Also, both the wider use of computers and more sophisticated cash management systems have facilitated the collection and transfer of large volumes of funds, contributing to the growth of the market. 1979 other sources of funds are plentiful, they supply RP funds. In addition, dealers act as financial intermedi aries or brokers between demanders and suppliers of funds. By “matching” an RP transaction and a re verse RP transaction, they may profit by the differ ence in interest rates. Dealers sometimes use reverse RPs to acquire securities in order to make a short sale; the net cost of obtaining the securities through reverse RPs is frequently less than the cost of bor rowed securities. Many types of institutions supply immediately available funds in this market, including nonfinancial corporations and state and local governments. Busi ness firms and municipalities with large cash balances are able to earn sizable returns (averaging about a 10 percent annual rate in the first nine months of 1979) on these funds by arranging RPs. Such a trans action converts the corporation’s or municipality’s de mand deposit balance into an interest-earning asset. Yet, since the funds are committed for only a brief period, they are still readily available for transactions purposes. Another advantage of RPs to certain nonbank in vestors is the flexibility in recording these transactions on their books. Some investors choose to record the ownership of U.S. government securities rather than the ownership of RPs. This reporting feature is par ticularly appealing to those institutions, such as state and local governments, that are required to invest in Treasury securities. Participants in the RP Market Over the past decade, banks have obtained increas ing proportions of their total resources in the open market. This activity, which involves bidding for li abilities in order to expand assets, is commonly called liability management. This behavior differs from the traditional role of banks, which is to receive deposits from customers and to use these funds to lend or invest. Devices now used to obtain such funds in clude federal funds borrowing, negotiable CDs, com mercial paper, Eurodollar borrowing, and RPs. Large banks usually are borrowers of funds in the RP market. These institutions typically seek funds and have portfolios of government-issued and -guar anteed securities with which to transact. Government security dealers also are important par ticipants in the RP market. They are generally net borrowers of funds so that they can carry their size able portfolios of government securities. Occasion ally, however, when their inventories are low or when Business firms and others hold cash primarily to bridge the periods when expenditures exceed receipts. Traditionally, most of these cash balances have been held in the form of demand deposits in commercial banks and, as a result, have not earned explicit inter est since 1933. This was not of great significance dur ing the low interest rate periods prior to the 1960s. The higher interest rates in the sixties and seventies, however, created the incentive for businesses to de velop better cash management techniques just as banks were developing liability management tech niques. A variety of procedures were adopted to achieve a reduction in cash balances, including speed ing up receipts, slowing down disbursements, and converting cash into interest-bearing liquid assets. RPs have become a particularly useful tool of cash management. They generate sizable income during periods of relatively high short-term interest rates. Moreover, they are relatively secure and liquid. In short, some corporations and municipalities treat RPs as income-earning “demand deposits.” Page 19 SEPTEMBER FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve System regulations play an impor tant role in the RP market by limiting the type of transactions member banks may undertake. Federal Reserve actions also influence federal funds rates which, in turn, dominate the interest rates available on RPs from day to day. The Fed is also a large direct participant in the RP market. Although the Federal Reserve supplies reserves primarily through outright purchases of government securities, it also uses RPs and reverse RPs to temporarily supply or absorb reserves. Implications of RPs for Financial Markets The development of the RP instrument has led to increased efficiency in financial markets. Both inves tors and borrowers are provided with an instrument which has a combination of risk, maturity, flexibility, and liquidity characteristics not previously offered by other money market instruments. RPs are backed by government securities and are adaptable for a wider range of uses than most other financial instruments. While participation in the federal funds market is limited to commercial banks, mutual savings banks, savings and loan associations, and federal agencies, any person or firm may deal in RPs. RPs can be ar ranged for any maturity, while CDs, an alternate source of funds to banks, cannot be issued for less than 30 days, and commercial paper is