Assessing the Role of Household Specialization and Time Usage Decisions:
Sexual Orientation and Earnings in the Labour Market
Greg Henry 10824898 Dr. Erik Plug greg.henry@student.uva.nl MSc Development Economics
Abstract
Using time usage data provided by the United States Bureau of Labor Statistics, this paper explores the impact that household specialization and time usage decisions have in explaining the apparent earnings differential found in past literature between individuals of different sexual orientation and gender. First, this paper estimates time usage variables which help illustrate differences in household specialization between heterosexual and gay and lesbian couples. Heterosexual females are found to allocate more time to household production (85 minutes per day) and less to work (73 minutes per day) when compared to heterosexual males. Contrarily, evidence suggests that lesbian females allocate more time towards work than heterosexual females (50 minutes per day). Second, this paper investigates the role that time usage decisions by heterosexual and homosexual individuals have in explaining the earnings differential found empirically in past literature. Utilizing an ordinary least squares empirical strategy, time usage decisions by both heterosexual and homosexual individuals involving work, leisure, and home production are found to have no impact on their earnings, a statistically significant result for two of the time variables in the most comprehensive specification.
1. Introduction
What causes the earnings differential found between individuals of different sexual orientation and gender? Although research regarding this subject has expanded rapidly recently, a complete understanding of this disparity and its causes have still yet to be fully explored. Consensus in past literature regarding the direction and magnitude of this earnings differential between individuals suggests that gay males are at an earnings disadvantage versus heterosexual males. Contrarily, lesbian females are at an earnings advantage over heterosexual females.
A full understanding of why this differential exists remains the relevant question. A variety of reasons can influence an individual’s value in the labor market. Specifically, educational attainment, occupation type, frequency of work, location of work, among many other factors, can influence the amount of wage received for a particular job. Building upon previous literature, this paper will investigate the role that time usage decisions and household specialization have in explaining the earnings differential found between individuals of different sexual orientation.
First, an investigation into how individuals utilize their time on a daily basis will be conducted. This paper will draw from past literature in its construction of time variables to sort
daily activities into categories representing work, leisure, and home production (Connolly, 2008). Following that, this paper will attempt to assess how much role, if any, those time usage decisions have in causing the disparity in earnings found between individuals of different sexual orientation and gender.
Utilizing an ordinary least squares empirical strategy, this paper finds that time usage decisions have no impact on the earnings differential found between individuals of different sexual orientation. Specifically, this paper finds statistically significant results that homosexual males are at an earnings disadvantage of approximately 18 percent relative to heterosexual males. Also confirming previous literature, lesbian females are found to be at an earnings advantage of approximately 20 percent over heterosexual females. In the most preferred OLS specification, time usage decisions are found to have no impact on these earnings gaps, a statistically significant result for the total work time and leisure variables.
This paper contributes to the growing literature on the earnings differential found between individuals of different sexual orientation by exploring a potential channel of causation. Time usage decisions and household specialization do not have any role in explaining the gap in earnings found between individuals of different sexual orientation. Despite preliminary research into potential causes of this differential, further research is needed to develop a comprehensive understanding of the entire dynamic of earnings in the labor market for individuals of different sexual orientation.
The rest of this paper will be organized as follows: In section 2, a background of previous literature will be given. Section 3 provides basic theoretical models of human capital and time usage which assist in guiding predictions to be investigated later on. Section 4 will give an explanation of the data used throughout this paper. In section 5, the empirical strategy used for analysis will be explained. Sections 6 and 7 follow with the results, first with the results related to household specialization and time usage, then with the results related to earnings. Finally, in section 8, this paper will conclude with the main findings of relevance.
2. Literature Review
This section will provide an overview of past literature by investigating two separate branches of focus. First, a review and summary of past literature which deals with time usage decisions and household specialization by individuals of different sexual orientation will be conducted. Second, this section will investigate past literature which focuses on potential links
between sexual orientation and earnings. Specifically, literature will be grouped by possible factors which may be the cause of differences in earnings between individuals of different sexual orientation.
i. Time Usage and Sexual Orientation
The topic of household specialization between individuals of different sexual orientation has been the subject of significant focus. Past literature has focused on the hypothesis that heterosexual partners are more likely to specialize within a household than their gay and lesbian counterparts (Becker, 1991; Leppel, 2009). Heterosexual couples are thought more likely to specialize for a variety of reasons including, among others, the more frequent presence of children, traditional gender roles in labour and household production, and differences in earnings. Contrarily, gay and lesbian couples are believed to divide labour and household production more evenly, with less overall specialization evident in the household (Kurdek, 2007). Empirically, past literature shows that these hypotheses largely hold true.
With respect to labour and household production of heterosexual couples, intuitively, it is mainly earnings in the labour market which drives the decision on how to allocate time resources of a couple between home and labour. Literature indicates that it is predominantly heterosexual females who will specialize toward home production when only one partner works (Black, Sanders, and Taylor, 2007). This is the typical traditional gender division where the breadwinner male works in order to provide for the family financially. Even when both partners work, heterosexual males are found to work approximately 10 more hours per week than their female partner (Black, Sanders, and Taylor, 2007). Gay and lesbian couples, however, are not seen to behave in this manner. Highlighting the lack of specialization and overall increase in labour supply among gay and lesbian couples, the empirical results suggest that the chance of both partners working full time are fifteen percent higher in same sex couples (Giddings et Al, 2010).
