Show Summary Details

Page of

PRINTED FROM the OXFORD RESEARCH ENCYCLOPEDIA, ECONOMICS AND FINANCE (oxfordre.com/economics). (c) Oxford University Press USA, 2020. All Rights Reserved. Personal use only; commercial use is strictly prohibited (for details see Privacy Policy and Legal Notice).

date: 18 February 2020

Marriage and Labor Market Outcomes

Summary and Keywords

Marriage and labor market outcomes are deeply related, particularly for women. A large literature finds that the labor supply decisions of married women respond to their husbands’ employment status, wages, and job characteristics. There is also evidence that the effects of spouse characteristics on labor market outcomes operate not just through standard neoclassical cross-wage and income effects but also through household bargaining and gender norm effects, in which the relative incomes of husband and wife affect the distribution of marital surplus, marital satisfaction, and marital stability.

Marriage market characteristics affect marital status and spouse characteristics, as well as the outside option, and therefore bargaining power, within marriage. Marriage market characteristics can therefore affect premarital investments, which ultimately affect labor market outcomes within marriage and also affect labor supply decisions within marriage conditional on these premarital investments.

Keywords: marriage, labor supply, household bargaining, marriage markets, displaced workers, tied movers, breadwinner norm, marriage premium

Marriage, Work, and Wages: An Overview

The effects of marital status, spouse characteristics, and marriage market characteristics on labor supply and earnings depend on how married-couple households jointly determine labor market outcomes. In a simple neoclassical model of household labor supply, labor supply decisions are a function of spouse’s productive characteristics through a cross-wage elasticity. Shocks to earnings of one spouse, such as through unemployment or displacement, can increase the labor supply of the other spouse. Labor market effort and outcomes may be affected by other spouse job characteristics besides earnings, such as demand for long hours or location or mobility requirements.

In a household bargaining model, spouse characteristics affect labor supply not just through standard cross-wage and income affects but by shifting the distribution of marital surplus. An increase in wage may act to increase labor supply through standard neoclassical effects but may also act to reduce labor supply by increasing own bargaining power and consumption conditional on contemporaneous labor supply.

Marriage market conditions affect marital status and spouse characteristics. They also affect the outside option, and therefore bargaining power, within marriage. A less favorable marriage market may increase the labor supply of married individuals both due to matching with a lower quality spouse and due to lower bargaining power within marriage. Among unmarried individuals, a less favorable marriage market may encourage premarital investment in human capital both to increase earnings outside of marriage and to increase future bargaining power within marriage.

Much of the literature finds asymmetric labor supply and earnings effects by gender within marriage. There is a question of whether these asymmetries simply reflect the effect of differences between men and women in productive characteristics, or whether there are gender norms which further influence these outcomes. Gender norms may affect human capital investment if the marriage market rewards or penalizes these investments differently by gender. Within marriage, gender norms may affect labor supply and earnings if marital satisfaction and marital stability are affected by the relative earnings of husband and wife.

While much of the literature on marriage and labor market outcomes focuses on married women, the marriage premium literature focuses on the wage gap between married and unmarried men. This literature considers whether the positive wage gap between married and unmarried men is due to selection into marriage, returns to household specialization within marriage, or a higher valuation of married men by employers.

Finally, it must be recognized that the vast majority of research on marriage and labor market outcomes to date has focused exclusively on heterosexual married couples. An emerging literature considers whether joint labor market decision-making within same-sex couples differs from heterosexual couples.

The Effect of Husband’s Earnings on Wife’s Labor Supply

The Cross-Wage Elasticity

An extensive literature in economics studies the labor supply of married women, including estimating the cross-wage elasticity, the effect of husband’s wage on wife’s hours of work. Many papers in the literature attempt to address biases due to measurement error and omitted variables using some combination of group averages, instrumental variables, and first-differences strategies. Some papers identify elasticities using cross-sectional variation in group average wages while others use differential changes over time in group average wages. With cross-sectional estimates, even those generated using group averages or using group indicators as instruments, omitted characteristics that are correlated with both group wages and group labor supply will bias the estimates. Differences estimates rely on the assumption that any such omitted characteristics are time-constant.

Pencavel (1998) analyzes the labor supply of married women ages 25 to 60 in birth cohorts 1920–1924 to 1960–1964 using March Current Population Survey (CPS) data for the years 1976 to 1995. To estimate own and cross-wage elasticities, he instruments hourly earnings of the woman and her husband with age, education, and cohort variables, as well as two time-varying trade variables (the U.S. real trade balance and real merchandise imports). His estimates indicate a sizable, negative, and statistically significant cross-wage elasticity. While his analysis includes two time-varying instruments, his estimates are most likely largely identified by cross-sectional comparisons across education and birth cohort groups

Blundell, Duncan, and Meghir (1998) instead estimate wage and income elasticities using changes over time in average wages and average income for groups of married and cohabitating women stratified by education and birth cohort. Analysis is conducted using the U.K. Family Expenditures Survey the for years 1978 to 1992 on the sample of married and cohabitating women ages 20 to 50 with an employed husband or partner. Variation over time in average after-tax wages and average other income (measured as household consumption minus after-tax earnings) is generated by both changes in wage dispersion and by tax code changes. The resulting income elasticities are negative but small and imprecisely estimated. For married and cohabitating women without children, the estimated income elasticity is zero.

Devereux (2004) similarly uses changes in group averages to estimate cross-wage elasticities for both husbands and wives. His analysis uses changes over time in relative wages for different husband–wife couple types, grouped based on age, education, and region, in 1980 and 1990 U.S. Decennial Census data. His differences specification includes indicators for husband age and education groups and wife age and education groups. Unlike Blundell et al. (1998), this controls for any unobserved characteristics of husband and wife age and education groups that affect both changes in average wages and changes in average hours. The wage elasticity estimates are therefore identified using differences in how average relative wages evolve within husband or wife type based on spouse characteristics. For example, the predicted wage growth from 1980 to 1990 for college-educated married men in their 30s differs for those married to college-educated women in their 30s compared to those married to high school graduates in their 20s. The estimates are also identified using regional variation in wage growth across groups. In contrast to Blundell et al. (1998), Devereux finds that a sizeable and significant cross-wage elasticity of roughly –0.4 for married women. Consistent with the prior literature he finds a negligible cross-wage elasticity for married men.

