Marriage markets as explanation for why heavier people work more hours

2.1 Marriage market mechanisms

Labor supply may be a function of body weight via the effect of body weight on two outcomes related to marriage markets. First, the literature reviewed in the previous section suggests that high-BMI people will be less likely to be married and they are less likely to be married to higher income spouses than people who are low-BMI. In heterosexual marriage markets, individuals with higher body weight may be less in demand on the part of members of the other sex and thus less likely to become part of a couple.

Second, lower demand in marriage markets may also lead to lower price. According to Becker (1973, 1981), to the extent that prices are established in multiple interrelated hedonic marriage markets, lower prices for married men or women may take the form of less access to consumption goods. This association between price in marriage market and access to consumption also follows from other theories of marriage such as McElroy and Horney (1981), Grossbard-Shechtman (1984), and Chiappori (1992). Becker (1973), McElroy and Horney (1981), and Chiappori (1992) assume that couples pool their resources in marriage and then decide on how to allocate the resources to individual consumption. In contrast, Grossbard-Shechtman (1984) assumes that individuals keep control of their time and personal income whether married or not.³ Consequently, prices in marriage markets may explain in-marriage financial transfers from one spouse to the other. The higher a person’s price in the marriage market, the higher the intra-household financial transfers they are likely to obtain from a spouse or the lower the financial transfers they are likely to make to a spouse (depending on whether they are a net financial beneficiary or a net financial contributor in the marriage). In turn, the more access a married person has to the spouse’s financial resources, the less they are likely to supply labor to the labor market (income effect).

Price in marriage is related to the (pre-transfer) income of a spouse. The higher one’s income the more one can afford to spend on a spouse. The transfer obtained in marriage (and the transfer a single person expects to obtain after marriage) are functions of both the spouse’s income and the spouse’s willingness to transfer some of that income in the marriage. Likewise, the price one pays for marrying a particular partner will be related to own income and own willingness to transfer some of that income to a spouse. Furthermore, single individuals planning to obtain a particular price in marriage may consider both a potential spouse’s income and their willingness to transfer as elements in their choice of mate. In the same vein, singles planning to pay a price in marriage will be influenced by their own income and prices in marriage markets.

High BMI is a factor related to relative demand in marriage markets. To the extent that they are in lower demand, high-BMI women (men) will command a lower price in marriage markets relative to that of low-BMI women (men) and consequently will be likely to obtain less personal access to consumption. Furthermore, they may either obtain fewer financial transfers from a spouse or may have to transfer more to a spouse once they are married. Consequently, we hypothesize that high-BMI married individuals will work more in the labor market than low-BMI people.

Spouses of high-BMI individuals are also likely to have a lower income than those married to low-BMI people. As for high-BMI people who are paying a price in marriage, they are likely to pay a higher price and may need more income compared to their low-BMI counterparts (Chiappori 1992, Oreffice and Quintana-Domeque 2012). Therefore, they may need to work more hours. High-BMI people may choose to marry individuals with attributes associated with lower prices (including higher weight, older age, and lower education) in order to avoid excessive financial obligations. This discussion suggests that BMI may also affect the hours of work of single people in the labor market. Hours of work in the labor market will vary with body weight regardless of marital status, to the extent that value in marriage markets affects both those who are already “employed” in marriage (or employing someone else) and those who are single and looking for a spouse. (This is similar to the idea that low wages affect the allocation of time of both the employed and the unemployed looking for a job.) In addition, to the extent that higher body weight has a negative effect on single individuals’ probability of marriage, this entails lower expected future access to spousal income as well. This is also likely to increase their hours of work prior to marriage. We thus hypothesize a positive effect of BMI on individual hours of work by both married and single individuals. In other words, effect of BMI would be similar in single and married women.

Body weight is likely to affect women’s hours of work more than men’s to the extent that women’s marriage prospects vary more with body weight than men’s (Mukhopadhyay 2008) and that intra-marriage income transfers are more likely to flow from men to women than in the opposite direction. The latter gender difference is related to the observation that relative to married men, married women tend to earn less in the labor force and to do more household work, and to the idea that in-marriage income transfers may partially be compensations for household production work (Grossbard 2015).

