How Does Collective Bargaining Affect the Gender Pay Gap?

WORK AND OCCUPATIONSElvira, Saporta / COLLECTIVE BARGAININGThe authors study the effect of unionization on gender wage differentials for production workersin nine U.S. manufacturing industries. They find that the wage gap is significantly smaller inunionized establishments for six of the industries, even after controlling for occupation andestablishment gender composition. But this union effect does not hold within three industries.The authors conclude that unionization generally reduces wage inequality between blue-collarmen and women, but the effect might be contingent both on the overall proportion of women in anindustry and on union characteristics. The authors discuss the implications of these findings forincome inequality and union policies.

How Does Collective BargainingAffect the Gender Pay Gap?

MARTA M. ELVIRAUniversity of California, Irvine

ISHAK SAPORTATel Aviv University

Salary differences between women and men in the United States remainsubstantial, despite the similarity in their skill levels (Blau & Kahn,

1995). The ratio of women’s to men’s earnings, which had been rising sincethe 1970s, stabilized during the 1990s, with women’s weekly wages repre-senting approximately 76% of men’s (Blau & Ehrenberg, 1997). This earn-ings gap has been explained by different mechanisms, including human capi-tal and seniority differences between men and women; gender segregation byoccupation, firm, and industry; and employer discrimination against women

469

Authors’ Note: We thank Trond Petersen for access to Bureau of Labor Statistics dataand Lisa Barron, Tom Buchmueller, Lisa Cohen, Ha Hoang, Deborah Kidder, Jona-than Leonard, David Levine, Jone Pearce, George Strauss, and Ian Taplin for helpfulcomments. We also thank the editor and two anonymous reviewers for very construc-tive feedback. Correspondence concerning this article should be addressed to MartaM. Elvira, Graduate School of Management, University of California, Irvine, Irvine,CA 92717-3125; tel: (949) 824-5178; fax: (949) 824-8469; e-mail: [email protected]; or to Ishak Saporta, Faculty of Management, Leon Recanaty GraduateSchool of Business, Tel Aviv University, Ramat Aviv, Tel Aviv, Israel; e-mail:[email protected].

WORK AND OCCUPATIONS, Vol. 28 No. 4, November 2001 469-490© 2001 Sage Publications

(Reskin, 1993). Human capital explanations focus on workers’ educationand accumulated experience. Because women traditionally have differenteducational backgrounds, shorter tenures, and more interrupted careers thanmen, these variables explain part of the wage gap (Blau & Beller, 1988;Gunderson, 1989). Gender segregation across occupations (Sorensen, 1989),industries (Hodson & England, 1986), and firms (Groshen, 1991) accountsfor another large component of the gap. The still “unexplained” portion ofwage differentials is usually interpreted as evidence of employerdiscrimination.

Looking specifically at employer-based inequality, one research streamemphasizes organizational characteristics (Baron, Mittman, & Newman,1991; Reskin, 2000). This work has focused mostly on the negative effect ofjob and establishment gender segregation on earnings (Huffman & Velasco,1997; Tomaskovic-Devey, 1993, 1995). For example, Nelson and Bridges(1999), in several case studies of between-job wage differences, concludedthat the gender compositions of occupations and the gender of individualemployees are more predictive of pay disparities than are market pay rates.Through organizational processes such as hiring and promotion, men andwomen are allocated to gender-segregated work units or job titles, with pre-dominantly female jobs receiving lower pay than predominantly male jobs(Bielby & Baron, 1984; Bridges & Villemez, 1994; Maume, 1999). Theselower earnings of women relative to men because of gender segregationwithin employers cannot be explained by lower productivity (Tomaskovic-Devey & Skaggs, 1999) and are not mitigated by employer policies such asformalization of employment procedures (Huffman & Velasco, 1997).

Adding to this research on organizational characteristics, we explore theeffects of unionization on wage differentials by gender within nine manufac-turing industries. Unionization has rarely entered studies of employer-basedgender inequality, except as a control variable (e.g., Tomaskovic-Devey &Skaggs, 1999). This is surprising because perhaps the most important factorinfluencing wage determination within firms is whether wages are subject tocollective bargaining. Statistics indicate that union membership helps raiseworkers’ pay in general and narrows the income gap that leaves women andminorities at a disadvantage. For example, the Economic Policy Institute(1999) reported that union workers overall earn 32% more than those not inunions. The union wage benefit is most pronounced among women andminorities: In 1998, women in unions earned 39% more than their nonunioncounterparts. These statistics suggest the potential relevance of unions inreducing the gender wage gap, but research evidence is scant (Gander, 1997;Nelson & Bridges, 1999).

470 WORK AND OCCUPATIONS

In this article, we suggest that unionization reduces the gender wage gapfor several reasons. First, unions tend to establish wage-setting, bureaucraticprocedures that reduce wage dispersion among employees covered by thesame collective bargaining agreement, especially those working in the sameoccupation (Freeman & Medoff, 1984). Therefore, unionization shouldreduce the wage gap for women working alongside men in the same estab-lishments and jobs. Second, unions tend to reduce wage differentials withinestablishments regardless of occupation, as well as across establishments, onthe basis of set pay for specific jobs (Freeman, 1980). This wage homogeni-zation tendency would reduce wage differences between segregated femaleand male jobs. Third, the management of unionized establishments is morelikely to adhere to such bureaucratic wage rules, reducing arbitrariness inwage rates and consequently reducing the potential for discrimination (Corn-field, 1987). Finally, a more direct mechanism through which unions mightreduce the gender gap in pay is explicit efforts to restore pay equity, that is, toraise pay in mostly female jobs relative to predominantly male jobs (Acker,1989).

To examine whether unions help reduce the pay gap between men andwomen working in similar occupations, our study looks at within-establish-ment union effects in production jobs, using individual-level data from nineindustries. Because collective bargaining takes place at the establishmentlevel, one needs information on both individual and establishment attributes,and such data are difficult to find. The Industry Wage Surveys (IWS) of theBureau of Labor Statistics (BLS) provide these data. We include in the studynine manufacturing industries that vary substantially in the proportions ofmen and women employed, from 5% to 92% female. Within each industry,surveys include a representative sample of several hundred establishmentsand several thousand employees, facilitating the generalization of results toeach particular industry and providing a unique opportunity to test the effectsof collective bargaining on wages.

