Sie sind auf Seite 1von 7

American Economic Association

The Gender Earnings Gap: Learning from International Comparisons Author(s): Francine D. Blau and Lawrence M. Kahn Source: The American Economic Review, Vol. 82, No. 2, Papers and Proceedings of the Hundred and Fourth Annual Meeting of the American Economic Association (May, 1992), pp. 533-538 Published by: American Economic Association Stable URL: http://www.jstor.org/stable/2117457 . Accessed: 11/11/2013 01:55
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

American Economic Association is collaborating with JSTOR to digitize, preserve and extend access to The American Economic Review.

http://www.jstor.org

This content downloaded from 202.57.58.94 on Mon, 11 Nov 2013 01:55:15 AM All use subject to JSTOR Terms and Conditions

The Gender Earnings Gap: Learningfrom International Comparisons


By
FRANCINE D. BLAU AND LAWRENCE M. KAHN*

Despite some dramatic reductions in the male-female pay gap since the 1950's, gender differentials persist in all industrialized nations. However, the size of these differentials varies considerably. This paper uses micro data to analyze international differences in the gender pay gap among a sample of eight industrialized nations (see Table 1 for included countries). A major theme in our analysis is that intercountry differences in the gender gap are affected by two processes. First are gender-specific factors including differences in the qualifications of men and women and differences in the treatment of equally qualified women (i.e., labor-market discrimination). Second is wage inequality, the prices the labor market of each country places on various labormarket skills, both observed and unobserved. For example, suppose that in two countries women have lower levels of skills than men but that the differences in skills (somehow measured) is the same in the two countries. If the return to skill is higher in one country, then that nation will have a larger gender pay gap. Skill prices can be affected by relative supplies, by technology (e.g., high-tech industries place a premium on highly trained workers) or by the wage-setting institutions of each country. Specifically, centralized wage-setting institutions which tend to reduce interfirm and interindustry wage variation and are often associated with conscious

* Blau: Institute of Labor and Industrial Relations, University of Illinois, Champaign, IL 61820, and NBER; Kahn: Institute of Labor and Industrial Relations, University of Illinois. We thank participants at the NBER Conference on "Differences and Changes in Wage Structures" for helpful comments. We are grateful to David Blanchflower for making the International Social Survey Programme (ISSP) data available to us. This research was supported by a grant from the Ford Foundation to the NBER.

policies to raise the relative pay of low-wage workers (regardless of gender) may indirectly reduce the gender pay gap. U.S. paysetting is far less centralized than that in the other countries in this study, with the possible exception of Switzerland, a factor contributing to gender pay differences in the United States.1 We adapt a framework used by Chinhui Juhn et al. (1989) in their analysis of black-white wage trends in the United States to estimate the contribution of gender-specific factors versus wage structure in explaining international differences in the gender gap. The striking finding of our study is that the higher level of wage inequality in the United States works to increase the gender differential in the United States relative to all the other countries in our sample and fully accounts for the lower gender earnings ratio in the United States compared to the Scandinavian countries and Australia (the countries with the smallest gaps). This approach helps us understand the seemingly paradoxical position of U.S. women compared to women elsewhere. U.S. women compare favorably with women in other countries on several measures of skills relative to men. Moreover, the United States has had a longer commitment to policies of equal pay and equal employment opportunity for women than have the other countries in our sample. Yet the United States does not rank among the countries with the smallest pay gaps; indeed the gender pay differential in the United States is above average for our sample of eight countries. An important part of the explanation of this

'See our longer paper (Blau and Kahn, 1991) for a description of differences across these countries in wage-setting institutions, gender-specific policies, and gender differences in qualifications. 533

This content downloaded from 202.57.58.94 on Mon, 11 Nov 2013 01:55:15 AM All use subject to JSTOR Terms and Conditions

534

ALA PAPERS AND PROCEEDINGS


TABLE 1-FEMALE/MALE

MAY 1992
EARNINGS RATIOS

pattern is that the U.S. labor market places a much larger penalty on those with lower levels of labor-market skills (both measured and unmeasured).
I. Earnings Ratios in the Micro Data

CORRECTED FOR HOURS

Country (earnings measure) Germany (monthly) United Kingdom (annual) United States (annual) Austria (monthly) Switzerland (monthly) Sweden (annual) Norway (annual) Australia (annual)

