WAGE COMPARABILITY JEFFREY

M. PERLOFF

IN THE U.S. POSTAL SERVICE and MICHAEL

L. WACHTER*

Federal law requiresthatthe U.S. Postal Service pay wages comparable to thosepaid in the privatesector.The authorsargue thattheappropriate testof thatconcept is a comparison of the wages paid to all comparably skilledworkers,not only to whitemale workersin the privatesector. By that standard, the authors conclude, froman analysisof CPS data, that in 1978 postal workersenjoyed a 21 percentwage advantage. They also point to the low quit rates and long employmentqueues in the Postal Service as evidence confirmingthat the wage of postal workersis well above that of comparable workerselsewhere. The authors predict that since the mail transmissionmarkethas become increasinglycompetitive, either the wage advantage of postal workerswill decline or postal jobs will continue to shiftto the privatesector.

FEDERAL

lawrequiresthattheU.S. Postal

Service pay wages comparable to those paid in the privatesector: "It shall be thepolicy of the Postal Service to maintain compensation and benefitsfor all officers and employees on a standard of comparability to the compensation and benefits paid for comparable levels of work in the privatesectorof theeconomy."' This paper investigateswhetherthe PostalServicepays a wage comparable to that paid in the private sector. The basic method of comparison is the classical one: Is the wage paid *JeffreyPerloffis Associate Professorof Agricultural and Resource Economics at the Universityof California,Berkeley; and Michael Wachteris Professor of Economics,Law, and Management at the Universityof Pennsylvania.A grant fromthe U.S. Postal Service supported thisresearch.The authors wish to thank Peter D. Linneman, Vincent W. Drumb, D. Richard Froelke, J. Elizabeth Callison, William L. Wascher, and Nancy Zurich for their helpful comments. 139 U.S.C. ?1003(a).

workers by the Postal Service higher or lower than the wage thatthe privatesector offers to similar individuals? In other words, is the "opportunitywage" of the workerequal to thatofferedby the Postal Service? The resultsof thisstudyindicatethatthe Postal Service pays wages far in excess of those paid in the private sector and that the PostalService'swage premiumhas been growing over time. Indirect evidence on quit rates and job queues indicates that workers view employment in the Postal Service as unusually attractive. Theory and Data In perfectlycompetitivelabor markets, the marketguaranteeswage comparability. Firms pay a wage dictated by marketconditions.So long as the firmsact to maximize profits,the wage is sufficientfor the firm to hire the necessaryworkers.Individuals withthe same skillsare paid the same wage.

Industrialand LaborRelationsReview,Vol. 38, No. 1 (October 1984). ? 1984 by Cornell University. 0019-7939/84/3801$01.00

26 Sage Publications, Inc. is collaborating with JSTOR to digitize, preserve, and extend access to Industrial and Labor Relations Review www.jstor.org

