Between 1960 and 1980, the annual divorce rate in the United States more than doubled, reaching approximately 50% of all first marriages before declining somewhat since 1980. This rise in marital instability generated concerns about the changing nature of the American family and its implications for children’s wellbeing. Understanding the causes of increased marital instability is crucial to understanding what policies might promote marital stability and protect children’s welfare in the event of their parents’ separation, as well as indicating the changes in family patterns we might expect in coming years. This paper seeks to explain how financial considerations affect the likelihood of divorce.

Considerable theoretical ambiguity exists among social scientists regarding rising divorce rates. The most prominent explanation, the economic independence hypothesis, suggests that women’s increased employment and earnings provided them with the resources to leave unhappy marriages. In contrast, other theories have posited that declining household specialization raised divorce rates by reducing the gains to marriage, or that increased financial strain in low-income couples’ marriages facilitated divorce and a divergence in divorce rates by race and class.

We argue that prior research has suffered from mismatch between theory and operationalization. In particular, the economic independence hypothesis has commonly been tested by examining the association between wives’ current earnings and their risk of divorce. Given that it is women’s expected financial well-being in the event of divorce that is hypothesized to affect risk of divorce, we employ a more theoretically-grounded measure of women’s economic independence by modeling their expected wages, were they to divorce. We can therefore distinguish the effects of current specialization and current financial strain from those of women’s economic independence.

In the economic independence model, women’s willingness to divorce should depend on the difference between their expected financial well-being in the event of divorce and their financial well-being if they remain married. However, financial losses of equivalent magnitude may have a larger effect on well-being at lower income levels (assuming diminishing marginal utility gains from money). To capture this possibility, we extend prior research by interacting measures of the expected financial position in the event of a divorce and the expected decline in financial wellbeing.

We use data from the Panel Study of Income Dynamics (PSID) to estimate a couple’s hazard of divorce. We use decennial census data to estimate rich models of predicted outcomes in the event of divorce. We then use these model estimates to generate hypothetical outcomes for spouses in the PSID.

Each theory of divorce suggests different policy interventions. If divorce is best predicted by women’s economic independence, programs promoting marriage quality, such as the recent Supporting Healthy Marriage project, may be most be beneficial, by reducing women’s desire to exercise their option to leave. Conversely, if financial stress is most predictive, economic support to families in financial distress might lower the divorce rate. Thus, our analyses both improve the quality and clarity of social science evidence on the risks of divorce and have implications for future family support policy.