Marriage and divorce: patterns by gender, race, and educational attainment
The current study differs from Stevenson and Wolfers’ 2007 study in that the current study examines a younger birth cohort of Americans. This paper considers differences by gender and by racial/ethnic group but focuses on differences across education groups and by age of marriage. The trends of declining marriage rates and increasing divorce rates, shown by Stevenson and Wolfers, continue with the 1957–1964 NLSY79 cohort. The longitudinal survey shows the same patterns regarding differences between racial/ethnic groups and education groups as did the SIPP—though the NLSY79 differences between college graduates and the other education groups are even starker. While the marriage rate for the NLSY79 cohort fell to 86.8 percent compared with 89.5 percent for the 1950–1955 cohort, the rate among college graduates slipped only slightly, from 89.5 percent to 89.0 percent, between the two cohorts. In addition, though the rate of divorce rose to 44.8 percent in the NLSY79 cohort compared with 40.8 percent in the 1950–1955 cohort, the rate of divorce among college graduates fell from 34.8 percent to 29.7 percent.
The National Longitudinal Survey of Youth 1979 is particularly well suited for studying marriage and divorce patterns. The NLSY79 is a nationally representative sample of men and women who were ages 14 to 22 when they were first interviewed in 1979. Respondents were interviewed annually until 1994, and since then they have continued to be interviewed on a biennial basis. The NLSY79 collects detailed information on fertility, marital transitions, and employment in a format that allows one to determine the dating of the specific events.
Because the NLSY79 contains a longitudinal marital history for each respondent, the survey permits the study of marriage and divorce over the life cycle. For a specific cohort, the NLSY79 can provide statistics on the percentage of marriages that end in divorce. In contrast, official statistics on marriage and divorce rates from Vital Statistics Records are based on counts of marriages and divorces reported by the states from registration records. The rates are calculated by dividing the marriage and divorce totals by population estimates from the decennial census. These rates tell us what percentage of the U.S. population experiences a marriage or divorce in a given year but cannot provide information on what percentage of marriages end in divorce for the U.S. population.4
Because the NLSY79 collects data on many aspects of respondents’ lives—including employment, fertility, and income—many researchers have used the NLSY79 to look at marriage in conjunction with a variety of outcomes. For instance, by estimating the relationships among marriage, divorce, work effort, and wage rates, researchers found that being married and having high earnings reinforce each other over time.5 Others looked at the how income affects the marriage and divorce decisions of young Americans; they found that high earnings capacity increases the probability of marriage and decreases the probability of divorce for young men, but decreases the probability of marriage for young women and has no effect on the likelihood of divorce.6 A different study used the NLSY79 to identify causal effects of marriage and cohabitation on total family income.7 This study found that women who enter a cohabiting relationship gain roughly 55 percent in needs-adjusted family income, defined as income per adult equivalent, regardless of whether or not they marry; for men, the level of needs-adjusted family income does not change when they make the same transitions.8 In addition, a 2009 study found that marriage lowers female wages by 2 to 4 percent in the year of marriage and lowers the wage growth of men by 2 percentage points and of women by about 4 percentage points.9
4 Calculating the ratio of the marriage and divorce rates can provide the proportion of marriages that end in divorce only if the marriages that occur in a given year have the same probability of ending in divorce as the past marriages that are generating the divorces recorded in the same year. This is not the case. Divorce rates peaked in 1981. Of marriages that began in the 1970s, approximately half ended in divorce within 25 years. An insufficient time has passed to calculate the percentage of marriages that began in the 1980s that will have ended 25 years later, but of marriages that began during the 1980s, the proportion that had ended by each anniversary was lower compared with those that began in the 1970s. The proportion ending by each anniversary is lower still for marriages that began during the 1990s. See Stevenson and Wolfers, “Marriage and divorce.”
5 Avner Ahituv and Robert Lerman, “How do marital status, work effort, and wage rates interact?” Demography, August 2007, pp. 623–647.
6 Simon Burgess, Carol Propper, and Arnstein Aassve “The role of income in marriage and divorce transitions among young Americans,” Journal of Population Economics, August 2003, pages 455–475.
7 Audrey Light, “Gender differences in the marriage and cohabitation income premium,” Demography, May 2004, pp. 263–284.
8 In “Gender differences,” Light calculates adult equivalence by assigning weights to family members: one for the first adult, 0.8 for the second adult, 0.4 for the first child, and 0.3 for subsequent children.
9 David Loughran and Julie Zissimopoulos, “Why wait? The effect of marriage and childbearing on the wages of men and women,” Journal of Human Resources, spring 2009, pp. 326–349.