The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
FEDERAL RESERVE BANK OF RICHM OND MONTHLY REVIEW Cyclical Indicators of Economic Activity: Part I INTRODUCTION Techniques developed for the purpose of pre dicting changes in economic activity date back as far as 1862 when Clement Juglar, a French statis tician, observed that time series data on “ prices and finance” appeared to suggest cyclical movements in business conditions. Since that time, significant pro gress has been made in business forecasting and prediction. Today, the important types of forecast ing techniques include (1 ) comprehensive economic models, of varying degrees of formality, that postu late certain relationships between key variables; (2 ) extrapolative methods that depend upon past and current values of particular variables in de termining future values; and (3 ) analysis of selected business cycle indicators chosen for their consistency in signaling turning points in aggregate economic activity. This article is the first in a two-part series de signed to familiarize the reader with the major uses and limitations of business cycle indicators, a group of statistical series that have proved useful to analysts o f cyclical or cycle-like swings in business. Analysts generally group these indicators into three classes: (1 ) the so-called leading indicators, which move in advance of changes in the direction of gen eral business activity; (2 ) the coincident indicators, which move coincidentally with general business; and (3 ) the lagging indicators, which usually change direction after general business has turned around. The first article concentrates on the leading indi cators. A subsequent article will discuss coincident and lagging indicators, along with the so-called dif fusion indexes that are useful in confirming cyclical turns in economic conditions. The first group of widely-known economic indi cators was published in 1919 by the Harvard U ni versity Committee on Economic Research under the direction of Warren M. Persons. The Harvard In dex Chart consisted of three basic categories of ac tivity: ( A ) speculation (stock p rices); (B ) busi ness the development of techniques used in the selection of modern business indicators. The identification of particular statistical series that are especially useful in studying business fluctua tions has been largely the work of the National Bureau of Economic Research (N B E R ). Selection is based on economic significance, statistical ade quacy, historical conformity to general business fluctuations, smoothness, currency of publication, and cyclical timing. Some of the indicators are classified according to timing. Generally speaking, these are of greatest immediate interest to business analysts, although all the selected series can be useful in ex plaining business fluctuations. Other indicators that play an important role in the explanation of cyclical movements are also selected, but these are not gen erally as reliable as those classified by timing. The business cycle indicators used by today’s analysts are an outgrowth of the work of the N B E R . Since its inception in 1920, this organization has made extensive studies of massive amounts of sta tistical data bearing on business activity. Over 500 series were investigated before its first list of indi cators was published in 1938. The initial list con sisted of twenty-one indicators of cyclical recovery. M ajor reviews and updates of the series were con ducted in 1950, 1960, and 1966. A t the present time, the N B E R classifies and re ports seventy-three individual economic series by timing and economic process. These series are re ported monthly in Business Conditions Digest (B C D ) — formerly called Business Cycle D evelop ments— a publication of the U. S. Department of Commerce. Selection of the cyclical indicators re ported in BCD is the responsibility of the N B E R , although the Department of Commerce includes several additional measures of economic activity. O f the seventy-three indicators currently reported by B C D , thirty-seven are leaders, twenty-five are coinciders, and eleven are laggers. They are cross classified into six economic processes as shown in Table I. (wholesale prices, later bank debits) ; and (C ) money market (short-term interest rates). The A -B -C sequence, as it was commonly called, has been developed because of its outstanding conformity to business conditions prior to W orld W ar I. A l though the index performed well in the 1920’s, its failure to anticipate the Great Depression led to its demise in the thirties. 2 Its major contribution was LEADING IN DICATO RS Leading indicators are those indicators that, in the past, have generally led changes in aggregate busi ness conditions. They are used mainly to decrease the recognition lag of cyclical turning points. Be cause the leads vary considerably in length and changes in individual series are often erratic, their use in the determination of significant turning points in the general economy usually requires the ex ercise of considerable caution. Successive increases or declines in leading indicators have occasionally occurred in the past without subsequent turning points in aggregate business conditions. Generally speaking, leading indicators include “ flow ” series as opposed to “ stock” series since flows change direction before their corresponding stocks. For example, investment in plant and equip ment at a rate greater than capital depreciation adds to the stock of capital. Hence, the stock of capital can continue to grow even though the current flow of investment expenditure may be declining. Other typical examples of leading indicators include build ing permits or contracts that precede actual con struction, and job vacancies that precede changes in employment. In addition to their value in anticipating future turning points, study of the leading indicators has helped to develop and explain links between different types of economic activity. Reasonably consistent patterns have developed between many of the leaders and the coincident and lagging indicators that follow. Although the selection is based primarily on his torical timing, economic logic also plays a significant role in the selection process. N B E R Short L ist and C om posite Index T he N B E R specifies a short list of twelve indicators from its complete list of leading indicators. They purport to provide a current view of substantially undupli cated economic processes that have been reasonably consistent in leading previous cyclical turning points. Eleven of the series are reported monthly by the source agency; the twelfth— corporate profits after taxes— is reported quarterly. A convenient summary measure of the twelve leaders is computed monthly and reported as a com posite index. In computing the composite index, each series is standardized so that all have an equal opportunity to influence the total index. Then, each of the twelve indicators is weighted according to its score (past performance) as an economic indicator.1 The composite index of twelve leading indicators has been criticized by forecasters on several grounds. One of the major weaknesses lies in the extreme variability of its leads for business peaks and its extremely short leads at business troughs. Lead times of the composite index were 6, 22, 13, and 7 months at the 1953, 1957, 1960, and 1969 peaks. 