THE THEORY DEVELOPED by Becker et al. (1977) contends that "surprises," whether positive or negative, increase the probability of divorce. In Becket et al.'s framework, household income volatility is a proxy for these unexpected events. Their model predicts that household income volatility increases the risk of divorce because unexpected changes alter the couples' expected returns from marriage. Negative shocks to household income could lower the returns from marriage below a particular threshold level, which may lead to divorce. Positive earnings shocks could induce a self-reliance effect, which may also increase the risk of divorce.
Economists have begun to question whether the traditional model of marriage and divorce (e.g., Becker 1973, 1974; Becker et al. 1977) remains valid in the presence of numerous changes to family life. For example, Stevenson and Wolfers (2007) suggest that labor-saving technologies, the ability to buy household services in the market, and the rising opportunity costs of specializing in household work for women have altered the returns from marriage. Today, it appears that marriage and its benefits are based on shared interests, i.e., consumption and leisure, rather than the gains from specialization in market and domestic spheres. In this context, negative earnings shocks reduce the gains from marriage, while positive earnings shocks increase the gains from marriage. The former may increase the risk of divorce, while the latter may lower the risk of divorce.
A number of studies examine the effects of earnings shocks on household consumption and other economic outcomes. (1) However, there have been few studies that examine the effects of earnings shocks on divorce. Previous attempts to measure the effects of earnings shocks on divorce have used actual minus predicted earnings (Becker et al. 1977), changes in predicted earnings capacities (Weiss and Willis 1997), and job displacement (Charles and Stephens 2004).
In this article, we investigate the effect of household income volatility on divorce by separating the household income shocks into transitory and permanent components. We use panel data from the 1979 cohort of the National Longitudinal Survey of Youth (NLSY79) to investigate this question. We also use a technique developed by Meghir and Pistaferri (2004), which is novel to this literature, to construct the transitory and permanent household income shocks. The approach developed by Meghir and Pistaferri (2004) emphasizes the importance of allowing the household income shocks to be generated by heterogeneous processes. Our analysis of household income fluctuations and divorce differs from previous work in three ways: (i) spousal incomes are jointly considered, (ii) transitory and permanent household income shocks are identified separately, and (iii) the household income shocks are generated separately by SES, for which individual educational attainments proxy, and race.
We find that negative (positive) transitory household income shocks increase (decrease) the probability of divorce, while there is weak evidence that positive (negative) permanent household income shocks increase (decrease) divorce propensities. The effects differ somewhat by SES and race. For example, the divorce propensities of high-school dropouts (low SES) are unaffected by transitory household income shocks but are affected positively by permanent household income shocks. The effects differ for those who graduate from high school and/or attend some college (mid SES): their divorce propensities are negatively affected by transitory household income shocks but are unaffected by permanent household income shocks. Neither the transitory nor permanent household income shocks affect the divorce propensities of college graduates (high SES). The divorce propensities of nonblacks/non-Hispanics and Hispanics are affected negatively by transitory household income shocks. …