Bryan Seegmiller and David Sims, BYU Economics Department
Introduction
It has been well-documented that the latter half of the 20th century and the early 21st century has seen vast upheavals in the societal roles of males and females in Western society. One noted result of this gender revolution is the striking change in college gender ratios—in the US, for example, the male-to-female college gender ratio changed from 1.6 in 1960 to .74 in 2003 (see also Figure 1). These changes in college gender ratios can have profound impacts on marriage and family patterns seen in society. For example, a recent Time Magazine Article examined the effects of a lowered male-to-female ratio on marriage patterns among Mormons in Utah, finding that a dearth of males has led to a higher average age at marriage and a more sexually permissive culture.1 At the same time, economists and sociologists have documented an increase in educational assortative marriages, or marriages between individuals of similar educational background, over the last several decades2 (see Greenwood et al, 2014, and Schwartz and Mare, 2003 for example). Most similar studies have focused on gender ratios in the overall population or within specific ethnic groups; this study, however, examines whether or not changes in the gender ratio within a specific subpopulation of note, namely among college students, has an effect on particular marriage patterns observed in society as a whole.
Methodology
To perform the analysis I gathered U.S. Census IPUMS data on individuals’ education levels, marital status, state of birth, and various other demographic features. From the National Center for Education Statistics (NCES) I gathered yearly data on male and female college enrollment numbers for each state. Then I matched individuals to the average college gender ratio they would have faced in their state of birth in the years that they were between the ages of 19 and 21. Combining the datasets in this manner provides a way to leverage within-state variations in the male-to-female college gender ratio me to perform the econometric analysis necessary to connect individuals’ marriage decisions with the ratio of males to females in college at the time they were of college age.
The basic econometric analysis consists of a estimating a set of linear regressions of a similar form. For each regression the independent variables are the same: the husband’s education, the average male-to-female ratio in the husband’s state of birth when he was of college age, the square of the male-to-female ratio, an intercept, and a set of state and year of college fixed effects. Since the male-to-female ratio in the state of birth is averaged for the years that the husband was between 19 and 21, the college year fixed effect corresponds to the year that the husband was 20 years old. Only males that are currently married are included in any of the regressions. The square of the male-to-female ratio is included in order to allow for nonlinearity in the relationship between marital outcomes and the male-to-female ratio. I estimate the effects of these variables on two dependent variables: whether or not the husband has married a college graduate wife, and the total years of education of the wife. These regressions are estimated separately for each of the U.S. Census years from 1960 to 2010.
Results
Table 1 below gives the results of the regressions.
Looking at the first two sets of independent variables we see some mixed evidence of the effects of the male-to-female college gender ratio on assortative marriage. There are strikingly different coefficients on the male-to-female ratio in the early census years versus the later census years, with the M/F ratio coefficient being positive until later in the sample period and then flipping to negative, while the square term follows the exact opposite pattern. In general, the coefficients are statistically significant. Also note that the coefficient on education increases as the Census year increases. This itself is evidence of increased assortative marriage due to changes in the college gender ratio, as the husband’s education tends to have a more powerful effect on the wife’s education even as the male-to-female ratio has tended to decrease over time (as demonstrated in Figure 1 below). The coefficients on the male-to-female ratio are a little more difficult to interpret, particularly in light of the fact that they change sign from the early to later Census years. However, in the later Census years (1990, 2000, 2010), the coefficients have the interpretation that as the number of males decreases relative to the number of females, the propensity for males to marry more educated women increases. Combined with the increasing education coefficients over time, this would illustrate that lower male-to-female ratio has had the effect of increasing assortative marriage in recent decades.
Conclusion
Though the results are a little mixed, the general magnitude and statistical significance of the male-to-female ratio on the wife’s education points to the importance of college gender ratios in determining some of the trends and patterns we see in marriages today. Further research on the subject would be greatly improved if individuals could be matched directly with their own college’s male-to-female ratio, rather than just the male-to-female ratio within their state. However, this study does provide evidence that the male-to-female ratio has indeed had some impact on educational assortative marriage in the U.S. over the last several decades. Because of the potential wide-reaching effects of college gender ratios on families and marriages in society as a whole (as evidenced by the Time magazine article referenced earlier), this subject should be of contemporary interest.
- Birger, Jon. “What 2 Religions Say About the Modern Dating Crisis,” Time Magazine Aug. 2015
- See Schwartz, Christine R. and Mare, Robert D. 2005. “Trends in Educational Assortative Marriage From 1940 to 2003,” Demography, Volume 42, Issue 4, pages 621-646, for example.