1 For a more detailed description o f the construction o f composite indexes, see Business Conditions D igest (Septem ber 1969 ), p. 104. For discussion o f the scoring system, see Geoffrey H . M oore and Julius Shiskin, Indicators o f Business E xpan sions and Contractions (N ew Y o r k : N B E R , 1 9 6 7 ). Table I CROSS-CLASSIFICATION OF CYCLICAL INDICATORS BY ECONOMIC PROCESS AND CYCLICAL TIMING N. Cyclical Timing Economic Process 1. E M P L O Y M E N T AND U N EM P L O Y M E N T (15 series) II. PRODUCTION, INCOM E, C O N SU M P T IO N , AND TRADE (8 series) III. F IX E D CAPITAL IN V E ST M E N T (14 series) IV. IN V E N T O R IE S AND IN VEN TO R Y IN V E ST M E N T (9 series) V. PRICES, COSTS, A N D PR O FIT S (10 series) VI. M O N EY A N D C R E D IT (17 series) Formation of business enterprises (2 series) New investment com mitments (8 series) Inventory investment and purchasing (7 series) Sensitive commodity prices (1 series) Stock prices (1 series) Profits and profit margins (4 series) Flows of money and credit (6 series) Credit difficulties (2 series) Comprehensive wholesale prices (2 series) Bank reserves (1 series) Money market interest rates (4 series) Unit labor costs (2 series) Outstanding debt (2 series) Interest rates on business loans and mortgages (2 series) N. Marginal employment adjustments (6 series) L EA DIN G IN D IC A T O R S (3 7 series) RO U G H LY C O IN C ID EN T IN D IC A TO R S (25 series) Job vacancies (2 series) Comprehensive employment (3 series) Comprehensive unemployment (3 series) Long-duration unemployment (1 series) Backlog of investment Comprehensive production com mitments (3 series) (2 series) Comprehensive income (2 series) Comprehensive consumption and trade (3 series) Investment expenditures (2 series) LAGGING IN D IC A TO R S (11 series) Source: U. S. Departm ent of Commerce, Business Conditions Digest, June 1971. Inventories (2 series) Lead times were 5, 0, 2, and 0 months respectively for the 1954, 1958, 1961, and 1970 cyclical troughs. Although the leads are highly variable, generally speaking they have been too long in expansionary periods to be helpful in signaling a need for policy change to prevent an oncoming recession. Further more, contractions have frequently come to an end and recovery has begun before a change in direction of the leading indicators is confirmed. Adjustments in the Composite Because m ost o f the components in the composite index are flows or changes in stock components that exhibit no secular trend, the composite index itself exhibits no secular upward trend. On the other hand, the components of the composite coincident index are primarily pro duction and employment series that follow a secular trend similar to that of aggregate economic output. T o facilitate comparison of the leading composite with the coincident and lagging composites, the lead ing index has been adjusted. The adjustment tech nique was developed under the direction of Julius Shiskin, Chief Economic Statistician at the Bureau of the Census. In brief, the technique involves re moving whatever trend exists in the leading com posite and adding the trend exhibited by the coinci dent composite to the leading index. The statistical method is called “ reverse trend adjustment” since a trend is added to the series rather than removed from the series in the usual statistical sense.2 Reverse trend adjustment of the composite index has also reduced the lead variability at cyclical peaks and troughs and has diminished the likelihood of false signals of recessions. Trend adjustment de creases lead times in expansions and increases lead times around troughs. It also moderates false signals of recessions that occur in periods of sustained ex pansion. In summarizing the advantages and limita tions of reverse-trend adjustment, Shiskin states: “ Reverse-trend adjustment promises to be another advance in the developm ent of . . . techniques [that make statistical data serve practical ends]. No one should expect it to make the leading indicators error-proof forecasting tools, nor to eliminate the difficulties of interpreting current changes.” 3 Preliminary and Revised Composite Index A prelim inary m onthly report of the com posite leading indicators is issued b y the Com m erce D e partment. A t the time of release, only eight of the tw elve individual series are usually available. Three of the four m issing com ponents are re 2 Julius Shiskin, “ Reverse Trend Adjustm ent of Leading Indicators,” The R eview o f Econom ics and Statistics, 49, N o. 1 (February 1 9 6 7 ), 45-49. 3 Ibid., p. 49. 4 ported on a m onthly basis at a later date. A s these figures becom e available, the com posite in dex is revised to show the influence of these factors. Th e fourth com ponent, corporate p ro fits after taxes, is available on a quarterly basis only. This figure is linearly interpolated into a m onthly series when it is reported by the source agency, and the com posite index is again revised. A dditional revisions are som etim es made as a re sult of changes in seasonal adjustm ent factors or when other more com plete inform ation is fur nished by the source agency of the com ponent series. Initial reports for the composite index are issued with approximately a one month lag. The largest revisions for a given monthly index usually occur in the next one to two months. A s a result, lead times should be at least two months, preferably longer, to be useful in anticipating turning points. Performance Record P erform ance of the leading indicators has been measured by several different tests. The most common criteria include determina tion of average lead times and variability of lead times. Other common criteria a re : the percentage of times an indicator or group of indicators within the composite group actually led the cycle, the percentage of times a series or group turned at a business cycle turning point, and the percentage of times a series or group turned without a turn in general business conditions. Individual leading indicators, as a rule, are not as reliable for purposes of prediction as the com posite group of indicators. Single components often have lead times that vary considerably from one cycle, or a phase of a cycle, to another. The index for the group as a whole is generally much smoother since erratic swings in individual components often offset one another. A s a result, the primary usefulness in analyzing individual components is to determine sectors of the economy that are likely to weaken in the near future. For example, a decline in contracts and orders for plant and equipment for several months will lead a decline in business expenditures for plant and equip ment. Careful study of the components may also be helpful in determining other less obvious links be tween different types of economic activity, as sug gested earlier in this article. The performance of thirty leading indicators has been studied by Michael K. Evans over the post w a r period through 1965.4 His requirements were 4 Michael K. Evans, Macroeconom ic A c tiv ity : Theory, Forecasting, and Control (N ew Y o rk : H arper & Row , Publishers, Inc., 1 9 6 9 ), pp. 455-60. rather restrictive in that he considered all series with mean leads of less than four months of little value in anticipating turning points. This procedure elimi nated six series at the peaks and twenty-four series at the troughs. He also eliminated series with highly variable lead times by deleting those series with high variability compared to the average (mean) lead over the previous peaks and troughs. For example, a series that led one downturn by twenty months and another downturn by two months would have been eliminated by Evans since the variation in lead times from cycle to cycle was so large compared with the average lead for that series. High variance in lead times eliminated nineteen of the remaining twentyfour series at peaks and three of the six series re maining at troughs. O f those passing both tests, only two were efficient in predicting both peaks and troughs. Evans’ procedure illustrates one of the major rea sons why the forecaster should be wary of using in dividual series in predicting future economic con