®

POSTAL SERVICE WAGE COMPARABILITY

27

This equality not only leads to economic demographic variables; Area Wage Surefficiency,but also has obvious appeal in veysof detailed occupationalwage ratesfor termsof equity. each Standard MetropolitanStatisticalArea In the governmentsector, the market (SMSA); quit rates and employmentmechanismhas been replaced by statutory registerdata; and establishmenttime-series mechanismsthatestablishwage and salary data on relativewages between the Postal scales or by collectivebargainingcontracts. Serviceand otherindustries.Althougheach Employerscovered by federalcomparabil- data set alone has its limitations,together itylawsare also clearlydirectedto pay wages they rule out the likelyerrors that could comparable to wages paid for similarjobs arise from analyses based on only one of in the privatesector.Actual wage comparthem. ability,however, is not always the result. Where agencies or firmsare unionized, as Wage Comparisons Controlling for in the case of the Postal Service (or in the Individual Characteristics private sector, generally), the parties intercedein the marketprocess and introThis sectioncompares the wages of fullduce theirown comparisons,whichmaybe timeworkersin the Postal Service to those quite differentfroma competitivemarket's of workerselsewhere,controllingforcharimplicit comparison. Although collective acteristicsof individual workers.Standard bargaining can yield wages that are nonstatisticaltechniquesmake it possibleto calcomparableacrossindustries,marketforces culate comparable wages for occupations, are not repealed and a negative trade-off such as thosethatpredominatein the Postal resultsbetween wages and employment Serviceand thathave no directcounterpart Whatever the outcome of bargainingin in competitivelabor markets.In thisstudy, the Postal Service,itswage positionin relawe accomplish these comparisons with tion to marketwages paid elsewhere is an multivariateregressionsusingtheCPS data importantissue. First,the Postal Reorganfor all industriesin May 1978.3 ization Act (PRA) requires comparability. Obviously,the size of the estimatedwage Second, the Postal Service is not immune premium will depend on the specification to marketforces.Althoughitis a regulated used and on the demographic charactermonopoly,the Serviceoperatesin a broader isticsof the group studied. We report two "message transmission"industryin which types of equations. In the first,"basic," key sectors are increasinglycontested by specification,we use only industrydummy firmsin the privatesector.High union wage variables to capture wage differentials.In premiumshinder the Postal Service'sabilthe second, "interactive,"specification,in ity to compete with the prices offered by additionto industrydummyvariablesalone, other message-transmission companies. we include interactionsbetween those varConsequently, some postal functionswill iables and dummy variables for race, gencontinueto shiftto the privatesector.2The der, and membershipin a union.4 result will be an erosion of Postal Service Both specificationscontain variables for employmentand a shiftof postal jobs to privatesector firms. 3At the time our original studywas completed for To analyze comparabilitybetween the the 1981 round of collectivebargaining (see Editor's Postal Service and the private sector, we note, p. 16), the 1978 data were the most current available. Those data are still relevant for our puremploy four data sets: the Current Popuposes here. The May CPS data filesafter 1978 conlation Survey (CPS) of individuals' usual tained wage data on only one-thirdof the sample. earnings,education,occupation,and other 2The electronicmail industryis one example of a potentialmarketthe Postal Servicehas lostto the private sector.Withinthe traditionalmail industry,mail pre-sortingby large firmsand pre-sortbureaus that performthe same functionforsmallerfirmsare substitutingprivatesectorlabor for Postal Service labor.

Comparisons involvingpostal workersdisaggregated by race, gender,and union membershipare thusseriously limitedby sample size. In addition,as indicated below, time-series establishment data suggest that postalwage premiumshave been relativelystablesince 1978. 4Weviewthe regressionsreportedbelow as descriptiveonly.We thereforetreatthe industry,occupation, and union dummyvariables as exogenous.

28

INDUSTRIAL AND LABOR RELATIONS REVIEW

SMSA size, region, marital status, race, gender, union membership, education, education squared, experience,experience squared, occupation, overtimehours, and the SMSA or state unemploymentrate. In the interactivespecification,we also interact educationand experiencevariableswith race and gender dummy variables. Table 1 presentsthe resultsof these regressions. Since our primaryinterestis in interindustrywage differentials,Table 2 reports the wage premiums in Percentages based on thetwospecifications.'These premiums represent the average hourly earnings in the PostalServiceminustheaverage hourly earnings in a particularindustry,divided by that industry'saverage hourlyearnings (times 100). The firstcolumn of Table 2 shows the wage premiums based on the basic specification.Parentheses around a number indicate that the relevant coefficient was not statisticallysignificantly differentfrom zero at the .05 level and was thus set equal to zero when calculatingthe wage differentials. The resultspresentedin Table 2 indicate that the Postal Service pays a wage substantiallyabove that paid by privatesector employersforcomparablyskilledworkers. Two specificwage comparisons are especially relevant,given the PRA standard of "comparable levels of work in the private sector." The firstcompares Postal Service wages withthose paid in the privatesector forcomparable workers(as definedby the CPS controlvariables).As shownin the last rowof column 1,overallwages in the Postal Service are 21.1 percenthigherthan those in the privatesector as a whole. (This statisticis calculated by weightingeach of the private sector industry categories by its share of privatesectoremployment.) 5PeterE. Kennedy has argued in "Estimationwith CorrectlyInterpretedDummyVariables in SemilogarithmicEquations,"American EconomicReview,Vol. 7 1, No. 4 (September 1981), p. 801, that estimatingthe wage premiumby merelycreatingthe exponentialof the relevantcoefficientand subtractingone resultsin a biased estimate.A less biased approach is to estimate the wage premium by exp[c

-

(?/2)

Var(c)]

-

1, where

c is theestimatedcoefficientand Var(c)is theestimated variance of that coefficient.When more than one dummyvariableare included,as discussed below,the generalizationof this procedure involvescovariance termsas well.

