Reading through an excellent 2005 paper by Ferraro, Pfeffer and Sutton* on the way that social science theories in general, and economic theories in particular, can make themselves true when previously they were not, I came across a fabulous social psychology article they cited. The 1981 article, Economists Free Ride, Does Anyone Else?, by Marwell and Ames (M&A) analyzes the results of 11 different variations on a single experiment involving the provisioning of a public good. The experiment was designed to maximize the likelihood of free riding – participants were given detailed explanation several times about the options for the investment of a small amount of tokens (worth a few pennies each, depending on the experiment) into either a collective or individual fund, such that investing 100% into the collective fund produced the maximal total payout.
M&A then asked 6 economists and 1 sociologist to make predictions, and to verify that the experimental design conformed nicely to the classic free rider problem. The economists willing to make predictions mostly agreed that the problem conformed to the strong version of the free rider hypothesis, and thus theory predicted 0% investment in the public fund, but they personally predicted between 0 and 20% investment in the public fund.
M&A ran this experiment on various groups of high school students (by mail and phone, with different variations allowing for feedback, higher stakes, etc.) and some college students. In almost all cases, participants contributed around 40-50%. Raising the stakes decreased this amount a little – to 35% probably. Decreasing the group size (i.e. telling participants that only 3 other people were investing in the fund) increased participation somewhat. Interestingly, telling participants that the collective good was going to be something nondivisible (spending the collective fund money on the common area of a floor, for entering college freshmen in the example) doubled contributions to about 80%, despite theoretical predictions that this sort of collective good would be less valuable than simply getting money, and thus would reduce contributions.
Only one group reacted very differently to the experimental conditions: first-year economics graduate students. They contributed, on average, 20% to the collective fund and many more tried to free ride completely.
The experimenters additionally asked the participants about fairness, in particular, whether they took fairness into account when making their decisions, and what they thought a fair contribution was. Most though around 50% was a fair contribution, and there was some correlation between the amount people thought was fair and what they actually contributed. However, “More than one third of the economics graduate students either refused to answer the question regarding what is fair, or gave very complex, uncodable responses. It seems that the meaning of ‘fairness’ in this context was somewhat alien for this group.”
The questions remain though: why do people give the amount they do? 100% would be socially optimal, 0% is the prediction of the classic version of the free-rider paradox. Feedback mechanisms (where participants had two opportunities to invest, and got to see each others’ behavior in between) had little or no impact, suggesting against Elster’s conditional altruism hypothesis (though that could be explored more). Why close to half? Why any? And are economists such free riders because they’ve been taught that way, or did they become economists because of their existing preferences?
This last question is taken up a bit in, amongst other places, Carter and Irons (1991) article in the Journal of Economic Perspectives, Are Economists Different, and If So, Why?, tries to tackle precisely this question by running similar experiments (well, actually a variation of the ultimatum game) on freshmen and seniors in economics and other disciplines. C&I found small but significant differences between econ students and non-econ students, but did not find that those differences came from learning economics – senior econ students actually kept less money than freshmen. The more interesting finding to me – in line with M&A – is that the econ students were not all that far from their peers, and were quite far from the predictions, “It seems even economists sometimes fall short of behavior expected of all good homines economici.” I love the language of that sentence – that this rational, self-interested, infinitely calculating, atomistic person is some sort of ideal from which even its most devout adherents fall short.
I’m sure there’s been plenty more evidence generated in the last 15-20 years on the subject of the effects of learning economics, and I am pretty sure some of these studies have reported significant findings for the learning effect, not just the selection effect. When I get a bit more time, perhaps I’ll track some down (maybe in Stephen Marglin’s new book, The Dismal Science: How Thinking Like an Economist Undermines Community, which I have not yet tracked down a copy of).
To return to Ferraro, Pfeffer and Sutton – it certainly does seem that the free rider problem is an example of an economic model that does not accurately describe most situations (at least in its strong version), but M&A’s study does not really support the hypothesis that studying economics leads to enacting the principle – econ grad students may have already been less inclined to provision public goods before they ever took up the subject, and M&A do not try to claim otherwise. I think that point is very much elided in the FPS paper, in its discussion of both M&A and C&I, which reserves the discussion of the learning hypothesis (which is not supported in the C&I study, nor even in play in M&A) for other work.
I still very much want to argue that ideas matter, but I want better data, that’s all.
* Economic Language and Assumptions: How Theories Can Become Self-Fulfilling, in the Academy of Management Review.