Effects of body weight on hours of work may differ by ethnic group. Lin et al. (2016) show that a relative worsening of Black women’s marriage market conditions helps explain changes in the racial gap in female obesity. Black women’s price in marriage may be lower (see Grossbard et al. 2014), possibly due to discrimination against dark skin (Goldsmith et al. 2007). It is also possible that on average Black women pay the price of marriage to men (Cherry 1998). This may have implications for marriage market effects of body weight. Averett and Korenman (1999) found that self-esteem is associated with body weight for White women but not for Black women. To the extent that the premium for being thin is higher in marriage markets for White women than in those for Black women, one expects more effects of body weight on intra-marriage transfers and thus on women’s hours of work for White women relative to those effects for Black women.

2.2 Labor market mechanisms

It is well-established (Cawley 2004 among others) that individuals (particularly White women) with higher BMI have lower wages compared to their lower-BMI counterparts. This can generate a substitution effect and an income effect. Therefore, theoretically, the effect of BMI on market hours of work, via lower market wages, is ambiguous. At the same time, most of the empirical literature suggests that women’s labor supply elasticity is positive (Blundell and MaCurdy 1999). In other words, a lower market wage (due to higher BMI) would induce women to work less. Therefore, this particular labor market mechanism and the marriage market mechanisms we mentioned above have different predicted effects regarding the association between BMI and hours worked.

There may be other ways that the labor market can affect the relationship between BMI and hours of work. If high-BMI workers have more health limitations, they may work less. However, if high-BMI workers anticipate a shorter working life, then they may work more during the early part of their life to compensate for an earlier exit from the labor force. Another possible factor may be health insurance. To the extent high-BMI people value insurance more than low-BMI people do (presumably, because high-BMI people have or expect more health problems), they have more of an incentive to get health insurance and they may work more hours to qualify for an employer-provided insurance. Employers typically provide health insurance to full-time employees, and the Bureau of Labor Statistics defines full-time workers as someone working for more than 34 h per week. (However, the Affordable Care Act of 2009 defines a full-time employee as someone working more than 30 h per week.)

3.1 Data

We use data from two cohorts of National Longitudinal Survey of Youth (NLSY): the 1979 cohort (NLSY79) and the 1997 cohort (NLSY97). The NLSY79 started interviewing 12,686 respondents in 1979 when they were 14–22 years old. These individuals were interviewed every year until 1994 and after that on a biannual basis. The most recent wave (2014) included 9964 of the original respondents. The NLSY97 cohort started with 8984 respondents who were between the ages of 12 and 17 when they were first interviewed in 1997. Since then, they have been interviewed annually until 2011 and biannually thereafter. The most recent round of interviews (2013) included 7141 of the original respondents.

NLSY contains information about height, weight, wages, marital status, and a plethora of other information about the respondents, including our principal outcome of interest: total hours worked by a respondent during the year before interview. Since NLSY interviews all eligible youths in an eligible household, we know the relevant information for the siblings of respondents as well, if they were also eligible for interview.

In our baseline analysis, we restrict our attention to adult respondents (≥ 18 years) who have a same-sex sibling and for whom all relevant variables are available. In the case of women, they are included if they have not been pregnant during the year before an interview, given that pregnancy can cause weight changes irrelevant to our analysis. We discard observations with weight below 70 lbs or above 400 lbs. We also discard observations with height below 4 ft or above 8 ft. We exclude data from marriages that end up in divorce, since our focus is on stable marriages. In our hours worked regressions, following the previous literature, we exclude respondents who worked 250 h or less in the previous year. If they are married, we exclude observations when the spouse of the respondent reported less than $5000 in annual income.

We use self-reported height and weight to compute BMI. In our data (both genders and all races combined), the 1st percentile of the BMI distribution is 17.2 and the 99th percentile is 45.5 (the 95th percentile is 39.7). These numbers are consistent with Sturm and Hattori (2013) who reported that 6.6% of US adults have a BMI of 40 or above. Then we Winsorize⁴ the BMI distribution at 1% to reduce the effect of outliers. Our decision to Winsorize the BMI distribution (as opposed to dropping the extreme observations) is driven by two factors. First, a sizeable fraction (6.6%) of the US adult population have a BMI of 40 or above, and second, hours of work for respondents between a BMI of 30 and 40 and those with BMI above 40 are not significantly different (1865 h/year vs. 1896 h/year in our sample).⁵ We also Winsorize (at 1%) hourly wages of respondents to reduce the effects of outliers.

