We examine the impact of a decline in the ratio of men to women on women's labor market outcomes. We do this for Mexico, where the sex and age imbalance of large-scale international migration has led to a significant decline in the male/female ratio. For example, according to Mexican census data, in the year 2000 the aggregate ratio of men to women 18–25 years old was 0.91, and for some states it was as low as 0.8. Given that the natural ratio of men to women is approximately 1:1 (Sen, 1990), this translates into 1–2 fewer men per 10 women than in contexts, such as in the United States or Brazil, with balanced sex ratios (see Figure 1).
Sex ratios. Panel A: By census year and age group, Mexico. Panel B: Comparison of Mexico, United States, and Brazil.
The specific question we raise is whether the relative scarcity of men results in Mexican women entering high-skilled and better paying jobs over time. This question is relevant as there has been an increase in women's education and labor force participation across the developing world, but less evidence of improvements in the gender wage gap (Heath and Jayachandran, 2018; Deshpande et al., 2018; Lee and Wie, 2017; Yahmed, 2018). This suggests that the job quality for women has improved more slowly than education, and that occupational change may be key to understand if and when women's labor market outcomes and earnings improve. From an aggregate perspective these improvements are important given that higher female income may benefit the status, education, and health of both women and children, which in turn increases their country's development and growth (Duflo, 2012).
We focus on sex ratios as previous literature has found that declining number of men can affect women's schooling and the likelihood of employment, two precursors to changes in job type. In the context of the United States, Acemoglu et al. (2004) found that the large-scale deployment of men in the United States during World War II led to higher female labor force participation after the war, while Goldin (1991, 2006) documented the period as a tipping point with continued increases in the labor force participation of women later. Angrist (2002) found that decreases in the relative number of men among specific immigrant groups in the United States led to an increase in labor force participation among women in those same immigrant groups, while Mechoulan (2011) found that high incarceration rates for black men increase education and labor force participation rates for young, black women. More generally, Grossbard and Amuedo-Dorantes (2007) examined differences by region and age cohort due to baby booms and busts and found that lower sex ratios are associated with higher labor force participation rates for women.
There is less evidence on the impact of declining sex ratios on female labor market outcomes in developing countries, despite the fact that many of the drivers of changing sex ratios, such as international migration, wars, and violence, are more prevalent than in developed countries. There is a literature on the impact on marriage, fertility, and bargaining power in the household.
We advance the current literature by providing more evidence of the impact of changes in sex ratios on women's labor market outcomes in a developing country setting and by examining impacts beyond labor force participation. To do this, we use Mexican census and inter-censal data from 1990 to 2015. We exploit variation across Mexican states and age groups, and start by documenting the dramatic decline in the sex ratios in Mexico, particularly for younger age groups. We also show that international migration is negatively and significantly related to the sex ratio, and explain a large portion in the variation across states.
In estimating the impact of declining sex ratios on women's labor market outcomes, our main empirical concern is over the endogeneity of sex ratio changes due to local labor demand shocks. Here we differ from Raphael (2013) whose main concern is that women with particular marital and fertility preferences may move across states in response to the scarcity of men. He addressed this by using the relative supply of men in the state of birth as an instrument for the relative supply of men in the state of residence.
We find that declines in the relative number of men significantly improve women's labor market outcomes in Mexico. A one standard deviation decline in the sex ratio leads to 3–4 percentage points increase in women's labor force participation, 3–4 percentage points increase of the percentage of working women in white collar jobs, and 5–6 percentage points increase in the percentage in “brain”-intensive jobs. We also find that for some women average wages increase between 3% and 5%, while the female share in the top 25% of earners goes up by 1–2 percentage points. All these results constitute large changes relative to their respective means, and they are robust to different model specifications.
A driving force behind our results is likely human capital attainment. We end by exploring education as a key mechanism through which the absence of men could impact women's labor market outcomes. In the Mexican context high-skilled jobs do not require a college degree, as approximately 60% of white collar workers have an upper secondary education (12 years) or less.
The rest of this article is structured as follows: Section 2 describes the main data sources and shows descriptive statistics; Section 3 discusses our empirical strategy; Section 4 presents the results; Section 5 explores education as a driving mechanism; and Section 6 provides some discussion and concludes.