seldom written for a period as brief as a day or two. Judging by the rapid growth of RPs since 1970, the demand for financial instruments with the combination of characteristics of RPs has been substantial. By providing a mechanism through which large supplies of liquid balances are “auctioned” daily, the RP market has improved the ability of the economy to channel short-term funds into areas of greatest de mand. Thus, they enhance the efficient allocation of resources. Moreover, RPs have bolstered the liquidity of the instruments eligible as collateral against RP borrow ings, including longer-term government and agency obligations. The securities become more liquid since they may be used to raise funds on short notice with out incurring the risk of capital loss that might occur if an outright sale were the sole option. RPs also facilitate arbitrage between financial in struments, as well as provide a straightforward source of cash or investment outlet for short-term funds. For example, RPs have broadened the sources of securities Page 20 1979 available to traders active in short selling. In addi tion, RPs can be used during a period of rising inter est rates to supplement the yield on portfolios by allowing investors to purchase higher-yielding instru ments without having to sell outright and take a cap ital loss on the securities in the portfolio. Those securities may be sold under an RP and the funds used to buy new securities. Implications of RPs for Monetary Policy Whether RPs should be included in measures of the money supply is still an unsettled issue. Unlike currency and demand deposits, RPs are not used di rectly as a medium of exchange, nor are they likely to be used for transactions in the near future. Never theless, the maturities of RPs are so short (frequently just overnight) that owners of the funds can treat them virtually as demand deposits. They may even write checks on the funds since the RPs will become available as deposits before the checks clear. In fact, the RP mechanism permits demand deposits to be much larger during the day, when transactions are conducted, than at the close of the day when the money stock is measured and reserve requirements are imposed.3 The accounting entries on the opposite page illus trate the balance sheet effects of a typical $1 million bank demand deposit with and without the use of RPs. By using the overnight RP, the customer has access to the funds (deposit) during the banking day and earns a return on the funds for the overnight loan ( RP). Also, with the RP the bank is able to extend more credit from the funds supplied. In addition, the reserves sup plied to the banking system are then available to sup port further bank credit and deposit expansion. In practice, although many RP contracts are writ ten for very short terms, a sizable portion of the funds used remains in RPs for relatively long periods. A number of RP contracts initially are written for longer terms, and some are made under continuing contracts, with the provision that either party can withdraw from the arrangement at any time.4 In many cases, the daily contracts are simply “rolled over” ( extended for 3However, many customers are likely to perceive their demand deposits to be reduced at the time the RP is arranged, usu ally early in the day. Thus, the timing of the bookkeeping may have little significance. 4 According to a Federal Reserve survey in the week ending December 7, 1977, about 10 percent of the RPs with nonbank customers had maturities of over 30 days, 22 percent had maturities of 8 to 30 days, and 17 percent had maturities of FEDERAL RESERVE BANK OF ST. LOUIS B an k D epo sit o f Funds (n o RPs u se d ) $ 8 5 0 ,0 0 0 Lo an s an d In vestm ents $1 M illio n D eposit SEPTEMBER N o n b a n k Cu stom er $1 M illio n D epo sit $1 M illio n N e t W o rth $ 1 5 0 ,0 0 0 R equired C ash R eserves (1 5 % assum ed re q u ire m ent) D epo sit o f Funds U sin g R Ps: D uring D ay M illio n Lo an s an d In vestm en ts $1 M illio n N e t W o rth $1 M illio n D eposit $1 M illio n D epo sit $1 M illio n RP B o rro w in g $1 M illio n $1 M illio n G o ve rn m e n t N et W o rth S e cu rity o r $1 M illio n C o lla te r a l ized Loans (R P ) N o rese rve s req u ire d sin ce not im posed u ntil close of d a y A t C lo se of Bank D a y $1 M illio n lo a n s & (R P 'n e ffe ct) In vestm en ts N o rese rve s req u ire d on RP fu n d s a longer period). Corporations use these longer-term arrangements when they are accumulating funds in anticipation of payments for taxes, dividends, pay rolls, construction, or other large outlays. To the extent that RPs are used to accumulate liquid balances over a period for some anticipated future outlay, they may be more appropriately classi fied as time deposits rather than demand deposits; such balances would be more appropriately included in the M2 concept of money which includes liquid savings, rather than the Mi concept which does not. Even if it is concluded that RPs are not money ( M ^ , however, the rapid growth of this highly-liquid asset has almost certainly affected the velocity of demand deposits by permitting corporations to obtain desired liquidity with fewer demand deposits than otherwise. The existence of the RP market, as well as the RPs themselves, have probably reduced the demand to hold demand deposits. For instance, the greater abil ity of nonbanking firms to borrow on short notice by 2 to 7 days, and slightly over half were 1 day or continuing contract. Wayne Smith, “ Repurchase Agreements and Fed eral Funds,” Federal Reserve Bulletin (May 1978), pp. 353-60. 