The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
1r (G) if©\lffi CC n CC (G) September 14, 1979 DistortedData? The unemployment rate rose sharply in August to 6.0 percent of the civilian labor force - the first time it reached that level in over a year's time - and Washington's policymakers sat up to take notice. But past Congressional decisions have already dictated that billions in taxpayers' money will be transferred to various local ities on the basis of such shifts in the jobless rate. That-raises the obvious question: Are the figures good enough for the purposes for which they're used? The National Commission on Employment and Unemployment Statistics - which appropriately released its report on Labor Day - has some interesting (and at times disquieting) things to say on that subject. The commission members, like most other economists, believe that the present procedures are adequate for measuring the national jobless rate. (Whether that figure is correctly interpreted is a completely different question, as we'll see in a moment.) But in contrast, they believe that the jobless data for smaller local areas are seriously inadequate - that the data which Congress uses to allocate billions of dollars annually to 6,100 local areas may be "extremely inaccurate." Interpreting national data The key question concerning the national jobless rate is one of interpretation rather than simple accuracy. What is the "fullemployment" unemployment rate - the lowest point to which the jobless rate can fall without generating inflationary pressures on the labor market? Two decades ago, that rate generally appeared to be in the neighborhood of 4 percent. Today, however, most economists believe that full employment is reached in the neighborhood of 6 percent, because of the demographic and institutional changes that have occurred since the mid1950's. • The composition of the labor force has shifted, with a sharp expansion in the number of teenaged and women workers - individuals who enter and re-enter the job market more often than average, and who consequently exhibit higher-than-average jobless rates. The unemployment-insurance program has been liberalized several times since the mid1950's, with coverage extended and benefits increased, and these changes have tended to increase the duration of unemployment and hence raise the overall unemployment rate. fit • Legislated increases in the minimum wage have tended to increase joblessness among young arid unskilled workers, because their efforts are not worth the higher mandated wage to prospective employers. • The work-registration requirement for welfare eligibility also has tended to boostthe jobless figure; many such individuals wouldn't otherwise be counted as jobless because they wouldn't be looking for work. Another question concerns the number of discouraged workers who remain outside the labor force because they believe no jobs are available forthem. Some analysts believe that discouraged workers should be counted in the official unemployment rate because they, like the unemployed, are available for regular jobs. Yet the data indicate that large numbers of discouraged workers remain discouraged even during periods of tight labor markets, when inflationary pressures mount amid substantial shortages of workers. (There were more than 650,000 discouraged workers in 1967-69 and again in 1973 - both periods of labor-market tightness.) Thus the national unemployment rate, which is popularly regarded as an indicator of cyclical joblessness, would be distorted if all discouraged workers were included in the official jobless count. To help improve our analysis of this problem, the National Commission has recommended that more information be Opinions r1C)r ()f expressed in this newsletter (10 not the f3c)arclc)f (If the 1 4.4-1 6.2 percent range in Julyand 'vvithin the 1 5.6-1 7.4 percent range in August. The width of this range is much larger than for the national unemployment rate, and thus supports the National Commission's conclusion that the Current Population Survey provides inadequate precision for small demographic groups. gathered to ascertain the labor-force attachment of discouraged workers and of jobmarket nonparticipants in general. Measuringnational data Questions of interpretation aside, the national jobless figure appears to be quite accurate, and its accuracy should increase even more later this year as the national monthly sample increases from 56,000 to 70,000 households. As the National Commission says, "The Current Population Survey (CPS) . . . provides highly reliable current estimates of the employed, the unemployed and those not in the labor force." Other statistical problems affect the national figures, such as the inclusion of part-time as well as full-time job seekers among the total unemployed. The commonly reported national figure makes no distinction on the basis of the amount of work sought by an individual. According to the official definition, employed persons include all individuals (16 years old and over) who were available but performed no work during the week the population survey was taken, and who had looked for pC!.idwork within the preceding four weeks. Hence, a teenager whose major occupation is attending school but who wants several hours work a week is considered just as unemployed as the head of a household seeking full-time work. However, BLS has developed some alternative jobless measures which consider only fulltime workers, such as the ratio of unemployed full-time jobseekers as a percentage of the full-time laborforce - which averaged 5.2 percent in the first quarter of 1 979, compared with the 5.7-percent official jobless rate. Still, the important point to note is that this measure of full-time unemployme"nt moves closely with the official jobless rate overtime. Yet, as with any sample estimate, the reported national jobless figure for all civilian workers should not be considered precise - should not be cited as exactly 6.0 percent in August, for example. The most we can say is that, 9 times out of 10, we can be certain that the estimate from the sample is within the range of 0.1 8 percentage points of the true employment figure. So instead of saying that the national jobless rate was 6.0 percent in August, we'd be more correct to say that with 90-percent confidence, the range of 5.8-6.2 percent incorporates the figure that would be obtained from a complete census of the