The presence of children is shown to have a significant impact on specialization within a household (Dalmia and Sicilian, 2008). Intuitively, children require time, money, and focus which dictate alterations in allocation of time by couples. Heterosexual partners are found to have children much more frequently than lesbian and gay partners (Black, Sanders, and Taylor, 2007). The impact of children is particularly strong in the labour market. Lesbian and gay partners, whom less frequently have children, are often able to allocate more of their time to work while heterosexual
couples frequently adopt the aforementioned traditional gender roles with the male working and the female focusing on home production. Interestingly, when lesbian and gay partners do have children, they are seen to behave in a strikingly similar manner to heterosexual couples (Giddings et Al, 2010). The result is a shift towards specialization in gay and lesbian households.
Time usage and household specialization among couples of different sexual orientation is principally shown to be driven by a few factors. Overall, gay and lesbian couples are found to specialize less, with more of their time allocated towards labour supply. When children are present within the household, heterosexual and same-sex couples are found to behave similarly. Ultimately, in an intuitive economic finding, earnings in the labour market are a strong indicator of where time will be allocated by partners who specialize in terms of labour and household production.
ii. Sexual Orientation and Earnings
Past empirical evidence shows that by and large, heterosexual males earn more than gay males (Ahmed and Hammarstedt, 2009; Ahmed, Andersson, and Hammarstedt, 2012; Badgett, 1995; Heineck 2009; Plug and Berkhout, 2004) while lesbian females earn more than heterosexual females (Ahmed, Andersson, and Hammarstedt, 2012; Baumle, 2009; Plug and Berkhout, 2004). The cause or factors behind this apparent earnings differential is the subject of significant research. In this section, focus will be paid to three potential causes for the earnings differential between individuals of different sexual orientation commonly brought up in past literature: discrimination, differences in human capital, and differences in preferences.
Discrimination
For the purposes of this paper, discrimination refers to employers treating gay and lesbian individuals differently than their equally qualified heterosexual counterparts. Labour discrimination towards gay and lesbian individuals can take form in two ways, both of which were introduced in past literature. Employers treat gay and lesbian individuals negatively solely on the basis of their sexual orientation (Becker 1971) or employers use known qualities of gay and lesbian individuals to speculate about their unobserved qualities (Arrow 1973; Phelps 1972). Given the aforementioned wage dynamics associated with gay and lesbian individuals, it is plausible that both branches of thought assist in developing our understanding of how discrimination may impact the earnings gap found between individuals of different sexual orientation. Gay male workers may be
penalized and paid less based on their sexual orientation while at the same time, lesbian females may be viewed favourably by employers based on a potential perception that they are more career focused (Buser, Geijtenbeek, and Plug, 2015).
Empirically, the issue of cleanly isolating the impact of discrimination on the wages of lesbian and gay workers has been a complicated matter. Researchers have thus been forced to develop several diverse strategies attempting to suss out the impact that discrimination has on the labour market outcomes of gay and lesbian individuals. Using identical fictitious resumes as a means to measure labour market discrimination, gay males in seven states in the United States received less calls than identical resumes of heterosexual males (Tilcsik 2011). Using the same strategy, similar discrimination can be found in Austria, Germany, Greece, and Sweden (Ahmed, Andersson, and Hammarstedt, 2013; Drydakis, 2009; Weichselbauer, 2003, 2015). The extent to which labour market discrimination of this manner impacts wages, through a lack of work experience for example, is less clear. Other empirical evidence suggests that only gay males face a discriminatory wage penalty while there is no effect on lesbian female wages (Klawitter, 2015).
Measuring discrimination in terms of wages is largely an inexact science and requires additional focus. Although these papers provide evidence that discrimination is a reality and does play some role in explaining the earnings differential between individuals of different sexual orientation, it is almost surely not the sole reason for the differential. Certainly, past literature suggests that there are other factors which also directly impact the difference in earnings found between individuals.
Differences in Human Capital
Differences in human capital, most notably work experience and education obtained, are potentially valid reasons which would drive an earnings differential between lesbian, gay, and heterosexual individuals. Early work on the impact of education and experience on earnings was captured in the Mincer earnings function (Mincer, 1958; Mincer, 1974). The Mincer regression simplifies earnings into educational attainment and potential experience, with the intuitive concept that more education and experience creates a more qualified candidate. The qualified candidate is thus able to parlay their skills and knowledge into higher earnings.