These IWS data overcome three important shortcomings common to priorstudies of organizational determinants of the gender gap. According toHuffman and Velasco (1997), prior data have (a) concentrated on censusoccupational categories rather than specific work settings, (b) downplayedorganizational attributes, and (c) focused on small samples of establishmentsin specific labor markets. By contrast, the IWS provide data on manufactur-ing industries with information on a large proportion of each employer’sworkforce and extensive detail on establishment characteristics. Besidesunionization, variables include organizational attributes such as region,establishment size, and gender composition. Data on individuals include

Elvira, Saporta / COLLECTIVE BARGAINING 471

detailed occupation and pay method, well-known variables affecting thewage gap.

In sum, our study adds to recent research on organizational determinantsof the earnings gap between men and women in several ways. First, we focuson a critical determinant of wages: unionization. Second, we test our hypoth-esis across occupations, establishments, and industries that vary in the gendercomposition of the labor force and in the levels of unionization. These fea-tures allow a more precise study of the relationship between wage differen-tials and unionization than other data sets in which employees are notmatched with their firms, only a few firms are surveyed, or only informationon broadly defined occupations is included (Anderson, Doyle, & Schwenk,1990; Huffman & Velasco, 1997). Before presenting our results, we elaborateon the theoretical rationale for our proposition.

EFFECTS OF UNIONIZATION ON THEGENDER WAGE GAP: THEORY AND EVIDENCE

How does unionization affect the gender gap in pay? Research suggeststhat wage determination differs in unionized and nonunionized establish-ments. We look at how these wage determination processes affect genderwage inequality, distinguishing between indirect and direct effects of unions.

THEORY: INDIRECT EFFECTS

Indirect effects occur because unions put in place bureaucratic systemsthat tend to lower pay variation for all employees. First, union policies gener-ally aim to obtain pay equality across and within establishments, standardiz-ing wage levels and thus reducing wage dispersion among workers coveredby union contracts (Card, 1992; Freeman & Medoff, 1984). Though not uni-versally applied, standard-rate wage policies can be defined as “uniformpiece or time rates among comparable workers across establishments andimpersonal rates or ranges of rates in a given occupational class within estab-lishments” (Freeman, 1980, p. 4). A consequence of standard wage rates isthe minimization of the impact of individual differences among workers insimilar jobs within establishments. Some evidence exists that unions reducethe gap between White and Black men (Freeman & Medoff, 1984). Likewise,unions may reduce the gender gap through adherence to wage rates assignedto job classification systems. Therefore, within a given establishment andoccupation, wage variation for men and women will be reduced, assumingother individual characteristics such as seniority remain constant. As long as


both men and women occupy similar jobs and are covered by similar collec-tive bargaining agreements, wage differences between them should be lower.

These collective bargaining wages are associated with bureaucratic proce-dures set up over time to regulate labor-management relationships and reducesubjective pay allocations by supervisors (Slichter, Healy, & Livernash,1960). Previous research on ethnic differences in layoff chances suggests theimportance of these procedures: In unionized workplaces, management’sadherence to bureaucratic rules such as seniority makes these rules better pre-dictors of layoff chances across ethnic groups (Cornfield, 1987). Amongnonunionized workers, by contrast, the determinants of ethnic inequality inlayoffs are less clear. Applying the same logic, we expect the management ofa unionized firm to follow collective bargaining outcomes in wage setting,thereby generating more predictable wages for male and female employeesthan would be the case in nonunionized workplaces, where managerial dis-cretion is greater.

THEORY: DIRECT EFFECTS

In addition to indirect effects, we expect unionization to have a directinfluence on the wage gap. Because more women are currently organizedthan ever before, some unions have adopted pay equity as a strategy to elimi-nate the pay gap. They have also adopted policies to combat the gender segre-gation of occupations, which is related to the wage gap (Hallock, 1993).These strategies do not always arise spontaneously, but at the request offemale members (Howell & Mahon, 1996). Regardless of the motivation,such pay equity policies would increase the earnings of women more thanthose of men, thereby narrowing the gender gap. Although this effect seemsrather obvious, empirical research is scant and inconclusive, with some stud-ies suggesting that unions have neglected women’s issues (Cook, 1991).

In summary, theory suggests that unions would reduce gender gap by vir-tue of indirect effects in reducing wage dispersion within firms and a poten-tially direct effect in setting pay equity policies.

EMPIRICAL EVIDENCE

Evidence for these unionization effects seems to exist only for male work-ers. The effect of unions on female wage inequality is relatively understudied(Pfeffer & Ross, 1981), and little research exists on the gap itself. Studies ofwage policies under collective bargaining agreements indeed indicate thatunionization reduces wage dispersion among union members by equalizingwages across and within establishments. This evidence comes primarily


from all-male data sets (Lewis, 1986). For example, Freeman’s (1980) studyfocused on men and the wage differential between white- and blue-collarworkers within establishments. Pfeffer and Ross (1980) also showed thatamong men, unions operate to reduce wage differentiation and individualcharacteristic effects. Evidence for women is inconclusive at best. A fewstudies have looked at union wage premiums for women only (Main & Reilly,1992) and found that unions have no effect on wage dispersion amongwomen (Pfeffer & Ross, 1981). In a recent study based on Current PopulationSurvey data for individuals between 1973 and 1992, DiNardo, Fortin, andLemieux (1996) found that the presence of unions explains an important por-tion of wage inequality among men only and that changing unionization pat-terns explain relatively little of the recent rise in female wage dispersion.

Although they are important, these studies are applicable to comparisonsof within-gender samples. Research focusing on how unions affect the wagedifferential between men and women is even scarcer. Unlike our study, thefew articles that have looked at unions and the gender gap have focused on thepublic sector because most unionized women work there (Baron & Newman,1990; Bridges & Nelson, 1989) or used individual-level data aggregatedacross national economies without considering establishment-level informa-tion on collective bargaining coverage (Doiron & Riddell, 1994; Even &Macpherson, 1993; Freeman & Leonard, 1987).

Because the theoretical mechanisms reviewed above operate at the estab-lishment level, so do the data we use in this study. Our choice is important forseveral reasons. First, collective bargaining agreements usually differ amongestablishments and specify wage rates for different jobs so that everyemployee working in a given job earns approximately the same amount. Sec-ond, evidence highlights the role of employer policies in shaping the wagegap (Groshen, 1991; Milkman, 1987). Interemployer wage differentialsaccount for a great share of the gap between men and women because mentend to work in higher pay firms (Gunderson, 1989; Halaby, 1979). Third,gender segregation is strongly related to the wage gap, especially segregationwithin jobs and establishments (Huffman & Velasco, 1997; Petersen & Mor-gan, 1995). Therefore, firm-level data with information on wage-setting pro-cesses are critical to understanding gender-related wage differentials.