Summary statistics for Black and Hispanic single women are presented in columns 2 and 3 respectively. Single Black (Hispanic) women work on average about 1667 (1597) hours per year. The average wage for single Black (Hispanic) women is $9.79/h ($10.44/h). The average BMI for single Black (Hispanic) women is 27.5 (25.4). In both cases, the average BMI falls in the “overweight” category. Single Black and Hispanic women have less education and are more likely to have children relative to single White women.

The last three columns of Table 1 (panel A) present summary statistics for married women. Average hours of work per year are similar for single and married White women. Married Black and Hispanic women work more than their single counterparts do. Average age in the married White women sample is 34.8 years, and as expected, they are more likely to have children (74%) than their single counterparts. They also have higher BMI (25.08). We also use spousal characteristics in our analysis. Unfortunately, NLSY does not have information on the BMI of spouses, but it does include information about spouses’ age, education, and annual income. For married White women, spouses’ average age is 36.8 years and they have 13.8 years of education. Average annual earnings is $57,000/year.

Table 1 (panel B) presents the summary statistics and sample sizes for variables not used throughout the analysis. In NLSY, some questions were not asked in each round, or they are not consistent across surveys (NLSY79 vs. NLSY97). An example of the former is whether the employer of the respondent offers health insurance as a benefit. In reply, 59% (80%) of single (married) White women answered yes to this question. An example of the latter is whether a woman believes in traditional gender roles. The NLSY79 respondents were asked whether they agree or not with the following statement: “A woman's place is in the home, not in the office or shop.” If the answer was “strongly agree” or “agree,” then the variable was coded as one and zero otherwise. We refer to this variable as “trad_NLSY79.” In those cases, we use subsamples of the abovementioned samples for our analysis. About 9% (8%) of single (married) White NLSY79 women replied that they believe in traditional gender roles.

Unfortunately, this question was not asked in the NLSY97 cohort. However, the NLSY97 respondents were asked whether they agree or disagree with the following statement: “I support long-established rules and traditions.” If they answered “Agree a little,” “Agree moderately,” or “Agree strongly,” then the variable was coded as one and zero otherwise. We refer to this variable as “Trad_NLSY97.” Trad_NLSY79 was coded as zero for all NLSY97 respondents and vice versa. Using this definition, about 23% (26%) of single (married) White NLSY97 respondents replied that they support rules and traditions.

In NLSY79, respondents were asked whether the respondent has any health issues that limits their ability to work. About 4% (3%) of single (married) White NLSY79 women replied that they have a health issue that limits their ability to work. In NLSY97, respondents were asked about their general health. The possible answers were Excellent, Very good, Good, Fair, or Poor. About 5% (4%) of single (married) White NLSY97 women replied that they are in fair or poor health.

We also use a unique question about marriage expectations that was asked to the single NLSY97 respondents in some of the waves. The respondents were asked “Now think about five years from now, you will be [{AGE IN 5YRS}]. What is the percent chance that you will be married?” This question was not asked in the NLSY79, and therefore, this analysis can only be performed on NLSY97 women. Answers indicate that single White women in our sample expect that the chance that they will be married within next 5 years is 54.9%.

3.2 Methods

We use three types of regression methods: ordinary least squares (OLS), IV, and sibling fixed effect (FE) to establish a relation between BMI and hours of work. First, we use OLS regressions to show the association between BMI and hours of work. To decipher whether the relation between BMI and hours of work is driven by labor markets or marriage markets, we run regressions that either include market wage or not. Wage tends to be lower for high-BMI people (Cawley 2004). Own wage effects can generate both a substitution effect and an income effect on hours of work. If wage is included in the regression, we can expect that the entire effect of body weight operates via marriage markets. In contrast, if own wage is not controlled for, body weight can affect hours of work via marriage market effects as well as wage.

Then, to establish causality, we pursue two strategies: IV regression and sibling FE regression. In our IV regression models, we use sibling BMI as an instrument for own BMI in line with the work of Cawley (2004). OLS estimates may not be consistent in this context. For example, there may be unobserved traits (such as value of leisure) that affect both BMI and hours of work. For example, individuals who value leisure more may work less and have higher BMI. Another potential problem is reverse causality. Individuals who choose sedentary types of jobs may have high BMI over time (Lakdawalla and Philipson 2007). We use sibling BMI as an instrument to address the endogeneity problem. While sibling BMI has been used as an instrument for own BMI in the past (e.g., by Cawley 2004), there are some concerns about its validity because sibling BMI represents the household environment that both the respondent and her sibling shared. However, current evidence suggests that the role of shared environment may not be very important, especially given that the association between non-biological siblings’ BMI (i.e., the correlation between an individual and her step/adopted siblings) is insignificant (Cawley 2004; Lindeboom et al. 2010; Cawley and Meyerhoefer 2012; Cawley 2015). Cawley and Meyerhoefer (2012) provides an extended discussion on this issue, including this citation: “Contrary to widespread assumptions about the influence of the family environment, living in the same home in childhood appears to confer little similarity in adult BMI beyond that expected from the degree of genetic resemblance.” (Wardle et al. 2008, p. 398).