The data on sex ratios and labor market outcomes come from the 1990, 2000, and 2010 Mexican Census and the 2015 Mexican Intercensal Survey, all accessed through IPUMS International (Minnesota Population Center 2020). We do not use the 1995 Mexican population and dwelling counts, as there is no corresponding U.S. dataset in 1995 for our instrumental variable (the ACS began in 2001). Meanwhile the 2005 intercensal survey does not have information on hours, occupation, or income.
Panel A in Figure 1 shows the average sex ratios in Mexico for each of the years in our sample. The figure shows that similar to most countries, the sex ratio for younger age groups (ages 0–9) is over 1, reflecting the slightly higher natural number of male births. The sex ratio declines sharply in the 18–25 age group, where the ratio falls below 1 for all years. It remains below 1 for all remaining age groups and worsens over time for the older age groups. This shows that the problem of “absent men” became more, rather than less, acute over time. These results echo those of other papers, which find that Mexican migration is male dominated in all periods and has become increasingly male over time (Durand et al., 2001; Hanson and McIntosh, 2010; Raphael, 2013).
To show that the lower sex ratios in Mexico are distinctive, we compare them with those from the United States and Brazil using their respective Census data from IPUMS (Minnesota Population Center 2020). We choose Brazil because it is a Latin American country with similar income levels to Mexico. The results, presented in Panel B in Figure 1, confirm that Mexico's trajectory is unique. In the year 2000, in the United States the sex ratio for the 18–25 age group was 1.04, while that for Brazil it was 0.999. For Mexico the ratio was 0.9, which means that relative to the United States or Brazil there was one fewer man for every 10 women in that age group.
In a simple theoretical framework it can be shown that if internal migration rates are similar for men and women, the sex ratio in any period is a function of the international out and return migration rates for men. Mexican Census and labor force survey data show similar internal migration rates for men and women. More detail on the theoretical framework, which is similar to that of Raphael (2013), is available upon request.
To further test if international migration explains sex ratio imbalances, we examine the correlation with an alternative cause: homicides. In theory, homicides could be another determinant of changes in sex ratios, as they increased dramatically in Mexico after 2007, were concentrated among young men, and rose in a small number of states (authors’ calculations using data from the National Institute of Statistics and Geography, INEGI). Data used from INEGI are available in INEGI's website:
Summary statistics for the sex ratios and labor market outcomes are shown in Table 1. Here we briefly discuss our outcome variables. Labor force participation is defined as the percentage of women in an age group, state and year which reports having an occupation. We choose this over employment status as the coverage is higher. The overlap between the two is high, with over 98% of those who are employed reporting an occupation. For example, one data point is the percentage of 18–25-year-old working women in the state of Aguascalientes in 2015 who are in white collar jobs.
Summary statistics
Sex ratio (men/women) | 640 | 0.93 | 0.05 | 0.81 | 1.13 |
Predicted migration, EMIF | 640 | 0.21 | 0.17 | 0.01 | 0.95 |
Predicted migration, ENE | 640 | 0.23 | 0.21 | 0.01 | 1.14 |
Predicted migration, MC | 640 | 0.21 | 0.17 | 0.01 | 0.84 |
Predicted migration, historical | 640 | 0.17 | 0.26 | 0.00 | 2.08 |
Women's labor force participation (%) | 640 | 0.35 | 0.12 | 0.05 | 0.62 |
Women in white collar jobs (%) | 640 | 0.47 | 0.14 | 0.08 | 0.78 |
Women in “brain” jobs (%) | 640 | 0.37 | 0.11 | 0.08 | 0.72 |
Women in male-dominated professions (%) | 640 | 0.06 | 0.03 | 0.00 | 0.19 |
Log (average earned monthly income), women | 640 | 7.99 | 0.26 | 7.27 | 8.91 |
Women in top 25 of income (%) | 640 | 0.20 | 0.09 | 0.03 | 0.55 |
Weekly hours worked, for women | 480 | 40.34 | 2.10 | 32.11 | 56.55 |
Women in blue collar jobs (%) | 640 | 0.53 | 0.14 | 0.22 | 0.92 |
Women in “brawn” jobs (%) | 640 | 0.63 | 0.11 | 0.28 | 0.89 |
Years of schooling, women | 640 | 8.00 | 2.43 | 1.57 | 12.98 |
Lower secondary educated women (%) | 640 | 0.21 | 0.09 | 0.01 | 0.44 |
Upper secondary educated women (%) | 640 | 0.11 | 0.07 | 0.002 | 0.29 |
College-educated women (%) | 640 | 0.09 | 0.06 | 0.00 | 0.34 |
Never married women (%) | 640 | 0.20 | 0.17 | 0.02 | 0.72 |
We also look at hours worked and wages, measured by the log of average monthly earned income. Income is restricted to those who worked more than zero hours, had nonzero income, and did not have a top-coded income.