Blau and Kahn (2007) examine how the cross-wage elasticity for married women has evolved over time using March CPS data for three 3-year intervals: a 1980 interval (1979–1981), a 1990 interval (1989–1991), and a 2000 interval (1999–2001). As the labor force attachment of married women has increased, occurring jointly with trends in delayed marriage and increased divorce, women may have become less sensitive to their husband’s earnings, perhaps motivated to investment in labor market skills given lower expected lifetime consumption that will be realized out of husband’s income.

Blau and Kahn (2007) generate estimates using both cross-sectional and differences approaches. Their cross-sectional approach instruments own and husband’s wages using wage decile and, in some specifications, education. With this approach, they obtain cross-wage elasticities for the 1980 interval of –0.33 to –0.39, decreasing to estimates of –0.19 to –0.22 for the 2000 interval. They also confirm negligible cross-wage elasticities for married men in all time periods.

The differences approach used by Blau and Kahn (2007) is similar to that of Devereux (2004). Three age groups for own age and husband age and four education groups for own and husband education generates 144 couple-type cells, each of which is used to generate changes in group averages from the 1980 interval to the 1990 interval and from the 1990 interval to the 2000 interval. Similar to Devereux, indicators for husband and wife age and education groups are included in the differences regression. The resulting cross-wage elasticity using changes from the 1980 interval to the 1990 interval is –0.65, somewhat larger than the Devereux’s estimates obtained using changes from 1980 and 1990 in Decennial Census data. The Blau and Kahn cross-wage elasticity estimate using changes from the 1990 to 2000 interval is –0.19, indicating a decline in the cross-wage elasticity over time, consistent with their cross-sectional results. Heim (2007) likewise documents a decline over time in income elasticities for married women ages 25 to 55 in cross-sectional estimates produced using CPS data from 1979 to 2003.

Added Worker and Displaced Worker Effects

A more specialized literature estimates the “added worker effect,” the effect of husband’s unemployment on wife’s labor supply. Most estimates in this literature suggest a relatively small added worker effect, but Cullen and Gruber (2000) argue that many of these estimates are likely biased, both due to the fact that wives of unemployed husbands may themselves have lower labor force attachment and because husband’s unemployment may result from diminished local labor market prospects that also affect the wife.

Using the Survey of Income and Program Participation (SIPP) data for 1984–1988 and 1990–1999 and leveraging variation in the generosity of potential unemployment insurance (UI) benefits by state, year, and husband’s prior earnings, they find that more generous UI benefits are associated with substantial crowdout of wife’s labor supply response. These results suggest a larger added worker effect than previously documented in the literature.

Most unemployment spells have a relatively small effect on household lifetime income, which, in a lifecycle context, should have a relatively small effect on spouse’s labor supply. Stephens (2002) points out that, in contrast, long-term earnings losses are well documented for displaced workers, those who experience plant closings or layoffs. He argues that the “displaced worker effect” on wife’s labor supply is therefore likely to be larger than the added worker effect. Using the 1968–1992 waves of the Panel Study of Income Dynamics (PSID), Stephens estimates the effects of husband’s displacement on wife’s labor supply using the sample of married couples in which both husband and wife are ages 25 to 65. The longitudinal data allow the inclusion of household-level fixed-effects in the model. Results indicate that wife’s average hours of work increase modestly up to three years prior to the year of displacement, with large, persistent post-displacement effects. Estimates suggest an 11% increase in wife’s annual hours post-displacement, with effects roughly 50% larger for the subsample of households in which husbands experience negative wage growth in the year after displacement. These estimates likely understate the true displaced worker effect, as husband’s displacement likely signals a negative local labor market shock, particularly in the case of plant closings, that diminishes the wife’s labor supply response.

Finally, Stephens (2002) finds that the displaced worker effect is much smaller and statistically insignificant for displacements that occurred in the 1970s, so that the overall estimates reflect even larger displaced worker effects during the 1980s. Unfortunately, estimates were not produced for the 1990s, so a comparison to the diminished cross-wage elasticities reported in Blau and Kahn (2007) is not possible.

The Labor Market Effects of Spouse Job Characteristics

Spouse Overwork

While a large literature estimates the effect of husband’s earnings on wife’s labor market outcomes, the effects of husband’s job characteristics besides earnings have received much less attention. For example, long hours of work in husband’s job, conditional on earnings, may affect wife’s labor market effort, particularly if children are present in the household.

Cha (2010) analyzes the effect of spousal overwork in a sample of dual-earner couples in the 1996 panel of the SIPP. She finds that having a husband who works 50 or more hours a week increases the probability a married woman with children quits her job, but the wife’s overwork does not affect the probability that a husband quits his job. Cha cannot rule out the possibility that men choose to longer work hours in response to their wife’s lower labor market attachment.

Because work hours can be chosen endogenously in response to spousal characteristics, McKinnish (2019) estimates instead the effect of overwork in husband’s undergraduate degree field on wife’s labor market outcomes using a sample of college-educated married couples from the 2009–2014 American Community Survey (ACS). Unlike work hours or occupation, undergraduate degree field is time constant and in most cases chosen prior to spouse characteristics. She finds that overwork in husband’s degree field, as measured by the fraction of college-educated prime-aged men who report 50 or more hours of work a week, has a negative effect on earnings for married women with children.

In contrast to the negative effect of spouse overwork on earnings for married women with children, McKinnish (2019) finds a positive effect of overwork in spouse’s degree field on earnings for married men with children and married women without children. These positive coefficient estimates are consistent with positive assortative matching. Overwork-prone degree fields tend to be higher paying and higher skilled than degree fields with lower rates of overwork. Through positive assortative matching, individuals from overwork-prone degree fields will tend to marry higher-wage, higher-skilled spouses, generating a positive association between spouse overwork and labor market outcomes. This suggests that even the negative effects found for married women with children may be subject to positive bias.

These results also highlight the fact that most any attempt to estimate the effects of spouse characteristics on labor market outcomes will be subject to empirical bias due to positive assortative matching affects. Conditional on observed characteristics, matching with a higher-quality spouse suggests positive selection on unobserved characteristics. For example, a college-educated woman married to a college-educated man will tend to be positively selected on unobserved characteristics relative to a college-educated woman married to a husband without a college degree.

