Endogenous Cheating

Chicago Sun-Times has an article on the increased amount of cheating on tests at universities. This came up in a faculty meeting here several months ago. Students had been found to be texting answers to one another and other students outside of class. A friend of mine works with judicial affairs, or something like that, here on campus and told the story of a grad student selling physics answers to tests to students. All of it involved technology, and I pointed out in the meeting that probably cheating is rising because the probability of detection is falling. I then mentioned Becker's "Crime and Punishment" paper, and specifically the footnote in which he points out that when probabilities of detection are low, the correct punishment is to increase the fine, at which point he cites a Vietnamese practice at the time of cutting the hands off certain farmers who speculated on rice.

But the article also notes that not only is it rising, but that it seems to be more accepted among students too. And I think I can explain why. I suspect what is going on is this: people get some utility from doing well on the test, but also from doing other things with their time, and so for some, cheating is a solution to their time constraints. But, cheating is also costly in terms of reputational stigma. If you're caught, you'll incur some reputational cost - perhaps by a faculty member or a family member. Over time, the probability of detection has fallen causing the expected value from cheating to rise causing students to sort into cheating who previously couldn't "afford" it due to having a valuation on their reputation.

The Akerlof models on reputation also incorporate social norms into cheating, and it's very simple. Assume that the reputational cost is a function of how many other people cheat. As the probability of detection falls, other people cheat. But as other people cheat, the reputational cost itself falls since reputational cost is partly due to the rareness of the activity in the first place. Thus, you see two effects occurring - the marginal cheater shifts towards historically more honest people due to the effects of technology on the probabilities of detection, and secondly, social norms weaken causing even people who mostly concerned about the ethics to no longer be as much concerned and therefore more likely to cheat.