To estimate the effect of changes in the male–female sex ratio on female labor market outcomes, we use the following specification:
The dependent variable are measures of labor market outcomes for women in an 8-year-age group
The main concern over the endogeneity of the sex ratio stems from local economic shocks, as negative shocks may incentivize men to leave seeking work elsewhere and may change labor market opportunities for women who remain. For example, Hanson and Spilimbergo (1999) found that declines in Mexican wages are associated with increased Mexican migration, while Orrenius and Zavodny (2005) found that worse economic conditions in Mexico increase migration rates, particularly of lower skilled individuals. Meanwhile, Cerrutti and Massey (2001) found that Mexican men are significantly more likely to migrate for employment than women.
There also are concerns that the sex ratio could be influenced by women's re-location decisions, if for instance, women with characteristics that impact their labor market outcomes move to states with more (or less) favorable sex ratios. We therefore limit the sample to non-movers, defined as those who reside in their state of birth. This is the closest we can come to eliminating movers, as the census does not include location histories. We test the robustness of our results to this sample restriction in Table A3 in the Appendix.
We create the following measure of predicted male migration that was developed by Card (2001):
To calculate
To calculate
The instrument captures stocks rather than flows of migrants, predicting the number of Mexican men in an age group living in the United States as of any given year rather than the number who have left over a certain time period. To provide more clarity, we calculate the average predicted male migration to the United States across the states for the year 2000. These means by age group are shown in Table A4 in the Appendix. For interpretation, a value of 0.195, found in the first row in column (1), means that 19.5% of the men between the ages of 18 and 25 from a given state, on average, are expected to reside in the United States. Approximately 11% of the total Mexican population is estimated to live abroad. Some reasons why the predicted migration estimates are higher than 11% are that women, the elderly, and the very young were less likely to migrate during this time period. Furthermore, we use stocks rather than flows of Mexican born men.
The first stage results are shown in column (1) in Table 2, and are strong. For all three weighting sources, the predicted migration value is negative and significant, showing that higher number of men predicted to be in the United States are associated with lower male-to-female ratios in the sending states in Mexico. Furthermore, as shown in Table 3, the
First stage results
Predicted migration, EMIF | −0.228*** (0.028) | −0.034*** (0.011) | −0.295*** (0.030) | −0.228*** (0.028) | −0.460*** (0.063) | −0.301*** (0.037) | −0.211*** (0.038) |
0.61 | 0.57 | 0.62 | 0.61 | 0.59 | 0.61 | 0.55 | |
Predicted migration, ENE | −0.155*** (0.022) | −0.033*** (0.011) | −0.175*** (0.028) | −0.154*** (0.022) | −0.356*** (0.051) | −0.194*** (0.033) | −0.162*** (0.030) |
0.59 | 0.57 | 0.59 | 0.59 | 0.59 | 0.58 | 0.55 | |
Predicted migration, MC | −0.206*** (0.025) | −0.033*** (0.011) | −0.225*** (0.029) | −0.206*** (0.025) | −0.360*** (0.063) | −0.244*** (0.034) | −0.251*** (0.035) |
0.60 | 0.57 | 0.60 | 0.60 | 0.58 | 0.59 | 0.56 | |
Observations | 640 | 640 | 640 | 640 | 640 | 600 | 640 |
Second stage IV results
Sex ratio (men/women) | −0.419*** (0.064) | −0.232*** (0.053) | −0.457*** (0.046) | −0.042** (0.018) | −0.355*** (0.129) | −0.063 (0.040) | −1.241 (1.871) |
0.89 | 0.90 | 0.85 | 0.81 | 0.84 | 0.88 | 0.61 | |
Sex ratio (men/women) | −0.693*** (0.145) | −0.676*** (0.152) | −1.065*** (0.141) | −0.020 (0.035) | −0.645** (0.292) | −0.125 (0.085) | 8.865* (5.039) |
0.89 | 0.88 | 0.81 | 0.81 | 0.84 | 0.88 | 0.58 | |
A-P |
66.12 | 66.12 | 66.12 | 66.12 | 66.12 | 66.12 | 41.60 |
K-P chi squared | 52.40 | 52.40 | 52.40 | 52.40 | 52.40 | 52.40 | 41.47 |