The second importantwage comparison is that between the Postal Service and the service and trade sectorsof the economy. Under the basic specificationof column 1 in Table 2, the postal wage is 32.7 percent above the wage in the servicesector(which employs 26.7 percent of the sample) and 32.3 percent higher than the wage in the trade sector (17.2 percent of the sample). The comparison with the "competitive" serviceand trade sectorsof the economyis useful because wages in those sectorsprovide an approximationto the opportunity wage of postalworkers.Not onlydo service and trade employ 44 percentof all private sector workers,but their share of private sector employmentis growing. Since new in thosesectors, jobs are disproportionately a workerlooking for a newjob or a postal worker seeking differentemployment is more likelyto find a job in those sectors thanin the more stronglyunionized sectors of the economy.Althoughunions typically seek to compare theirwages withthosepaid in otherunionized industries,thedeclining employmentopportunitiesin those industriesmake theirwages inappropriatemeasures of the opportunitywage. Columns 2 through 5 of Table 2 depict the resultsof the interactiverace-genderunion specification, where the industry premiumwas made a functionof race, gender, and union membership. The Postal Servicewage premiumscalculated in Table 2 include all of the interactiveterms that were statistically significantat the .05 level. This equation also includes an interactive effectbetween the race-gender and education-experiencevariablesis Although race has a statisticallysignificant overall effecton wages, there are no statistically significant (at the .05 level) raceindustryeffects,and they were therefore omitted from the calculation of industry wage premiums. That is, although white workershave higher wages than do nonwhite workersoverall, this premium does not varyin a statistically significantmanner across industries. All but one gender6The model including age-gender-indCLustry inter-

activevariableswas originally presentedat thearbitrationbetweentheU.S. PostalServiceand theMail HandlersUnionin December1981.

POSTAL SERVICE WAGE COMPARABILITY

29

Table 1. RegressionCoefficients,fromMay 1978 CPS Data. Interactive Specification Coefficients Basic Specification Variable

Interactions with:

Coefficient1-Statistic Coefficients

Race

Gender

Intercept Northeast North Central South Large SMSA Medium SMSA Education Education Squared Experience

0.798 -0.0920 -0.0284 - 0.0932 0.100 0.119 0.000465 0.00209 0.0230

22.8 -15.8 -5.0 - 17.0 16.3 24.8 0.1 13.1 44.6

1.402*** -0.0892*** -0.0280*** - 0.0907*** 0.0976*** 0.116*** 0.0216** 0.0125 -0.0281** 0.00314*** -0.000869** -0.000281 0.00870*** 0.0150*** 0.00672***

Experience Squared

- 0.000388

-35.2

-0.000110**

White Single Male Union Overtime Clerk (Bookkeeper) Clerk (Typist) Clerk (Other) Auto Mechanic Other Mechanic Other Craftsman Operator

(Nontransportation) Operator (Trans., Driver) Laborer Professional(Technical) Manager (Administrator) Farmer UnemploymentRate Construction Manufacturing,Durable Manufacturing, Nondurable Transportation Trade Finance, Insurance, Real Estate Mining Service Agriculture Federal Government State Government Local Government R2

0.0748 -0.0629 0.284 0.165 -0.310 0.0914 0.113 0.0305 0.110 0.128 0.125

11.8

Union Membership

-0.000159***-0.000241***

58.3 34.7 -20.3 6.2 11.3 3.8 6.6 9.9 14.3

- 0.129** -0.0547*** -0.227** 0.110*** -0.310*** 0.114*** 0.108*** 0.0337*** 0.117*** 0.124*** 0.120***

-0.0315 0.0153 -0.0759 0.231 0.274 -0.242 0.027 -0.0564 -0.0913

-3.7 1.4 -7.4 28.7 33.5 -7.9 17.3 -2.6 -4.4

-0.0157 0.0104 -0.0707*** 0.240*** 0.268*** -0.209*** 0.0271*** -0.525*** -0.381***