In sum, in this study, we ask: How does unionization affect the genderwage gap among individual men and women working in the same establish-ment who are covered by similar collective bargaining agreements? Is the gapsmaller in unionized than in nonunionized establishments? We investigatethis question by adding an interaction term of unionization and gender to anequation containing individual and establishment characteristics. We expectthis interaction to be positive.


DATA

The IWS of the BLS provide nationwide data on establishment and pro-duction workers’ characteristics. We obtained access to nine manufacturingindustries that vary widely in their proportions of men and women employed:industrial chemicals, nonferrous foundries, paints and varnishes, textile dye-ing and finishing, wood household furniture, cotton fiber textiles, wool tex-tiles, miscellaneous plastics, and men’s and boys’ shirts. Our main criterionin selecting these industries was to cover the spectrum from low to high pro-portions of women in an industry’s workforce, that is, to obtain broad varia-tion in our independent variable while varying also the type of product manu-factured. These are almost all of the manufacturing industries included in theIWS between 1974 and 1985 (U.S. Department of Labor, 1976a, 1976b,1977a, 1977b, 1977c, 1977d, 1977e, 1977f, 1982).1 In 1985, the IWS stoppedincluding questions about employee gender, so later surveys are not valid forour research question. Therefore, we have access to most of the manufactur-ing industries surveyed during the period, precisely the same industries stud-ied in other research on the wage gap (Groshen, 1991; Petersen & Morgan,1995).2

In each industry, the BLS drew a stratified random sample of establish-ments, including a sufficient mix of union and nonunion establishments totest our prediction. The number of establishments surveyed ranges from 57 inthe wool textiles industry to 876 in the miscellaneous plastics industry. Dataon establishment characteristics include size, region, urban area, and tech-nology. These data are combined with information on individuals, includinggender, occupation, establishment, method of wage payment, and hourlyearnings. Our unit of analysis is the individual employee.

Apart from the contributions to research already explained, there are sev-eral methodological advantages to using these data. First, wage data used forour dependent variable are in hourly earnings form. Because one confound-ing factor in explaining gender wage differences is that men and women workdifferent numbers of hours, our data avoid the drawback of having to controlfor the estimated number of hours worked.

Second, potential gender bias in the sample of occupations surveyed isreduced because only production workers are included. Moreover, the dataon hourly earnings are collected for narrowly defined occupations represen-tative of the range of activities performed in the industry (Anderson et al.,1990). Thus, results are generalizable within industries. The surveys providedetailed occupational coding, which often corresponds to the listing in theDictionary of Occupational Titles (U.S. Department of Labor, 1977a). Forexample, in the miscellaneous plastics industry, the list of detailed


occupations includes titles such as “extrusion-press operators (set up andoperate)”; “extrusion-press operators (operate only)”; “plastic cutters,machine”; and “scrap-preparing operators.” Similarly, in the textile dyeingand finishing industry, occupational titles distinguish among “printers,machine”; “printers, screen (automatic flat)”; “printers, screen (automaticrotary)”; and “printers, screen (hand).” With this level of detail, as Groshen(1991) explained, “classifications are not aggregated to avoid combiningsimilar male and female jobs into apparently integrated occupations” (p. 460).By controlling for job title, we can also assume that skill level within occupa-tion is close to constant, thus controlling for different returns to education andskill for men and women (Lemieux, 1993). The number of occupations sur-veyed per industry ranges from 19 in paint and varnishes to 71 in wooltextiles.

Our key independent variables are gender and unionization. Gender ismeasured by an indicator variable that takes the value of 1 when the employeeis female and 0 otherwise. Regarding unionization, the IWS classify a workeras unionized if the majority of production workers in the establishment arecovered by collective bargaining agreements. Therefore, the definition ofunion status is at the establishment level: A worker scores 1 for the unioniza-tion variable if most workers in his or her establishment are covered by a col-lective bargaining agreement, whether or not that worker is a union member.In manufacturing, this measure is practically the same as unionization rate.

Descriptive statistics for the key variables used are presented in Table 1.We first report average wages for all employees in each industry and then formen and women separately. In all industries, women’s mean hourly wage islower than men’s. The ratio of women’s to men’s pay ranges from 0.78 inmiscellaneous plastics to 0.93 in cotton and fiber textiles. Hourly wages inindustrial chemicals are the highest among both men and women: 6.47 and6.20, respectively. The extent of union coverage ranges from approximately21% in cotton textiles to about 80% in industrial chemicals. Aside from cot-ton, the industries at the lower end of the union representation scale tend to betextiles: wool textiles (26.2%) and men’s and boys’ shirts (33.5%). On aver-age, more men than women work in unionized establishments.

The percentage of women in these industries varies widely from 5.6% inindustrial chemicals to 92.7% in men’s and boys’ shirts. Not surprisingly,textiles employ greater proportions of women than other industries. Segrega-tion seems common in textiles: Descriptive statistics show that womenalways work with more women in the establishment than men do.

Control variables at the establishment level include size and geographicallocation because previous research has shown that union workers tend to beconcentrated in large firms, which often pay more than small ones, and in


TABLE 1: Variable Means by Industry and Gender

Percentage WomenHourly Wage Union in Establishment N

N N

Occu- Establish-Industry All Men Women All Men Women All Men Women All Men Women pations ments

Industrial chemicals 6.460 6.470 6.220 0.80 0.810 0.640 2.470 2.070 9.080 71,400 67,371 4,029 27 266(1.02) (1.01) (1.17) (0.40) (0.39) (0.48) (0.06) (0.05) (0.14)

Paint and varnishes 5.210 5.260 4.770 0.710 0.710 0.780 5.550 4.560 15.60 10,939 9,956 983 19 296(1.03) (1.02) (1.02) (0.45) (0.46) (0.42) (0.09) (0.07) (0.16)

Nonferrous foundries 4.690 4.780 4.030 0.640 0.650 0.580 9.410 7.120 25.570 18,218 15,959 2,259 43 364(1.20) (1.21) (0.88) (0.48) (0.48) (0.49) (0.14) (0.11) (0.17)

Textile dyeing and finishing 3.960 4.110 3.420 0.450 0.490 0.310 19.820 13.110 43.60 19,571 15,267 4,304 49 148(0.98) (1.02) (0.53) (0.50) (0.50) (0.46) (0.22) (0.16) (0.25)

Wood household furniture 3.040 3.190 2.810 0.370 0.380 0.360 35.920 29.640 46.190 36,140 22,417 13,723 33 332(0.81) (0.82) (0.73) (0.48) (0.48) (0.48) (0.20) (0.19) (0.18)

