The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
Working Paper 74-3 AN EMPLOYMENT PRESSURE INDEX AS AN ALTERNATIVE MEASURE OF LABOR MARKET CONDITIONS* William E. Cullison Presented at the 49th Annual Conference of the Western Economic Association, Las Vegas, Nevada, June 9-12, 1974. The views expressed here are solely those of the author and do not necessarily reflect the views of the Federal Reserve Bank of Richmond. AN EMPLOYKEXI PRESSURE INDEX AS AX ALTERNATIVE ME4SURE OF LABOR MARKET CONDITIONS* William E. Cullison Federal Reserve Bank of Richmond Much attentj.onhas been devoted to the peculfar behavi.orof the unemployment rate from 1969 to 1973. In p&t recessions, the unemployment rate reached definite turning yolnts and showed fairly consistent recovery shortly thereafter. Its behavior during the 1970 recession and the subsequent recovery, however, was noticeably ditferent. The unemployment rate rose to around 6 pe,rcenti.nNovember 3.970and remained very close to that level until June 1972, setting a record for the longest peak in the history of the series, Moreover 9 through the summer and fall of 1973, when other economic indicators had been signaling that recovery was weli under way and that the economy was approaching full capacity: the unemployment rate continued to indicate a relatively slack labor market. Geoffrey Moore, former Commissioner of the Bureau of Labor Statistics, has suggested (1973) that the unemployment rate may be overemphasized as'a target variable for economic policy. After analyzing several labor market indicators, Moore concluded that " ...the evidence indicates that in recent months wc have been closer to full employment than the unemployment rate by itself suggest8.(1 (1973). Moore had advocated earlier that employment data should be given at least as much attention as unemployment data in analyzing *The author wishes to thank the Eureau of Labor Statistics for providing data; the Federal Reserve Ean!iof Richmond for its support; and Geoffrey PSooreof the NBER, and the Economic Staff of the Federal Reserve Bank of Kfc!lmondfor helpful comments. The analysis and conclusrons of th%s paper are those of tix author and not o f the.Federai Reserve System, -2- labor market statistics. He suggested that the employment data were supericr to the unemployment figures, not only because of problems in defining involuntary unemployment, but also because the employment scrl.escontains relatively less sampling error (1972). This paper presents a new data series that relates employment data to an historical standard. The new series is referred to as an employment pressure index (EPI); its purpose is to transform raw employment figures into a series that can bc used to measure labor market conditions, Since the EPI does not rely upon either unemployment or civilian labor force d;lta, it is not affected by the definitional problems inherent in the unemployment statistics. The new series is compared with the unemplo-ymentrate as a labor market indicator, Figure 1 shows the EPI (defined below) plotted with the uncmploymeritrate for all civi.l.ian workers. is equated to an E)?Iof 100. A 5 percentl unemployment rate (inverted) There is clearly a high degree of overall corre- spondence between the two series from 1955 through 1969. the two series have begun to diverge, Since then, however, This recent disparity is discussed at length. The Employment Pressure Index The employment pressure index was dcrlved by dividing actual employment figures by a population-adjusted trend value. Theoretically, the index measures excess demand or supply, assuming actual employment as a proxy for labor demand and that the trend measures long-term labor supply. 'The average from 1955 to 1973. All Workers Age 16 to 64, January 1955--April 1974 . UNEMPLO~NT RATE (Right Scale) EMPLOYMENT INDEX (Left Scale) : i a 2 c- i t 1956 i 1957 5 1959 k 1959 k 1960 ii 1561 1962 1963 $ 1964 a 1965 t 1966 ? 1967 9 k 1966 1969 k 1970 h 1971 tr Ii72 5 1973 197Y -4- The trend is derived by (1) regressing population and a time trend on employment; (2) generating trend data from the regression equations; and (3) summing across subgroups to obtain the trend for total employment. Figure 2 shows the summary trend plotted against actual employment from 1955 to 1973. The estimating equations for the employment trend are: ET = ai f hi(T) + Cam -tdi(Pi), where 1";= computed employment in the ith employment group,2 T = time (January 1954 = J.:), and Pi = U. S. civilian resident.population in the ith group. The employment preF:WX: index plotted in Figure 1 is thus defined as E'/(B~~l I$), where E' is total employment. In developing the EPI, some adjustments in the basic technique were made to account for changes in armed forces personnel. Since the physical and mental abflitj.