Divorce as a risky prospect

In 1910, the divorce rate per 1000 members of the US population stood at 0.9. ^{1 1}Histroical divorce rates can be found in the Statistical Abstract for the US (1981). . This rate showed a slow upward trend for the next 50 years, and by 1960 had more than doubled to 2.2. It took only 20 years for the rate to more than double again so that by 1980, the rate was 5.3. For the last 20 years, the marriage rate, by contrast, has experienced mild fluctuations between 10.0 and 11.0 per 1000 in the population with no discernible trend. If the same pattern for both rates holds until the year 2000, the annual number of divorces will exceed the annual number of marriages.

Although sociologists have researched divorce extensively, only a few economic studies exist. This is unfortunate since divorce is likely to have considerable impact on economic vaiables such as hours of work, labour force participation, human capital accumulation, work performance and earnings. ^{2 2}For a recent study of hours at work and labour supply, see Green and Quester (1982); for studies on earnings and work performance, see Santos (1975) or Hoffman and Holmes (1976). King (1982) argues that couples anticipating divorce will individually invest more heavily in human capital since the costs of any current investment are at least partially absorbed by the spouse. Without denying the influences of peer groups, social norms and role models, it seems reasonable to suggest that pecuniary considerations may also help to explain divorce.

A search through the economics literature uncovered only two studies of the determinants as opposed to the implications of divorce: one by Orcutt, Caldwell and Wertheimer (1976) and another by Becker, Landes and Michael (BLM) (1977). The study by BLM is by far the most widely cited of the two. The authors of both studies argue that the current state of marriage is the primary determinant of divorce. BLM, for example, assert that ‘the probability of divorce is smaller the greater the expected gain from marriage, and the smaller the variance of the distribution of unanticipated gains from marriage’. BLM, in other words, view marriage as a risky investment with a distribution of returns. The alternative is divorce which, by implication in BLM, involves a certain return.

The first contribution of this study is to draw the implications for an alternative view in which the investment in marraige is certain, but the investment in divorce is risky.

The second contribution lies in presenting formal expected utility-maximizing models of an individual and/or a couple contemplating divorce which can be tested empirically. The thrid contribution is the method developed to test the predictions of the models.

The paper of follows a simple format. Section I presents the models. Section II provides an explanation of the data used in the empirical tests. Methodology and results are presented in sections III and IV. Caveats are observed in section V. The final section closed with a summary of the arguments and evidence.