Wool textiles 3.180 3.290 3.030 0.260 0.280 0.240 38.890 35.380 44.170 10,228 6,141 4,087 71 57(0.61) (0.65) (0.49) (0.44) (0.45) (0.43) (0.15) (0.12) (0.16)

Cotton fiber textiles 3.110 3.230 2.970 0.210 0.220 0.210 45.640 43.690 47.90 163,842 87,886 75,956 43 340(0.52) (0.57) (0.42) (0.41) (0.41) (0.41) (0.10) (0.09) (0.11)

Miscellaneous plastics 3.310 3.750 2.940 0.520 0.560 0.490 51.930 35.260 65.880 70,355 32,057 38,298 42 876(0.88) (0.98) (0.58) (0.50) (0.50) (0.50) (0.28) (0.28) (0.20)

Men’s and boys’ shirts 3.340 3.920 3.290 0.330 0.330 0.340 92.50 88.70 92.790 40,068 2,928 37,140 30 220(0.75) (1.13) (0.69) (0.47) (0.47) (0.47) (0.05) (0.09) (0.05)

NOTE: Data were collected between 1974 and 1981, depending on the industry (U.S. Department of Labor, 1976a, 1976b, 1977a, 1977b, 1977c,1977d, 1977e, 1977f, 1982). Numbers in parentheses are standard deviations.

477

urban areas, which have higher pay than rural areas. Establishment size ismeasured by six dummy variables, region by four dummy variables, and citysize by one dummy variable taking a value of 1 if a firm is located in a metro-politan area. Because gender segregation within firms explains a large por-tion of earnings inequality between men and women (Bielby & Baron, 1984;Bridges & Nelson, 1989), we control for the proportion of women in an estab-lishment’s workforce.

Variables used as controls at the individual level include detailed occupa-tion dummies (which vary by industry) and incentive pay, a dummy that takesa value of 1 if an employee receives incentive compensation. This last controlis important because historical research has shown that the method of pay-ment tends to vary between men and women (Goldin, 1986).

METHOD

We estimated three standard log-wage equations using ordinary leastsquares (OLS) regression analysis. The dependent variable in these equations


TABLE 2: Ordinary Least Squares Estimates of Gender and Union Effects onthe Logarithm of Hourly Wage

PercentageWomen in

Female Establish- City RegionIndustry Female Union × Union Incentive ment Size 2

Industrial chemicals –0.065** –0.018** 0.035** –0.109** –0.411** 0.060** 0.029**(0.004) (0.001) (0.005) (0.004) (0.009) (0.001) (0.001)

Paint and varnishes –0.088** 0.053** 0.087** — –0.653** 0.179** –0.122**(0.011) (0.004) (0.013) — (0.018) (0.009) (0.005)

Nonferrous foundries –0.057** 0.056** 0.023** 0.085** –0.264** 0.019** —(0.007) (0.003) (0.008) (0.004) (0.011) (0.003) —

Textile dyeing and –0.029** 0.012** –0.098** 0.024** –0.178** 0.033** –0.127**finishing (0.004) (0.003) (0.005) (0.003) (0.007) (0.002) (0.003)

Wood household –0.047** 0.005 0.020** 0.096** –0.316** 0.073** –0.116**furniture (0.003) (0.003) (0.004) (0.002) (0.005) (0.002) (0.003)

Wool textiles –0.059** 0.121** 0.015* 0.080** –0.273** 0.091** —(0.004) (0.004) (0.007) (0.004) (0.012) (0.003) —

Cotton fiber textiles –0.065** 0.008** 0.012** 0.071** –0.105** 0.012** –0.047**(0.001) (0.001) (0.002) (0.001) (0.003) (0.001) (0.001)

Miscellaneous –0.042** 0.033** 0.000 0.106** –0.175** 0.048** –0.054**plastics (0.002) (0.002) (0.003) (0.003) (0.003) (0.002) (0.002)

Men’s and boys’ –0.010 0.170** –0.065** 0.089** 0.086** 0.004 –0.059**shirts (0.006) (0.007) (0.007) (0.004) (0.018) (0.002) (0.003)

NOTE: All models include occupation control dummies, proper to each industry. Num-bers in parentheses are standard errors.*p < .05. **p < .01.

is the natural logarithm of hourly earnings. This semilogarithmic form ofwage equation facilitates interpretation of estimated effects: The coefficientsapproximate the percentage change in wage level resulting from a unitincrease in the independent variable.

We assume the same wage determination model for men and women. Thebasis for this premise is the fact that within an establishment, one can reason-ably assume that all employees are covered by the same collective bargainingagreement and therefore that wages are set for union as well as nonunionworkers. Thus, collective bargaining coverage is the appropriate measure ofunion status to determine union effects on the earnings of men and women ina given work unit.

Table 2 shows the estimates of an OLS model where the log of hourlyearnings (W) is regressed on a gender dummy (1 = Female), a union dummy(1 = Union), an interaction variable (Female × Union), and the vector of con-trol variables. The model is as follows:

W = α + β1F + β2U + β3(F × U) + β4(X)


Depen-dent

Region Region Variable3 4 Size 2 Size 3 Size 4 Size 5 Size 6 Size 7 R 2 Mean

— — –0.242** –0.207** –0.135** –0.100** –0.050** –0.006** 0.400 1.850— — (0.011) (0.004) (0.002) (0.002) (0.002) (0.002)

0.043** 0.095** 0.139** 0.211** 0.281** 0.291** 0.376** — 0.470 1.630(0.004) (0.005) (0.017) (0.017) (0.016) (0.017) (0.017) —0.005 –0.046** 0.007 0.050** 0.074** 0.147** 0.286** 0.391** 0.570 1.510(0.003) (0.004) (0.009) (0.009) (0.008) (0.009) (0.009) (0.009)

— –0.064** –0.220** –0.126** –0.195** –0.147** –0.142** –0.036** 0.620 1.350— (0.005) (0.013) (0.007) (0.006) (0.006) (0.006) (0.007)

0.092** 0.110** 0.084** 0.058** 0.110** 0.163** 0.152** 0.176** 0.460 1.080(0.003) (0.005) (0.023) (0.022) (0.022) (0.022) (0.022) (0.022)

— — 0.006 –0.118** –0.040** –0.085** — — 0.450 1.140— — (0.010) (0.007) (0.004) (0.004) — —— — –0.022 –0.077** –0.057** –0.042** –0.037** –0.025** 0.420 1.120— — (0.015) (0.003) (0.002) (0.001) (0.001) (0.001)

0.052** 0.006* 0.034* 0.037** 0.088** 0.154** 0.129** 0.262** 0.570 1.160(0.002) (0.002) (0.013) (0.013) (0.013) (0.013) (0.013) (0.013)

— –0.011** –0.297** –0.304** –0.274** –0.249** –0.244** –0.166 0.260 1.180— (0.004) (0.088) (0.088) (0.088) (0.088) (0.088) (0.088)

In this model, β1 measures the gender gap in nonunion establishments andβ1 + β3 measures the gender gap in union establishments. The reference groupis men working in nonunionized establishments. This specification empiri-cally addresses the research question of whether the gender gap differsbetween union and nonunion establishments (measured by β3). This interac-tion evaluates whether collective bargaining not only affects wages but alsoincreases wages more for women than men (i.e., if the gender effect is lessnegative in unionized establishments), thereby lowering the gender wagegap.