Sex ratio (men/women) | −0.666*** (0.167) | −0.815*** (0.192) | −1.167*** (0.176) | −0.048 (0.042) | −1.092** (0.465) | −0.396*** (0.152) | 11.499* (6.706) |
0.89 | 0.87 | 0.79 | 0.81 | 0.83 | 0.86 | 0.56 | |
A-P |
49.09 | 49.09 | 49.09 | 49.09 | 49.09 | 49.09 | 23.15 |
K-P chi squared | 35.33 | 35.33 | 35.33 | 35.33 | 35.33 | 35.33 | 20.89 |
Sex ratio (men/women) | −0.822*** (0.155) | −0.764*** (0.173) | −1.197*** (0.172) | −0.047 (0.039) | −0.871** (0.358) | −0.244** (0.103) | 13.474* (7.201) |
0.88 | 0.88 | 0.79 | 0.81 | 0.84 | 0.87 | 0.54 | |
A-P |
65.68 | 65.68 | 65.68 | 65.68 | 65.68 | 65.68 | 32.31 |
K-P chi squared | 44.19 | 44.19 | 44.19 | 44.19 | 44.19 | 44.19 | 28.70 |
Observations | 640 | 640 | 640 | 640 | 640 | 640 | 480 |
The exclusion restriction holds if the predicted number of male migrants is independent of factors that determine women's labor market outcomes that we do not control for. This exogeneity in turn rests on the argument that time invariant weights capture historic trends rather than contemporaneous supply shocks, leading to predicted migration rates that differ from actual ones. Given the unobservable nature of these shocks, we cannot directly test these assumptions. Instead, we provide several indirect tests of whether alternative forms of the instrument are good predictors of the sex ratio. The goal is to rule out arbitrary potential correlations, which may indicate that our instrument violates the exclusion restriction.
First we examine the assumption that the weighting system reflects historic migrant networks and that it is through these networks that information about demand conditions is received. If this assumption holds, an increase in employment opportunities in U.S. states where men from a given Mexican state historically have not gone should not impact the sex ratio in that Mexican state. To test this, we randomly assign stocks of men between the observed minimum and maximum numbers in U.S. states from the EMIF. We repeat this exercise 1,000 times and use the mean value to construct a new instrument. The results are shown in column (2) in Table 2. Although the coefficients remain negative and significant, they are 15% of the size of those from the original instrument (shown in column (1)). This suggests that male out-migration responds strongly to conditions in U.S. states where there is an established network, but weakly where these networks are negligible.
Second we address concerns that migration networks are so strong that the stock of Mexican men in the receiving U.S. state reflects supply conditions in the sending state. We create a new instrument that removes strong connections, defined as one where the weights for both the sending and receive state exceed 15% (above the 90th percentile) or where the weights for a sending Mexican state are at or exceed 50% in any U.S. state. In other words, more than 15% of migrants from a given Mexican state go to a particular U.S. state, and more than 15% of migrants in that U.S. state are from a particular Mexican state. The double weighting ensures that U.S. states that have small Mexican migrant populations remain.
Third, we address concerns that the first stage is driven by a few receiving states in the United States or a few sending states in Mexico. We remove the largest receiving states in the United States, defined as those in the EMIF where either all or all but one Mexican state has a presence. This removes California, Texas, Arizona, and Florida. These results are shown in column (5) in Table 2. We next remove Mexican states with the highest level of international migration in the year 2000 (the first census where this information was included). We define high migration states as those where more than 10% of households said they had an international migrant (95th percentile), which removes Michoacán and Zacatecas. The results are shown in column (6) in Table 2. In both cases, the conclusions about the strength of the instrument remain. These confirm that the first stage results are not driven by a small set of states.