0.0968 0.0718

0.354*** 0.257***

0.251*** 0.0375

- 0.129 -0.0407 -0.280

- 6.1 - 1.9

- 0.473*** -0.308*** -0.658***

0.0868 0.0691 0.0965

0.231*** 0.209*** 0.346***

0.0411 0.0816** 0.161***

-(0.479*** -0.350** - 0.460*** -0.449*** -0.196** -0.370*** -0.471***

0.0849 0.200 - 0.00138 -0.0675 -0.0237 0.0440 0.0221

0.343*** 0.298*** 0.179*** 0.0986 0.244*** 0.177*** 0.268** 0.5104

-0.107** 0.0639 - 0.0388 0.254*** -0.0354 -0.0146 0.0701**

-14.3

-13.4

-6.6 -0.142 0.108 4.2 - 13.7 - 0.283 -8.4 -0.269 -0.00186 -0.1 -0.174 -6.7 -9.0 -0.212 0.4946

at the.05 levelin a two-tailed test. **Significant at the.001 levelin a two-tailed test. ***Significant

30

INDUSTRIAL AND LABOR RELATIONS REVIEW

Table2. 1978 U.S. Postal Service Average Hourly Wage Premiums,in Percentages.a Basic Specification Construction Manufacturing,Durable Manufacturing,Nondurable Transportation Trade Finance, Insurance, and Real Estate Mining Service Agriculture Federal Government State Government Local Government Weighted Average of PrivateSector

5.98 9.58 13.77 4.18 32.29 15.29 -8.98 32.74 30.92 (0.0) 19.09 23.68 21.07

Female Nonunion 69.43 46.62 60.78 36.30 93.36 61.83 42.73 58.67 57.27 21.87 45.24 60.53 68.98

Interactive Specification Male Female Nonunion Union 18.77 13.31 16.59 10.58 36.72 14.80 5.82 32.64 (57.27) - 4.51 21.64 22.74 25.35

31.88 (46.62)b (60.78) 25.60 64.52 80.12 (42.73) (58.67) 22.10 (21.87) (45.24) 49.69 53.73

Male Union -7.56 (13.31) (16.59) 1.89 16.33 27.76 (5.82) (32.64) (22.10) (- 4.51) (21.64) 14.43 10.09

'The premium equals 100 x [U.S. PostalServicewageminustheindustry wage]/industry wage(adjustedforvariances). wassetequal to zero. bParenthesesindicate thata statistically coefficient insignificant Source:Based on regressions shownin Table 1.

industryinteractiontermsand all but four union-industryinteractiontermswere statisticallysignificant. The resultsindicate that postal workers average 10.1 percenthigherwages thando unionized, male workers (white or nonwhite) in the private sector. The average wage premiums for other postal workers are as follows:25.4 percentmorethanother male, nonunionworkers;53.7 percentmore than otherfemale,union workers;and 69.0 percentmore thanotherfemale,nonunion workers. In their study,Asher and Popkin use a similarregressiontechniquewith1979 CPS data and conclude that "the average wage paid whitemen in the Postal Service fellin the middle of the range of wages paid to comparable white male workers in other industries."7The differencebetween our results and these of Asher and Popkin is primarilythe result of a differentdefinition of comparabilityand a differenteconometricapproach. We base our comparison on postal wages vis-at-vis those received by comparable workersin the privatesector. Asher and Popkin base theirs on postal 7MartinAsherand Joel Popkin,"The Effectof Gender and Race Differentialson Public-PrivateWage Comparisons: A Study of Postal Workers," herein, p. 22.

