Unit 5 The rules of the game: Who gets what and why

How economists learn from facts Divorce law and domestic violence

We have modelled employment relationships as a game between employer and employee, in which the outcome depends on their bargaining power (shaped by the rules of the game) and their structural power (the value of walking away). Economists have used a similar approach to model relationships within the household. Couples can share the benefits from their relationship in different ways. When they decide how they will each spend time on childcare, household work, and employment, and what to spend money on, the resulting benefits to each party may likewise depend on their bargaining and structural power: how they make decisions, and what happens if the relationship breaks down. In the context of the household, both dimensions of power are influenced by social norms and by the legal framework governing individual behaviour, gender rights, property rights, and divorce or separation.

We will focus on one of these laws, divorce law, because it bears directly on the reservation option: it sets the terms on which a woman can legally end the marriage. Imagine a regime in which a woman can only divorce if her husband agrees. How good is her reservation option? Very poor. She can physically leave him, but without his consent she cannot legally end the marriage, cannot remarry, and cannot claim any share of the assets the couple built up. Her situation parallels Angela’s in Case 1 (forced labour): her reservation option is so poor that her partner can extract almost anything from the relationship.

Now imagine the rules change, and she can end the marriage without her husband’s consent. This is called unilateral divorce. Those who gain most from this change are the women with the strongest reasons to exit. Violent marriages are one obvious case. By giving women a credible way out, the reform has two possible effects on these marriages. First, some violent marriages may end, as women leave abusive partners. Second, even in marriages that continue, the threat of exit may deter abusive behaviour. Either way, the prediction is the same: fewer women experiencing violence from their partners.

Economists Betsey Stevenson and Justin Wolfers tested this prediction using a natural experiment from the United States. Until the late 1960s, US divorce law required either both spouses to consent or one to prove the other guilty of misconduct (such as adultery, desertion, or documented cruelty) in court—a high bar that often required witnesses or physical evidence. In 1969, California became the first state to adopt unilateral divorce. Through the 1970s and 1980s, most other US states followed.

How could they tell whether the reform actually reduced violence? They couldn’t simply look at what happened in reforming states before and after the reform, because lots of other things were changing in American marriages over these decades. Attitudes were shifting, more women were working, the economy went through ups and downs. Any of these could have reduced violence on its own.

The trick is to also look at the states that did not reform. Before the reforms, violence had been evolving similarly in the two groups of states. There is little reason to think it would have evolved differently in one group than the other in the absence of the law. Both were exposed to the same shifting attitudes, the same labour market changes, the same economy. The only systematic difference between them was that one group changed its divorce law and the other did not. So whatever change in violence happened in non-reforming states is roughly what would have happened in the reforming states anyway. If violence fell much more in reforming states than in non-reforming states, the extra fall can be attributed to the change in the law. Figure 5.17b illustrates the comparison between reforming and non-reforming states. This way of comparing two groups—one exposed to a change, one not—before and after the change is called difference-in-differences. It is one of the most widely used techniques in empirical economics.

Fullscreen

https://books.core-econ.org/microeconomics/05-the-rules-of-the-game-08-bargaining-in-democracy.html#figure-5-17b

Figure 5.17b Estimating the causal effect of unilateral divorce on domestic violence.

Stevenson and Wolfers applied this approach to several different outcomes, drawing on different datasets. They looked at self-reported domestic violence from a national survey of women, at female suicide rates, and at the rate of women murdered by their partners. Each measure has its limitations—survey responses depend on what women are willing to report; suicide and homicide are extreme outcomes that capture only the worst cases—so finding the same pattern across all three is more convincing than any single measure on its own.

The three measures pointed in the same direction. Take domestic violence, for example. In non-reforming states it rose by around 2.5 percentage points, and the assumption is that reforming states would have risen by the same amount had the law not changed. Instead, violence in reforming states fell by 1.7 percentage points. The gap between the two (around 4 percentage points) is the effect attributed to the reform: a fall of around 30%. On the same basis, female suicide fell by around 10% over the two decades following the reform. Partner homicides of women also fell in reforming states, though some of the decline appeared to begin before the law changed, which makes the homicide result harder to attribute cleanly to the reform. Although our discussion has taken the woman’s perspective, the survey recorded violence by and against both spouses, and the reform reduced domestic violence by roughly the same amount for husbands and wives.^{1 2}