Finally, we consider an alternative measure of the demand for Mexican born labor using annual data from the Annual Social and Economic Supplement of the Current Population Survey (CPS), as accessed through IPUMS (Flood et al. 2020). We use changes in the median number of Mexican born and Hispanic men (this includes other Latin American immigrants) in a given age group employed in different U.S. states in 5 years previous to the year in question. The CPS is not our preferred measure given that it is smaller than the Census and ACS, and does not have sufficient sample size to use only Mexican born men.
The second stage results are shown in Table 3. Each cell contains the estimated coefficient on the sex ratio for an age group
The results show that declines in the number of men to women have strong impacts on the labor market outcomes of women. Starting with labor force participation in column (1), all of the coefficients are negative and significant, with the IV coefficients ranging from −0.666 to −0.822. This means a one standard deviation decrease in the male-to-female ratio (approximately 0.05) results in an increase in women's labor force participation of 3–4 percentage points. Given that a women's labor force participation rate averages 35%, this constitutes an increase in a non-trivial amount of 10–12%.
We also find significant impacts on occupation among women in the labor market. Declines in sex ratios women in the labor force are more likely to be in white collar or “brain” jobs. Specifically, according to the coefficients in column (2), a one standard deviation decrease in the male-to-female ratio leads to an increase in the percentage of women in white collar jobs which ranges from 3 to 4 percentage points (7–9% of the mean). Results for the percentage of women in “brain” jobs are predicted to increase between 14% and 16% of the mean. We therefore find strong evidence that “absent men” increase women's participation in high-skilled jobs.
We also find evidence of gains on earnings. On average, a one standard deviation decline in the ratio of men to women is associated with an increase in log monthly earned income that ranges from 3% to 5%. Using ENE and MC migration weights, we note 1–2 percentage points increase (6–10%) in the percentage of women in the top 25% of earners. This suggests that the gains from high-skilled jobs might be concentrated at the upper end of the earnings distribution. Finally, for hours worked we find a robust positive coefficient, indicating that as men become relatively scarce, the average number of hours women work in a week falls by 1–2%.
In sum, we find that declines in the number of men relative to women lead to a significant increase in women's labor force participation and their number in high-skilled professions. We also find an increase in average wages likely coming from the top end of the earnings distribution, and a decline in hours worked. These results confirm the findings of Bhalotra et al. (2015), and provide evidence of an environment in which women's labor market outcomes have improved significantly over time.
In this section, we address concerns that our results might be driven by endogeneity of the instrument. A concern about the Card style instrument is that if economic shocks in the sending locations are correlated over time, weights from the recent past might be correlated with current local economic conditions. We therefore consider weights from an earlier period, when the structure of the economy was different enough from the time period of our study that such correlations are unlikely. We use two sources of historic data that cover a small number of U.S. states to construct these weights. The first data were collected by Foerster (1925) in April 1924 on Mexicans arriving at border ports and at districts in Los Angeles, San Antonio, and El Paso. The second data are a representative sample collected by Taylor on Mexican migrants in different locations in California, Texas, and Illinois (Taylor, 1930 and 1932). The second stage results from an IV model that uses these historic weights are shown in the top panel in Table A3 in the Appendix. Although the instrument is weaker, the coefficients have the same sign as those in our original model. This suggests that our results do not simply reflect unobserved economic shocks. The correlations between these weights and the EMIF ones for Texas, California, and Illinois are 0.52, 0.24, and 0.45, respectively.
We further address concerns over the extent to which variables in our model control for contemporaneous shocks or preexisting trends determine migration and local labor market outcomes. To do this, we include: (a) state and year trends; (b) state and age group year trends; and (c) foreign direct investment and public expenditure per capita by state and year. These results are shown in the last three panels in Table A3 in the Appendix. The sign and size of the coefficients are such that our main conclusions remain. This provides further evidence that local economic shocks are unlikely to explain our findings. We also estimate a model with lagged industry controls. These results, available upon request, are similar to our original ones, providing further evidence that the contemporaneous industry controls are not driving our results.
Finally, we address the concern about potential bias from selective migration, given that we limit the sample to those who reside in their state of birth. We re-estimate the model using the full sample of individuals and present the results in the second panel in Table A3 in the Appendix. Our conclusions remain for all of salient variables, suggesting that selective migration does not drive our findings.