As an alternative identification strategy, we use sibling FE regressions.⁷ Sibling FE strategy has been used by Averett and Korenman (1996) and Baum and Ford (2004) to establish a causal relation between BMI and wage. We are not aware of any studies that have used this strategy to establish causality between BMI and hours of work. One advantage of the sibling FE strategy over IV is that identification in a sibling FE model does not require an exclusion restriction (which is not testable). However, a sibling FE strategy cannot address reverse causality (Lakdawalla and Philipson 2007). In addition, given relatively high correlation between siblings’ BMI, any measurement error in BMI (possibly due to reporting error) is likely to be magnified during differencing. This limitation of FE models that has been noted in other contexts (e.g., by Deaton 1995) applies to our setting as well. Each method thus has its advantages and disadvantages.

We check whether our results are robust to changes in sample construction criteria, functional form assumptions, and methods. Our robustness tests include tests for non-linearity of the relationship between hours of work and BMI. A number of papers (e.g., Kline and Tobias 2008; Gregory and Ruhm 2011; Caliendo and Gehrsitz 2016) have found evidence of non-linearity in the relationship between wage and BMI. Therefore, at the end of Section 4, we test whether our results are robust to a semiparametric specification where BMI enters the wage equation non-parametrically.

We also estimate unconditional quantile regressions (UQR). This method, developed by Firpo et al. (2009), allows us to test whether the effect of BMI varies across the unconditional hours of work distribution.⁸ In addition, UQR can help determine whether the positive association between BMI and hours of work is related to a health insurance motive. If that were the case, we are likely to see a stronger association for people who are right below the cutoff of health insurance eligibility (which is about 30 h/week or about 1500 h/year). In contrast, if the association between is driven by marriage markets, we are likely to see this association at all levels of hours of work.

4.1 Baseline analysis: single and married respondents

4.2 Are marriage markets driving the relation between BMI and hours of work in women?

4.2.2 Single women

Our results so far suggest that the association between BMI and hours of work in single women stems from how BMI affects expectations about marriage. For single women, we cannot pursue the direct strategy employed in Table 5 above. Instead, we investigate the rational expectations argument presented above by performing two more tests: (1) we use a unique question about marriage expectations and (2) we investigate how the association between BMI and single women’s hours of work changes when we control for predicted cumulative spousal income.

First, we use a question that was asked to the single NLSY97 respondents in some of the waves (2000, 2001, 2009, 2010, and 2011). The respondents were asked “Now think about five years from now, you will be [{AGE IN 5YRS}]. What is the percent chance that you will be married?” This question was not asked in the NLSY79, and therefore, this analysis can only be performed on NLSY97 women. As we discussed earlier, our hypothesis is that single women with high BMI expect smaller future income transfer from husbands, and therefore, they work more in the labor market even when they are single. It follows that the effect of BMI will be stronger in single women with higher self-assessed probability of getting married in the next 5 years, compared to their counterparts who have a lower self-assessed probability of getting married in the next 5 years. In a different but relevant context, Kureishi and Wakabayashi (2013) showed that marriage expectations could affect single women’s savings decisions: they found that single women who expect to get married save significantly less than single women who do not expect to get married in the next 3 years.

Accordingly, we re-estimate the OLS regressions for single women, controlling for probability of marriage. We add an interaction term between the self-assessed probability of marriage and BMI. This regression has the same set of control variables (including log wage) as in Table 3. However, sample sizes are smaller given that we only used the NLSY97 and that only selected waves of NLSY97 included the question about marriage expectations.