All workers 0.6880 0.6337 0.6849 0.7256 0.6174 0.7673 0.7308 0.7489

Married workers 0.5730 0.5966 0.5944 0.6558 0.5768 0.7242 0.7160 0.6909

Single workers 1.0270 0.9489 0.9552 0.9703 0.9449 0.9350 0.9158 0.9144

The following countries and time periods were used in this study: Austria [1985-1987], West Germany [1985-1988], the United Kingdom [1985-1988], the United States [1985-1988], and Switzerland [1987] (International Social Survey Programme [ISSP] data); Sweden [1980] and Norway [1982] (Class Structure and Class Consciousness [CSCC] data); and Australia [1986] (Income Distribution Survey [IDS] data).2 The sample was restricted to workers aged 18-65. To maximize sample size, we pooled years of data for those countries in the ISSP surveyed more than once. Table 1 gives estimated female/male earnings ratios by marital status for log earnings corrected for hours. To obtain these estimates, the following regression was run separately by sex for each country:3 (1) ln(EARN)
=

Notes: The earnings ratios were evaluated at 40 hours. The earnings ratios for married workers are for married workers with one person other than the spouse in the household (for Sweden, Norway, and Australia, one child); those for single workers are for nonmarried people with no other persons in the household.

bo + bjPART+ b2HRPART + b3HRFULL+ b4MARSP + b5OTHERS + e

where ln(EARN) is the natural log of earnings; PART is a dummy variable for parttime employment; HRPART and HRFULL are interactions of weekly work hours with part- and full-time status; MARSP is a dummy variable for married, spouse pres-

2The ISSP data are described in David Blanchflower and Richard Freeman (1992). See Rachel Rosenfeld and Arne Kalleberg (1990) for a description of the CSCC data, originally compiled by Erik Wright; and see McKinley Blackburn and David Bloom (1991) for a description of the IDS data. 3For countries with more than one year of data, the log-earnings variable was obtained by transforming each observation into its 1988 (or end year) equivalent on the basis of regressions including only gender and year dummy variables.

ent; OTHERS is number of people other than the spouse in the household (ISSP countries), or number of children in the household (Scandinavia and Australia); and e is an error term. The coefficients from (1) were used to compute a gender ratio for each country, assuming a 40-hour work week. The first column in Table 1 indicates the hours-corrected gender earnings ratio for all workers. These estimates are obtained by evaluating the family-composition variables at their mean values for each sex in each country. The ratio is highest in the Scandinavian countries, Australia, and Austria, ranging from 73 to 78 percent, and lowest for Switzerland and the United Kingdom, ranging from 62 to 63 percent.4 The

4While our estimates for Australia and the Scandinavian countries are below those in published sources (ILO, various issues), they are consistent with other estimates from micro data (Blau and Kahn, 1991). Published ratios for these countries are most likely

This content downloaded from 202.57.58.94 on Mon, 11 Nov 2013 01:55:15 AM All use subject to JSTOR Terms and Conditions

VOL. 82 NO. 2

GENDER AND LABOR-MARKETOUTCOMES

535

United States and Germany comprise an intermediatecategory. In the next two columns,we use equation (1) to estimate the earningsratio separately for married workers with one nonspouse "other"(or child) in the household5and for single (i.e., nonmarried) people with no others (children) in the household. The pay ratio is uniformlyvery high among single workers,rangingfrom 0.91 to 1.03. In contrast,the pay gap is muchlargerfor married workers, with the rankingsfor this group fairly similar to the overall rankings.(The most notable differences are that Austria falls into the intermediategroup and Germany falls into the lowest group.) These findings suggest that intercountry differences in the earnings ratios of married workers drive the internationalpattern of genderpay gaps and that in orderto explain this pattern, we need to concentrate on married workers.6
II. Analysisof the Intercountry GenderGap

where Yij is the log of wages;Xi is a vector of explanatoryvariables;Bi is a vector of coefficients; Oij is a standardizedresidual (i.e., with mean zero and variance1 for each country);and oj is the country's residual standarddeviation of wages (i.e., its unexplained level of male wage inequality). Then, the male-female wage gap for country j is
(3) Dj-Ymj -Yfj = 8 Xj Bj +0 y01

where the m and f subscriptsrefer to male and female averages,respectively;and a 8 prefix signifies the average male-female differencefor the variableimmediatelyfollowing.Equation(3) states that the country's pay gap can be decomposedinto differences in measured qualifications(5X1), and differences in the standardizedresidual (60j) multipliedby the log money value per unit differencein the standardizedresidual(o). The pay-gap difference between two countriesj and k can then be decomposed as follows:
(4)
Dj - Dk
-