1979 “selling” government securities on RPs reduces the price risk which otherwise exists if the securities were sold outright when cash needs arise. Hence, business firms are better able to substitute government secu rities for some demand deposits. Money plays a unique role in economic stabiliza tion. Since the demand for money relative to total spending has usually been relatively stable, changes in the money stock produce similar changes in total spending for goods and services. A number of re searchers working with economic models of the econ omy, however, have found an apparent shift in the money demand equation in the mid-1970s.8 Prelimi nary analyses suggest that the apparent shift in the money demand function would have been somewhat smaller if RPs are included in the stock of money. The evidence does not settle the conceptual issue of whether RPs are money, but the studies do provide some empirical support for including RPs in the defi nition of money for policy purposes. Even if RPs are not judged to be money, they are closer substitutes to it than other near monies and help explain the prob lems in estimating the money demand function.8 On the other hand, it has been argued that RPs are not money (Mi) .7 Most RP transactions are actually made by early afternoon, and thus do not represent an automatic investment of end-of-day balances. Also, if the shift in money demand in 1974 was a one-time exogenous shock, the more stable relationship by in cluding RPs in the money demand function would not necessarily last, and money-demand specifications without RPs would be correct. In any case, the effect of RPs on the money stock (or money demand) is less than indicated by the total RPs outstanding. Interbank RPs are offsetting and have no net effect on the demand to hold money. Transactions among the nonbank public — for exam ple, an RP between a business concern and a govern 5Jared Enzler, Lewis Johnson, and John Paulus, “ Some Prob lems of Money Demand,” Brookings Papers on Economic Activity (1: 1976), pp. 261-80. eGillian Garcia and Simon Pak, “ Some Clues in the Case of the Missing Money,” American Economic Review (May 1979), pp. 330-34. See also John Wenninger and Charles Sivesind, “ Defining Money for a Changing Financial Sys tem,” Quarterly Review, Federal Reserve Bank of New York (Spring 1979), pp. 1-8, and John Scadding, “Are RPs Money?” Federal Reserve Bank of San Francisco Weekly Letter, Federal Reserve Bank of San Francisco (June 29, 1979), pp. 1-3. 7Richard Porter, Thomas Simpson, and Eileen Mauskopf, “Financial Innovation and the Monetary Aggregates,” Brook ings Papers on Economic Activity (1 : 1979), pp. 213-29. Page 21 FEDERAL RESERVE BANK OF ST. LOUIS ment bond dealer — also have little net effect, since what one firm gains the other forgoes. Similarly, when a bank engages in a reverse RP in order to obtain government securities to facilitate an RP, these matched transactions become largely counterbalanc ing. Transactions that have the largest monetary effects are those in which a commercial bank deals with the nonbank public, securing the transaction with government securities from its own portfolio. Conclusions Repurchase agreements have grown markedly in recent years, attaining the status of a major financial instrument. Since RPs provide a market yield on funds available virtually on demand, they are useful for pri Digitized forPage FRASER 22 SEPTEMBER 1979 vate businesses and municipalities which must hold large highly-liquid balances. RPs are also valuable to banks and government bond dealers as a source of short-term funds. The public generally has also bene fited since funds and resources have been more effi ciently channelled by RPs. Everything else being equal, the pronounced ex pansion in RPs has stimulated total demand for goods and services. Because of data limitations and many offsetting impacts, however, knowledge of the net extent of this stimulative effect is limited. It is un settled whether analysis of the impact of RPs would be improved by considering them (or a portion of them) as money or as a force affecting the demand for money. FEDERAL RESERVE BANK OF ST. LOUIS SEPTEMBER 1979 Publications of This Bank Include: Weekly U. S. FINANCIAL DATA Monthly REVIEW MONETARY TRENDS NATIONAL ECONOMIC TRENDS Quarterly CENTRAL MISSISSIPPI VALLEY ECONOMIC INDICATORS INTERNATIONAL ECONOMIC CONDITIONS Annually ANNUAL U. S. ECONOMIC DATA Single copies of these publications are available to the public without charge. For information write: Research Department, Federal Reserve Bank of St. Louis, P. O. Box 442, St. Louis, Missouri 63166. Page 23