ditions. Many analysts have cautioned against the use of individual time series for the purpose of an ticipating turning points. A s mentioned earlier, the primary research value in studying separate com ponents is to discover and analyze links between dif ferent types of economic processes. Evans continued his study of the thirty leading indicators with an investigation of false signals. and 1962), he found that only four of the thirty lead ing indicators in his sample gave no false signals; three others had relatively small declines compared to their average declines in actual recessions. teen of the indicators signaled one or less false Adding the 1967 false downturn to Evans’ results changes them very little. Three showed no ap preciable turn, and eight other indicators declined C h art In d e x : 1950 Sou rce: 1952 1954 (July) (A p r.) P T 1956 1958 P 1 1967=100 ( M a y ) (Feb.) T 1960 (N o v .) (N o v .) P T 1962 U. S. D e p a r tm e n t o f C o m m e rce , B u sin e ss C o n d it io n s D ig e st, Ju n e 1971. F if turns. COMPOSITE INDEX OF TWELVE LEADING IN DICATO RS (July) (A u g . ) P T In his sample of four false downturns (1952, 1956, 1959, 1964 19 66 1968 1970 The 1952 and 1967 downturns in business activity were the weakest non-recessionary periods in the post-W ar era. Indeed, declines in real G N P were re corded in two quarters in the 1951-1952 slowdown and in one quarter in 1967. O f the non-recessionary slowdowns since the W ar, only the 1962 period failed to experience at least one quarterly decline in real GN P. Real growth did, however, fall to an an nual rate of 2.2 percent by the end of 1962. Hence, it is clear that the false signals given by the leading indicators in the non-recessionary post-W ar periods did in fact signal slowdowns in economic activity, even though these slowdowns were not severe enough to be classified as recessions. less than their average decline in actual recessions. (T w o are no longer reported in B C D .) Since components of the N B E R short list of twelve leading indicators were selected because of their better-than-average performance record and economic significance in previous cyclical periods, the short list has been more reliable than Evans’ list of thirty indicators. The performance record of the twelve leading indicators is summarized in Table II. Average lead times at peaks varied between sixteen months for new housing permits to six months for both plant and equipment contracts and orders and change in value of inventory for manufacturing and trade establishments. Average lead times at troughs were considerably shorter. Again, new housing per mits exhibited the longest lead time with an average of seven months. The ratio of price to unit labor cost in manufacturing was the only leading indicator in the group that failed to show a positive mean lead time over the five post-W ar recessions. False signals were registered by eight of the twelve indicators in non-recessionary downturns. One in dicator exhibited four false signals, and two others exhibited three false signals. The only indicator that failed to signal a recession was contracts and orders for plant and equipment. The indicator fluctuated around a flat trend line prior to the 1960-1961 recession but failed to show any appreciable decline prior to the turning point. The composite index of twelve leading indicators Sum m ary T he m ajor contribution of the leading indicators is to help forecasters and policymakers to recognize turning points in general economic con ditions before they actually occur. Their record of performance suggests that they have been reasonably reliable in anticipating downturns. One of the ob vious weaknesses of the indicators, however, has been their inability to determine the magnitude of expected slowdowns. The percentage decline in the leading indicators is not highly correlated with the severity of subsequent declines in economic activity. Inconsistent lead times also pose a problem to users of leading indicators. Lead times at peaks are usually longer than lead times at troughs, and spe cific indicators are not consistent from peak to peak and trough to trough. There is little doubt, how ever, that the leading indicators are useful— where model builders have had little success— in determin ing the timing of prospective changes in economic activity. Clyde H . Farnszvorth, Jr. had a mean lead time of five months around peaks and a mean lead time of four months around troughs. It did not fail to signal any of the post-W ar reces sions, but it did falsely signal a recession in 1952 and 1967. T a b le II PERFORMANCE RECORD OF TWELVE LEADING INDICATORS, 1945-1971 M e a n Lead T im e,* Se rie s P e a k s (m os.) Nam e A v e r a g e w o r k w e e k , p ro d u c tio n w o rk e rs, m a n u f a c tu r in g A v e r a g e w e e k ly in itia l c la im s, S t a t e u n e m p lo y m e n t in su ra n c e N e w b u ild in g p erm its, p r iv a te h o u s in g units N e t b u sin e s s fo r m a t io n N e w o rd e rs, d u r a b le g o o d s in d u strie s C o n t ra c t s a n d o rd e rs, p la n t a n d e q u ip m e n t C h a n g e in b o o k v a lu e , m a n u f a c t u r in g a n d t r a d e in v e n to rie s *T h e 1948-49, 1953-54, 19 57 -5 8, 1960-61, 6 6 12 In d u s tr ia l m a t e ria ls prices S to ck prices, 5 0 0 co m m o n stocks C o r p o r a t e p ro fits a fte r t a x e s R a tio, price to un it la b o r cost, m a n u f a c tu r in g C h a n g e in c o n su m e r in sta llm e n t d e b t C o m p o s ite in d e x , reverse tren d a d ju ste d * * B a s e d on a r a n g e o f 0 to 1 0 0 % . Y o rk : N B E R , 1967). 11 15 16 1 5 *** 7 9 9 14 13 5 19 6 9 -7 0 re c e ssio n s w e r e M e a n Lead T im e,* T r o u g h s (m os.) False S ig n a ls (N o .) 2 3 1 2 7 3 2 2 2 2 2 1 4 4 5 2 2 2 2 2 0 1 3 4 3 2 Source: 0 0 0 0 0 1 0 0 0 0 0 0 0 A v e r a g e -* * Sc o re (percen t) 66 73 67 68 78 64 65 67 81 63 69 63 u se d to d e te rm in e le a d tim es. Se e G e o f f r e y H. M o o r e a n d J u liu s S h isk in , In d ic a t o r s o f B u sin e ss E x p a n s io n s a n d C o n t r a c t io n s (N e w * * * l n d e x o f net b u sin e s s fo r m a t io n s w a s n ot a v a i la b le in d e t e r m in in g le a d tim e p rio r to the 1948 p e a k . U. S. D e p a rtm e n t o f C o m m e rce , B u s in e ss C o n d it io n s D ig e st, Ju n e 1971. F a ilu re to S i g n a l (N o .) Income Distribution and Its Measurement PART I= DISTRIBUTION AMONG THE FACTORS OF PRODUCTION Economic analysis in recent years has focused in creasingly on the question of income distribution. There are a number of reasons for this resurgence of interest in a subject which occupied much of the attention of nineteenth century economists. In the first place, public concern over the problem of poverty has stimulated efforts to determine whether the gap between the poorest stratum and the rest of an in creasingly affluent society is narrowing or widening. Second, recent experience with inflation and unem ployment has generated a suspicion in some quarters that these two economic evils may have resulted in a significant redistribution of purchasing power among socioeconomic groupings. Then, too, the steadily increasing emphasis on human capital in eco nomic analysis has pointed up the connection between education and productivity on the one hand and in come on the other, suggesting that wide disparities in income levels might indicate large long term losses of output for society. Finally, the increasing quantity and quality of national income data has enabled re searchers to undertake empirical evaluation of long accepted but largely untested theoretical models of distributive shares. Analysts tackling the subject generally distinguish between size distribution and functional distribution of income. Size