Based on the ideas put forth by Mincer and the previously mentioned earnings dynamic among gay and lesbian individuals, one would expect that lesbian females, based on their wage
advantage, are more qualified in terms of education and experience than heterosexual females. Contrarily, one would expect that gay males would be less qualified than heterosexual males based on their wage deficiency. Empirically, the raw data shows this prediction to only be partially true. While lesbian females indeed are more educated on average than heterosexual females, gay males are also more educated than heterosexual males (Black et Al, 2000; Zavodny, 2007). Utilizing an Oaxaca-Blinder decomposition, lesbian females are found to have generated their wage premium over heterosexual females using their human capital advantage, particularly their increased educational attainment (Antecol, Jong, and Steinberger, 2008). Thus, while differences in human capital excel in explaining the wages of lesbian females, it struggles to account for the case of the gay male, whom are well qualified in their own right.
Overall, empirical evidence and past literature suggests that some of the difference in earnings among individuals with different sexual orientation is captured by differences in human capital. Lesbian females, with an average overall higher standard of human capital, do see a return in terms of their earnings due to their higher education attainment and experience, however, the results of gay males are less convincing as a result of their unexplained wage penalty. Beyond discrimination and human capital, there still remains several plausible reasons which help contribute to the understanding of not only the wages of gay males, but lesbian females and their heterosexual counterparts as well.
Differences in Preferences
Differences in preferences for the purpose of this paper refers to the concept that individuals with different sexual orientation may select into different occupations for a variety of reasons, including differences in personality traits, types of career, and location, among others. It is plausible that selecting into different occupations based upon these reasons may work in one way or another to influence the earnings differential found between individuals of different sexual orientation. Although scarce overall, researchers have gradually begun exploring these topics in literature. In terms of location, past literature shows that gay and lesbian individuals locate to specific neighbourhoods, often in wealthier and more socially accepting cities (Black et Al, 2002). Past literature has also shown that gay and lesbian individuals select more open-minded occupations (Plug, Webbink, and Martin (2014). In terms of personality traits, appetite for competition has been offered as a potential channel through which gay and lesbian individuals would create a wage
premium or penalty. Research indicates that indeed there is a lack of competition for gay males, but the taste for competition is not necessarily present for lesbian females (Buser, Geijtenbeek, and Plug, 2015). This absence in appetite for competition could assist in explaining why gay males find themselves with an earnings penalty compared to heterosexual males. However, one would expect that lesbian females would be more competitive due to their wage premium. This dynamic is not evident in the literature.
Overall, differences in preferences may account for some of the earnings differential found between individuals of different sexual orientation. If gay and lesbian individuals are selecting into certain occupations based on criteria or conditions they value, this may produce a natural earnings gap between themselves and their heterosexual counterparts. It is clear that in some regard, differences in human capital, preferences, and discrimination do work to explain part of the differential in earnings. However, what is evident from past literature is that significant research is still required in order to complete our understanding of this dynamic. This paper builds upon previous literature by investigating the topic of household specialization and time usage decisions by individuals of different sexual orientation. It then investigates what role those time usage decisions, if any, have in explaining the earnings differential between gay and lesbian individuals and their heterosexual counterparts.
3. Basic Models of Household Specialization and Earnings
i. Household Specialization
The manner in which couples and partners divide their time has been the focus of past researcher’s keen on exploring the concept of household specialization. In constructing a basis for how gay, lesbian, and heterosexual couples make time usage decisions, this paper uses the basic models for time usage and utility provided in earlier work by Grossbard-Shechtman1.
Individuals are assumed to divide their time between three categories: work, personal, and household production, in a manner which satisfies the following equation:
𝑇𝑖=𝑙𝑖+ ℎ𝑖 + 𝑧𝑖
where 𝑇𝑖 is equal to the total amount of time allocated to an individual i in a given day (24 hours),
𝑙𝑖 is equal to the amount of time allocated towards labour by individual i, ℎ𝑖 is equal to the amount of time allocated towards household production by individual i, and 𝑧𝑖 is equal to the amount of time allocated to general personal time by individual i. Couples and partners are assumed to allocate their time in a manner which maximizes their individual utility. The utility of an individual is a function expressed as the following:
𝑈𝑖 = 𝑈𝑖(𝑙𝑖, ℎ𝑖, 𝑧𝑖, 𝑥𝑖, ℎ𝑗)
where 𝑈𝑖 is equal to the utility of individual i, 𝑙𝑖 is equal to the amount of time allocated towards
labour by individual i, ℎ𝑖 is equal to the amount of time allocated towards household production by individual i, 𝑧𝑖 is equal to the amount of time allocated to general personal time by individual i, 𝑥𝑖 is equal to the amount of goods consumed by individual i, and ℎ𝑗 is equal to the amount of time
allocated to household production individual j, the partner or spouse of individual i. Thus, an individual derives utility from their partner working and ensuring the household is well maintained. These two expressions offered by Grossbard-Shechtman help shape basic predictions on how gay, lesbian, and heterosexual partners may specialize within the household, and how certain factors may influence that specialization decision. First, both individuals derive utility from consumption. Thus, within couples, divisions of time will be made to ensure that the partner whom earns the highest wages also allocates the most time towards labour, thereby maximizing consumption. Given what is known empirically about wages, in heterosexual couples, one would predict that males on average would allocate more of their time to labour than females. Similarly, due to their presumed lack of specialization, one would predict that gay and lesbian couples would divide labour allocation relatively even within the partnership.