wages vis-a-visthoseof whitemale workers. As discussed in the conclusion below, we do not believe thattheircomparison is relevant to the problem of wage comparability,eitherin termsof the PRA or in terms of economic concepts of efficiency. Ignoring for a moment the issue of a standard of comparability,we find that postal workershave a positivepremiumof 10.1 percent with respect to male unionized workers,whetherwhiteor nonwhite, and 25.3 percent with respect to nonunionized male workers (Table 2). The weighted average premium for postal all male workers,whiteor workersvis-at-vis nonwhite,is 12.7 percent. The econometricdifferencesthat seem pivotal in Asher and Popkin's conclusion of a near-zero premium for white male workersare the following.First,Asher and Popkin calculate the average wage premium by countingall one-digitSIC industries equally. But the industriesthat they show as paying more than the Postal Service are those industries that employ relativelyfew workers (i.e., mining and transportation),while the industries that they show as paying lower wages are the industriesthat employ most of the labor force (service and trade). If one weights Asher and Popkin'sindustrycoefficients by the percentage of white male workers in

POSTAL SERVICE WAGE COMPARABILITY

31

each industry,theircoefficientsalso show a postal wage premium-one of almost 3 percent. Second, Asher and Popkin adopt a very differentspecificationof theimpactof race on industrywage differentials.As noted above, we include race interactivelywith educationand experiencebutfindthatraceare insignificant. They industryinteractions do notinclude therace-educationand racebut do includeraceexperienceinteractions industryinteractions.Most of their raceindustryinteractivevariables are insignificant, however. Hence, it is largelystatistivariables that lead them cally insignificant to the conclusion of a 3 percent premium for whitemen. To explore furtherthe extent to which differenceswithrespectto race specification affectour differingassessmentof the postal wage premium, we reestimatedthe wage premiums for a sample limited to white male workers.We find,based on the point estimates (that is, including insignificant coefficients),that the Postal Service wage premiumsforwhitemale workersvis-at-vis their counterparts in the private sector range from8.7 to 14.7 percent,depending upon the specificationused. There are a few other differences in thatcontributeto the equation specification differencesin our results.Asher and Popkin include an employer-sizevariable, for example, and thatvariabledoes narrowthe industrywage differentials.Our calculations (based on the 1979 tape, which includes that variable) suggest that their wage premiums are lowered by approximatelytwo percentage pointsby including that variable. If, however, we adopt the approach used by Mellow8and make firm size interactive withunion membership,the overall effectof the interactivespecification of firmsize and union statuswould be even smaller (that is, the postal premium would be larger). Althoughall relevantvariablesincluding firmsize should be included in the regression (if available), the appropriate wage comparison is not one based on white

unionized males in large industriesonly. Rather, the true opportunitywage for an individual with given demographic characteristics,such as experience and education, should reflectthe weighted average across all industries,firmsizes, and other nondemographicvariables.To say thatthe Postal Service pays more and should pay more because it is a large, unionized firm misses the point of wage comparability.If the PostalServicewereonlypayingthesame as other large, unionized firms,it would stillbe paying more than the opportunity wage of the "private sector of the economy,"as required by law.

8WesleyMellow,"EmployerSize and Wages,"Review ofEconomics and Statistics, Vol. 64, No. 3 (August 1982), pp. 495-501.

9To conserve space, these data are not shown here in detail. They are available from the authors upon request.

Comparisons Using the Area Wage Surveys One of the major advantages of multivariate regressionwage studies using the CPS data is that theyallow comparison of Postal Service wages with those paid to otherwisesimilarindividuals employed in the privatesector.The CPS estimatesmay be biased downward,however,to the extent that wage premiums attractworkerswho are more qualifiedthan required bythejob itself.The wage premium on similarjobs, that is, may be differentfrom the wage premium forsimilarworkers. For some PostalServicejobs, such as rural and citylettercarriers,similarprivatesector counterpartsdo not exist.Jobs similar to thoseof mail handlersdo exist,however, namely,material handlers. Data from the Area Wage Surveysshow that Postal Service mail handlersreceivea wage thatis 21 percenthigherthan thatreceivedby material handlersin the SMSAs surveyedbythe U.S. Bureau of Labor Statistics.: Employment Queues, Quit Rates, and Other Job Separations Althoughthe Postal Service pays a wage premium of approximately 21.1 percent over wages in theprivatesectorof theeconomy, it is possible that this premium is attributable to "compensating differen-