distribution refers to the division of income among families and individuals classified by income brackets. Functional distribution denotes the division of the national income among the factors of production— land, labor, capital, and enterpreneurship— that combine to produce it. The paragraphs that follow outline the evolution of distributive share analysis in economic thought, discuss the behavior of the functional distribution of income in the United States, and describe some of the methods and measures employed by researchers who study it. A second article, to appear in a future issue of the Monthly Review, will discuss the size distribution of income. EARLY DISTRIBUTIVE SHARE A N A LYSIS Traditionally, economists have devoted more at tention to the functional than to the size distribution of income. Early nineteenth century economic analysis was dominated by the view, associated largely with David Ricardo, that the study of dis tributive shares held the key to the understanding of the entire economic mechanism, including the forces determining the rate and character of eco nomic growth. T o the Classical Economists of nineteenth century England, who took their cue largely from Ricardo, the distribution of income served three purposes. It divided the recipients into mutually exclusive economic groups, identified by their function in the production process; it served as an indicator of the relative welfare of the re spective grou ps; and it defined the social classes that would play key roles in the economic evolution of the nation. Economic development was looked upon as a drama in which the actors were grouped by economic function, serving specified socioeconomic roles. For example, the working class not only sup plied labor but, through procreation, insured the existence of labor supplies in perpeturity. The in dustrial class was associated with accumulation and the capital-supplying function, while the landed aristocracy exercised stewardship over land, a scarce and increasingly remunerative resource. On the basis of this model, British economists predicted that excessive procreation by the laboring class would combine with diminishing returns in land cultivation to bring bare minimum subsistence wages to labor, zero profits to capitalists, riches to landowners, and eventually cessation of growth for the economy as a whole. Karl Marx, writing later in the century, also identified each factor of production with a distinct social class. Following the classical tradition, his analysis assumed that no income recipient could be long to more than one economic group, supply more than one type of productive resource, nor receive more than one type of factor income. A laborer could not simultaneously be a capitalist, nor a capital ist a laborer. In M arx’s scenario, accelerating antagonism between an ever growing laboring class doomed to subsistence wages and an increasingly exclusive and wealthy capitalist class meant the eventual end of traditional capitalist socioeconomic organization, along with its political superstructure. By the time M arx systematized his model of class conflict, however, a new breed of classicists were weaving an intricate analysis demonstrating that the free market would achieve distributive justice and (Continued on page 10) 7 C H A R T S 1 A N D 2 Consum er credit, w hich fi nances nearly 20% of the value of all consumer purchases, is comprised of short- or intermediateterm loans for automobiles, household appliances, and other personal debts. Although consumer spend ing is determined primarily by current and expected levels of income, changes in the volume of consumer credit, along with changes in saving, have often contributed to some short-run movements in con sumption. Normally, as income rises, so does spend ing, saving, and the accumulation of debt. Sudden changes in income, however, often result in adjust ments in saving and debt accumulation in order to maintain spending. On the other hand, during periods of uncertain economic conditions, consumers may postpone current spending and increase saving while curtailing debt accumulation. All these phe nomena have occurred in recent years. From 1965 through the first half of 1968, rapid economic expansion was accompanied by similar in creases in consumer spending, saving, and credit. Following the surtax of July 1968, the expected slowdown in consumer spending did not materialize. Instead, consumers relied more heavily on consumer CONSUMER CREDIT TRENDS credit and decreased their rate of saving in order to maintain spending levels. This pattern continued into the tight money period of 1969. The saving rate increased sharply in the third quarter of 1969 and continued to rise in 1970. Faced with un certainty over inflation, unemployment, and political problems, consumers reduced their borrowing in re lation to disposable income. During the first quarter of 1971, however, consumer spending regained some of its lost strength, despite a drop in consumer credit outstanding. Again consumers appeared to rely on a reduced saving rate to finance current spending. C H A R T 5 Installm ent credit extended reached a peak of 16.8% of disposable income in the second quarter of 1969 and then decreased slowly until the first quarter of 1970. The decline in this percentage was caused primarily by the decrease in automobile sales and the consumer’s desire for cheaper, smaller cars. C H A R T 3 Consum er credit is com posed o f in stallment and noninstallment credit. Noninstallment credit, the smaller of the two components, consists C H A R T 4 M ore than 80% of all consum er credit outstanding is extended on an installment basis, with repayment scheduled in more or less equal monthly installments running over a specified period. Automobile paper, other consumer goods paper, re pair and modernization loans, and personal loans are the major components of installment credit. Automobile loans, the largest and most volatile com ponent, make up approximately 38% of all install ment credit. Automobile credit expanded rapidly throughout 1968, tapered off in 1969 and 1970, then of single-payment loans, charge accounts, and serv declined during the General Motors strike in the fourth quarter of 1970, primarily because of the wrork ice credit. Single-payment loans and charge accounts, fourth quarter of 1970. stoppage in the industry in that quarter. although decreasing slightly during the economic stallment credit continued to expand throughout this Ch art 1 C h art C O M P O N E N T S OF CO NSUM ER CREDIT O U TSTAN D IN G slowdown of 1969-70, have shown substantial net in creases over the past three years. Service credit was singularly unaffected by the slowdown and has continued to increase steadily since 1968. period; however, their small increase was not suf ficient to offset the drop in automobile paper in the fourth quarter of 1970. In the first quarter of 1971 the surge in automobile sales following the strike was not accompanied by an increase in automobile paper outstanding, suggesting that many new car sales were financed from past savings rather than credit. la r t 2 PERSONAL S A V IN G A S A PERCENTAGE OF DISPOSABLE PERSONAL IN CO M E Other components of in income in the second quarter of 1970, repayments of installment debt rose steadily through the first quarter of this year and exceeded extensions in the Jane N. Haws C h art 4 C h art 5 C O M P O N EN T S OF INSTALLMENT LOANS O U TSTAN DIN G INSTALLMENT CREDIT A S A PERCENTAGE OF DISPOSABLE IN C O M E 3 C O M P O N E N T S *'fN O N IN S T A L L M E N T LOAN S ST A N D IN G After hitting a low point of 14.6% of disposable ( S e a s o n a lly A d ju s t e d A n n u a l Per C e n t $ B illio n s $ B illio n s $ B illio n s 301---------- Rate) Per C e n t 120 TOTAL IN S TA L LM E N T L O A N S TOTAL N O N I jTALLMENT L O A N S ■ .yment A u to m o b ile Loans Paper s R e p a id O th er C o n su m e r G o o d s P aper A c c o u n ts 14 R e p a ir & M o d e r n iz a tio n Loans Serie C re d it P e rso n a l Loans - i ______I_____ i______i_____ i 1968 Sou rce: 1969 F e d e ra l R e se rv e Bu lle tin . 