RESULTS

The three main coefficients reported in Table 2 are those for the variablesfemale, union, and the interaction between the two. In all industries but one,the coefficient for female is negative and significant, indicating that being awoman in a nonunionized establishment is associated with lower wages by afactor ranging from 8.8% in paints and varnishes to 2.9% in textile dyeingand finishing. Only in men’s and boys’ shirts is the gender coefficient not sig-nificant, if negative, probably because the overwhelming majority of workersin this industry are women.

Estimated union coefficients suggest that collective bargaining coverageis associated with higher wages in almost all the industries studied. Thesecoefficients are positive and significant in seven out of nine industries, rang-ing from 0.8% in cotton and fiber textiles to 17% in men’s and boys’ shirts. Inwood and household furniture, the coefficient is positive but not statisticallysignificant. The only negative union effect is for industrial chemicals; asAnderson et al. (1990) also reported, the wage difference between union andnonunion workers is only 2%, even though 61% of the production workers inthis industry are unionized.

As predicted, the interaction between the union and female variables ispositive and significant in six of the nine industries, with coefficient valuesranging from 1.2% in cotton and fiber textiles to 8.7% in paint and varnishes.For the remaining three industries, one coefficient is not significant at con-ventional levels (miscellaneous plastics), and two are negative and signifi-cant (textile dyeing and finishing and men’s and boys’ shirts). These resultssuggest that in most cases, unionized establishments have lower wage gapsbetween men and women. Consider the nonferrous foundries industry as anexample. The estimated gender coefficient indicates that being a woman in anonunionized establishment in this industry is associated with an approxi-mately 5.7% lower pay rate.3 For women working in a union establishment,


however, the net effect results from adding the female, union, and interactioncoefficients (–5.7% + 5.6% + 2.3% = 2.2%). A wage advantage of about2.2% accrues to women covered by collective bargaining, compared with awage gap of about –5.7% in nonunion establishments, all other factors keptconstant. In sum, in six industries, unionization is associated with a narrowerwage gap between unionized men and women relative to men and women insimilar jobs in nonunionized establishments.

The two industries with a negative and significant interaction of thefemale and unionization variables are textiles: men’s and boys’ shirts andtextile dyeing and finishing. In these industries, the gender gap in pay isgreater in unionized than nonunionized establishments.

Note also that women receive higher wages when unionized, althoughcomparing women in unionized and nonunionized establishments is not thefocus of this article. For example, in industrial chemicals, the predicted logwage difference between the two groups of women is 0.035 – 0.018 = 0.017.That is, women in union establishments earn about 1.7% higher wages thannonunion women. If similar calculations are done for all industries, womenin unionized establishments do earn somewhat more than other women in allbut one of the industries, with a gain ranging from about 1.7% to 14%. Insum, women gain from unionization both relative to men (as shown by thesignificant interactions) and in absolute terms (the coefficient for union plusthe coefficient for the interaction add to a positive effect).

The estimated effects of control variables are consistent with priorresearch. For example, the percentage of women in an establishment has anegative effect on wages in all industries except men’s and boys’ shirts (anindustry where 92% of establishments’ workforces are female). Incentivepay has a positive and statistically significant effect on workers’ wages in allbut one industry. The size of the city where an establishment is located and anestablishment’s size are positively related to pay levels.

Overall, our estimates are consistent with previously reported studiesusing similar data, but also reveal that unionization is a significant determi-nant of the gender wage gap.

DISCUSSION

Our purpose was to determine whether collective bargaining coveragenarrows the gender wage gap among manufacturing employees. Our findingsprovide compelling evidence that the estimated pay disparity between menand women is smaller in unionized establishments in six of the nine indus-tries studied, encompassing a large portion of the manufacturing workforce.


That is, blue-collar women generally benefit from working in unionizedplants.

Our findings also corroborate previous research showing the negativeimpact of establishment gender composition on wages and the need to con-sider organizational attributes in studies of wage inequality (Huffman &Velasco, 1997; Tomaskovic-Devey, 1993, 1995). Our study expands this lit-erature by showing the influence of labor unions in reducing gender wage dif-ferentials and theorizing about the direct and indirect effects of collectivebargaining on the gender wage gap.

Several reasons may explain the lack of significance or negative interac-tions found in three industries. These reasons include industry characteristicssuch as the percentage of women in the workforce, the percentage of womenorganized, and union characteristics. The first potential explanation is the dif-ferent proportion of women across manufacturing industries. In fact, two ofthe three industries (miscellaneous plastics and men’s and boys’ shirts) havethe largest percentages of women in our sample. In both men’s and boys’shirts and textile finishing, women’s premium for being unionized is smallerthan men’s, whereas in all other industries, women reap even higher percent-age gains from unionization than men. In short, unionization seems to reducewithin-establishment gender gaps in pay in most industries, but not in themost female-dominated industries. Moreover, the negative interactionbetween the union and female variables may indicate a relative weakness ofpredominantly female unions.

The different findings across industries may also reflect varied levels ofcommitment by different unions to pay equity issues. This commitment hashistorically been less consistent across U.S. unions than among unions inother countries such as the United Kingdom and Canada (Hallock, 1993).Specifically, U.S. unions have placed less emphasis on obtaining equitablewages for all workers, and fewer have set up special women’s divisions(Cook, 1991). The AFL-CIO has endorsed but not necessarily bargained forpay equity, and that at the request of its female members (England, 1992).This attitude among U.S. unions dates back at least one century and seemsrelated to two factors. On one hand, women’s unions united in leagues thattook a political approach, allying themselves more with the women’s move-ment than with the labor movement (Jacoby, 1994). On the other hand, maleunions either paid less attention to gender wage inequality or were con-strained in their demands by managerial conditions. For example, when theissue of wage inequality with women was raised, management often pro-posed to limit men’s wage increases to pay for the equity differential, thusdiscouraging male workers from effectively demanding equity (Acker, 1989;Hallock, 1993). These two forces might have hindered some unions’ ability


to close the gender gap, especially in those industries where women make upthe majority of the workforce.