32

INDUSTRIAL AND LABOR RELATIONS REVIEW

tials."If workersviewed Postal Servicejobs manufacturing.In other words, approxias unattractive, matelyfivetimesas manyworkersquit their highwageswouldbe needed to compensate them for taking the jobs. jobs in manufacturing as in the Postal This is not the case, however;workersview Service. 1 Postal Servicejobs as unusuallyattractive. These quit rates among postal workers In competitive labormarkets,a firmraises are reflectedin a recentBLS studyon occuits wages when it cannot easily attractand pational separation rates. The BLS found retain sufficient numbers of qualified that the occupational separation rates of workers.The attractiveness of a job depends postal mail carriersand postal clerkswere on such factorsas the relativeprobability considerablybelow thoseexhibitedby other of a layoff,the relativevalue of benefits, clericaland blue-collarworkers.In fact,mail nonpecuniary job attributes, as well as carriers ranked below only eight other wages.l occupations,all of whichwere professional Although anecdotal evidence has to be or managerial.'3 used with caution, it is at least suggestive One reason for the low quit rate in the that numerous newspaper articles have Postal Service is its relativelyhigh wages. indicated that the Postal Service has no Another is its traditionalpolicy of no laytrouble findingnew workers.At the time offs. Although postal employment has of the mail handlers'arbitrationhearingin declined over the past several years, this 1981, for example, newspapers reported decline was achieved through attrition. thata handful of New York Cityopenings Similarly,seasonal peaks in demand, such attractedmore than 100,000 applications as the Christmas season, are absorbed and that the Baltimore police had to be through the employmentof part-timeor called in to controla crowdof 7,000 people seasonal workers. formed after the local Postal Service The labor literatureclearly shows that announced 20 job openings." job securityhas a value and thathighunion These responses, although more exuwages in the private sector are, in part, a berant than most, are not atypical. The "compensatingdifferential'forthe higher Postal Service maintains an employment riskof layoffthere. Abowd and Ashenfelter estimate,forexample, thatcompensatregisterfor some of itsjobs, and that register is always filled.The data maintained ing differentials are "over fourteenpercent by the Postal Service indicate that it vir- in industrieswhich experience substantial tuallynever has troublefillingpositions. anticipatedunemploymentand unemployMoreover, the Postal Service has no dif- mentrisk."'4Given the Postal Service'spolficultyretainingworkers;indeed, it has an icyof no layoffs,however,the postal wage premiums are not explainable even parextraordinarilylow quit rate. Accordingto the Bureau of Labor Statistics(BLS), the tiallyas compensation for a high risk of layoff. monthlyquit rate per 100 workersin 1980 was 0.3 for the Postal Service and 1.5 for Time-Series Evidence on Relative Wages "0The differentialsreported above are for hourly The positive wage differentialpaid by earnings and hence ignore fringebenefitsand working conditions.Douglas K. Adie, in An Evaluationof the Postal Service has risen substantially PostalServiceWageRates(Washington,D.C.: American over the past decade, as shown in Table 3. EnterpriseInstituteforPolicyResearch, 1977), p. 53, estimates that fringebenefitswere 32.1 percent of payrollin the Postal Service (1976-77) and 30.0 percent in the privatesector (1975). "1San Francisco Chronicle,May 10, 1981 ("This World," p. 5). The New YorkTimes,April 2, 1981, p. A24, reported that when the Postal Service listed 20 positionsat the main post officein Baltimore(paying more than$17,000 per year),15,000people picked up applications.Job seekerswaitedin block-longlines for the applications for examinations for jobs that mightopen up later thatyear or early in 1982.

"2The postal quit rates are based on unpublished Postal Service data. See also Adie, An Evaluation. "3AllanEck, "New Occupational Separation Data Improve Estimates of Job Replacement Needs," MonthlyLabor Review,Vol. 107, No. 3 (March 1984), pp. 3-10. "John M. Abowd and Orley Ashenfelter,"Unemployment,Layoffs,and Wage Differentials,"in Sherwin Rosen, ed., Studiesin Labor Markets(Chicago: Universityof Chicago Press, 1981), p. 162.