1970 1968 S o u rc e : 1969 19 70 S u r v e y o f C u rre n t B u sin e ss. 1971 II I II 1968 So u rc e : IV | | I II 19< IV I Fed e ra l Resve Bulle tin . mm II I II 1 9 70 IV I 1971 I II I II 1968 S ou rce: IV I II I II 1969 I_____ L . IV I F e d e ra l R e se rve Bulle tin . II I II 1970 IV I 1971 I II I II 1968 Source: IV I II III 1969 IV I F e d e ra l R e se rve Bulle tin . II I II 1970 IV I 1971 Income Distribution and Its Measurement (Continued fi-om page 7) harmony by providing each factor of production with a reward just equal to its contribution to total output. Each of these nineteenth century doctrines implied that the lines separating the factors of production also marked the division of social classes. Only later, with an increasingly widespread ownership of pro perty and a growing degree of social mobility did this identification of social classes and economic function disappear from professional analysis. T o day economists find it useful to retain the original division of the factors of production, but without the presumption of social class identification. In recent decades the focus of distributive share analysis has shifted away from discussions of wel fare. The blurring of factor ownership classes has forced the virtual abandonment of functional dis tribution as a welfare indicator. The factors of pro duction, although analytically separate and distinct, are now seen as overlapping at the ownership level. Modern economists, unlike their classical predeces sors, recognize that individuals often own and sup ply several types of productive resources. For ex ample, it is not unusual to find the same individual receiving wage income from his employer, rent in come from property leased to tenants, interest income from bonds and savings deposits, and dividend in come from equity shares in the capital assets of corporations. Contemporary income distribution analysis focuses on explanations of the alleged constancy of relative shares. This focus derives largely from economists’ study of the Cobb-Douglas aggregate production function. A production function expresses the tech nological relationship between output and the as sociated inputs used in the production process. The Cobb-Douglas production function relates national output to only two factors of production, labor and capital, and implies that factor-income shares will be constant regardless of the amounts of the two inputs existing in the economy. In the Cobb-Douglas model, changes in the ratio of labor to capital re sulting from dissimilar growth rates of the two inputs would have no effect on factor shares. The wide spread acceptance of this model among economists has helped to foster the presumption of constant factor shares. DISTRIBUTIVE SHARES IN 1970 The statistical series which most closely corres ponds to the economist’s concept of factor shares is 10 published by the Department of Commerce in the Survey of Current Business. This series shows the distribution of the national income (prior to govern ment taxes and transfers) by type of payment. The percentage breakdown for 1970 is as follow s: Employee Compensation Proprietors’ Income Corporate Profit Interest Rental Income of Persons Total 75.0% 8.4% 9.6% 4.2% 2.8% 100.0% The lion’s share of national income goes to em ployees, with corporate profits and proprietors’ in come running a distant second and third, respectively, and interest accounting for most of the remainder. The relative size of the slice of the national income pie claimed by labor resources is especially note worthy in view of the vital and conspicuous role played by capital resources in the production pro cess. One might expect capital resources to claim a large part of the income generated by the world’s most “ capitalistic” economy. Flowever, a quick calculation from the above figures indicates that ap proximately 82% of the national income pie was dis tributed to labor resources, leaving only 18% to be claimed by capital resources. This estimate was made by counting employee compensation as labor-resource income and all profits, interest, and rent as capital-resource income, and by dividing proprietors’ income into labor and capital income in the propor tion which the share of employee compensation bears to the combined shares of profits, interest, and rent (75.0 to 16.6). T oo much faith should not be placed on the accuracy of these figures. For example, probably 3 or 4 percentage points of the 75 per centage point employee compensation share consists of salaries of corporation executives, not usually considered as labor income in the ordinary sense. Furthermore, the allocation of proprietors’ income is arbitrary. Nevertheless, the order of magnitude of the estimates is correct and it may safely be said that, in 1970, between three-fourths and four-fifths of the national income pie went to sellers of labor services. M EASUREMENT A N D INTERPRETATION Although the Department of Commerce’s classifica tion of distributive shares is the best the economist has to work with, it is imperfectly suited to his needs. For example, two of the income claims, proprietors’ income and corporate profits, are classified by type of business institution rather than b y type of economic resource to which payment is made. This and other discrepancies between theoretical concepts and empirical measures create a host of problems for the researcher in his analysis of the behavior of rela tive shares. Some of the major problems are dis cussed below. Im pure In com e Categories The Com m erce D e partment’s measures of employee compensation, pro prietors’ income and property income (corporate pro fits, interest, and rent) are comprised of heterogenous income elements whereas the wage, rent, interest, and profit components of economic theory are conceptually homogeneous and distinct. Economic theory defines wages as the payment for human ef fort exerted in the production process; rent as the return to non-reproducible resources supplied in fixed amounts by nature; interest as the return to non-human, reproducible means of production; and profits as the residual reward to entrepreneurship for risk-bearing, coordinating, and innovating ac tivity. In the national income accounts, however, the employee compensation category includes in determinable amounts of “ interest” yield on invest ment in education and training plus rent on unique ability, in addition to pure wage income. The em ployee compensation category may also include some entrepreneurial type income because the salaries of top corporation executives are included. The other national income categories are also a mixture of income elements and thus do not cor respond precisely to their theoretical counterparts. The rent share recorded in the national income ac counts is comprised mainly of rental income on housing and other leased structures. Very little of it represents the return to scarce natural resources, the theoretical concept of rent. Moreover, only rental income going to persons is recorded. Rental income received by corporations is excluded. Some of the measured interest income consists of interest on consumer debt as well as the yield on tangible capital equipment— the interest concept of economic theory. Finally, neither the corporate profits nor proprietors’ income categories consist solely of pure economic profit. The former includes some rent and interest income received by corpora tions, and the latter includes the implicit wage, rent, and