Another relevant union characteristic that may vary by industry is thenumber of women occupying leadership positions. Women are often absentin executive union jobs (Cook, 1991), and even when they are present, theytend to be leaders at the local level and in relatively marginal positions such aschairs of education committees (Melcher, Eichstedt, Eriksen, & Clawson,1992). Thus, women are still underrepresented in influential positions wherethey could participate in collective bargaining negotiations.

Although our data do not allow us to test the above explanations, the his-torical evolution of unions and gender patterns in the men’s and boys’ shirtsindustry offers a fruitful illustration. The garment trades form a labor-inten-sive and highly competitive industry because of easy entry. Firms are usuallyin low-wage areas and use new technology that reduces the number of high-paying jobs. Low wages, piecework, and occupational segregation havealways characterized this industry with a largely female labor force (Carpen-ter, 1972). During the period covered by our data, the three main unions in thegarment industries were the craft-oriented United Garment Workers, theAmalgamated Clothing Workers of America (ACWA), which organized themen’s clothing branch of the industry, and the International Ladies GarmentWorkers Union, which organized the women’s garment industry. The ACWAfinally merged with the Textile Workers Union of America, forming theAmalgamated Clothing and Textile Clothing Union in 1976. Partly becausewomen were isolated from the predominantly male labor force, these tradeunions remained ambivalent toward women (Acker, 1989). Union goals ofhigher wages, shorter hours, and better working conditions seem to have lostout for years in this industry (Carpenter, 1972).

In sum, industry gender composition and union characteristics help inter-pret our results: The women in mostly female establishments within mostlyfemale industries do not benefit as much from union representation aswomen in more gender-balanced industries. But the effect of unions on thewage gap remains a complex issue.

FUTURE RESEARCH

Future research should explore other reasons for the different effect acrossindustries. For example, detailed comparisons of union structures and bar-gaining power in different industries could be examined (Fiorito &Hendricks, 1987). Similarly, understanding the barriers to full integration ofwomen into union offices might clarify and solve gender differences in unioneffects (Cornfield, 1993). Research could also explore the percentage of


women in a union, the existence of caucus or women’s groups within a union,and the proportion of women in positions of leadership.

Additional research is suggested by our study’s limitations. For one thing,the IWS lack data on some worker characteristics, such as education andseniority, both important variables linked to union and gender wage effects(Cornfield, 1987). However, our study compares men and women working inthe same detailed occupations, where differences in terms of education orhuman capital are likely to be small (Levy & Murnane, 1992).

A second limitation is that our data are two decades old. Yet, we believewe can generalize our findings through the present for two reasons. First, weare not looking at changes over time in the union effect, nor are we studyingthe effects of changing unionization rates on wage inequality. Rather, we esti-mate the effects of unions on women’s wages where they are covered by col-lective bargaining. From this perspective, the role of unions continues to berelevant in reducing wage inequality. Second, recent research using aggre-gate data confirms that the overall union wage premium has been relativelystable between 1980 (approximately when our data were collected) and 1992(Wunnava & Peled, 1999). Moreover, as Card (1998) affirmed, “Despite theshifts in union membership, the structure of union relative wage effects isabout the same in the mid 1990s as in the mid 1970s” (p. 20). Wunnava andPeled (1999) also reported that the union premium across different demo-graphic groups during the same period seems to have converged, reinforcingthe applicability of our argument. The authors suggest that this convergencerelates to the “equalization hypothesis,” that is, unions’ tendency to raisewages more for workers with lower measured skills, thus helping close gen-der and racial wage gaps. If the union premium has remained relatively stabledespite the decline of overall unionization, our prediction that unions helpreduce the wage gap should still hold.

A final limitation lies precisely in our static perspective. Several recentstudies have examined the long-term effects on wage inequality of changingunionization rates among men and women. These studies looked at effectsfor men and women separately, but something similar could be done for thegender gap. For example, Card (1998) found that among women, unionmembership has declined among low-wage workers but risen for high-wageworkers. When looked at separately within sectors and genders, the emergingpattern suggests that unions have strongly deterred increasing wage inequal-ity among public sector workers for both men and women. Future researchshould consider whether this finding applies to the private sector as well.


CONCLUSION

Past research on the earnings gap between men and women has addressedindividual-level variables (e.g., human capital), occupational gender compo-sition, and industry and organizational characteristics. Our work extends pre-vious organizational-level research by examining how unionization affectswithin-employer gender wage differences in nine manufacturing industries.Our most relevant finding is that unionization is associated with smaller gen-der gaps, especially in industries where women do not dominate theworkforce. The implications of this finding are paramount because a largenumber of women still work in low-paying jobs in predominantly femaleindustries (Milkman, 1987).

Our findings have implications for research on inequality as well as forunion research and practice. First, collective bargaining structures must beconsidered in studies of the wage gap that tend to focus mostly on segrega-tion: Unions’ impact on the gender gap remains even after controlling forestablishment gender segregation. Second, important differences exist, evenwithin manufacturing, on how unions affect the wage gap. These differencesare sometimes overlooked when studies compare workers across economicsectors.

From a practical standpoint, if unions lower within-establishment sexgaps in pay, then holding constant how segregated men and women arebetween establishments, more unionization will lower the economy-widegender gap in pay. This highlights the importance of continuing efforts atorganizing sectors where unionization has declined. Service-sector unions,with their higher female membership bases, have actively negotiatedwomen’s issues such as affirmative action, family policies, and pay equity(Hartmann, Spalter-Roth, & Collins, 1994). Success in these negotiationscan affect the extent of later female labor activism (Cornfield, Cavalcanti, &Chun, 1990). What the service sector unions have done to understand andorganize the female workforce could be applied to the blue-collar sector,because women appear to be at least as interested as men in joining unions,and very committed once they have joined (Crain, 1994).

Finally, our findings suggest that although unions help reduce the genderwage gap, they can do more to incorporate equity concerns in their agendas.As Cook (1991) affirmed,

While national unions and officers have given importance to integration ofwomen and minorities, especially in view of the fact that they will be a majority


of members by next century, little has been done to impress on the rank and filethat the problem of integration is a major one on which the vitality of the unionsmay well depend in the future. (p. 253)

With the erosion of affirmative action policies, women must assume leader-ship to ensure fair treatment that may no longer be enforced by governmentagencies (Kirton & Healy, 1999). This is particularly important in the privatesector, where enforcement or attention to pay equity issues is less keen(Creese, 1995).