Figure 1 shows the marginal effect of BMI for single women and how it varies with the self-assessed probability of marriage, along with the 95% confidence interval. Panel A suggests that for single White women, the marginal effect of BMI on hours of work is insignificant when the expected probability of marriage is zero. The association between BMI and hours of work remains statistically insignificant as long the chance of getting married remains below 40%. However, the marginal effect is positive and significant for single White women when the chance of getting married in the next 5 years is at 50% or above. Panel B (panel C) shows the marginal effect of BMI for Black (Hispanic) single women. Results here are never statistically significant.

Open image in new window

Next, we investigate how including predicted cumulative spousal income affects the association between BMI and single women’s hours of work. To do this, first, we compute the cumulative spousal income of all NLSY79 women at the time of the last survey, when all women were 50 or older. Cumulative spousal income depends on the annual income of the spouse and number of years the woman was married. If a woman was never married by the time of the last interview, then her cumulative spousal income is zero. We regress cumulative spousal income on the characteristics of women including their BMIs. Results (not shown) suggest that for White women, a one-unit increase in BMI is associated with $11,610 less in cumulative spousal income,¹³ but we do not find any association for Black women. We then use the estimated coefficients to obtain the predicted cumulative spousal income of single women. Then we run regressions for single women in the entire NLSY data sets, with and without including this predicted cumulative spousal income.

Table 6 presents OLS regressions for single women. All regressions include all the controls used in Table 3, including their own wage. Thus, columns 1, 3, and 5 in this table reproduce the results presented in columns 2, 4, and 6 of Table 3 (panel A). In the regression results reported in columns 2, 4, and 6 of Table 6, we have added predicted cumulative spousal income as an additional control. Results show that for White single women, the coefficient of BMI goes down from 0.00581 to 0.00373 when we add predicted cumulative spousal income as an additional control. This suggests that one of the mechanisms by which BMI influences single White women’s hours of work is via BMI’s effect on predicted spousal income. However, this is not the case for Black single women: here, the effect of BMI does not seem to act via its effect on predicted spousal income: the coefficients of BMI in columns 3 and 4 of Table 6 are very similar. In the case of Hispanic women, we find no association in both columns 5 and 6.

Table 6

BMI and log hours of work for single women: with and without predicted cumulative spousal income

	White	White	Black	Black	Hispanic	Hispanic
BMI	0.00581*** (3.246)	0.00373* (1.825)	0.00682*** (4.496)	0.00634*** (4.086)	−0.00033 (−0.105)	−0.00059 (−0.182)
Control for own wage	Yes	Yes	Yes	Yes	Yes	Yes
Control for predicted cumulative spousal income	No	Yes	No	Yes	No	Yes
Observations	6469	6469	4671	4671	2020	2020

Note 1: Control variables include work experience (quadratic), educational categories, dummies for whether the woman believes in traditional gender roles, whether the woman has any children, if the youngest child is below six, yearly age dummies, region of residence dummies, and year dummies. Note 2: All t-stats reported are based on standard errors clustered at the individual level. *** p < 0.01, ** p < 0.05, and * p < 0.10

Adding up the results reported in Fig. 1 and Table 6, it appears that one reason why higher-BMI White single women work more in the labor market relative to their lower-BMI counterparts is that they expect lower future in-marriage income transfers. In turn, these expected future in-marriage transfers are lower because either heavier women are likely to get married to men with fewer resources or heavier women have less access to spousal resources¹⁴ (i.e., a lower bargaining power). It seems that in the case of White women, both of these channels operate.

4.3 Can labor market mechanisms explain the relation between BMI and hours of work in single women?

In our Section 2, we discussed two additional labor market-based mechanisms that can explain the positive association between BMI and hours of work in single women. First, is it possible that high-BMI women are less healthy, and unhealthy women expect to spend fewer years in the labor force at older ages and therefore they work more hours at an early age to compensate for a shorter expected working span. Since the single sample is substantially younger than the married sample, we are most likely to observe this in the single sample. A second possibility is that high-BMI singles are more likely to work full time in order to qualify for health insurance benefits.¹⁵

4.4 Robustness checks

We perform a number of robustness checks. First, all results reported so far were estimated for samples excluding respondents without siblings. Therefore, we re-estimate OLS regressions in Tables 2 and 3 for full samples, including those without any siblings. The results (presented in Table 10 in the Appendix) are qualitatively similar to those reported in panel A of Tables 2 and 3.