interest income on the labor, land, and capital owned by proprietors and employed in their own enterprises. A pportion ing P roprietors’ Incom e System atic study of functional shares also encounters a difficulty in splitting proprietors’ income into its labor and capital income components. One of the key ob jectives of empirical research on income distribution is to test the frequently stated hypothesis of the “ re markable constancy of relative factor shares.” As previously mentioned, the hypothesis of constant relative shares is based upon certain theoretical eco nomic models embracing only two factors of produc tion labor and capital. In order to test the hypothesis of factor-share constancy, researchers must consoli date the national income categories into the two groups recognized by the theory. Little difficulty is experienced in consolidating rent, interest, and cor porate profits into a capital income component, and assigning employee compensation to the labor income component. But the allocation of proprietors’ in come, which is an amalgam of labor and capital (in cluding profits) returns, is a different matter. The question of how the proprietors’ share should be divided is one of the most vexing and controversial in the study of income distribution. Three positions have been taken regarding the disposition of pro prietors’ income. According to one view, it is vir tually impossible to identify the labor and capital components and therefore, any separation must be completely arbitrary. Advocates of this view hold that the analysis of relative shares should be limited to those sectors of the economy not dominated by unincorporated forms of business enterprise. A second view argues that proprietors’ income is too important to ignore and that it should all be as signed to the labor share category. Proponents of this view rationalize that a large part of proprietors’ income goes to self-employed professionals (doctors, accountants, architects, lawyers) and to proprietors of retail trade establishments, all engaged in pre dominantly labor type activities. A third view, adopted by the majority of re searchers in the field of income distribution, is op posed to the complete allocation of proprietors’ in come to labor because such a procedure implies that the property used by the self-employed in their work has a zero yield. Advocates of this position hold that alternative procedures can be used to obtain a rea sonably accurate disentanglement of the constituent parts of proprietor income. Suggested techniques in clude (1 ) dividing the shares in the same ratio as they are divided in the corporate sector, and (2 ) imputing to each self-employed person a labor in come equal to the annual wages of a worker and a capital income equivalent to the market yield on assets similar to those owned by the self-employed. Im putation of G overnm ent-Sector Incom e An other problem arises from the way in which product and income originating in the government sector is measured. Government output cannot be valued at 11 market price because, unlike private output, it is not sold on the market. Instead, it is valued at labor cost of production. That is, in the official statistics, the value of output and income produced in the public sector consists solely of the public payroll. The out put contributions of government-owned land and capital are not measured. In short, government product, as officially measured, is 100% labor-in tensive. The overstatement of employee contribution to public output may exert an upward bias in labor’s share as the relative importance of the government sector in the total economy increases. The conceptual problems discussed in this section make the task of interpreting movements in income shares treacherous. Observed changes in the shares may be the result of measurement bias rather than of real forces. These hazards weaken the reliability of empirical investigation. It may be hard to de termine from the discrepancies between measured and predicted movements of factor shares whether it is the theory or the measurement that is in error. T a b le I DISTRIBUTIVE SHARES (PER CENT) OF TOTAL IN U. S. N A T IO N A L INCOM E, 1900-1970 (Decade Averages of Shares for Individual Years) Decade Em p lo y e e Com p en sa tion P ro p rie to rs' In co m e C o r p o r a te P ro fits In tere st Rent 19 00 -1 90 9 1 9 1 0 -1 9 1 9 1 9 2 0 -1 9 2 9 19 30 -1 93 9 55.0 53.6 60.0 67.5 23.7 23.8 17.5 14.8 6.8 9.1 7.8 4.0 5.5 5.4 6.2 8.7 9.0 8.1 7.7 5.0 100 100 100 100 19 39 -1 94 8 19 49 -1 95 8 19 54 -1 96 3 1 9 6 3 -1 9 7 0 64.6 67.3 69.9 71.7 17.2 13.9 11.9 9.6 11.9 12.5 11.2 12.1 3.1 2.9 4.0 3.5 3.3 3.4 3.0 3.2 100 100 100 100 So u rc e : To tal Ir v in g K r a v is , " In c o m e D istrib u tio n : F u n ctio n a l S h a r e , " In te r n a t io n a l E n c y c lo p e d ia of S o c ia l Scie nces, V o lu m e 7 ( N e w Y o rk : M a c M i ll a n a n d Free Press, 1968), p. 134. R e p rin te d w ith p e r m issio n o f the P u b lish e r fr o m TH E IN T E R N A T I O N A L E N C Y C L O P E D IA OF THE S O C IA L S C IE N C E S , D a v id L. S ills, E ditor. C o p y r ig h t 1968 b y C r o w e ll C o llie r a n d M a c M i ll a n , Inc.; B u s in e ss C o n d it io n s D ige st. LONG-TERM TRENDS Formidable measurement problems notwithstand ing, the bulk of the research on functional income distribution has been devoted to explaining the secular behavior of relative factor shares. Table I shows estimates of the percentage distribution of na tional income since 1900. The data for the period since 1929 were developed by economists in the De partment of Commerce. Data for earlier years are the estimates of several scholars, including Simon Kuznets of Harvard, D. Gale Johnson of the U ni versity of Chicago, and Irving Kravis of the U ni versity of Pennsylvania. The table indicates that over the century the measured wage share has risen substantially, largely at the expense of the proprietor share and only slightly at the expense of the combined shares of in terest, rent, and corporate profits. Although the relative shares, as measured, display a moderate de gree of stability over the 25-year post-W orld W ar II period, the figures in Table I do not reveal the “ re markable constancy” which economists often pro claim as the most conspicuous characteristic of dis have tended to consolidate all non-labor shares into a “ property income” category whose overall stability conceals the divergent behavior of its constituent parts. Explanations of the Trend of Labor’s Share H ow do the experts account for the secular rise in labor’s relative share as measured in the national income accounts? Tw o alternative explanations have been offered. The first emphasizes structural alterations in the product-mix and industry-mix of the economy. This explanation, which appears in the work of E. F. Denison of the Brookings Institution and D. Gale Johnson, implies that the data can be reconciled with the theory of constant shares by showing that income distribution would remain unchanged in the absence of shifts in the composition of output. The second explanation, advanced by Irving Kravis, stresses the differing supply and demand conditions in the mark ets for labor and capital. Each of these explanations is discussed in greater detail below. Structural Changes The major structural shifts tributive shares. Most of the research effort has been devoted to affecting labor’s share include (1 ) the rise in the investigation of the trend in labor income. The dis corporate form of enterprise increasingly supplanted parate trends of corporate profits (up from 7 % to the individual proprietorship, (2 ) the growth in the 1 2 % ), interest (dow n from 5.5% to 3 .5 % ), and importance of the government sector, and (3 ) the proportion of wage