The presence, or absence, of children is believed to have a significant impact on household specialization. Children require resources in terms of both time and money. In the absence of children, one would predict that lesbian and gay couples would operate as previously predicted, with a relatively even split of duties. However, in the presence of children, one would expect to see partners begin specializing in a relatively similar manner to heterosexual couples with one higher earning partner focused on labour and the other partner focused on household production and child care.
The expressions offered here were meant to provide a theoretical basis for how gay, lesbian, and heterosexual couples may use their time in the face of changing conditions. In the sections that follow, the merit of these intuitive basic predictions will be tested empirically.
ii. Earnings
The preceding section focused on how individuals within a marriage or partnership utilize their time in terms of labour, household production, and leisure. This section will build upon it by discussing potential ways in which time usage decisions may impact earnings, specifically channels which may help explain the earnings differential found between gay, lesbian, and heterosexual individuals. In constructing an expression for wage earnings, this paper draws influence from Mincer, Acemoğlu, and Autor. The earnings of an individual is defined by the following expression:
𝑌𝑖=𝛽0+ 𝛽1𝐸𝑖 + 𝛽2𝐿𝑖+ 𝛽3𝑇𝑖 + 𝛽4𝑂𝑖 + 𝛽5𝐴𝑖+ 𝜀𝑖
where 𝑌𝑖 is the earnings of individual i, 𝛽0 represents the baseline earnings for an individual, 𝐸𝑖 is
the educational attainment of individual i, 𝐿𝑖 is the work experience of individual i, 𝑇𝑖 is the skills
training of individual i, 𝑂𝑖 is the amount of opportunity available to individual i, 𝐴𝑖 is the natural,
innate ability of individual i, and 𝜀𝑖 represents a general error term which captures potential unobservable factors in wages. Unobservable factors for the purpose of this paper may be things such as discrimination or labour imperfections.
The earnings of an individual are impacted by the time usage decisions of that individual. In deriving predictions on how potential time usage decisions may help explain the gap in wages among those of different sexual orientation, it is important to assess each indicator separately. For the purposes of this paper, the innate born ability of both heterosexual and homosexual individuals is assumed to be the same on average and unaffected by their daily time usage decisions. In terms of opportunity, as previously mentioned, it is plausible that gay and lesbian individuals select into different locations or occupations which may serve to increase or decrease their opportunity for advancement and higher wages. However, these opportunities would not be influenced by the daily time usage decisions of individuals.
Time usage decisions may help explain the difference in earnings found between individuals of different sexual orientation in three areas of interest: Education, Experience, and
Training. In heterosexual couples, one would expect that the breadwinner male partner would have significantly more experience and training as a result of his allocation of time towards labour and away from household production. The presence of children in the household would only serve to increase this gap by magnifying this specialization dynamic. Thus, like in past empirical research, one would expect to see that heterosexual males earn more than heterosexual females.
Predictions for gay and lesbian couples are less definitive. Gay and lesbian individuals are found in past literature to have significant educational attainment, so one would expect to see both gay and lesbian individuals with more years of schooling than their heterosexual counterparts. With less children in the household, specialization is less necessary and one would expect that lesbian females and gay males would devote additional time to labour force and less to household production. As a result of increased schooling and labour force supply, one would expect gay males and lesbian females to earn wages similar to that of heterosexual males. Predicting how patterns of time usage may explain the earnings dynamic found in past data and literature for homosexual individuals is less clear. It is certain that there are many factors at play in attempting to rationalize this gap in wages.
4. Data
The quantitative data used for analysis in this paper comes from the American Time Use Survey (ATUS) multi-year microdata file collected by the United States Bureau of Labor Statistics. The ATUS data utilized in this paper spans from 2003 to 2013 and consists of a multitude of variables that were collected from respondents in a questionnaire. Respondents are interviewed by the United States Census Bureau via the telephone and asked about which activities they utilize their time on. Households eligible for the ATUS represent a wide range of demographics, with respondents for the ATUS being randomly chosen members of selected households over the age of 15. Time use activities are identified via unique codes, which allows for an analysis of how individuals allocate their time during a given day.
i. Final Sample
The final sample used for analysis in this paper includes all those respondents who report a married or unmarried spouse living within the household. The sample is filtered to include only those respondents who reported weekly earnings. As the impact of time usage on earnings is the
main outcome of interest, time use decisions of respondents who choose not to report their earnings are of no use. In total, there were 48,026 respondents in the final sample used for analysis.
ii. Determining Sexual Orientation of Respondents
Sexual orientation of respondents is identified using a combination of sex of the individual and reported relationship to their married or unmarried spouse living within the same household. While this approach does not guarantee sexual orientation with verifiable accuracy, it is assumed that individuals reporting relationships of married or unmarried spouses to individuals of the same sex within the household are indeed homosexual.
Summary statistics for the final sample selected for analysis are presented in Table 1. The summary statistics are organized by sexual orientation and gender. Heterosexual males and females are presented in column 1 and 2, while gay males and lesbian females are presented in columns 3 and 4 respectively. Differences in educational attainment, measured in years of schooling completed, and weekly earnings mirror typical patterns found in past literature. Gay males and lesbian females complete 0.86 years and 0.97 years more education respectively. Gay males in the final sample are at an earnings disadvantage while lesbian females are at an earnings advantage.