Household Migration

Jobs may also come with location and mobility requirements that affect spouse labor market outcomes. A basic model of household migration predicts that a married couple will migrate if the total increase in household income from the move exceeds migration costs. Married individuals will therefore tend to be less mobile that unmarried individuals, as the effects of the household move on both spouses enter the decision. A married individual may make a move that decreases earnings, or be unable to make a move that would increase earnings, resulting in a “tied mover” or “tied stayer.” Because men are more likely to work in occupations in which larger returns to migration, married women are more likely to end up as tied movers.

The household migration literature finds that the earnings of married women typically decrease with migration. There are empirical challenges to estimating these migration effects, which involves a comparison of movers to non-movers. It is difficult to fully control for differences in returns to migration between the two groups. Additionally, if the household move is generated by a negative labor market shock to the husband, the wife may have increased her labor market effort (an added worker effect) just prior to the move. In this case, the earnings decrease associated with migration may just reflect the wife returning to an earlier level of labor market effort.

Instead of comparing migrants to nonmigrants, McKinnish (2008) uses samples of married couples in the 2000 Decennial Census to estimate the effect of the migration rate in spouse’s occupation on own labor market outcomes. She finds that for wives of college-educated men, conditional on wife’s occupation fixed effects and regardless of wife’s education level, the migration rate in husband’s occupation negatively affects wife’s earnings. Negative effects are stronger for married women with children. In contrast, college-educated men do not experience any negative effects from migration rate in wife’s occupation, suggesting gender asymmetries in how married couples respond to migration opportunities.

The empirical literature on marriage, migration, and labor market outcomes has largely focused on tied movers. It is even more difficult to identify tied stayers, married individuals who do not realize a potential gain from migration. This is because it is difficult to accurately measure potential returns to migration for nonmigrants. McKinnish (2008) shows that a regression of own migration on migration rate in own occupation produces a coefficient estimate close to 1 for both unmarried men and unmarried women, suggesting both men and women, when unmarried, are equally and highly responsive to migration rates within their occupations. However, when analyzing couple migration, coefficients on occupation migration rates are substantially less than 1 for both spouses and the coefficient estimate on wife’s occupation migration rate is about half that on her husband’s. This suggests that both married men and women are likely to be tied stayers but married women are disproportionately so.

Household Bargaining and Marriage Markets

Household Bargaining and Labor Supply

When household bargaining is incorporated into models of household labor supply, husband and wife characteristics affect household labor supply decisions not only through their effects on productivity and earnings but also through their effect on bargaining power. Chiappori, Fortin, and Lacroix (2002) develop and estimate a collective model of household labor supply in which the marriage market sex ratio and state divorce laws affect labor supply decisions by shifting bargaining power and therefore the distribution of marital surplus within a marriage. An increase in own bargaining power should reduce labor supply through a standard income effect in which consumption is increased conditional on contemporaneous labor supply.

Chiappori et al. (2002) use 1988 PSID data to estimate the effect of state divorce laws and sex ratios on the labor supply of married men and women. Sex ratios are calculated by state, age, and race. They find that divorce laws and sex ratios that are more favorable to the wife reduce wife’s labor supply and increase husband’s labor supply, consistent with model predictions.

In a household bargaining context, policies expected to increase women’s labor supply, such as those reducing gender wage discrimination or increasing control over fertility decisions, may be partially offset by their effect on the bargaining power of married women. Oreffice (2007) estimates the effect of legalized abortion on the labor supply of married women using CPS data from 1968 to 1979. Using state-level variation generated by the 1970 legalization by five states and the 1973 federal legalization, she finds that, among couples married prior to legalization, legalization is associated with a decrease in the labor supply of married women and an increase in the labor supply of married men. These effects are consistent with household bargaining effects but the opposite of what would be predicted by a labor supply model without household bargaining.

Marriage Markets and Labor Market Outcomes

Household bargaining must also be considered when studying how marriage market characteristics affect labor market outcomes. A more or less favorable marriage market can affect labor market outcomes though effects on marital status, spouse characteristics, and the bargaining power of married individuals. Marriage market characteristics can therefore affect premarital investments in human capital as well as labor supply decisions within marriage conditional on human capital.

The sex ratio, the ratio of men to women, in an individual’s preferred marital search market affects marriage market options. A low sex ratio reduces the probability of marriage for women, reduces the predicted quality of their spouse, and also reduces bargaining power within marriage by reducing the attractiveness of their outside option. An unfavorable sex ratio might therefore spur women to invest in human capital, both in anticipation of a lower probability of marriage and to increase bargaining power within marriage. It might increase labor supply of women through increased premarital human capital investment, reduced marriage rates, and reduced bargaining power within marriage.

Angrist (2002) and Lafortune (2013) both make use of the fact that second-generation immigrants to the United States exhibit high rates of ethnic endogamy to estimate effects of the marriage market sex ratio. Angrist (2002) uses U.S. Decennial Census data from 1910, 1920, and 1940 to examine changes in outcomes of adult children of immigrants in response to national-level changes in the sex composition of same-ethnicity immigration flows from 1910 to 1940. He finds that higher sex ratios (more men per woman) among immigrants in an ethnic group led to decreased female labor force participation and increased male earnings among second-generation immigrants in the same ethnic group, consistent with an oversupply of men in the marriage market leading to greater labor market effort by men and reduced labor market effort by women. He also finds that the sex-ratio effects are strongest for ethnic groups that exhibit higher rates of endogamy.

Like Angrist, (2002), Lafortune (2013) estimates the effects of variation in sex ratios generated by immigration flows, but she uses variation at the state-ethnic group level that is predicted by the historic location of immigrants. She additionally considers effects on premarital investments. Using U.S. Decennial Census data from 1900 to 1970, she finds that males respond to an unfavorable sex ratio through premarital investments, by increasing education levels, while females instead respond with increased labor supply.

Charles and Luoh (2010) study the impact of the rise in male incarceration in the United States from 1970 to 2000 on women through the marriage market. Increases in male incarceration for a particular race/age/location group should reduce marriage prospects for women in that group. Analysis with Decennial Census data confirms that women affected by higher male incarceration in their marriage market increase both schooling and employment.