earners to proprietors as the rent (down from 9 % to 3 % ) have received rela shift from land-intensive agriculture production to tively little study. Researchers, in their eagerness the production of labor-intensive services. Empirical to test the conclusions of two-factor economic models, techniques have been devised to measure the influence 12 of each of these factors on the growth of labor’s share. T o estimate the effect of the shift from proprietor ship to corporate form of enterprise, researchers divide proprietors’ income into its labor and non labor ( “ property” ) components, using any of several statistical procedures. The simplest procedure is to split proprietor income in the same ratio which labor income bears to property income in the rest of the economy. A more sophisticated method (a ) as signs a value to each proprietor’s labor equal to the annual wages of a hired worker, (b ) estimates the annual return on proprietors’ property from the ob served market yield on similar assets, and (c ) ad justs the total of estimated labor and property in come components proportionally to agree with re ported proprietor income. The shift out of self-em ployment into wage employment explains some of the rise in labor’s share but still leaves approximately seven to ten percentage points (depending on the method used to split proprietors’ income) of the rise unexplained.1 The growth of the government sector accounts for much of the remainder. The effect of the rising relative importance of the government on labor’s relative share is estimated by subtracting the govern ment’s contribution to national income from both na tional income and employee compensation. Identical amounts are excluded from both the income and the wages categories because, as previously mentioned, government output is valued solely at the cost of labor input and thus the wage share in the value of government output is 100%. After the exclusion of the government’s contri bution to income and the division of proprietors’ in come into its labor and property components, the relative shares of labor and capital do indeed display a high degree of long-term stability. Kravis’ esti mates indicate that since 1900, labor’s adjusted share be noted that part of the effect of the declining im portance of agriculture is also captured in the esti mate of the effect of the shift from unincorporated to incorporated business. In summary, significant adjustments must be made to estimates of factor shares derived from national income statistics before they will square with the theorists’ oft-proclaimed “ remarkable constancy of the relative shares,” which applies to a theoretical construct whose share definitions find no precise counterpart in national income statistics. importance of agriculture in the product-mix is Secular Changes in Input Supply and Demand An alternative explanation of labor’s rising share goes behind the facade of “ structural changes” to focus on the changing conditions of demand and supply in labor and capital markets. This explanation, as sociated chiefly with Irving Kravis, lumps all non labor income into one category (capital income) and emphasizes total demand for labor and capital rather than the structural changes discussed above. In addition to explaining the rise in labor’s share, Kravis’ demand-supply approach reconciles the fol lowing developments that have occurred in the American economy since the early 1900’s: (1 ) a six fold rise in the capital stock, (2 ) a doubling of the man-hour inputs, (3 ) a more than threefold rise in the real wage rate, and (4 ) a virtually unchanged real rate of return on capital. Kravis concludes that these trends are due primarily to differences in the responsiveness of the supplies of labor and capital to increases in demand as well as to the tendency for businessmen to substitute relatively low-price capital for relatively high-price labor. These conclusions are illustrated by the demand and supply diagrams in Chart 1. The long-run sup ply of labor-hours (S i) is depicted as less responsive to a rise in the price of labor-hours than is the longrun supply of capital ( S K) to a rise in its price. Economic growth over the century has increased the demand for labor and capital. However, because of the differing supply conditions, the increased demand for labor has greatly affected labor’s price, whereas the increased demand for capital is reflected in the rise in capital’s quantity. Because of the lack of re measured by comparing labor’s actual share with sponsiveness of the supply of labor-hours to changes what it would be if the relative importance of the in demand, businessmen have had to quadruple the different sectors had remained unchanged. real wage rate to induce the additional labor into (including proprietors’ labor income and excluding government) has remained within the narrow range of 69% to 76% .2 Finally, the effect on labor’s share of the declining It should employment. The rise in the wage rate relative to the 1 I. B. Kravis, “ Income D istribution: Functional Share,” Interna tional Encyclopedia o f the Social Sciences, Vol. 7 (N e w Y o rk : Mac Millan and Free Press, 1968) p. 134. Note that if all o f pro prietors’ income share were lumped with the employee compensa tion share in Table I, then the resulting “ labor’s share” would in deed exhibit virtual long-run constancy. A few economists, using this procedure, argue that the statistics strongly support the theory of constant shares. 2 1. B. Kravis, “ Relative Income Shares In Fact and Theory,” Am erican Econom ic R eview 49 (December 1959) p. 928. price of capital has also induced businessmen to sub stitute capital for labor in production and may have stimulated the search for new, capital-using (laborsaving) technology as well. The substitution of rela tively cheap capital for relatively dear labor is mani 13 fested by the sixfold rise in the amount of capital employed versus the mere doubling of man-hours employed. Labor income (price of x quantity of labor) has increased more than capital’s income (price of capital x quantity of capital) because the rise in labor’s price relative to capital’s price has exceeded the rise in the quantity of capital relative to the quantity of labor.3 THE CYCLICAL BEHAVIOR OF DISTRIBUTIVE SHARES Although relatively little research has been done on the short-run movements of distributive shares, many economists believe that the evidence is suf ficient to establish a definite cyclical pattern for the employee compensation, fixed income (rent and in terest), and corporate profit shares. The wage and fixed income shares appear to rise in periods of fall ing economic activity and to decline in periods of expansion. The share of profits, on the other hand, apparently rises in prosperity and falls in depression. 3 More precisely, labor’s income share relative to capital’s share may be expressed as L I / K I = ( P fxQ t) / ( P^.xQ^) = ( P j/P ^ ) ( Q j/Q ^) where L I and K I represent the incomes o f labor and capital, Pj and P^ their respective prices, and Q( and Q . their respective quantities. Labor’s relative share has risen because the price ratio ( P j /P ) has These cyclical patterns were most conspicuous in the 1930’s. During the contraction of 1929-1932, the shares of wages and interest spurted but the profits share fell sharply. W ith the progress of recovery after 1933, the wage and interest shares sagged and the share of corporate profits rose. These same cyclical patterns appear in the postW orld W ar II period, although with diminished intensity. Table II shows the percentage income shares at the peak and trough dates of all post-war cycles, as established by the National Bureau of Economic Research. The table clearly reveals the pro-cyclical behavior of corporate profits share and the counter-cyclical behavior of the employee com pensation, rent, and interest shares. In each cycle, the corporate profits share was higher at the peak than at the trough. In all but one of the cycles the employee compensation, interest, and rent shares were higher at the trough than at the peak. Overhead Costs, U nit P rofits, and D istributive Shares E con om ists have advanced several h y potheses to account for the observed cyclical be havior of the relative shares. The most plausible hypothesis holds that the p ro-cyclical behavior risen by a greater percent than the percentage fall in the quantity of profits’ share results from the relation between <Q(/ Q K) ratio. profits per unit of output and unit overhead costs. Ch art 1 INPUT D E M A N D A N D SUPPLY CURVES E co n o m ic g r o w t h sh ifts u p w a r d the d e m a n d s fo r la b o r a n d c a p ita l, in c r e a s in g the in c o m e s o f b oth . In c o m e s in 1900 a re sh o w n a s the sm a lle r, cro ss-h a tc h e d re c ta n gle s. In c o m e s in 1970 a s the la r g e r , g re e n re c ta n gle s. L a b o r in co m e h a s in c re a se d m o re th a n c a p it a l in com e. Table II DISTRIBUTION OF N A TIO N A L IN C O M E SHARES (PER CENT) AT THE CYCLICAL PEAK A N D CYCLICAL TROUGH DATES IN FIVE POST-WAR CYCLES Employee Compensation Proprietors' Income Corporate Profits Rental Income Net Interest Total Shares So u rc e : 1948 IV P 1949 IV T 1953 III P 1954 II T 1957 III P 1958 II T 1960 II P 1961 1 T 1969 IV P 1970 IV T 63.1 17.5 15.0 3.6 0.8 65.4 16.4 13.3 4.1 0.9 68.7 13.0 13.2 4.2 0.9 68.4 13.2 12.6 4.6 1.3 69.9 12.1 12.4 4.1 1.5 70.4 12.9 10.5 4.3 1.8 70.7 11.2 12.4 3.8 1.9 71.5 11.5 10.9 3.9 2.2 74.1 8.6 10.4 2.8 4.0 75.4 8.4 9.1 2.9 4.3 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 B u s in e ss C o n d it io n s D ig e st. This relation varies with changes in aggregate out put. During economic expansions, increased pro duction induced by rising aggregate demand enables firms to spread overhead (i.e. fixed) costs— including the wages of overhead labor as well as rent and in terest expenses— over a greater volume of output. Overhead costs per unit of output fall and profit margins rise, thereby enlarging the profits share and diminishing the labor, interest, and rent shares in the value of each unit of output. In recessions, out put falls and unit overhead costs rise, thereby squeezing the profits share and enlarging the other shares. This hypothesis implies that a substantial portion of labor income is a component of the overhead costs of hiring firms. Many economists believe that a sizeable part of the labor force is of the overhead variety. Overhead labor includes su pervisory and adm inistrative personnel as well as top m anagem ent officials whose salaries are largely independent of their firms’ output volume. Overhead labor also includes employees retained or “ hoarded” by employers in the face of cutbacks in output, either because they possess specialized skills or because employers have made contractual com mitments (such as a guaranteed annual wage) to them. W orkers may also be retained because em ployers wish to avoid costs (e.g., severance pay) of laying-off labor as well as the costs of rehiring and retraining workers when business conditions improve. More precisely, the relation between profits, price, and costs per unit of output can be expressed a s : charges as well as the cost of overhead labor. A s previously mentioned, unit overhead costs fall as output increases. The category labeled “ other unit costs” consists largely of the costs of materials and of unskilled and semi-skilled labor. These unit costs are virtually constant because the firm can adjust its material and unskilled labor inputs in roughly the same proportion as output changes. When out put contracts by, say 10%, firms can lay off 10% of their unskilled labor. Employers are less averse to laying off unskilled and untrained workers than pro fessional and highly skilled workers, because the latter are harder to find and the costs of training their replacements would be very high. C h art 2 UNIT COSTS, SELLING PRICE, A N D UNIT PROFITS AT DIFFERENT OUTPUT LEVELS FOR A HYPOTHETICAL FIRM D o lla r s per U nit o f O u tp u t selling price per unit of output equals unit overhead costs plus other unit costs plus unit profits. Over head costs consist of certain contractually fixed costs (such as insurance, property taxes, rent, and interest on bonded indebtedness) and fixed depreciation Level Level 15 The remaining component of selling price is pro fits per unit of output. A s illustrated in Chart 2, this unit profit component is larger at higher outputs where the cost component of price is smaller. The chart shows the unit cost curve of a hypothetical firm, as well as the price at which it sells its product. The hypothetical unit cost curve is composed of (falling) unit overhead costs plus (constant) other unit costs. As depicted in the chart, profits per unit of output, i.e., the difference between price and unit costs, are larger at the prosperity level of out put than at the recession output level. Several qualifications to the preceding analysis should be noted. T he discussion assumed that selling price, w age rates, and produ ctivity re main unchanged over the cycle. In actuality, prices, w ages, and productivity tend to expand, albeit at different rates, during the upswing. Changes in those variables cause the curves shown in Chart 2 to shift. Rising prices shift the price line. Rising wages and productivity shift the cost curve. Rising wages, which tend to raise the cost curve, may be offset by rising productivity, which tends to lower the cost curve. During the first half of an upswing, productivity growth tends to more than offset wage increases, thereby shifting the cost curve downward. Moreover, firms enjoying some degree of monopoly power may respond to the increase in aggregate de mand by raising prices as well as output. Thus, rising prices may combine with falling unit costs (both a shift in the curve and a rightward move ment along it) in prosperity to raise unit profits and profits’ share. 16 The favorable influence of prosperity on unit pro fits is likely to diminish as the expansion proceeds, however. During the later stages of prolonged booms, several forces combine to raise unit costs. As the labor market becomes tight, wage increases accelerate, productivity growth slows, and the unit cost curve shifts upward. Operating rates of plant and equipment reach and then surpass their most efficient levels. These two factors— capacity limita tion and wage increases in excess of productivity growth— cause unit costs to rise faster than selling price, thereby encroaching on unit profits. The pro portion of profits to the value of output declines. The erosion of profits’ share relative to labor’s share during the later stages of a business expansion is well-documented. Official figures show that in all but one of the post-W orld W ar II cycles the share of corporate profits was lower and the share of em ployee’s compensation higher in the quarter of peak economic activity than in the three quarters im mediately preceding the peak. The W a g e L a g H ypoth esis A n alternative h y pothesis that has been advanced in explanation of the shift in favor of profits’ share in the upswing and in favor of labor’s share in the downswing is the so-called wage lag hypothesis. According to this hypothesis, sticky money wages lag behind price in creases during booms and price decreases during slumps. Twenty years ago, economists thought the wage lag was the most important factor accounting for the counter-cyclical behavior of labor’s share. Recent empirical work has cast doubt on the strength of this effect, however. Thomas M . Humphrey