"Retail therapy" has a bad name: we associate it with vacuous Carrie Bradshaw-types splashing out on their hundredth pair of Jimmy Choos. However, experimental evidence suggests that this image is wrong. Retail therapy actually works.

Researchers at the University of Michigan got female students to watch a film clip intended to make them sad or angry. They then gave them the option of choosing or not to buy one of six snacks. They found that women who chose to buy became less sad than those who didn't. In this sense, buying stuff cheers people up.

This could be because the act of buying simply distracts people from what makes them miserable. Two things, however, suggest this is not the case. First, women who became angry at seeing the flimp clip stayed angry whether they bought a snack or not.This is not what you'd expect if the act of buying distracted people. Second, another experiment found that people who spent time just browsing did not experience any decline in sadness.

Nor is it the case that sadness declined because people anticipated a consumer surplus from eating the snack. If this were the case, you'd expect happy people to become happier after buying. But this is not so.

Instead, something else is happening. This is that sadness is associated with feelings that one is not in control of one's. Shopping helps to restore a sense of personal control, say the authors, and so relieves depression.

This raises a question. Is retail therapy really therapy, or is it instead more like a drug which offers short-term relief without solving the underlying problem? If spending leads to higher levels of personal debt - and thus perhaps (pdf) to greater feeling that one is not in control of one's life - it's the latter.

And this in turn raises other questions. Are there not better ways in which people might regain a feeling of control than by shopping? If so, why aren't they used? And if not, why not? Those people who perceive a link between capitalism and poor mental health will, of course, have answers to these.