~:s of armsd forces recruits 2nd draftees mcdc them **-n ....Lw likely to have been employed than the average member of their age-race--se;,: population group, changes in armed forces personnel affected employment more than proportionally in some groups. therefore, were The data for males 20-24 and 25-34, adjusted by regressing armed forces personnel on the difference between the EPI and the unemployment rate (inverted) of the group, and then adjusting the emplo*ymenttrend for variation explained by the regression. The R21s for those regressions were 0.64 and 0.13 respectively. The Differences Between the Emmloyment Pressure Index and the -- Unemployment Rate The differences between the employment pressure index and the unemployment rate are classified into two categories: (1) differences ---- 'There were 16 groups, by sex, race, and age (16-19, 20-24, 25-31;; and 3$e16). A table is available from the author upon request detailing the rcgress.tcnresults. Figure 2.-Total EmpIoy& Persons, Age 15 to 64 Actual EmpIoyment and Popukkon-Adjusted 1 -b‘ I 1956' -4. 1 # i958 I IIt 1960 .. I ’ 1962 1964 Trend, 1955-1973 I I 1966 I 1 1968 I I 1970 I 1 1972 -6- attributable to "discouraged workers" and (2) differences attributable to other factors. The "Discouraged Worker" Effect One reason for developing an employment-based measure to analyze labor market conditions is to eliminate the so-called "discouraged worker" effect. The measured unemployment rate, it has been argued, has been seriously affected during economic contractions and recoveries by the exit and re-entry of discouraged workers from the labor force. During the 1969-1970 recession, when the unemployment rate leveled off at around 6 percent, the EPL continued to fall, reaching a definite lower turning point in June 1971. time. It has shown a persistenl:recovery s:Lncethat By contrast, the unemployment rate remained at around 6 percent:for aimost l-112 years and was continuing to indic:-:te reiati-.*ely ~1.~1:labcr markets as late as fall 1973. At least for the 1969-1972 time period, a strongly operative discouraged worker effect could have explained the seemingly erratic behavicr of the unemployment rate; for 5f workers dropped out of the labor force in 1970 and early 1971 and re-entered in late 1971 and 1972, the peak in the unemployment rate and the subsequent recovery could have been disguised. Alfred Tella (1964 and 1965) and others have estimated the discouraged worker effect. Tella used trend variables to estimate participation rates for various age-sex breakdowns. He wa.sthen able to estimate a potential full employment labor force, which, when compared with the actual data, enabled him to estimate the number of discouraged workers. Tella found that the discouraged worker phenomenon affected females more strongly than it did males, Since.females are less likely to be bread- winners than males, this finding is consistent with a priori judgment. In -7- fact, a priori reasoning could lead a step further. Since white females are less likely to be breadwinners than nonwhite females, they should be more strongly subject to the discouraged worker effect than nonwhites. The female participation rate for the civilian labor force in 1960 was 33.4 percent. In 1969, it was 37.8 percent. Much of this increased female .participationwas in the white female group. The changing composition of the civilian labor force between 1960 and 1969 thus may have lent itself to a stronger discouraged worker effect. Moreover, as can be seen from a comparison of F'igilres 3 and 4, there is a greater difference between the BP1 and the unemployment rate for females than for Tales. The 1970 recession, whj.chfell relatively heavily on highly skilIed and highly educa.tedwhite-collar workers, may have produced a stronger discouraged worker effect for males as well since highly skilled workers, having relatively more job information than tl~eaverage worker, may be more inclined to postpone their job search when prospects lock meager. These hypotheses provide plausible explanations of some divergency between the EPI and the unemployment rate, but the evidence weighs against discouraged workers as the root cause of the differences. According to the Labor Department's estimates of discouraged workers, their numbers have been quite small relative to the total number of unemployed workers. The BLS survey put the average number of discouraged workers at 574,000 in 1969 and 774,003 in 1971. almost 5 million. By contrast, the total number of unemployed workers averaged Thus, if the Labor Department data are accurate, only a small part of the divergence between the unemployment rate and the BP1 could be explained by changes in the numbers of discouraged workers. Paul .??laim of the BLS has noted: However, as "Given the subjective and elusive nature of 'discouragement,' the extent of its possible overstatement and u;u:erstatenent cannot be measured." (1973). _ _ _ -lO- Tella's estimates, derived indirectly by methods that are similar to those used in estimating the employment trend values for the EPI, have been consistently higher than those of the Labor Department. Unfortunately, his estimates would not provide an independent test of the EPI. Preliminary findings, however, suggest that the discouraged worker effect cannot completely explain the recent divergence. For example, the EPI had matched the 1967-1968 levels, equivalent to an unemplo-ymentrate of 4 percent, by Jnn:?1973. around 4.8 percent. The measured unemployment rate, however, was The EPI for males moved above 101.0 in early 1973, and it was indicating extraordinary labor market pressure by September (101.9). Tha male unemployment rate, however, was 4.2 percent during Ja.nuary1973 and 4.0 percent in September, equivalent to an EPI of only XlO.8. Such large discrepancies, particularly when the economy was near full capacity, IJLzkkee z.hediscouraged t:crkerexplanation seem implausible. Other Differences i-2ett.veen t'he and the Unemplo~e;~t m-.