TABLE 1: SUMMARY STATISTICS
Summary Statistics
Hetero Males Hetero Females Gay Males Lesbian Females
Outcomes Work Time 306.36 235.43 309.20 294.20 Leisure Time 305.76 269.72 301.21 249.59 Home Time 193.23 274.94 191.50 217.73 Covariates Age 43.42 42.33 41.93 41.28 Education 14.02 14.18 14.88 15.15 Earnings $1104.89 $729.65 $1062.67 $991.11 Children <18 1.33 1.17 0.26 0.59 Labor Force 61.09% 54.25% 62.31% 60.38% Observations 24933 22804 130 159
Note: Data are taken from the American Time Use Survey (ATUS) spanning 2003-2013. The time use variables are constructed using the activity
codes outlined in the paper by Connolly (2008). Time variables are measured in minutes per day. Education is measured in total years completed. Earnings are measured weekly. Labor force measures participation. An active labor force participant is defined as any individual whom logged work time of greater than 0 minutes.
5. Empirical Strategy
Using an Ordinary Least Squares (OLS) strategy, this paper will first estimate time usage of individuals by sexual orientation and gender. To estimate time allocation by sexual orientation and gender, this paper will draw from the paper by Connolly (2008) in its construction of variables representing total work time, total leisure time, and total home production time. Specifically, firstly, this paper will estimate the following regression three times, with the outcomes of interest being total work, total leisure, and total household production time usage:
1. 𝑌𝑤ℎ𝑙=𝛽0 + 𝛽1𝐻𝑀 + 𝛽2𝑆𝐹 + 𝛽3𝐻𝐹 + 𝛼𝑋 + 𝜀𝑤ℎ𝑙
where 𝑌𝑤ℎ𝑙 is the outcome variable of interest representing total work time, total leisure time, or total home production time, measured in minutes per day. 𝛽1 is an indicator for a homosexual male, 𝛽2 is an indicator for a heterosexual female, and 𝛽3 is an indicator for a
homosexual female. A heterosexual male is thus the reference point for the regression results. 𝛼𝑋 represents a vector of covariates including age, age squared, years of education, number of children, and sample year, while 𝜀𝑤ℎ𝑙 are robust standard errors.
Following that, again using an OLS strategy, this paper will run a regression which uses the time usage variables to estimate the role of household specialization and time allocation on earnings. Specifically, the following regression will be estimated:
2. 𝑌𝑖=𝛽0+ 𝛽1𝐻𝑀 + 𝛽2𝑆𝐹 + 𝛽3𝐻𝐹 + 𝛾1𝑇𝑊 + 𝛾2𝑇𝐿 + 𝛾3𝑇𝐻 + 𝛼𝑋 + 𝜀𝑖
where 𝑌𝑖, the outcome variable of interest, is the log of weekly earnings. 𝛽1 is an indicator for a homosexual male, 𝛽2 is an indicator for a heterosexual female, and 𝛽3 is an indicator for a homosexual female. A straight male is thus the reference point for the regression results. 𝛾1
measures the role of total work time, 𝛾2 measures the role of total leisure time, and 𝛾3 measures the role of total home production time. 𝛼𝑋 again represents a vector of covariates while 𝜀𝑖 are robust standard errors.
6. Results: Household Specialization
To measure the impact of sexual orientation and gender on total work, total leisure, and total home production time, I measure total time variables by including indicators for gender and
sexual orientation, along with controls for age, age squared, years of education, number of children, and sample year. The regressions are run on the full sample, a filtered male only sample, and a filtered female only sample. Results from the OLS estimates for total work, total leisure, and total home production by gender and sexual orientation are reported in Tables 2A, 2B, and 2C respectively. Regression results for the full sample are located in Panel A (Columns 1-3), while regression results for the filtered male and female samples are located in Panel B (Columns 4-6) and Panel C (Columns 7-9) respectively.
i. Total Work Time
Columns 1-3 report the total work time results for the full sample, with the reference point being total work time for a heterosexual male. The estimate for heterosexual females is highly statistically significant, with total work time of approximately 73 minutes less per day than a heterosexual male. Estimates for gay males and lesbian females are insignificant, however, estimates do exhibit some rationality and retain their expected negative signs. Of interest, in column 8, statistically significant estimates indicate that lesbian females work approximately 58 minutes more per day than heterosexual females. It is plausible that lesbian females, with higher earnings and less family restrictions, are more apt to work longer hours than heterosexual females. When children are added as a covariate as in column 9, this figure shrinks to only 50 minutes more per day, a possible sign of lesbian females withdrawing slightly from the labour market due to the presence of children. For the full sample, listed in column 3, the estimates for gay males and lesbian females decrease by 6 to 8 minutes per day when children are added as a covariate.