One question that arises is whether or not women move into any type of job once men leave, or instead move more intensively into high-skilled jobs. To answer this question, we estimate our model using the percentage of women in the labor force in blue collar or “brawn” jobs as outcome variables. These results are shown columns 1 and 2 in Table 4, and the positive coefficients indicate that women are 3–6 percentage points less likely to be employed in either type of job when men are relatively scarce (6–10% of the mean). The impact of absent men therefore results in women entering high-skilled and higher paying jobs, but not lower skilled ones. We also estimated the effects of declining sex ratios on men and found that men who stay behind are significantly more likely to enter “brawn” jobs show no evidence of entering white collar jobs and show weak evidence of entering “brain” jobs. Therefore, our results for women do not reflect uniform changes in the composition of low- and high-skilled jobs for men and women.
Other outcomes, second stage, IV results
Sex ratio (men/women) | 0.232*** (0.053) | 0.460*** (0.046) | −6.669*** (0.736) | −0.371*** (0.062) | 0.067** (0.028) | −0.136*** (0.028) | 0.033 (0.032) |
0.90 | 0.85 | 0.96 | 0.82 | 0.92 | 0.90 | 0.99 | |
Sex ratio (men/women) | 0.676*** (0.152) | 1.077*** (0.141) | −12.733*** (1.614) | −0.708*** (0.140) | 0.381*** (0.070) | −0.034 (0.067) | −0.023 (0.065) |
0.88 | 0.81 | 0.95 | 0.80 | 0.89 | 0.89 | 0.99 | |
Sex ratio (men/women) | 0.815*** (0.192) | 1.168*** (0.176) | −17.306*** (2.528) | −1.108*** (0.193) | 0.281*** (0.074) | −0.109 (0.077) | 0.063 (0.085) |
0.87 | 0.79 | 0.93 | 0.73 | 0.90 | 0.90 | 0.99 | |
Sex ratio (men/women) | 0.763*** (0.173) | 1.207*** (0.173) | −16.072*** (1.965) | −0.875*** (0.145) | 0.303*** (0.068) | −0.172*** (0.065) | 0.006 (0.073) |
0.88 | 0.79 | 0.94 | 0.78 | 0.90 | 0.90 | 0.99 | |
Observations | 640 | 640 | 640 | 640 | 640 | 640 | 640 |
One of the main channels through which the absence of men can impact women's labor market outcomes is through human capital accumulation. Human capital accumulation can interact or reinforce other mechanisms that may influence women's ability to get high-skilled and better paid jobs. For instance, as a result of fewer men, women may seek more schooling due to fewer marriage market opportunities (Raphael, 2013); an increase in time since women have less to do in the home (Goldin, 1991); an increase in remittances that provide resources to invest in education (Antman, 2015); and an increase in job vacancies for high-skilled jobs that motivate women to achieve the necessary level of schooling to fill-in these positions. While we cannot investigate each of these channels with our data, we can examine what happens to educational attainment as a result of missing men. We start with total years of schooling. The results shown in Table 4 indicate that a decline in the sex ratio of one standard deviation (approximately 0.05) leads to an increase in education of 0.6–0.9 years. Given that women get on average 8 years of education (less than lower secondary school), this constitutes an increase of 8–11%.
The next question is whether an increase in years of schooling falls at key points in the distribution that would push women into educational attainment levels deemed necessary for white collar jobs. To answer this, we examine three different attainment levels: a lower secondary education (9 years of schooling); an upper secondary education (12 years of schooling); or a college education (16 years of more). The results shown in Table 4 indicate that lower secondary school is a key point in the education distribution affected by declining sex ratios. A one standard deviation decrease in the sex ratio leads to an 8–13% increase in women with a lower secondary education. We also find an indication of some gains for college education, as all the coefficients are negative, but the results are not consistently statistically significant. Meanwhile we find positive coefficients at the upper secondary schooling level, meaning that declining sex ratios lowers the percentage of women with 12 years of schooling. This could be due to some women achieving more education by going on to college, or to some women achieving less education by stopping at lower secondary school. While we cannot rule out the possibility that some women are leaving school earlier to enter the labor market, the overall increase in years of schooling and the relative magnitude of the coefficients indicate that this is not likely the dominant story.