Juhn et al. (1989) have devised a method that allowsus to decomposethe cross-country differences in the gender gap into a portiondue to gender-specific factors and a portion due to differences in the overall level of wage inequality.Followingtheir notation,supposethatwe havefor maleworker i and countryj a male wage equation:
(2) Yij= XijBj + o-jOij

(Xj

Xk)Bk + 5Xj(Bj -Bk)


-

+ (89 - 50k )o'k + 50j(o,i

Ck )

overstated in that the Scandinavian data are limited to manufacturing workers, and the Australian data exclude supervisory personnel. 5The average number of nonspouse others in the household is roughly 1. 6The very small gaps for single workers may be due in part to their being disproportionately young (the pay gap -is lower for young workers); in addition, single males are less productive than married males (see Sanders Korenman and David Neumark, 1991), whereas the opposite is likely to be the case for women. In our longer paper (Blau and Kahn, 1991), we show that differences in the composition of the labor force with respect to marital status do not explain differences in the pay gap across countries, again suggesting that we must concentrate on differences among married workers.

The first term in (4) reflects the contribution of intercountry differencesin observed labor-market qualifications(X) to the gender gap. The second term reflects the impact of different measured prices across countries for observed labor-marketqualifications. The third term measures the efdifferencesin the relafect of cross-country tive residual wage positions of men and women (i.e., whetherwomen rankhigheror lower within the male residualwage distribution). Such differences in rankingsmay reflect gender differences in unmeasured or the impactof labor-market characteristics discrimination againstwomen. We label this term the "gap" effect. Finally, the fourth term of (4) reflects intercountry differences in residualinequality.Suppose that unmeasured deficits in female relative skills or

This content downloaded from 202.57.58.94 on Mon, 11 Nov 2013 01:55:15 AM All use subject to JSTOR Terms and Conditions

536

ALA PAPERS AND PROCEEDINGS

MA4Y 1992

discrimination lower women's position in the male distribution of wage residuals as is likely. The larger the penalty a country places on being below average in wages, the larger will be its pay gap. In the empirical work below, we label this the effect of "unobserved prices." Following Juhn et al. (1989), the third term, (80Q 0k)dok, can be obtained empirically by assigning each woman a percentile number corresponding to her position in her country's distribution of male wage residuals. We can then find what the country-j mean female wage residual would have been given the percentile rankings in country j and the distribution of male wage residuals in country k (note that the mean male residual is always zero). The difference between this imputed wage residual for country j and the actual female mean wage residual for country k is used to find the estimate of (80 - 0k)o-k. That is, we find the contribution to the cross-country difference in the gender gap that would result if two countries had the same levels of residual male wage inequality and differed only in their percentile rankings of the female wage residuals. The fourth term of (4),
0j(oj - ok),

may be obtained analogously

and measures the contribution to the crosscountry difference in the gender gap that would result if two countries had the same percentile rankings of the female wage residuals and differed only in the extent of male residual wage inequality. According to (4), the full impact of gender-specific factors is reflected in the first and third terms, the effect of gender differences in qualifications and of gender differences in wage rankings at a given level of measured characteristics. Labor-market structure is reflected in the second and fourth terms, the impact of intercountry differences in returns to measured and unmeasured characteristics. In a traditional decomposition, the sum of the third and fourth terms represents the impact of intercountry differences in the "unexplained" differential, which is commonly taken as an estimate of discrimination. The possibility of discrimination complicates the interpretation of the last term of

(4). With labor-market discrimination, the last term in part reflects the interaction between country j's level of discrimination (pushing women down the distribution of wages) and intercountry differences in the overall level of inequality which determine how large the penalty is for that lower position in the distribution (Juhn et al., 1989). The observed price effect may also reflect discrimination if, for example, women "crowd" into certain sectors lowering relative earnings there even for men (Barbara Bergmann, 1974). In light of the evidence presented above that intercountry differences in the pay gap are driven by differentials for married workers, we implement this decomposition using as the dependent variable YMCH, hourscorrected earnings simulated for married people with one nonspouse other (or child) in the household using equation (1).7 An alternative to using wages for married workers (with one other or child) as the dependent variable in equation (2) would be to control for marital status and number of children (others) on the right-hand side. However, such an approach is problematic in that these variables may measure higher skills for men and lower skills for women. The decomposition of YMCH is shown in Table 2. The underlying regressions include the traditional human capital variables (education, potential experience, and its square) and a vector of dummy variables for union status, one-digit industry, and onedigit occupation.8 The structural variables may reflect both worker skills and rents received by workers with these characteristics. Unfortunately, the data sets available
7For each worker i, we have
YMCHi = ln(EARN)i - bjPART1 - b2HRPART1

