Meta‐Analysis of the Effects of Sexual Orientation on Earnings

• Date: 01 January 2015
• Published date: 01 January 2015
• Author: Marieka Klawitter
• DOI: http://doi.org/10.1111/irel.12075

Meta-Analysis of the Effects of Sexual

Orientation on Earnings

*

MARIEKA KLAWITTER

Most studies show that gay men earn less and lesbians earn more than their

heterosexual counterparts, but the size of estimated sexual orientation differences

varies greatly across studies. Using studies published between 1995 and 2012, a

meta-regression shows that the gay sample size, sexual orientation measure, and

controls for work intensity explain variation in estimates for men. For women,

there are few conclusive influences although controlling for work intensity seems

to be most important.

Introduction

In 1995, Badgett published the first econometric study of the effects of sex-

ual orientation on earnings. The work followed a body of literature estimating

similar models for differences by race and gender using traditional regression

techniques (Altonji and Blank 1999). Badgett’s study was made possible by

the emerging confluence of data on sexual orientation and earnings, a

researcher willing to work on a stigmatized subpopulation, and an economics

journal willing to publish the results—all elements that were rare or nonexis-

tent prior to that time (Klawitter 1998).

Since Badgett’sfirst article appeared, researchers have used a growing num-

ber of datasets and models to explore sexual orientation earnings differences.

The body of work that has accumulated addresses the key scientific questions

with important policy implications: Do earnings show evidence of labor

market discrimination on the basis of sexual orientation?, What other factors

interact with sexual orientation in determining earnings?, and What are the

contributions of human capital and intrahousehold influences? However, the

estimated sexual orientation differences and the answers to these questions are

less than uniform across studies, which suggests the necessity of a systematic

*The author’saffiliation is Evans School of Public Affairs, University of Washington, Seattle, Washington.

Email: marieka@uw.edu.

The author appreciates Samatha Bannon and Danielle Fumia for their able and cheerful assistance with

data collection and the helpful advice from anonymous referees, seminar participants at University of

Washington, and participants in the University of Paris workshop on sexual orientation discrimination.

INDUSTRIAL RELATIONS, Vol. 54, No. 1 (January 2015). ©2014 Regents of the University of California

Published by Wiley Periodicals, Inc., 350 Main Street, Malden, MA 02148, USA, and 9600 Garsington

Road, Oxford, OX4 2DQ, UK.

4

assessment of the literature. Differences in the measures of earnings and sexual

orientation, choice of dataset, sample limitations, control variables, and model-

ing choices might all contribute to the variation in estimates and to conclu-

sions about how sexual orientation affects earnings. Although the number of

published studies is still small at thirty-one, a meta-analysis will provide a

much-needed summary of the patterns in the findings and guidance for

researchers and those interested in collecting new data.

Sexual Orientation and Earnings

Recent studies have found that, before controlling for any explanatory fac-

tors, lesbians earn significantly more than do heterosexual women and gay

men earn less than heterosexual men. For example, Cushing-Daniels and

Yeung (2009) show unadjusted earnings differences equal to a 26 percent pre-

mium for lesbians over heterosexual women and a –7 percent penalty for gay

men compared to heterosexual men using General Social Survey (GSS) data.

1

The gender gaps are also large, with lesbians and heterosexual women earning

13 and 38 percent less than heterosexual men, respectively. The gender differ-

ences in earnings explain the household income patterns: Lesbian couples have

household incomes lower than those of married heterosexual couples, while

gay male couples have incomes similar despite having individual earnings

lower than those of married men (Klawitter 2011: 353).

Following earlier studies of labor market differences on the basis of gender

or race, researchers have used multivariate analysis to parcel out variation in

earnings levels to that related to expected productivity (through education and

experience, occupation, industry, geographic location), tastes (other observed

characteristics), and the unexplained component that could be attributable to

discrimination. Beginning with Badgett (1995), almost all studies have found

that gay men earn less than heterosexual men after controlling for other char-

acteristics (e.g., Carpenter 2007; Elmslie and Tebaldi 2007; Klawitter 2011;

Martell 2012). And most, but not all, studies find that lesbians earn more than

their heterosexual counterparts (e.g., Antecol, Jong, and Steinberger 2008;

Black et al. 2003; Jepsen 2007). However, the size of estimates of sexual

orientation differences for both men and women range widely. For example,

1

Author calculations using unadjusted means from Cushing-Daniels and Yeung (2009). Similarly,

Antecol, Jong, and Steinberger (2008), using the 2000 U.S. Census, reported unadjusted hourly earnings

lower by 4.6 percent for men and 18 percent higher for women in same-sex couples compared to men and

women in married different-sex couples. Both gay men and lesbians had earnings about 25 percent higher

than those in unmarried different-sex couples.

Meta-analysis of Sexual Orientation effects /5