NOTES

1. The combination of establishment- and individual-level data is difficult to find in currentdata because the BLS discontinued the IWS in 1990.

2. We did not have access to the IWS for a few manufacturing industries: basic iron and steel,men’s and boys’ coats, women’s and misses’ dresses, footwear, and hosiery. Although we do nothave the data, these industries’ products, union coverage, and gender composition are relativelysimilar to those in the industries already included in our study. For example, research on the gar-ment industries suggests that the gender composition and union coverage for the men’s andboys’ shirts industry are similar to those for men’s and boys’ coats and women’s and misses’dresses industries. Similarly, the gender composition of the nonferrous foundries industry in oursample likely resembles that of the iron and steel industry. We did have data on the fabricatedstructural steel industry, but we decided against including it in the study; only 0.6% of this indus-try’s workers are female, making it very difficult to conduct appropriate statistical analyses. Inother words, although we do not have the complete industry set, our study encompasses morethan two thirds of all industries surveyed between 1975 and 1985, before the gender variable waseliminated from the questionnaire, and it includes industries that vary widely in gender composi-tion and union coverage.

3. The coefficient in a semilog regression multiplied by 100 approximates the percentagechange in the dependent variable associated with a one-unit change in the independent variable.To be precise, one must take the exponentiated coefficient minus 1 and multiply it by 100. We usethe simpler approximation to facilitate relating the text to the tables.

REFERENCES

Acker, J. (1989). Doing comparable worth: Gender, class, and pay equity. Philadelphia: TempleUniversity Press.

Anderson, K. A., Doyle, P. M., & Schwenk, A. E. (1990). Measuring union-nonunion earningsdifferences. Monthly Labor Review, 113(6), 26-38.

Baron, J., Mittman, B., & Newman, A. (1991). Targets of opportunity: Organizational and envi-ronmental determinants of gender integration within the California civil service, 1979-1985.American Journal of Sociology, 96(6), 1362-1401.


Baron, J. N., & Newman, A. E. (1990). For what it’s worth: Organizations, occupations, and thevalue of work done by women and nonwhites. American Sociological Review, 55(2), 155-175.

Bielby, W., & Baron, J. (1984). A woman’s place is with other women: Sex segregation withinorganizations. In B. Reskin (Ed.), Sex segregation in the workplace: Trends, explanations,remedies (pp. 27-55). Washington, DC: National Academy Press.

Blau, F. D., & Beller, A. H. (1988). Trends in earnings differentials by gender, 1971-1981. Indus-trial and Labor Relations Review, 41(4), 513-529.

Blau, F. D., & Ehrenberg, R. G. (1997). Gender and family issues in the workplace. New York:Russell Sage.

Blau, F. D., & Kahn, L. (1995). The gender earnings gap: Some international evidence. InR. Freeman & L. Katz (Eds.), Differences and changes in wage structures (pp. 105-143).Chicago: University of Chicago Press.

Bridges, W. P., & Nelson, R. L. (1989). Markets and hierarchies: Organizational and market in-fluences on gender inequality in a state pay system. American Journal of Sociology, 95(3),616-658.

Bridges, W. P., & Villemez, W. J. (1994). The employment relationship: Causes and conse-quences of modern personnel administration. New York: Plenum.

Card, D. (1992). The effect of unions on the distribution of wages: Redistribution or relabelling?(NBER Working Paper No. 4195). Cambridge, MA: National Bureau of Economic Research.

Card, D. (1998). Falling union membership and rising wage inequality: What’s the connection?(NBER Working Paper No. 6520). Cambridge, MA: National Bureau of Economic Research.

Carpenter, J. (1972). Competition and collective bargaining in the needle trades, 1910-1967.Ithaca: New York State School of Industrial and Labor Relations.

Cook, A. H. (1991). Women and minorities. In G. Strauss, D. G. Gallagher, & J. Fiorito (Eds.),The state of the unions (pp. 237-257). Champaign, IL: Industrial Relations Research Associ-ation.

Cornfield, D. B. (1987). Ethnic inequality in layoff chances: The impact of unionization on lay-off procedures. In R. S. Lee (Ed.), Redundancy, layoffs, and plant closures: Their character,causes and consequences (pp. 116-140). London: Croom Helm.

Cornfield, D. B. (1993). Integrating U.S. labor leadership: Union democracy and the ascent ofethnic and racial minorities and women into national union offices. In S. B. Bacharach,R. Seeber, & D. Walsh (Eds.), Research in the sociology of organizations (pp. 51-74). Green-wich, CT: JAI.

Cornfield, D. B., Cavalcanti, H. B., & Chun, B. J. (1990). Household, work, and labor activism:Gender differences in the determinants of union membership participation. Work and Occu-pations, 17, 131-151.

Crain, M. (1994). Gender and union organizing. Industrial and Labor Relations Review, 47(2),227-248.

Creese, G. (1995). Gender equity or masculine privilege? Union strategies and economic re-structuring in a white-collar union. Canadian Journal of Sociology, 20(2), 143-166.

DiNardo, J., Fortin, N. M., & Lemieux, T. (1996). Labor market institutions and the distributionof wages, 1973-1992: A semiparametric approach. Econometrica, 64(5), 1001-1044.

Doiron, D. J., & Riddell, W. C. (1994). The impact of unionization on male-female earnings dif-ferences in Canada. Journal of Human Resources, 29(2), 504-534.

Economic Policy Institute. (1999, June 2). Economic snapshots: Unions and wages [Online].Washington, DC: Author. Available: http://www.epinet.org/webfeatures/snapshots/archive/060299/snapshots060299.html


England, P. (1992). Comparable worth: Theories and evidence. New York: Aldine de Gruyter.Even, W. E., & Macpherson, D. A. (1993). The decline of private-sector unionism and the gender

wage gap. Journal of Human Resources, 28(2), 279-296.Fiorito, J., & Hendricks, P. (1987). The characteristics of national unions. In D. Lewin, D. B. Lipsky, &

D. Sockell (Eds.), Advances in industrial and labor relations (pp. 1-42). Greenwich, CT:JAI.

Freeman, R. B. (1980). Unionism and the dispersion of wages. Industrial and Labor RelationsReview, 34(1), 3-23.

Freeman, R. B., & Leonard, J. S. (1987). Union maids: Unions and the female work force. InC. Brown & J. A. Pechman (Eds.), Gender in the workplace (pp. 189-216). Washington, DC:Brookings Institution.