-b3(HRFULLi
-

-40)-

b4(MARSPi -1)

b5(OTHERSi - 1) .

8For Australia, union status is unavailable, and the occupation and industry codes differ from those in the ISSP or CSCC data. For U.S-Australia comparisons, we estimated a U.S. equation that had the same variables as Australia's.

This content downloaded from 202.57.58.94 on Mon, 11 Nov 2013 01:55:15 AM All use subject to JSTOR Terms and Conditions

VOL.82 NO. 2
TABLE 2-ANALYSIS

GENDER AND LABOR-MARKET OUTCOMES


OF LOG WAGES (YMCH)
Mean percentile of female residuals in male distribution 15.78 20.98 30.44 21.61 25.37 30.91 27.88 31.45

537

Country Germany United Kingdom United States Austria Switzerland Sweden Norway Australia

Gender differential 0.5569 0.5165 0.5202 0.4220 0.5502 0.3227 0.3341 0.3698

Di-DUSA 0.0367 -0.0037


-

-0.0982 0.0300 - 0.1975 -0.1861 -0.1504

that these differencesin rankingsraise the differential relative to the United States, often substantially (the unweightedaverage 10 The column headed effect is 0.2374). "Unobservedprices" shows that the lower level of residualwage inequalityin each of the other countries has a negative effect, often quite considerable,on its gap relative to that in the United States (the unweighted
average effect is - 0.2791).

Decomposition Country Germany United Kingdom Austria Switzerland Sweden Norway Australia Observed X's 0.0240 0.0170 0.0633 0.0889 -0.0218 0.0229 0.0399 Observed prices -0.1038 -0.0302 -0.1475 - 0.0397 -0.0394 -0.0948 -0.0561

of D,-DUSA Gap 0.5739 0.3465 0.3412 0.2384 0.0145 0.1865 - 0.0389 Unobserved prices -0.4574 - 0.3370 -0.3552 -0.2576 -0.1508 -0.3007 - 0.0953

Notes: Log earnings were estimated at 40 hours per week for married individuals with one nonspouse other (or child). The regression includes controls for education, potential experience and its square, union status, and occupation and industry dummy variables.

Table 2 also provides estimates of the impact of measured skills and their prices on intercountry differencesin the pay gap. The "observedX's" effect is positive with the exception of Sweden, indicating that U.S. women have relativelyfavorablelevels of the measured characteristics(the unweightedaverageeffect is 0.0335).The "observed prices" effect is negative in every case, indicating that the male returns to measured characteristicsdecrease the pay gap in other countriesrelativeto the United States (the unweighted average effect is
-

0.0731).

to us lack information on actual labormarket experienceor weeks worked,which remainimportantomittedvariablesin these analyses.9
The column headed Di - DUSA shows the

difference between each country's gender differentialand that of the United States. The U.S. gender gap is higher than in the Scandinavian countries,Australia,and Austria, about the same as in the United Kingdom, and lower than in Germany and Switzerland.The mean percentile of the female residual in the male distribution rangesfrom 16 percentin Germany to 30-31 percent in Australia, Sweden, and the United States. It is noteworthy that U.S. women place near the top of the list. The column headed "Gap" shows the of each country'sfemale placecontribution ment in the male residualwage distribution to its relativepay gap. The figureis positive for all countriesexcept Australia,indicating
9To some degree,our controlsfor education,hours, industry,and occupationmay pick up the effects of such omissions.

