Tuesday, March 15, 2011

Stated vs. Revealed Preference in Abusive Relationships

It is always a good rule of thumb that when people say they want one thing and consistently do another, this should make you suspicious of whether they actually know what they want. Or in economists terms, when people's stated preferences and revealed preferences diverge, they are probably screwing something up. ("Screwing something up" is in fact a technical term :)  ).

Economists, for largely good reasons, tend to trust revealed preference. Talk is cheap, but when the chips are down, go with what people actually do. Usually this is a good way to bet - when a guy tells you he wants to see more opera but actually spends his weekends watching TV, it's a fair bet that he doesn't actually want to see opera, he just likes the idea of it.

But what about a woman who is in an abusive relationship who manages to leave and tells you she really wants to be done with the man, but then keeps going back again and again after he apologises?

In that case, it's not so simple. Right-minded people immediately jump to the conclusion that the woman actually wants to leave, and must somehow just be being prevented from this (such as by threats from the man).

The trouble with this view is that it has difficulty explaining why there is such cyclicality - women will leave, and then come back, many times over. It can't be that they never can find a way to leave. Even if you have all the sympathy in the world for such women, you're still left with a puzzle of trying to explain what the hell they're doing. In other words, something funny is going on.

One of my favourite papers in economics looks at this. They argue that women in abusive relationships have time-inconsistent preferences - in other words, they truly do want to leave when they leave, but they predictably change their mind and return to the guy. Hence their preferences are inconsistent over time.

They study a fascinating case that provides evidence for this - the case of 'no-drop' laws, whereby when a woman complains of domestic violence, prosecutors are obliged to proceed with the case even if the woman subsequently recants her testimony. This tends to happen a lot, which should also make you suspicious - if the guy is already in prison pending charges, what's the harm in going to trial?

When these laws were passed, they resulted in a drop in violence of men towards women - no surprises there. But here's where it gets interesting - the law also resulted in a drop of murders of men by abused women. The authors argue that people with time-inconsistent preferences need commitment mechanisms to stop them changing their mind. Murder is one such mechanism, albiet a very poor choice. These no-drop laws work because they substitute a much less costly commitment mechanism, helping women stop themselves from predictably going back to their man. They might be my favourite example ever of 'nudging' type laws, where you can stop people making bad mistakes by subtly crafted laws.

The reason this is such a fantastic paper is that it gets towards the heart of understanding why people end up in these crappy situations. The feminists of the world would tell you that the women are 100% victims, and that if they're returning to the men, it must be because of threats - we just need to be harsher on the men. In fact, believing this misplaced sympathy would cause you to completely miss the bigger picture of how these relationships persist, and what you can actually do to help end them.

Sympathy is not a substitute for analysis. And you can depend on it that when stated preference and revealed preference diverge, it's a situation worth your studying.

No comments:

Post a Comment