Freeman, R. B., & Medoff, J. L. (1984). What do unions do? New York: Basic Books.Gander, J. P. (1997). Gender-based faculty-pay difference in academe: A reduced-form ap-

proach. Journal of Labor Research, 18(3), 451-461.Goldin, C. (1986). Monitoring costs and occupational segregation by sex. Journal of Labor Eco-

nomics, 4(1), 1-27.Groshen, E. L. (1991). The structure of the female/male wage differential: Is it who you are, what

you do, or where you work? Journal of Human Resources, 26(3), 457-472.Gunderson, M. (1989). Male-female wage differentials and policy response. Journal of Eco-

nomic Literature, 27(1), 46-72.Halaby, C. N. (1979). Sexual inequality in the workplace: An employer specific analysis of pay

differences. Social Science Research, 8, 79-104.Hallock, M. (1993). Unions and the gender wage gap. In D. S. Cobble (Ed.), Women and unions:

Forging a partnership (pp. 27-42). Ithaca, NY: ILR.Hartmann, H., Spalter-Roth, R., & Collins, N. (1994). What do unions do for women? Chal-

lenge, 37, 11-18.Hodson, R., & England, P. (1986). Industrial structure and sex differences in earnings. Industrial

Relations, 25(1), 16-32.Howell, C., & Mahon, R. (1996). Editorial introduction. Economic and Industrial Democracy,

17(4), 499-510.Huffman, M. L., & Velasco, S. C. (1997). When more is less: Sex composition, organizations,

and earnings in U.S. firms. Work and Occupations, 24, 214-244.Jacoby, R. M. (1994). The British and American women’s trade union leagues, 1890-1925: A

case study of feminism and class. Brooklyn, NY: Carlson.Kirton, G., & Healy, G. (1999). Transforming union women: The role of women trade union offi-

cials in union renewal. Industrial Relations Journal, 30(1), 31-45.Lemieux, T. (1993). Unions and wage inequality in Canada and the United States. In D. Card &

R. B. Freeman (Eds.), Small differences that matter: Labor markets and income mainte-nance in Canada and the United States (pp. 69-107). Chicago: University of Chicago Press.

Levy, F., & Murnane, R. J. (1992). U.S. earnings levels and earnings inequality: A review of re-cent trends and proposed explanations. Journal of Economic Literature, 30(3), 1333-1381.

Lewis, G. H. (1986). Union relative wage effects: A survey. Chicago: University of ChicagoPress.

Main, B.G.M., & Reilly, B. (1992). Women and the union wage gap. The Economic Journal,102(1), 49-66.

Maume, D. J., Jr. (1999). Glass ceilings and glass escalators: Occupational segregation and raceand sex differences in managerial promotion. Work and Occupations, 26, 483-509.


Melcher, D., Eichstedt, J. L., Eriksen, S., & Clawson, D. (1992). Women’s participation in localunion leadership: The Massachusetts experience. Industrial and Labor Relations Review,45(2), 267-280.

Milkman, R. (1987). Gender at work: The dynamics of job segregation by sex during World WarII. Urbana: University of Illinois Press.

Nelson, R. L., & Bridges, W. P. (1999). Legalizing gender inequality: Courts, markets, and un-equal pay for women in America. New York: Cambridge University Press.

Petersen, T., & Morgan, L. (1995). Separate and unequal: Occupation-establishment sex segre-gation and the gender wage gap. American Journal of Sociology, 101(2), 329-365.

Pfeffer, J., & Ross, J. (1980). Union-nonunion effects on wage and status attainment. IndustrialRelations, 19(2), 140-151.

Pfeffer, J., & Ross, J. (1981). Unionization and female wage and status attainment. Industrial Re-lations, 20(2), 179-185.

Reskin, B. (1993). Sex segregation in the workplace. Annual Review of Sociology, 19, 241-270.Reskin, B. (2000). The proximate causes of employment discrimination. Contemporary Sociol-

ogy, 29(2), 319-328.Slichter, S. H., Healy, J. J., & Livernash, E. R. (1960). The impact of collective bargaining on

management. Washington, DC: Brookings Institution.Sorensen, E. (1989). Measuring the pay disparity between typically female occupations and

other jobs. Industrial and Labor Relations Review, 42(2), 625-639.Tomaskovic-Devey, D. (1993). Gender and racial inequality at work: The sources and conse-

quences of job segregation. Ithaca, NY: ILR.Tomaskovic-Devey, D. (1995). Sex composition and gendered earnings inequality: A compari-

son of job and occupational models. In J. Jacobs (Ed.), Gender inequality at work (pp. 23-56).Thousand Oaks, CA: Sage.

Tomaskovic-Devey, D., & Skaggs, S. (1999). An establishment-level test of the statistical dis-crimination hypothesis. Work and Occupations, 26, 422-445.

U.S. Department of Labor. (1976a). Industry wage survey: Miscellaneous plastics, September1974 (Bureau of Labor Statistics Bulletin 1914). Washington, DC: Government PrintingOffice.

U.S. Department of Labor. (1976b). Industry wage survey: Wood household furniture, June 1974(Bureau of Labor Statistics Bulletin 1930). Washington, DC: Government Printing Office.

U.S. Department of Labor. (1977a). Dictionary of occupational titles (4th ed.). Washington, DC:Government Printing Office.

U.S. Department of Labor. (1977b). Industry wage survey: Nonferrous foundries, May 1975(Bureau of Labor Statistics Bulletin 1952). Washington, DC: Government Printing Office.

U.S. Department of Labor. (1977c). Industry wage survey: Textile dyeing and finishing, June1976 (Bureau of Labor Statistics Bulletin 1967). Washington, DC: Government PrintingOffice.

U.S. Department of Labor. (1977d). Industry wage survey: Paint and varnishes, November 1976(Bureau of Labor Statistics Bulletin 1973). Washington, DC: Government Printing Office.

U.S. Department of Labor. (1977e). Industry wage survey: Industrial chemicals, June 1976 (Bu-reau of Labor Statistics Bulletin 1978). Washington, DC: Government Printing Office.

U.S. Department of Labor (1977f). Industry wage survey: Wool textiles, May 1975 (Bureau ofLabor Statistics Bulletin 1945). Washington, DC: Government Printing Office.

U.S. Department of Labor (1982). Industry wage survey: Men’s and boys’ shirts and nightwear,June 1981 (Bureau of Labor Statistics Bulletin 2131). Washington, DC: Government Print-ing Office.


Wunnava, P. V., & Peled, N. O. (1999). Union wage premiums by gender and race: Evidencefrom PSID 1980-1992. Journal of Labor Research, 20(3), 415-422.


Documents

How Does Collective Bargaining Affect the Gender Pay Gap?