33

POSTAL SERVICE WAGE COMPARABILITY Table3. Time Series of Relative Postal Service Wages.t (1969

=

Table 4. Relative Wages of Postal Workers,1983. (1969

100.0)

Year

Relative Wage

1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983

104.6 107.3 110.7 114.3 115.9 118.8 120.3 120.3 118.9 118.3 122.2 121.4 121.0 121.9

hantCalculated bydividingU.S. PostalServiceclerks'/mail dlers' average hourlyearningsbythe industry-weighted average hourlyearningsin the privatesector,multiplying by 100, and then dividingby 123.7 (the 1969 totalvalue). Source: U.S. Department of Labor, Bureau of Labor StaandEarnings,variousissues; and U.S. Postal tistics,Emiploynient Service, unpublished data.

Since these data do not control for skill mix,theycan onlybe used to depict trends in relativewages. The numbersin the table measure the average hourly earnings of postalclerks/mail handlersdivided byaverage hourly earnings in the private sector. (The othermajor craftshad wage increases similar to those of clerks and mail handlers.) That ratio is then indexed to equal 100 in 1969, the yearchosen as the starting point because it immediatelypreceded the Postal ReorganizationAct. As depicted in Table 3, between 1969 and 1983 wages for clerks and mail handlers increased 21.9 percent faster than wages for nonsupervisory, nonagricultural,privatesectorworkers.The index was at 118.9 in 1978, the timeof the CPS study. Relativepostal wages have increasedsomewhat in the years since then. Moreover, since 1969 the relativeearnings position of clerks and mail handlers has increased in relationto everyone-digit privatesectoryindustry,as shownin Table 4 (which shows the Postal Service wage divided by the relevantindustrywage and normalized to equal 100.0 in 1969). The postal wage increase over the manufactur-

=

100.0)

PrivateSector Comparison Group Total Mining Constructioll Manufacturing Transportation Wholesale and Retail Trade Finance Services

RelativeWage 121.9 102.4 129.2 115.9 107.9 126.7 129.0 115.5

Source:see Table 3.

ing wage was 15.9 percentas of 1983, and itwas 26.7 percentover the wage in wholesale and retail trade. The substantialincreasesin relativewages in the Postal Service can be attributedto two factors.First,politicalpressuresled to major postal wage increases. In particular, the Postal Service granted a large wage boost in 1970 to ensure union support of passage of the Postal Reorganization Act. In 1973, partlyin responseto politicalpressure, as Tierney describes it, management "gave the store away ...

in an extraordi-

narycapitulation.""5Whetherthisdescriptionis valid,thepackage was certainlymore generous than could be explained by economicconditionsor even bywage increases won by other labor unions. Second, the unionized sector in general has increased its wage premium over the nonunionizedsectorover the past tenyears. This has been particularlytruein thetransportation,communications,and public sectors,in mining,and in some manufacturing industries,such as steel and autos. In the major nonunion sectors-service, wholesale and retail trade, and finance-wage increaseswere considerablybelow those of the union sectors. In earlier periods, the wage premiums of high-wage or unionized industries Managing I5John T. Tierney, PostalReorganization: thePublic'sBusiness(Boston: Auburn House, 1981), p. 99.

34

INDUSTRIAL AND LABOR RELATIONS REVIEW

increased during recessions but declined duringexpansions. In contrast,the relative union wage gains duringthecyclicaldownturnsof the early 1970s were not reversed during the followingexpansion. This shift in patternappears to have been the result of cost-of-living adjustment(COLA) clauses indexed to the Consumer Price Index for Urban Wage Earners and ClericalWorkers (CPI-W). Indeed, in mostof the industries where relative wages increased significantly,long-termcontractsincludingCOLA clauses were in effect." The basic problems were the upward bias in the CPI-W measure of inflation and the difficulties in insuringwages againstan inflationprocess caused by an increase in importprices. Thus, even if the Postal Service had a zero wage differentialin 1970, the differential in 1983 would have been approximately22 percent.In fact,earlierresearch suggeststhatthe Postal Service had a positivewage differential in the late 1960s,even before postal reorganization.For example, using a sample of older whitemen, Quinn found thatin 1969 postal workersreceived 11 percent more than private sector employees, when not controllingfor their individual characteristics,or 12 percent more,when controllingfortheirindividual characteristics.17 Smith noted that regression estimatesbased on 1970 census data in Washington,D.C., Maryland,Virginia, and Delaware showed thatbothsexes in the Postal Service received significantly higher wages than their private sector counterparts.She wenton to show,using 1973 CPS data, that postal workers of both sexes "receive wages which are superior to the wages of nonunionized private sector workers of similar socioeconomic characteristicsand at least comparable to union8 ized privatesectorworkers."'