The results in Table 2 suggest that U.S. women fare well with respect to genderspecific factors (as measuredby the contribution of the observedX's and of gap) but are adverselyaffected by the high level of wage inequality in the United States (as measured by the contributionof observed and unobserved prices)." Indeed,the higher level of wage inequalityin the United States fully accountsfor the lower gender ratio in the United States in comparison to the Scandinavian countriesand Australia.However, as noted earlier,what we have labeled wage inequality could also reflect the impact of the interactionof differencesin wage structureand the country-i(non-U.S.) level
of discrimination.

t0Note that although the percentile ranking of women in Sweden is slightly higher than in the United States, the gap effect is small and positive. This is because the gap effect takes into account not simply the rankings, but also the associated wage rewards (penalties). 1"Further indication of U.S. women's high relative labor-market qualifications compared to women in other countries is the far greater incidence of full-time work among U.S. employed women (relative to men) than for those elsewhere (Blau and Kahn, 1991).

This content downloaded from 202.57.58.94 on Mon, 11 Nov 2013 01:55:15 AM All use subject to JSTOR Terms and Conditions

538

ALA PAPERS AND PROCEEDINGS

MAY 1992

What are we then to conclude about the role of labor-market structure? From a number of indirect indicatorswe conclude that it is important,even though it may not be possible to estimate its effect precisely. First,U.S. wage-settinginstitutionsare considerably less centralized than in other countries,thus makinga findingof the importance of wage structureplausible. Second, the U.S. commitment to policies of equal pay and equal employmentopportunity comparespositivelyto the other countries in our sample, although such policies may well be more effectively implemented under more centralizedwage-determination systems.12Further, U.S. women compare favorablyto women in the other countries in terms of their qualificationsrelative to men. Thus, it is crediblethat gender-specific factors do not explain the relatively high pay gap in the United States. Third, we found that the wage and residualwage variation of both men and women in the United exceed those of the same Statesconsiderably gender group in other countries(resultsnot shown),suggestingthat the same set of factors (measuredand unmeasuredprices and wage-settinginstitutions)affect the wages of both men and women in each countryin a similarway. Finally, and perhaps most importantly,even though the estimated wage inequalityeffect may include the impact of gender discriminationas it interacts with wage structure, our findings nonetheless role for wage suggestan extremelyimportant inequalityin affectingthe genderratio.
III. Conclusion

that to understandchanges in the gender pay gap fully, it would also be fruitful to examine the impact of changes in wage structure.Given growingwage inequalityin the United States in the 1970's and 1980's (Juhn et al., 1990), American women may have been swimmingupstream in a labor to low-wage unfavorable marketincreasingly workers.
REFERENCES Bergmann,BarbaraR., "Occupational Segregation, Wages and Profits When Employers Discriminate by Race or Sex," Eastern Economic Journal, April-July 1974, 1, 103-10. Blackburn,McKinleyL. and Bloom, David E., "Changes in the Structure of Family Income Inequality in the United States and Other Industrialized Nations During the 1980s," Working Paper, Columbia University, June 1991. Blanchflower, David and Freeman, Richard, "Unionism in the United States and Other Advanced OECD Countries," Industrial Relations, Winter 1992, 31, 56-79. M., "The Blau,FrancineD. and Kahn,Lawrence Gender Earnings Gap: Some International Evidence," unpublished manuscript presented at the NBER Conference on "Differences and Changes in Wages Structures," London, 1991. Juhn, Chinhui, Murphy,Kevin M. and Pierce, Brooks,"Accounting for the Slowdown in Black-White Wage Convergence," unpublished manuscript, University of Chicago, October 1989. _ , "Wage Inand equality and the Rise in Returns to Skill," unpublished manuscript, University of Chicago, January 1990. Korenman, Sanders and Neumark, David, "Does Marriage Really Make Men More Productive?" Journal of Human Resources, Spring 1991, 26, 282-307. Rosenfeld, Rachel and Kalleberg, Arne, "A Cross-National Comparison of the Gender Gap in Income," American Journal of Sociology, July 1990, 96, 69-106. ILO, Yearbook of Labour Statistics, Geneva: International Labor Organization, various issues.

The major finding of this paper is the importance of wage structure, specifically the higher level of wage inequality in the United States relative to other industrialized countries, in explaining international differences in the gender pay gap. Much attention has focused on women's growing relativelevels of skills and labor-forcecommitmentas causes of changesin the pay gap in the United States. Our researchsuggests
12In addition, the comparable-worth approach pursued in Australia might be expected to produce a larger immediate impact on wages.

This content downloaded from 202.57.58.94 on Mon, 11 Nov 2013 01:55:15 AM All use subject to JSTOR Terms and Conditions

Das könnte Ihnen auch gefallen