While immigration, incarceration, and wars can generate marriage market imbalances, variation in sex ratios can also result from simple variation in cohort size. Men tend to marry at somewhat older ages then women, and the typical marriage has a positive age gap between husband and wife. To the extent that men prefer to search for wives among slightly younger women and women prefer to search for husbands among slightly older men, women born during a baby boom will face an unfavorable sex ratio (too few older men) and women born during a baby bust will face a favorable sex ratio (oversupply of older men). Grossbard and Amuedo-Dorantes (2007) leverage variation in cohort size by U.S. region to estimate the effects of marriage market sex ratio using samples of women ages 25 to 44 in 1975–2005 CPS data. They find that women born during a regional baby boom (unfavorable sex ratio) exhibit higher labor force participation rates than women born during regional baby busts (favorable sex ratios) and that these effects are found both for the full sample of women and for the subsample of married women.

Because marriage markets tend to be fairly segmented by education, gender differences in educational attainment results in sex ratio imbalances within education category. Negrusa and Oreffice (2010) analyze the effects of sex ratios by education and metropolitan area using Decennial Census and CPS data from 1980, 1990, and 2000. They find that a shortage of women in education and location-specific search pools is associated with decreased labor supply of married women and increased labor supply of married men, consistent with a shortage of marriageable women increasing the bargaining power of women in marriage. They do not find effects of the sex ratio on the labor supply of unmarried men or unmarried women.

Marriage Market Effects of College Quality

For women, marriage market considerations can moderate the effect of ability on labor market outcomes. High aptitude women are more likely to attend competitive postsecondary institutions and work in occupational settings where their marital search pool will primarily consistent of high-ability men. This access to high-earning men in their marriage market may actually reduce their own labor market effort. Black and Smith (2004) estimate the effect of college quality on labor market outcomes separately for men and women using data from the National Longitudinal Survey (NLS) of Youth 1979 cohort. For men, they find a monotonic, positive relationship between college quality and adult wages. For women, however, they find that the wage returns to attending a college in the upper quartile of quality is statistically indistinguishable from the returns to attending a college in the second quartile of quality, suggesting that, conditional on a woman’s ability, attending a higher-quality college may reduce labor market effort by facilitating marriage to higher-earning men.

Ge, Isaac, and Miller (2018), however, find very different results for the effect of college quality on women’s labor market and marriage market outcomes. They extend Dale and Krueger’s (2002) analysis of college quality and labor market outcomes using the College and Beyond 1996 survey of individuals who entered one of 34 selective U.S. colleges and universities in 1976. A crucial difference between this analysis and the Black and Smith (2004) analysis is that all of the postsecondary institutions in the College and Beyond sample would likely be in the top quartile of college quality as defined in Black and Smith.

Dale and Krueger (2002) identify the returns to school quality by comparing students who were admitted and rejected from schools with similar average SAT scores but who ultimately attended schools with different average SAT scores. They find no evidence of returns to school quality. Extending the analysis, Ge et al. (2018) find that, unlike men, women experience a 14% earnings return to attending a school with a 100-point higher SAT average. They also find that this same increase in college quality is associated with a four percentage point reduction in the predicted probability of marriage and that the earnings returns are eliminated once controls for marital status are included in the model.

There are several different mechanisms that are consistent with these results. One is that, unlike men, women consider their preferences for marriage when choosing their college. Conditional on a set of college acceptances, women who choose the less selective college may do so based on preferences to marry earlier and engage in relatively less market production with marriage. Another mechanism is that women who attend elite institutions experience a marriage market penalty, and they respond to these diminished marriage market opportunities with increased labor market effort. Finally, if women experience labor market returns to attending an elite institution that men do not, this could raise their reservation quality for a spouse by improving their outside option. Ge et al. (2018) show that college selectivity is not associated with decreased husband quality, either based on educational attainment or earnings, which is not consistent with the marriage market penalty mechanism.

Gender Norms, Marriage, and Labor Market Outcomes

Another active area of research is how gender norms interact with marriage and marriage markets to affect labor market outcomes, particularly outcomes of women. While there is a much broader literature on the effect of gender norms on labor market outcomes, this particular line of study focuses more specifically on how gender norms affect how the labor market effort of women responds to different marriage market or spousal characteristics. These gender norm effects may counteract or moderate the responses to marriage market conditions and spouse characteristics predicted by a standard neoclassical model.

For example, standard models predict that a woman faced with unfavorable marriage market conditions will respond by increasing future earnings through greater human capital investment and/or labor market attachment. However, to the extent that the marriage market penalizes women for high levels of professional ambition and potential earnings, this may moderate their investment and labor market response to unfavorable marriage market conditions. Likewise, standard models predict that married women will respond to a low-earning husband by increasing their own labor market effort. But gender norms may predict that marital conflict and instability result when a married woman’s earnings increase relative to her husband’s, again moderating the wife’s labor market response to low husband earnings.

Marriage Markets, Signaling, and Human Capital Investment

Bursztyn, Fujiwara, and Pallais (2017) provide particularly unique evidence of adjustments women make in response to marriage market pressures. Studying grade records from an elite U.S. MBA program, they find that while married and unmarried women have similar grades on most coursework, unmarried women have significantly lower class participation grades than married women. This suggests that the unmarried women are avoiding public behavior that might send a negative marriage market signal regarding their ambition and potential earnings.

Bursztyn et al. (2017) also report results from an experiment in the same MBA program in which students are asked to rank hypothetical jobs that involved trading off higher pay and career advancement for longer hours and more travel. Students expected that their answers would be shared in small groups, and the gender composition of their small groups was manipulated. Single women were less likely to report preferences for the higher-paying jobs with longer hours when they were in small groups with male peers as opposed to an all-female group. Single women similarly reduced their reported preferences for jobs with more rapid career advancement but more travel when with male peers. Single women made no such adjustments when reporting preferences over jobs that traded off social impact with interaction with coworkers, a choice with fewer implications for marriage market signaling.