TABLE 2A: TOTAL WORK TIME
Full Sample Male Sample Only Female Sample Only
(1) (2) (3) (4) (5) (6) (7) (8) (9) Gay Male 2.716 0.10 1.989 0.07 -6.258 0.23 2.838 0.11 2.082 0.08 -0.970 0.04 Hetero Female -70.888 27.49** -71.128 27.53** -72.555 28.00** Lesbian Female -12.320 0.54 -12.783 0.56 -18.803 0.83 58.351 2.57* 57.830 2.55* 49.684 2.18* Controls Age, Age2, Education
Children No No Yes No No Yes No No Yes
Year Yes Yes Yes Yes Yes Yes Yes Yes Yes
R2 0.02 0.02 0.02 0.00 0.00 0.00 0.00 0.00 0.00
Observations 48,026 48,026 48,026 25,063 25,063 25,063 22,963 22,963 22,963
Note: Data are taken from the American Time Use Survey supplied by the United States Bureau of Labor Statistics, 2003-2013. Dependent
variable is total work time per day, measured in minutes. Total work time variable is constructed using the activity codes outlined in Connolly 2008. Reference point regarding estimates is a heterosexual male. Standard errors are listed in italics within the table.
*-Significance at the 5 percent level **-Significance at the 1 percent level
ii. Total Leisure Time
Columns 1-3 of Table 2B report the total leisure time results for the full sample, with the reference point being total leisure time for a heterosexual male. The estimates for heterosexual and lesbian females are both highly statistically significant, with total leisure time of approximately 34 and 49 minutes less per day respectively (column 2). Given our understanding of total work time, it is perhaps rational to expect that lesbian females would have less total leisure time than heterosexual females. Estimates for homosexual males do not differ substantially from heterosexual males and are not statistically significant. When children are added as a covariate in column 3, the point estimates for gay males, lesbian females, and heterosexual females all decrease anywhere from 3 to 14 minutes per day when compared to heterosexual males, an indication that they have more of a role in daily child care than heterosexual males.
TABLE 2B: TOTAL LEISURE TIME
Full Sample Male Sample Only Female Sample Only
(1) (2) (3) (4) (5) (6) (7) (8) (9) Gay Male -4.475 0.27 1.395 0.08 -13.448 0.81 -4.386 0.26 1.438 0.09 -11.773 0.71 Hetero Female -36.062 19.96** -33.730 18.77** -36.297 20.15** Lesbian Female -56.069 4.05** -48.691 3.48** -59.525 4.27** -20.165 1.46 -15.257 1.09 -24.632 1.77 Controls Age, Age2, Education
No Yes Yes No Yes Yes No Yes Yes
Children No No Yes No No Yes No No Yes
Year Yes Yes Yes Yes Yes Yes Yes Yes Yes
R2 0.01 0.02 0.03 0.00 0.01 0.01 0.00 0.01 0.02
Observations 48,026 48,026 48,026 25,063 25,063 25,063 22,963 22,963 22,963
Note: Data are taken from the American Time Use Survey supplied by the United States Bureau of Labor Statistics, 2003-2013. Dependent
variable is total leisure time per day, measured in minutes. Total leisure time variable is constructed using the activity codes outlined in Connolly 2008. Reference point regarding estimates is a heterosexual male. Standard errors are listed in italics within the table.
*-Significance at the 5 percent level **-Significance at the 1 percent level
iii. Total Home Production Time
Columns 1-3 of Table 2C report the total home production time results for the full sample, with the reference point being total home production time for a heterosexual male. The estimate for heterosexual females is highly statistically significant, with total home production time approximately 85 minutes more per day than a heterosexual male. However, in the restricted female only sample in column 8, lesbian females are found to have spent approximately 62 minutes less per day on home production, a statistically significant result. All of these estimates suggest that heterosexual females devote substantially more time to home production in a typical day, a usual result of parenthood and traditional family norms. When children are added as a covariate, specialization for gay and lesbian individuals begins to appear. In column 3, adding children as a covariate increases the point estimate of home production time for lesbian females and gay males by 25 minutes and 33 minutes, respectively. In the filtered female only sample, compared to heterosexual females, lesbian females decrease from 61 less minutes per day in home production time to only 35 minutes less when children are added as a covariate (column 9). These findings suggest that the presence of children force gay and lesbian couples to begin specializing in a manner similar to that of heterosexual couples.