The results presented in Table 4 indicate that the education channel is operating most strongly on the marginal woman for whom finishing lower secondary school opens job opportunities. The stronger effect on this group is consistent with the average level of schooling for the men who are most likely to migrate. Furthermore, in this study the proportion of women with lower secondary schooling for the time period who have white collar jobs is non-negligible, reaching as much as 27.5%, as documented in Table A1 in the Appendix. These are likely the women who are driving the employment results.
We investigate if impact of declining sex ratios differs across those with more and less education. Results in Table 5 show that where we saw the most gains in schooling, the lower secondary level (panel C), women have increased their labor force participation and attained white collar and “brain” jobs, but have not increased their wages. This suggests that while these women are moving into high-skilled jobs, they are not moving into the best paid ones. Meanwhile, for college-educated women (panel A) we find evidence of an increase in high-skilled jobs and in wages. While the first stage is weak for this group (given this group is the least likely to migrate), these results suggest that the overall gains we see in terms of salary are driven by college-educated women. Finally, for women with more than lower secondary schooling but less than college (between 10 and 12 years), we find little evidence of movement into high-skilled jobs or an improvement in wages. Thus the aggregate gains appear to be driven by women at the lower and higher ends of the educational distribution for high-skilled jobs.
Second stage results: by education level
Sex ratio (men/women) | −1.857* (0.994) | −1.668 (1.113) | −2.039 (1.304) | −0.447 (0.333) | −2.373 (1.551) | −1.052 (0.694) | 32.611 (49.891) |
−4.47 | −9.08 | −15.96 | −1.51 | −0.77 | −0.87 | −11.02 | |
A-P |
3.29 | 2.89 | 2.89 | 2.89 | 2.89 | 2.89 | 0.43 |
Sex ratio (men/women) | 0.006 (0.033) | 0.071* (0.042) | −0.059 (0.044) | 0.114*** (0.030) | 0.029 (0.094) | 0.077 (0.053) | −4.924*** (1.843) |
0.66 | 0.68 | 0.85 | 0.32 | 0.79 | 0.76 | 0.28 | |
A-P |
22.04 | 22.95 | 22.95 | 22.95 | 22.95 | 22.95 | 17.81 |
Sex ratio (men/women) | −0.297*** (0.065) | −0.235*** (0.081) | −0.325*** (0.060) | 0.046*** (0.011) | 0.650*** (0.186) | 0.182*** (0.039) | 5.324* (2.912) |
0.86 | 0.84 | 0.89 | 0.76 | 0.86 | 0.82 | 0.69 | |
A-P |
56.16 | 56.16 | 56.16 | 56.16 | 56.16 | 56.16 | 29.44 |
In this article, we explore the impact of declining ratios of men to women on the labor market outcomes of working age women in Mexico. Unlike other studies in Mexico, we explore outcomes beyond labor market participation, going further in the life cycle of women to look at job type, wages, and hours worked. We examine a more recent period in Mexico (1990–2015), when women's education and labor force participation have increased. These extensions shed light on factors that may change occupations for women, leading to better and higher paying jobs over time.
Overall we find that the decline in the relative number of men significantly improved labor market outcomes for women. We find that women are more likely to enter the labor force and are more likely to have high-skilled jobs once there. For women who achieve lower secondary schooling, we find that increased educational attainment may be a leading explanation for the job results seen. Interestingly, for this level of schooling we find no significant impacts on wages. This suggests that while these women move into high-skilled jobs, they may not be able to move into the highest paid ones. Meanwhile, the aggregate increase in wages that we find appears to be driven by college-educated women.
Despite gains in educational attainment and labor force participation, women have made muted improvements in job type or wages (Blau and Kahn, 2017, Heath and Jayachandran, 2018). Few studies in developing countries have found recent improvements in the gender wage gap (Ahmed and McGillivray, 2015). Recent literature has studied the role of gender norms as one potential explanation. Some of this work focuses on attitudes at the community level (Alesina et al., 2013; Fernández, 2013), while other has looked at intra-household dynamics (Bernhardt et al., 2018). Male migration in Mexico may have contributed to a break in those societal norms that defined the economic roles of women. Women may be observing high-skilled job opportunities becoming available and thus preparing themselves for these positions by seeking more education. Future research could explore whether the driving factor in the Mexican context was a change in social norms and exploit the reversal in migration trends to explore whether these changes are short-lived or long lasting.