Some human capital and labor market choices may have ambiguous effects on women’s marriage market prospects. Consider, for example, pursuing a higher level of education, specializing in a higher-paying field of study, or taking a job in a more demanding but higher-paying workplace. On one hand these activities may signal a higher level of ambition and potential earnings that may be penalized in the marriage market. On the other hand, these activities may give women access to better marital search pools by placing them in proximity with high-earning potential mates. Bursztyn et al.’s (2017) results suggest that, when concerned about marriage market options, women may be particularly likely to adjust on margins that do not change the marital search pool but affect the marriage market signal. For example, women may adjust competitive or ambitious behaviors that are observed by male peers in school or workplace settings.

The “Male Breadwinner” Norm

Bertrand, Kamenica, and Pan (2015) consider whether the male breadwinner norm causes women to adjust labor market effort in marriage to avoid out-earning their husband. Using 1970–2000 Decennial Census data as well as 2008–2011 ACS data to analyze the distribution of wife’s share of household income, they document a sharp drop at 0.5, where the wife starts to out-earn the husband. A discontinuity of this type is difficult explain with a standard economics model but is consistent with a male breadwinner norm.

This absence of couples in which the wife just barely out-earns the husband could be generated by selection into marriage, by adjustments to labor market effort within marriage, and by greater marital dissolution among female breadwinner couples. Bertrand et al. (2015) investigate these mechanisms. First, defining marriage markets based on age, race, education, and state, they generate marriage market specific predictions of the probability a randomly chosen woman out-earns a randomly chosen man. In some specifications, these predictions are generated using a Bartik-style instrument based on industrial composition. They show that marriage rates are lower in markets in which women are more likely to out-earn men, conditional on average woman’s income and average man’s income in the market.

Bertrand et al. (2015) then generate within-marriage predictions of the probability the wife out-earns the husband. They find that when the wife is likely to out-earn the husband based on race, age, education and state of residence, wife’s labor force participation is lower and the gap between wife’s realized and predicted earnings is wider. These results suggest that women adjust labor market effort in marriage to avoid out-earning their husband.

Additional analysis by Bertrand et al. (2015) with the 1987–1988 wave of the National Survey of Families and Households (NSFH) finds that, even controlling for both wife’s and husband’s income, there is a discontinuity in which couples in which the wife earns more than the husband report lower levels of marital satisfaction and are more likely to report having discussed marital separation. These female breadwinner couples are also more likely to be divorced in the second wave of the survey in 1992–1994. These results indicate that women who adjust the labor market effort to avoid violating the male breadwinner norm may be reacting rationally to improve the quality and stability of their marriage.

Foster and Stratton (2018) point out that the marital satisfaction and dissolution analysis in Bertrand et al. (2015) is based on a sample of married couples from 1987–1988 and that gender norms have likely evolved over time. They replicate the Bertrand et al.’s NSFH analysis and show that the marital satisfaction and dissolution effects estimated in that sample are being driven by women who were ages 35 and older at the time of the 1987–1988 interview. Foster and Stratton then perform similar analysis using more recent data from both the United States and Australia. In both 1998–2015 NLS year 1997 data from the United States and 2001–2016 Household Income and Labor Dynamics in Australia survey data, they find little evidence that female breadwinning affects marital quality or dissolution for married couples. These results suggest that gender norms with respect to male breadwinning may be evolving.

Blinder and Lam (2018) provide an important note of caution in interpreting the Bertrand et al. (2015) results. First, they show that if there is positive assortative matching on earnings in a marriage market in which men typically out-earn women, this will generate a skewed distribution of matches in which women rarely out-earn their husband, even in the absence of a preference on the part of either men or women for male breadwinner matches. Importantly, the positive assortative matching that generates this skewed distribution could result from a preference for equal earnings or even a preference that the wife out-earns the husband.

Given these results, the distribution of wife’s earnings shares reported in Bertrand et al. (2015) does not necessarily imply a male breadwinner norm, except for the sharp discontinuity just past 0.5 where the wife just starts out-earning her husband. Blinder and Lam (2018), however, show that this sharp discontinuity is largely an artifact of the larger point mass at 0.5, where husband and wife have equal earnings. In fact, they demonstrate that there is a sharp discontinuity just before 0.5, which would, using similar logic, imply that couples prefer equal earnings over a male breadwinner household. Eriksson and Stenberg (2015) perform similar analysis with Swedish population register data from 1990 to 2011 and also document that the discontinuity is driven by the large spike at 0.5. They further point out that wives at the spike have lower education relative to their husbands compared to wives immediately above and below the spike, which is inconsistent with the argument that these women have a disproportionate potential to out-earn their husband but reduce their effort in response to the breadwinner norm.

Blinder and Lam (2018) argue that therefore the earnings share distribution analysis in Bertrand et al. (2015) is not evidence of a male breadwinner norm but that their additional analysis of marital satisfaction and dissolution could still be interpreted as such. This could help reconcile why the earnings share analysis in Bertrand et al. documents a sharp discontinuity in even recent 2008–2011 ACS data, but the Foster and Stratton (2018) analysis of marital satisfaction and dissolution finds much diminished effects of a breadwinner wife for more recent cohorts. The marital satisfaction and dissolution results are a more direct test of the male breadwinner norm, which has diminished over time, while the earnings share distribution results merely reflect positive assortative matching and gender earnings gaps, which have persisted over time. Sprengholz and Wieber (2019) also document a diminishing breadwinner norm in Western Germany comparing the periods 1984–1990, 1997–2006, and 2007–2016.

Murray-Close and Heggeness (2018), however, provide evidence that the male breadwinner norm still persists to some degree. They study whether the earnings reported by married men and women in survey data are affected by whether the wife out-earns the husband. Using CPS–Annual Social and Economics Supplement data from 2003 to 2013 that is matched to administrative reports to the Internal Revenue Service, they compare survey reports of earnings to administrative reports of earnings in the sample of married couples ages 25 to 54 with at least one working spouse. They find that when the wife out-earns the husband, the gap between a husband’s survey and administrative earnings is 2.9 percentage points higher and the gap between if the wife survey and administrative earnings is 1.5 percentage points lower.

