Loss Aversion as applied Tax Ethics

Greg Mankiw linked to this article in the Washington Post by experimental philosopher Kwame Anthony Appiah. Appiah points out that whether you think a tax system is equitable is determined partly by whether it is framed as a loss or a gain:

In the 1970s, the Nobel Prize-winning economist Thomas Schelling used to put some questions to his students at Harvard when he wanted to show how people's ethical preferences on public policy can be turned around. Suppose, he said, that you were designing a tax code and wanted to provide a credit -- a rebate, in effect -- for couples with children. (I'm simplifying a bit.) In a progressive tax system such as ours, we try to ease the burden on the less well off, so it might make sense to adjust the child credit accordingly. Would it be fair, do you think, to give poor parents a bigger credit than rich parents? Schelling's students were inclined to think so. If the credit was going to vary with income, it seemed fair to award struggling families the bigger tax break. It would certainly be unfair, they agreed, for richer families to get a bigger one.

Then Schelling asked his students to think about things in a different way. Instead of giving families with children a credit, you'd impose a surcharge on couples with no children. Now then: Would it be fair to make the childless rich pay a bigger surcharge than the childless poor? Schelling's students thought so.

But -- hang on a sec -- a bonus for those who have a child amounts to a penalty for those who don't have one. (Saying that those with children should be taxed less than the childless is another way of saying that the childless should be taxed more than those with children.) So when poor parents receive a smaller credit than rich ones, that is, in effect, the same as the childless poor paying a smaller surcharge than the childless rich. To many, the first deal sounds unfair and the second sounds fair -- but they're the very same tax scheme.

That's a little disturbing, isn't it? Especially if your judgments about social justice and taxation are central to your moral and political beliefs. Intuitions about fairness are some of the most basic moral sentiments we have -- arising, developmental psychologists tell us, soon after we're toddlers. Stanford psychologist William Damon has conducted studies in which he asked groups of children to make bracelets. Afterward, the group got a pile of candies as a reward and was told to divvy them up among themselves. From the age of 5 on, the arguments kids had about dividing the loot were all about what was truly "fair" -- equal amounts for everybody or more candy for more productive bracelet-makers?

One thing we know is that our sense of what's fair can be hard to reconcile with a "rational" assessment of costs and benefits. Not least in the realm of taxation, policymakers ignore the workings of moral psychology at their peril.

All that Appiah is saying is true, but when I read this passage the first thing I though about was loss aversion.

Loss aversion is the psychological tendency of humans (and other animals) to overvalue losses as opposed to gains. A person dislikes an $100 dollar loss far more than they like an $100 dollar gain. This tendency also leads to risk aversion in which people will avoid risky scenarios that could potentially lead to large losses.

The neurological origin of this loss aversion is a little complex. I am no expert on the subject, but originally it was thought that losses are coded differently in the brain because they include activation in the amygdala. The amygdala is a brain region usually associated with negative affect learning like fear conditioning. This addition of negative affect learning to the evaluation of reward was imputed to make people tend more to avoid it.

This assertion has been contested though in other work. (The paper to read is this one.) This work has shown that rewards and losses are processed through the same brain regions, but that there is asymmetric activation or inactivation resulting from losses and gains. In this case, the inactivation in reward associated regions is much larger than the corresponding activation for a gain would have been. The origins of this asymmetry are not clear. The mechanisms for all this are still being worked out.

Basically, when I read examples like the one above I see two important experimental questions. First, why does framing affect our perceptions of whether a system is equitable or not? Second, and more basic, why are people predictably irrational in over-valuing losses as opposed to gains?

I don't think that we have solid answers to either question yet, but we are working on it.