-w-----.-.---~L,--- EPI -. -.-4-m- -- Rite The difference between the El?1and the unemployment rate for females may be partially attributable to biases in the EPX trend equations stemming from 1:herapidly rising female participation rates in the late sixties. the other hand, r+idly On rising participation rates also mske difficult the int:erprz&.tionof female unemploment, since a larger percentage may have been frictional and not of concern for aggregate economic policy. As shown in Figure 3, the EPI for males fluctuated between 101.0 and 101.5 during the 1967-1969 time period, while the male unemployment rate fell from around'3.5 percent to almost 2.5 percent. Thus, the two series, usually quite similar, were considerably different during that interval. The EPI and the unemployment rate at full employment, however, may diverge -ll- for technical reasons. As firms approach full capacity, there is some limit to the number of workers that they can absorb per time.period. The EPI would indicate that the limit for male workers was around 1.5 percent more than trend in 1967-1969. The unemployment rate, of course, would continue to fall whenever EPI > 100.0 because long-term labor supply is included in the trend. Finally, some of the divergence in 1972 and 1973 can probably be explained by a variation of the additional worker effect. Secondary bread- winners can enter the labor force to supplement the family income during recoveries as well as during downturns. During periods in which prices are increasing rapidly, as in 1972-1973, such an additional worker effect may be quite strong, If so, unemployment will not drop as rapidly as might be predicted from the increase in employment. ccIlcll~siol-.~-. --.a-- This article has presented a new index, based solely on employment data, a;ldhas used it to evaluate alleged deficiencies in the published unemplogr;:ect statistics, The new series, an employment pressure index, 1 generally corroborates until recently the accuracy of the male unemployment rate series. For females, however, some discrepancies cannot be reconciled without additional information, and they nay be attributable to bias in both series. Recent behavior of the unemployment rate and the EPI, however, seems to indicate that the latter is presently the more sensitive coincident indicator. Consistent with the preponderance of signals of the growing pressure of aggregate demand on economic capacity, the EPI showed increased tightness in labor markets throughout the first half of 1973. Ey the summer -12- of 1973, the EPI had regained the high employment levels of 1967-1968. In contrast, the unemployment rate gave a different and seemingly less accurate picture. The explanation of the recent divergence between the EPI and the unemployment rate is therefore of importance for purposes of economic policy, for if the unemploymeut rate measures something different from what it used to measure, the working definition of full employment should be modified. It appears, therefore, that although regular publication of the EPI series would involve periodic revisions (the trend equations should not be extrapolated far beyond the data from which they are estimated), publication of them would provide important additional information on labor market conditions, SELECT BIBLIOGRAPHY 1. Bowen and Finegan. "Labor Force Tarticipation and Unemployment," a>oyment Policy and the Labor -- Market (Berkeley and Los Angeles: University of California Press, 1$3m 2. Thomas Dernburg and Kenneth Strand, "Hidden Unemployment 1953-1962: A Quantitative Analysis by Age and Sex,' American Economic Review, Vol. LVI, ijo.1, March 1965. 3. F&aim, Paul."Discouraged Wor'iersand Changes in Unemployment," 1&.nthlyLabor Review, Xarch 1973. 4. Vail. Strcec Journal Moore, Geoffrey. .-B-,---9 5. Moore, Geoffrey. How _____. Full is ?::ll - ---. G- Emmloymcnt --._____ (Washington: American for hbiic F'olicy F;?seal.c>, 1973). Enterprise Substitute 6. Perry, George. "C!xx-@.ngI&or- i?arket:, and Inflation," Brookinf;sPapers on Economic Acti.vity, II::, Il.370. -M-w- 7. Tella, A. J. "Labor Force Eerzsitivitybo Emplo~meni;by Age, Sex," Industrial Rela-tions,Vol. 4, No. 2, Feb. 1965. 9. N. J. SiKLer and A. J. Tells. "Labor Reserves and t!lePhillips Curve," Review of -a Economics Vol. 50, Feb. 1968. -.u---.- a.:~% Statistics, 10, Feb. 3, 1972. Tella, A. J. "The Relation of Labor Force to Employment," Industrial and Labor Relations Rev&r, Vol. 17, April 1964. -,. .-