The Male Marital Wage Premium

While much of the literature on marriage and labor market outcomes studies how marriage market or spouse characteristics affect labor market outcomes, there is also a literature that studies whether simply the act of being married affects the wages of men. The descriptive finding in the literature that, conditional on observed characteristics, married men tend to earn a wage premium above unmarried men has been termed the “marriage premium.” There are several different potential explanations for this result. The first is that the wage difference is not a causal premium at all but rather an artifact of omitted variable bias, as men who select into a marriage tend to be positively selected on unobserved characteristics. The second is that there is a causal effect of marriage on male earnings that is generated by household specialization. If wives specialize in household production, this frees up husbands to focus on market production, making married men more productive in the market than unmarried men. The third is that marriage changes the costs and incentives facing men in ways that affect their workplace effort and stability and therefore their value to employers. Ribar (2004) provides an overview of mechanisms by which marriage might affect earnings, empirical issues in estimating the marriage premium, and empirical findings in the literature.

Researchers have attempted to determine the extent to which the marriage premium merely reflects positive selection into marriage through a variety of approaches. The most common is individual fixed-effects or differences analysis, which estimates within-person changes in wages in response to changes in marital status. Stratton (2002) uses the first (1987–1988) and second (1992–1994) waves of the NSFH to generate panel data estimates of the marriage premium for White men. She finds that wages increase more for men who marry between the two waves compared to men who experience no changes in marital status. However, she finds similarly positive wage growth for men who end a marriage between the two waves due to divorce, separation, or death of spouse. Rodgers and Stratton (2010) analyze NLS year 1979 data, finding that the marriage premium for White men disappears once individual fixed-effects are included in the model but that a significant marriage premium persists for African American men even controlling for individual fixed effects.

Krashinsky (2004) investigates whether married men experience higher wage growth even prior to marriage. In NLS year 1979 data, male respondents who are ever married during the years 1979 to 1994 exhibit no difference in annual wage growth during marriage compared to outside of marriage. This suggests that men who marry tend to have higher wage growth overall then men who do not marry but that this higher wage growth is not confined to years in which they are married. Loughran and Zissimopolous (2009) estimate a specification that differences out not just individual average wage levels but individual average wage growth. Using NLS year 1979 data, they find that models that condition on individual average wages find a positive marriage premium but specifications that condition on individual average wage growth indicate that marriage is associated with a decrease in wage trajectory.

Ginther and Zavodny (2001) take a somewhat novel approach to the selection problem by reasoning that “shotgun marriages,” cases in which a birth occurs soon after a marriage, are likely less selected on husband characteristics than marriages that occur at least nine months prior to a birth. If so, then the marriage premium should be smaller for men in these marriages. Their analysis with 1980 Decennial Census data and NLS Young Men Cohort data finds that the cross-sectional marriage premium estimated for husbands in shotgun marriages is only slightly smaller than that estimated for husbands in non-shotgun marriages. This suggests that the marriage premium is not driven by selection, at least not by selection into marriage beyond selection into initial partnering. They do find that the wage returns to marriage accrue quicker for non-shotgun marriages than shotgun marriages, consistent with suboptimal timing for shotgun marriages.

Antonovics and Town (2004) compare within monozygotic twins to correct for selection into marriage, using the Socioeconomic Survey of Twins to analyze a sample of twins born in Minnesota between 1936 and 1955. They find a cross-sectional wage premium of 19% in their sample, which surprisingly increases to 26% when the estimate is identified using within-twin variation in marital status. Krashinsky (2004) also conducts within-twin analysis of the male marriage premium. Analyzing data collected 1991 to 1995 at the Twinsburg Twins Festival in Twinsville, Ohio, he finds, in contrast to Antonovics and Town, that the cross-sectional marriage premium becomes small and insignificant when using within-twin comparisons.

Hersch and Stratton (2000) consider whether measures of household specialization can explain the marriage premium. They analyze changes in weekly housework hours between waves 1 (1987–1988) and 2 (1992–1994) of the NSFH data for the sample of employed, White, prime-aged men. They find that men who marry between waves 1 and 2 experience a decline in housework hours, consistent with household specialization in marriage. They also find that men who marry between waves 1 and 2 experience wage growth that is 9% greater than men with no change in marital status. However, when they include controls for housework hours in the wage regression, they find that the marriage coefficient is essentially unchanged. Therefore, to the extent that there is a marriage premium, it is not explained by household specialization within marriage that allows married men to reduce investment in household production and focus on market production.

Another hypothesis considered in the marriage premium literature is that employers may discriminate in favor of married men. But unfortunately these discussions in the literature are not always careful to distinguish between a productivity effect, in which employers value married men because marriage motivates them to exert greater effort and to seek out more stable employment, or true discrimination, in which wages differ from true productivity. There could be statistically discrimination in which employers, given that true effort and stability are hard to measure, use marriage as a signal. Statistical discrimination would imply that less productive married men would benefit from being assigned the productivity of the average married man (conditional on other observables), while more productive unmarried men would suffer from being assigned the productivity of the average unmarried men. Statistical discrimination would not, however, affect the average wage gap between married and unmarried men.

The direct causal effect of marriage on men’s workplace effort and stability is difficult to estimate. This would require time-varying measures of productivity that could be used to estimate changes in men’s productivity in unmarried and married states Korenman and Neumark (1991) analyze cross-sectional data on a sample of White, professional males working for a specific large manufacturing firm in 1976. The data includes information on worker’s job grade and his performance rating by his supervisor. The 11.9% cross-sectional difference in annual salary between married and unmarried men decreases to 2.5% once job grade indicators are included and decreases to 1.9% when performance ratings are added. Korenman and Neumark are able to rule out that differences in job grade are generated by differences in hiring but instead find that the concentration of married men in higher-paying grades is driven by higher rates of promotion, which are in turn explained by their higher job performance ratings.

Loughran and Zissimopolous (2009) point out that the marriage premium literature largely focuses on the effect of marriage on male wages, while the literature on female wages focuses primarily on the effects of childbearing. They find that marriage is associated with lower wage growth for women but that these effects are muted for women who delay childbearing after marriage or never have children. This suggests that this negative marriage premium for women may be primarily driven by anticipation of future childbearing.

Same-Sex Couples

Almost all existing research on marriage and labor market outcomes is conducted on samples of heterosexual married couples. This raises the question of the extent to which results in the existing literature apply to same-sex couples. Black, Sanders, and Taylor (2007) document in 2000 Decennial Census data that partnered same-sex couples are much less likely to engage in complete specialization, in which one member stays at home, than heterosexual couples.

