Rational cheating at self-service checkouts

I really enjoyed this article by David Glance (University of Western Australia) on cheating at self-service checkouts. Glance wrote:

Despite some technological safeguards, self-service checkout machines in supermarkets rely heavily on customer honesty to scan, and pay for, their shopping. It turns out however, that around a third of all customers “cheat” the machines in some way...

Nobel prize winning economist Gary Becker has proposed the “Simple Model of Rational Crime” to explain this type of behaviour. He put forward the view that people do a simple “cost-benefit analysis” of every given situation to decide whether they are going to be dishonest. In deciding whether to park illegally for example, they will weigh the benefits of free parking against the risk of getting caught and the consequences of a fine if that does, in fact, happen.

Under Becker's model, in the case of supermarket self-checkout cheating the chance of being caught is probably fairly low (the employees aren't watching everything you do), and the penalty is likely to be low as well (probably you get let off with a warning unless your cheating is particularly egregious). So the explicit costs of this behaviour are pretty low, relative to the benefits. Remember that a rational decision maker will do an action if the expected benefits outweigh the costs. Even if the monetary benefits outweigh the costs though, not everyone cheats because there are still moral incentives (based on what you believe about right and wrong) and social incentives (based on what other people perceive as right and wrong) that also affect our behaviour.

Glance continues:

What behavioural economist Dan Ariely has discovered however, is that cheating is an irrational process that a large number of us will actually do. However, this type of dishonesty is always for small amounts. Ariely calls this amount the “fudge factor”. It can be dismissed as being inconsequential in comparison to the overall amount of a transaction. This type of cheating is independent of the potential reward and the likelihood of being caught, undermining Becker’s rational model of crime...

This is exactly the same type of behaviour that is seen when people download movies, get around a new’s [sic] site’s firewall, or even cheat in an online test. 

I'm not convinced that Ariely's work necessarily undermines Becker's model. At least, not to the extent that the model needs to be thrown away. Moral and social incentives are important here. If many people are cheating, and few people see cheating as a problem, then the social incentives to avoid cheating are relatively weak, and more people will cheat. Observing a lot of cheating in lab experiments is not surprising - in the lab there are very weak moral and social incentives not to cheat (since lab experiments are like a game). I'm not knocking Ariely's work, which is important and credible. It just doesn't mean that the rational crime model is necessarily completely wrong. Really Ariely and Becker have provided complementary explanations.

However knowing this, what can be done? Glance says:

Because this behaviour is irrational, it can be manipulated to reduce its happening. Ariely has found that if you get people to simply sign a statement saying that they will behave morally and won’t cheat, they do in fact cheat less...

What is important with this practice however is that it needs to be done before the task is carried out and only has a limited time that its effect will last. Asking students to agree to act honestly before an online quiz is likely to be effective, whereas getting them to sign a statement after they have written an essay and attach it to their submitted work, is not. In the case of the self-service checkout systems, a simple introductory screen that asked shoppers to agree that they will be honest would likely be effective in reducing cheating at the checkout. Another way would be to have a staff member who greets every shopper as they come to the checkout and reminds them that they will be there to help if needed.

Reminding people about right and wrong impresses on them the moral and social incentives to 'do the right thing'. It makes those incentives more salient in the decision about whether to cheat or not, so both rational and quasi-rational people will be less likely to cheat. It won't work for everyone, but you shouldn't expect any solution to be a silver bullet.

[HT: New Zealand Herald]