TABLE 2C: TOTAL HOME PRODUCTION TIME
Full Sample Male Sample Only Female Sample Only
(1) (2) (3) (4) (5) (6) (7) (8) (9) Gay Male -1.419 0.09 -7.818 0.47 25.631 1.58 -1.613 0.10 -8.081 0.48 13.926 0.85 Hetero Female 81.614 46.88** 79.567 45.90** 85.352 49.68** Lesbian Female 24.903 1.65 16.957 1.12 41.371 2.79** -56.369 3.73** -61.735 4.04** -35.00 2.36* Controls Age, Age2, Education
No Yes Yes No Yes Yes No Yes Yes
Children No No Yes No No Yes No No Yes
Year Yes Yes Yes Yes Yes Yes Yes Yes Yes
R2 0.05 0.06 0.08 0.00 0.01 0.02 0.00 0.02 0.06
Observations 48,026 48,026 48,026 25,063 25,063 25,063 22,963 22,963 22,963
Note: Data are taken from the American Time Use Survey supplied by the United States Bureau of Labor Statistics, 2003-2013. Dependent
variable is total home production time per day, measured in minutes. Total home production time variable is constructed using the activity codes outlined in Connolly 2008. Reference point regarding estimates is a heterosexual male. Standard errors are listed in italics within the table. *-Significance at the 5 percent level
7. Results: Earnings
To measure the impact of time usage decisions by individuals on their earnings, I measure the log of weekly earnings by including previously calculated total time variables into an OLS regression complete with indicators for sexual orientation and gender. Controls for age, age squared, years of education, number of children, and sample year are included in the regression. The regressions are run on the entire sample, a filtered male only sample, and a filtered female only sample.
Results from the OLS estimates for log of weekly earnings by gender and sexual orientation are reported in Table 3. Regression results for the full sample are located in Panel A (Columns 1-4), while regression results for the male and female samples are located in Panel B (Columns 5-8) and Panel C (Columns 9-12) respectively. The reference point of regression results is the log of weekly earnings for a heterosexual male.
The estimates for heterosexual females, homosexual males, and lesbian females are all highly statistically significant. Homosexual males are estimated to make 18.5% less in weekly earnings when compared to heterosexual males, an estimate which resembles estimates found in past literature. Heterosexual females are estimated to make 50.1% less than heterosexual males, while lesbian females are estimated to make 24.6% less. Once again, this reinforces conventional findings from past literature that lesbian females earn more than heterosexual females. Of note, when children are added as a covariate for the filtered female sample (Column 12), the earnings premium for lesbian females shrinks to 19.5% from 24.3% over heterosexual females, a possible indication that the presence of children causes lesbian females to lose part of their advantage in the labour market.
To assess the role that time usage decisions and household specialization may have on the weekly earnings of individuals, previously constructed total time variables for home production, leisure, and work were included in the regression. In the most preferred specification in column 4, the estimates for total work time, total leisure time, and total home production time are all 0.000, two of which are statistically significant. These estimates imply that time usage decisions do not have any role in explaining the earnings differential found between individuals of different sexual orientation. Previous estimates for the magnitude of log weekly earnings by sexual orientation and gender remain unchanged and highly statistically significant when total time variables are included in the regression.
TABLE 3: LOG WEEKLY EARNINGS
Full Sample Male Sample Only Female Sample Only
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) Gay Male -0.035 0.59 -0.153 2.78** -0.153 2.82** -0.185 3.39** -0.034 0.57 -0.146 2.67** -0.145 2.67** -0.125 2.30* Hetero Female -0.484 67.92** -0.505 79.99** -0.493 77.59** -0.501 77.44** Lesbian Female -0.077 1.57 -0.222 5.58** -0.223 5.71** -0.246 6.31** 0.405 8.30** 0.271 6.70** 0.243 6.15** 0.195 4.92** Time Variables Total Work 0.000 8.47** 0.000 8.94** 0.000 6.87** 0.000 6.70** 0.000 6.52** 0.000 7.68** Total Leisure -0.000 2.66** -0.000 2.52* 0.000 0.29 0.000 0.28 -0.000 3.63** -0.000 3.18** Total Home Production -0.000 0.27 0.000 0.91 0.000 6.34** 0.000 5.89** -0.000 4.36** -0.000 1.50 Controls Age, Age2, Education
No Yes Yes Yes No Yes Yes Yes No Yes Yes Yes
Children No No No Yes No No No Yes No No No Yes
Year Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
R2 0.10 0.29 0.30 0.30 0.01 0.26 0.27 0.27 0.02 0.20 0.21 0.22
Observations 47,899 47,899 47,899 47,899 25,016 25,016 25,016 25,016 22,883 22,883 22,883 22,883 Note: Data are taken from the American Time Use Survey supplied by the United States Bureau of Labor Statistics, 2003-2013. Dependent
variable is the log of weekly earnings. Total time variables are constructed using the activity codes outlined in Connolly 2008. Reference point regarding estimates is a heterosexual male. Standard errors are listed in italics within the table.
*-Significance at the 5 percent level **-Significance at the 1 percent level
8. Conclusion
Using American Time Use Survey data, this paper investigated the role of a potential channel in explaining the earnings differential found in previous literature between individuals of different sexual orientation and gender. Utilizing an ordinary least squares empirical strategy, estimates for weekly earnings of homosexual males are 18.5% less than heterosexual males, while weekly earnings for heterosexual and lesbian females are 50.1% and 24.6% less than heterosexual males respectively. In the most preferred specification, estimates for the role of total time use variables remain 0.000 and two retain statistical significance. This finding emphasizes the fact that household specialization and time usage decisions are not responsible for the disparity in earnings between individuals of different gender and sexual orientation. More research is thus required to develop a more comprehensive understanding of what other potential channels may explain the disparity in earnings between individuals of different sexual orientation.