6See Michael L. Wachter and Susan M. Wachter, "InstitutionalFactorsin Domestic Inflation,"in After thePhillipsCurve: The Persistence ofHigh Inflationand High Unemiployment, Conference Series No. 19 (Boston: Federal Reserve Bank, 1978), pp. 124-55. 17Joseph F. Quinn, "Postal SectorWages,"Industrial Relations,Vol. 18, No. 1 (Winter 1979), pp. 92-96. 18Sharon P. Smith,"Are Postal WorkersOver- or Underpaid?" IndustrialRelations,Vol. 15, No. 2 (May 1976), pp. 175n and 176, respectively.

Conclusion The appropriate comparison to use in examining wage comparability is the opportunitywage: what a given individual employee could earn elsewhere.This standard has appeal on efficiencyand equity grounds and is consistentwith the legal mandate set for the Postal Service by Congress. Beyond theirinconsistency withthecomparabilitystatute,there are many reasons for rejecting direct comparisons with unionizedworkersor workersin large firms only.Postal Serviceworkersare not subject to cyclicalor seasonal layoffsas are most unionized workers.In addition,othernonpecuniary factors,such as those found in manufacturing,construction,or mining, may favor the Postal Service; lower quit rates among postal workers than among manufacturingworkersmay be an indication that these factorsare important. These factorsall point to the conclusion thatthe opportunitywage for postal workers is well below that of employed unionized workers.Most of the economy is not unionized, and, in fact,the unionized sectors have been declining in relative and absolute size forover a decade. The decline in overall union employmentmakes it very unlikely that alternative employment opportunities and thus the opportunity wage for Postal Service workersare in the high-wageunionized sectors. An interestingquestion is the choice of the appropriatecomparisongroup, in view of the extent to which race and sex discriminationexistin the privatesector.That the Postal Service does not discriminate seems beyond controversy (and is confirmedby formalstatisticaltests).This lack of discriminationdoes not mean, however, that the appropriate wage-comparison standard should be that of white male workers(or whitemale unionized workers in large firms). The law requires that the Postal Service treat workersthe same regardless of race or gender. We obviouslysupport thisview and the Postal Service's current nondiscriminatoryposture. Nonetheless, neither the law nor principlesof economics dictate that the Postal Service, as an employer,

POSTAL SERVICE WAGE COMPARABILITY should pay the highestwage in the economy. Indeed, to the extentthatlabor market discrimination causes wage differentials, the whitemale wage is "too high." That is, if there were no race-genderwage differentials,the wage forwhitesand men would decline, while that for nonwhites and women would increase.'9 Thus, the nondiscriminatory standard would be most closelyapproximatedby the average wage across race and gendergroups,and, bythat standard, the Postal Service pays a wage premium of 21 percent. Althoughwe believethatthecomparison withthe serviceindustryor the entireprivate sector provides an estimate of the opportunitywage of Postal Service work'9We referhere to the differencein the twogroups' nominal wages. We are ignoringthe general equilibriumissue of whetherracialdiscrimination causes both groups' wages to be higheror lower than theywould be in the absence of discrimination.Rather, we are simplyconcentratingon the wage differentialat the margin.

35

ers, our estimatesindicate that the Postal Service wage scale is high enough to more thancompensatewhitemale workersin the unionized sector. These very high postal wage rates, which are considerablyabove costof postalworkers,have the opportunity a cost. This cost is absorbed, not only by users of postal servicesbut also by workers who mightotherwisebe employedin a more competitivepostal industry. Indeed, there is even a potentialcost to currentemployees in the form of partial deregulation of certain postal functions. Current regulatoryinitiativesattempt to address inefficientor high-costregulated industriesby encouraging"contestedmarkets."Their intentis to limitthe regulated monopoly sector of an industry to the smallest feasible part. Priorityand electronic mal, private sector pre-sortfirms, package delivery,and contractstationsare only a few of the servicesthat are already causing a shiftof some postal jobs to the nonregulatedprivatesectoroftheeconomy.