population. For the components of the total, jobless estimates are subject to greater error because of their smaller sample size. Thus, the error for nonwhite workers is just over four times the size of that for white workers, while the error for adult females is about 1.5 times as large as that for adult males. For teenaged workers, the error is considerably higher than for other groups. So instead of saying thatthe teenaged jobless rate jumped from 15.3 percent in July to 16.5 percent in August, we'd be more correct to say that the chances were 9 out of 10 that teenaged employment was within the Measuringsmaller-areadata The problems with the national data, however, pale into insignificance compared with those of state and local areas. This is especially important because Congress has decreed that these weak estimates should be used to determine the allocation of Federal grants which totalled $10 billion in fiscal 1 979 - up more than tenfold since the beginning of this decade. The fundallocation procedures assume a high degree of data accuracy which is simply non- 2 Percent 10 8 Official 6 rate(U5) /o-._.-··_,._.!. Full-time rate (U4) 4 ./o//"- 2 Official rate IUC;I- persons 16 yearsand over as a percent of civilian labor torce 16 yeaisand over Full-time rate 1U41- unemployed full-time iobseekersas a percentof full-time labor force AUocai;nggrantfunds existent. For example, the formula used for allocating funds under the Comprehensive Employment and Training Act (CETA) would allocate substantial funds to an area with 6. S-percent estimated unemployment, but would allocate nothing to an area with 6.4percent estimated unemployment -despite the fact that those two figures statistically are probably identical. Congress has mandated the use of smallarea unemployment data as a basis for allocating Federal grants to some 6,1 00 separate localities throughout the country. In its final report, the National Commission stated flatly that there is no way, at reasonable cost, to produce accurate jobless statistics for so many areas every month. To obtain the type of information which Congress demands wou Id requ i re a massive expansion of the Current Population Survey, with the cost rising from the present $20 million a year to as much as $2,300 million a year. The unemployment data for most states are derived from both the state data on unemployment-compensation claims and the state data provided by the Current Population s.urvey (CPS). Neither source provides very accurate estimates. In particular, the claims data are highly unreliable, partly because many unemployed persons are not entitled to unemployment compensation, and partly because eligibility and benefit provisions vary widely by state. As a practical matter, the Commission has recommended certain incremental improvements which would raise the cost from $20 million to $54 million a year. The principal recommendation would bean increase in the CPS sample size, from 70,000 at the end of 1 979 to 1 20,000 in future years. This would improve the reliability of the annual unemployment estimates for each of the 50 states and for 11 major cities. But for smaller areas, the Commission sees no solution in sight for the data problems which lead to such substantial inequities in the allocation of the taxpayers' money. For the 1 0 largest states, the Bu reau of Labor Statistics estimates unemployment figures directly from Current Population Survey data. For smaller states, it uses a complex 70-step estimating method - the "handbook method" - to adjust claims data, and then combines those data with Current Population Survey data. For local areas, it uses the handbook method, adjusted for consistency with statewide estimates. RoseMcElhattan Because of the weakness of the estimates, state and local-area unemployment data frequently have been subject to substantial data revisions. Over the 1 974-77 period, for example, the year-end revisions exceeded 1 0 percent each year for more than half of the individual states. (A 1O-percent error in the unemployment rate for, say, Chicago would mean a $2S0-million shift in that city's allocation of Federal money.) In an attempt to limit the amount of year-end revisions, the Bureau of Labor Statistics last year introduced a procedure of adjusting the monthly claims data by CPS data for the preceding six-month period. Even so, the resultant data remain quite weak. 3 .lScH.:i U01SU!4seM· 4Pln • UoSaJO • epeAaN • o4PPI !!PMPH • E!UJOJ!lP:) • EUOZP'v' • e>jsEI'v' J:I U2S 'j!l1:!::> 'o:>sPU1:! C;SL'ON ll W1 Bd Ol'v'd :I!JVlS Od 's'n llVW SSV1J JJ, BANKINGDATA- TWELFrHFEDERAL RESERVE DISTRICT (Dollar amounts in millions) Selected AssetsandLiabilities largeCommercialBanks Loans(gross,adjusted)and investments* Loans(gross,adjusted)- total# Commercial and industrial Realestate Loansto individuals Securitiesloans U.s. Treasurysecurities* Other securities* Demand deposits - total# Demand deposits- adjusted Savingsdeposits - total Time deposits - total# Individuals, part. & corp. (LargenegotiableCD's) WeeklyAverages of Daily figures MemberBankReserve Position ExcessReserves(+ )/Deficiency (- ) Borrowings Net free reserves(+ )/Net borrowed(-) Amount Outstanding 8/29/79 131,587 108,608 31,521 39,869 22,425 2,086 7,481 15,498 41,945 30,891 30,401 52,641 44,233 19,163 Weekended 8/29/79 - Change from 8/22/79 - - Changefrom yearago@ Dollar Percent 18,503 17,276 4,272 8,194 582 459 144 146 184 89 57 180 213 395 162 375 307 237 NA - Weekended 8/22/79 33 147 114 NA 957 2,184 2,282 1,090 41 6,219 7,498 1,099 16.36 18.92 15.68 25.87 NA NA - 11.34 16.40 5.75 3.66 - 0.13 13.40 20.41 6.08 Comparable year-agoperiod 11 230 219 88 61 27 242 + 491 718 - 681 FederalFunds- Sevenlarge Banks Net interbank transactions [Purchases(+)/Sales(-)] Net, U.S. Securitiesdealer transactions [Loans (+ )/Borrowings (-)] 545 - 285 + * Excludestrading account securities. # Includes items not shown separately. @ Historicaldataarenot strictlycomparable dueto changes in the reportingpanel;however,adjustments havebeen to 1978 datato removeasmuchaspossibletheeffectsof the changes in coverage.In addition,for someitems,historicaldataarenotavailabledueto definitionalchanges. Editorialcommentsmaybeaddressed to theeditor(William Burke)or to the author.... Freecopiesof this andotherfederalReserve publications canbeobtainedbycallingor writingthe PubliclnfonnationSection, FederalReserveBankof SanFrancisco, P.O.Box7702, SanFrancisco 94120.Phone(415)544-2184.