Second, we experiment with alternative assumptions about extreme values of BMI. As we discussed in Section 3, Winsorizing BMI (at 1%) involves replacing all BMI below 17.2 by 17.2 and replacing all BMI above 45.5 by 45.5. Here, we follow two alternative strategies: first, we ignore all BMIs that are below 17.2 or above 45.5, and second, we ignore all BMIs that are below 18.5 or above 40.0 (a strategy followed by OQD). Table 11 in the Appendix shows results using our three methods of estimation (OLS, IV, and sibling FE) when using these two alternative strategies, along with our baseline results. The results show that our results are robust to these alternative approaches.

Third, we check whether the association between BMI and hours of work varies across the hours of work distribution by using the UQR method described in Section 3. To the extent unobserved characteristics are responsible for different women choosing different hours of work, UQR can be a check on whether the OLS results are driven by this type of unobservable factors. An added benefit of using UQR is that it can help us infer whether health insurance motive is driving the association between BMI and hours of work. If the health insurance motive is the driving factor behind this association, then we are likely to see the effect in people who are right below the cutoff of health insurance eligibility (which is about 30 h/week or about 1500 h/year). The 1500 h/year is between the 40th and 50th percentile, depending on the race/marital status group. In contrast, if the marriage market is the driving force behind the association between BMI and hours of work, then we are likely to see an effect at all levels of hours of work. Figure 2 shows the coefficient of BMI and how it varies across the unconditional log (BMI) distribution for White women. The left panel is for single women, and the right panel is for married women. Figure 2 shows that the effect of BMI is positive and significant over most of the distribution for both single and married White women. The coefficient of BMI seems to be highest well before the 40th percentile for both single and married White women. This suggests that the health insurance motive is unlikely to explain the pattern represented in Fig. 2. Figures 3 and 4 present the corresponding results for Black and Hispanic women. Consistent with our OLS results, we find that the coefficient of BMI is positive and significant for single Black women over most of the distribution, while it is never significant for married Black women or for Hispanic women.

Open image in new window

Open image in new window

Open image in new window

Fourth, in our analyses so far, we assumed that the relation between BMI and hours worked is linear. This assumption may not be accurate, and if the true relationship is non-linear, this may introduce bias in our estimates. To address this issue, we estimate a partially linear model where BMI enters the “hours worked” equation non-parametrically. We use Robinson’s (1988) double residual estimator.¹⁷ Figure 5 presents the results from semiparametric regressions in which we treat BMI as an exogenous variable. The left panel shows the results for single women and the right panel for married women. To show the difference in the semiparametric fit across races, we do not include the 95% confidence interval in this figure. The left panel in Fig. 5 indicates that the relationship is broadly linear (and monotonic) until a BMI of 40 in the case of both White and Black single women. Beyond that, an increase in BMI is associated with a decline in hours worked, most likely because the health effects dominate other considerations. The semiparametric estimates for married White women (right panel of Fig. 5) suggest more evidence of non-linearity. Even in this case, the estimated increase in hours of work when BMI doubles from 20 to 40 is not statistically different from the OLS estimate.

Open image in new window

Fifth, following a common practice in this literature, we created dummies for weight categories, to replace the linear BMI variable. We followed the literature and created four weight categories: underweight (BMI < 18.5), healthy weight (18.5 ≤ BMI < 25), overweight (25 ≤ BMI < 30), and obese (BMI ≥ 30). Then we estimated OLS regressions with weight categories as our independent variables of interest, with healthy weight as the omitted category and controlling for wage. The results from these regressions are presented in Table 12 in the Appendix. Results suggest White and Black women who are either overweight or obese work more hours than their healthy weight counterparts. This holds for single and married women. As for men, the overweight single men work more hours, but the obese do not.¹⁸

Finally, so far we have focused on hours of work as our outcome variable and ignored the extensive margin or labor supply. It is also conceivable that BMI affects the labor-force participation decision. The estimates reported in Table 13 in the Appendix are from OLS and IV regressions. We tried to estimate the corresponding probit and IV-probit models, but the IV-probit likelihood does not converge in two out of six cases (single Black women and married Hispanic women). Since Angrist and Pischke (2009) argue that “IV methods capture local average treatment effects regardless of whether the dependent variable is binary, non-negative, or continuously distributed on the real line,” we report the estimates from linear IV models. Results shown in Table 13 in the Appendix suggest that BMI increases probability of employment in single White women (using OLS) and in married White women (using IV). We do not find any consistent pattern of relationship between BMI and employment probability.