9. References
Acemoglu, Daron, and D. Autor. "The Basic Theory of Human Capital." Lectures in Labor
Economics (2011): 3-34.
Ahmed, Ali, Lina Andersson, and Mats Hammarstedt. "Sexual Orientation and Full-time
Monthly Earnings, by Public and Private Sector: Evidence from Swedish Register Data." Review
of Economics of the Household Rev Econ Household (2012): 83-108.
Ahmed, Ali M., Lina Andersson, and Mats Hammarstedt. "Are gay men and lesbians
discriminated against in the hiring process?." Southern Economic Journal 79.3 (2013): 565-585. Ahmed, Ali M., and Mats Hammarstedt. "Sexual Orientation and Earnings: A Register Data-based Approach to Identify Homosexuals." Journal of Population Economics J Popul Econ (2009): 835-49.
Antecol, Heather, Anneke Jong, and Michael Steinberger. "The sexual orientation wage gap: The role of occupational sorting and human capital." Industrial & Labor Relations Review 61.4 (2008): 518-543.
Arrow, Kenneth Joseph. "The theory of discrimination." (1973): 3.
Badgett, M. V. L. "The Wage Effects of Sexual Orientation Discrimination." ILR Review (1995): 726-39.
Baumle, Amanda K. "The Cost of Parenthood: Unraveling the Effects of Sexual Orientation and Gender on Income." Social Science Quarterly (2009): 983-1002.
Becker, Gary S. "The Economics of Discrimination." (1971). Becker, Gary S. "A Treatise on the Family." (1991).
Black, Dan A., Gary Gates, Seth G. Sanders, and Lowell J. Taylor. "Demographics of the Gay and Lesbian Population in the United States: Evidence from Available Systematic Data Sources."
Demography 37.2 (2000): 139-154.
Black, D., Gates, G., Sanders, S., & Taylor, L. “Why do gay men live in San Francisco?”.
Journal of Urban Economics, 51(1), (2002): 54-76.
Black, Dan A., Seth G. Sanders, and Lowell J. Taylor. "The Economics of Lesbian and Gay Families." Journal of Economic Perspectives 21.2 (2007): 53-70.
Buser, Thomas, Lydia Geijtenbeek, and Erik Plug. "Do Gays Shy Away from Competition? Do Lesbians Compete Too Much?." (2015).
Connolly, Marie. "Here Comes the Rain Again: Weather and the Intertemporal Substitution of Leisure." Journal of Labor Economics Journal of Labor Economics, The (2008): 73-100. Dalmia, Sonia, and Paul Sicilian. "Kids Cause Specialization: Evidence for Becker’s Household Division of Labor Hypothesis." Int Adv Econ Res International Advances in Economic Research 14.4 (2008): 448-59.
Drydakis, Nick. "Sexual orientation discrimination in the labour market." Labour Economics 16.4 (2009): 364-372.
"Frequently Asked Questions (FAQs)." U.S. Bureau of Labor Statistics. U.S. Bureau of Labor Statistics. Web. <http://www.bls.gov/tus/atusfaqs.htm>.
Giddings, Lisa, John Nunley, Alyssa Schneebaum, and Joachim Zietz. "Children, Family Size and Household Specialization: A Comparison of Different-Sex and Same-Sex Couples Using Exact Covariate Matching." ResearchGate (2010): 1-39.
Grossbard-Shechtman, Amyra. "A Theory of Allocation of Time in Markets for Labour and Marriage." The Economic Journal 94.376 (1984): 863-882.
Heineck, Guido. "Sexual Orientation and Earnings: Evidence from the ISSP." Applied Economics
Letters (2009): 1351-354.
Klawitter, Marieka. "Meta‐analysis of the effects of sexual orientation on earnings." Industrial
Relations: A Journal of Economy and Society 54.1 (2015): 4-32.
Kurdek, Lawrence A. "The Allocation of Household Labor by Partners in Gay and Lesbian Couples." Journal of Family Issues 28.1 (2007): 132-148.
Leppel, Karen. "Labour Force Status and Sexual Orientation." Economica 76.301 (2009): 197-207.
Mincer, Jacob. "Investment in human capital and personal income distribution." The Journal of
Political Economy (1958): 281-302.
Mincer, Jacob A. "Age and Experience Profiles of earnings." Schooling, Experience, and
Earnings. NBER, (1974). 64-82.
Phelps, Edmund S. "The statistical theory of racism and sexism." The American Economic
Review 62.4 (1972): 659-661.
Plug, Erik, and Peter Berkhout. "Effects of Sexual Preferences on Earnings in the Netherlands."
Journal of Population Economics (2004): 117-31.
Plug, Erik, Dinand Webbink, and Nick Martin. "Sexual orientation, prejudice, and segregation."
Tilcsik, Andras. "Pride and Prejudice: Employment Discrimination against Openly Gay Men in the United States1." American Journal of Sociology 117.2 (2011): 586-626.
Weichselbaumer, Doris. "Sexual orientation discrimination in hiring." Labour Economics 10.6 (2003): 629-642.
Zavodny, Madeline. "Is There A ‘Marriage Premium’For Gay Men?." Review of Economics of