Antecol and Steinberger (2013) investigate the extent to which the labor supply gap between partnered lesbian women and married heterosexual women can be explained by differences in the presence of children and differential responsiveness to children. They argue that it is important to differentiate between partnered lesbian women who are primary earners and partnered lesbian women who are secondary earners. To the extent that most married heterosexual women are most often secondary earners, comparisons to primary earner partnered lesbian women may not be appropriate.

Using 2000 Decennial Census data, Antecol and Steinberger (2013) compare married heterosexual women to partnered lesbian women in the sample of non-Hispanic White women ages 25 to 54. In this sample they first confirm that partnered lesbian women have higher participation rates and hours of work than married heterosexual women. They then categorize partnered lesbian women as either primary earners or secondary earners using two different decision rules: one based on which partner is reported as the householder and the other based on which partner has the higher earnings. They find that the labor supply of secondary earner partnered lesbians if more similar to that of married heterosexual women. They also find that the responsiveness of labor supply to the presence of children for secondary earner partnered lesbians is more similar to that of married heterosexual women. Part of the gap in labor supply between partnered lesbian women and married heterosexual women is due to the fact that the labor supply of primary earner partnered lesbian women is, similar to married men, less responsive to the presence of children.

Oreffice (2011) estimates a labor supply model with household bargaining for partnered same-sex couples in the 2000 Decennial Census. She uses the age gap and non-labor income gap as measures of bargaining power. Consistent with household bargaining, she finds a five-year increase in the age gap is associated, among partnered lesbian women, with a 21-hour decrease in annual hours for the younger partner and a 20-hour increase in annual hours for the older partner and, among partnered gay men, a 22-hour decrease for the younger partner and a 23-hour increase for the older partner. Estimates for the gap in non-labor income display similar patterns.

Challenges for Future Research

Research on marriage and labor market outcomes continues to face a number of empirical challenges. Estimating the effect of marriage or spouse characteristics on outcomes is difficult because marital status and spouse characteristics are endogenously chosen. For example, if high- wage men typically marry women who are themselves high skilled with high labor force attachment, this will generate a positive bias in estimates of the cross-wage elasticity. Increases in positive assortative matching over time will result in less negative estimates of the cross-wage elasticity. Studies that use exogenous shocks to marriage markets to study how the changes in expected spousal quality affect outcomes avoid endogenously chosen marital status and spouse characteristics, but there are fewer settings in which such exogenous shocks are available, and in some cases, these shocks also affect labor market outcomes directly.

Prior literature suggests that married women experience more negative consequences of household migration than their husbands and that married women experience negative consequences of outearning their spouse that married men do not. There remains an active debate whether such asymmetries are consistent with a standard neoclassical model or reflect some additional influence of cultural gender norms. Research that is designed to better distinguish between these competing explanations is still needed.

Finally, household structure is increasingly complex with changes in marriage, divorce, remarriage, cohabitation, and same-sex couples and spouses. There is room for more research on the associations between relationship status, partner characteristics, and labor market outcomes that considers more richly the full variety of household structures.

Further Reading

Bertrand, M., Kamenica, E., & Pan, J. (2015). Gender identity and relative income within households. The Quarterly Journal of Economics, 130(2), 571–614.Find this resource:

Blau, F. D., & Kahn, L. M. (2007). Changes in the labor supply behavior of married women: 1980–2000. Journal of Labor Economics, 25(3), 393–438.Find this resource:

Blinder, A. J., & Lam. D. (2018). Is there a male breadwinner norm? The hazards of inferring preferences from marriage market outcomes. IZA Discussion Paper No. 11693. Bonn, Germany: IZA Institute of Labor Economics.Find this resource:

Bursztyn, L., Fujiwara, T., & Pallais, A. (2017). “Acting wife”: Marriage market incentives and labor market investments. The American Economics Review, 107(11), 3288–3319.Find this resource:

Charles, K. K., & Luoh, M. L. (2010). Male incarceration, the marriage market and female outcomes. The Review of Economics and Statistics, 92(3), 614–627.Find this resource:

Chiappori, P.-A., Fortin, B., & Lacroix, G. (2002). Marriage market, divorce legislation, and household labor supply. Journal of Political Economy, 110(1), 37–72.Find this resource:

Ge, S., Isaac, E., & Miller, A. (2018). Elite schools and opting-in: Effects of college selectivity on career and family outcomes. NBER Working Paper No. 25315. Cambridge, MA: National Bureau of Economic Research.Find this resource:

Lafortune, J. (2013). Making yourself attractive: Pre-marital investments and the returns to education in the marriage market. American Economic Journal: Applied Economics, 5(2), 151–178.Find this resource:

Ribar, D. C. (2004). What do social scientists know about the benefits of marriage? A review of quantitative methodologies. IZA Discussion Paper No. 998. Bonn, Germany: IZA Institute of Labor Economics.Find this resource:

Stephens, M., Jr. (2002). Worker displacement and the added worker effect. Journal of Labor Economics, 20(3), 504–537.Find this resource:

References

Angrist, J. (2002). How do sex ratios affect marriage and labor markets? Evidence from America’s second generation. The Quarterly Journal of Economics, 117(3), 997–1038.Find this resource:

Antecol, H., & Steinberger, M. D. (2013). Labor supply differences between married heterosexual women and partnered lesbians: A semi-parametric decomposition approach. Economic Inquiry, 51(1), 783–805.Find this resource:

Antonovics, K., & Town, R. (2004). Are all the good men married? Uncovering the sources of the marital wage premium. The American Economic Review, 94(2), 317–321.Find this resource:

Bertrand, M., Kamenica, E., & Pan, J. (2015). Gender identity and relative income within households. The Quarterly Journal of Economics, 130(2), 571–614.Find this resource:

Black, D. A., Sanders, S. G., & Taylor, L. J. (2007). The economics of lesbian and gay families. The Journal of Economic Perspectives, 21(2), 53–71.Find this resource:

Black, D. A., & Smith, J. A. (2004). How robust is the evidence of the effects of college quality? Evidence from matching. Journal of Econometrics, 121, 99–124.Find this resource:

Blau, F. D., & Kahn, L. M. (2007). Changes in the labor supply behavior of married women: 1980–2000. Journal of Labor Economics, 25(3), 393–438.Find this resource:

Blinder, A. J., & Lam, D. (2018). Is there a male breadwinner norm? The hazards of inferring preferences from marriage market outcomes. IZA Discussion Paper No. 11693. Bonn, Germany: IZA Institute of Labor Economics.Find this resource:

Blundell, R., Duncan, A., & Meghir, C. (1998). Estimating labor supply responses using tax reforms. Econometrica, 66(4), 827–861.Find this resource:

Bursztyn, L., Fujiwara, T., & Pallais, A. (2017). “Acting wife”: Marriage market incentives and labor market investments. The American Economics Review, 107(11), 3288–3319.Find this resource:

Cha, Y. (2010). Reinforcing separate spheres: The effect of spousal overwork on men’s and women’s employment in dual-earner households. American Sociological Review, 75(2), 303–329.Find this resource:

Charles, K. K., & Luoh, M. L. (2010). Male incarceration, the marriage market and female outcomes. The Review of Economics and Statistics, 92(3), 614–627.Find this resource:

Chiappori, P.-A., Fortin, B., & Lacroix, G. (2002). Marriage market, divorce legislation, and household labor supply. Journal of Political Economy, 110(1), 37–72.Find this resource:

Cullen, J. B., & Gruber, J. (2000). Does unemployment insurance crowd out spousal labor supply? Journal of Labor Economics, 18(3), 546–572.Find this resource:

Dale, S. B., & Krueger, A. B. (2002). Estimating the payoff to attending a more selective college: An application of selection on observables and unobservables. The Quarterly Journal of Economics, 117(4), 1491–1527.Find this resource:

Devereux, P. J. (2004). Changes in relative wages and family labor supply. The Journal of Human Resources, 39(3), 696–722.Find this resource:

Eriksson, K. H., & Stenberg, A. (2015). Gender identity and relative income within households: Evidence from Sweden. IZA Discussion Paper No. 9533. Bonn, Germany: IZA Institute of Labor Economics.Find this resource:

Foster, G., & Stratton, L. S. (2018). Does female breadwinning make partnerships less healthy or less stable? IZA Discussion Paper No. 11938. Bonn, Germany: IZA Institute of Labor Economics.Find this resource:

Ge, S., Isaac, E., & Miller, A. (2018). Elite schools and opting-in: Effects of college selectivity on career and family outcomes. NBER Working Paper No. 25315. Cambridge, MA: National Bureau of Economics Research.Find this resource:

Ginther, D. K., & Zavodny M. (2001). Is the male marriage premium due to selection? The effect of shotgun marriages on the return to marriage. Journal of Population Economics, 14(2), 313–328.Find this resource:

Grossbard, S., & Amuedo-Dorantes, C. (2007). Cohort-level sex ratio effects on women’s labor force participation. Review of Economics of the Household, 5(3), 249–278.Find this resource:

Heim, B. T. (2007). The incredible shrinking elasticity: Married female labor supply, 1978–2002. The Journal of Human Resources, 42(2), 881–918.Find this resource:

Hersch, J., & Stratton L. S. (2000). Household specialization and the male marriage wage premium. Industrial and Labor Relations Review, 54(1), 78–94.Find this resource:

Korenman, S., & Neumark, D. (1991). Does marriage really make men more productive? The Journal of Human Resources, 26(2), 282–307.Find this resource:

Krashinsky, H. A. (2004). Do marital status and computer usage really change the wage structure? The Journal of Human Resources, 39(3), 774–791.Find this resource:

Lafortune, J. (2013). Making yourself attractive: Pre-marital investments and the returns to education in the marriage market. American Economic Journal: Applied Economics, 5(2), 151–178.Find this resource:

Loughran, D. S., & Zissimopoulos, J. M. (2009). Why wait? The effect of marriage and childbearing on the wages of men and women. The Journal of Human Resources, 44(2), 326–349.Find this resource:

McKinnish, T. (2008). Spousal mobility and earnings. Demography, 45(4), 829–849.Find this resource:

McKinnish, T. (2019). Overwork in spouse’s degree field and the labor market outcomes of skilled women. IZA Discussion Paper No. 12149. Bonn, Germany: IZA Institute of Labor Economics.Find this resource:

Murray-Close, M., & Heggeness, M. L. (2018). Manning up and womaning down: How husbands and wives report their earnings when she earns more. U.S Census Bureau Social, Economic, and Housing Statistics Division Working Paper No. 2018–20. Washington, DC: U.S. Census Bureau.Find this resource:

Negrusa, B., & Oreffice, S. (2010). Quality of available mates, education, and household labor supply. Economic Inquiry, 48(3), 558–574.Find this resource:

Oreffice, S. (2007). Did the legalization of abortion increase women’s household bargaining power? Evidence from labor supply. Review of Economics of the Household, 5(2), 181–207.Find this resource:

Oreffice, S. (2011). Sexual orientation and household decision making: Same-sex couples’ balance of power and labor supply choices. Labour Economics, 18(2), 145–158.Find this resource:

Pencavel, J. (1998). The market work behavior and wages of women: 1975–94. The Journal of Human Resources, 33(4), 771–804.Find this resource:

Ribar, D. C. (2004). What do social scientists know about the benefits of marriage? A review of quantitative methodologies. IZA Discussion Paper No. 998. Bonn, Germany: IZA Institute of Labor Economics.Find this resource:

Rodgers, W. M. III, & Stratton, L. S. (2010). Male marital wage differentials: Training, personal characteristics and fixed effects. The Journal of Human Resources, 48(3), 722–742.Find this resource:

Sprengholz, M., & Wieber, A. (2019). Gender identity and wives’ labor market outcomes in West and East Germany between 1984 and 2016. IZA Discussion Paper No. 12284. Bonn, Germany: IZA Institute of Labor Economics.Find this resource:

Stephens, M., Jr. (2002). Worker displacement and the added worker effect. Journal of Labor Economics, 20(3), 504–537.Find this resource:

Stratton, L. S. (2002). Examining the wage differential for married and cohabiting men. Economic Inquiry, 40(2), 199–212.Find this resource: