Do you treat money someone gives you and money you earn at work differently? I think I do.  This question is central to my dissertation research in which I plan to look at several methodological issues in economics.  One of these issues - legitimacy - coheres around the way in which we treat earned endowments differently to how we treat windfall endowments.  It goes back to an idea of Milton Friedman's in his permanent income hypothesis: we humans treat money differently depending on its source.  But, economists hadn't begun to use experimental methods back then, and only recently have economists begun to see that where money comes from (what the trade calls 'endowment source heterogeneity') affects what people do with money in experiments.  Today, I plan to start a new series of posts about earning and endowment 'legitimacy' in experimental economics and how it affects some of our conclusions from experimental economics and behavioral economics. 

Cherry, Frykblom and Shogren's 2002 paper 'Hardnose the Dictator' made a bit of a splash when it first came out.  Other papers before it dealt with payoff legitimacy, but not many showed as stark a result as theirs did.  The idea works as follows.  Consider a normal dictator game: two players one who is the Dictator the other the Receiver.  What happens if you make the Dictator do a task beforehand to earn their money, and, moreover, when there's differential achievement in that task?  Cherry, et al get some of their subjects to answer questions from the mathematics section of a GMAT (yes, that's problematic, but let's not worry about it for now).  Depending on how many questions they answer correctly, subjects are given $10 or $40. Another group are given an endowment of 'windfall' cash, as has been customary in experimental economics up to now, and they were given $10 and $40 to ensure a decent comparison. For some of the subjects 'distance' is increased when the experiment is done 'double blind'.  Customarily, in the dictator game many Dictators give substantial amounts of money to their Receivers (Camerer, 2003). The question here is: do those who earned their money give the same, more, or less than those who were given windfall endowments?

Cherry et al. showed that giving decreases dramatically when players earn their money, especially when players earned the most money (and perhaps feel the most entitled to their money). Here's the money quotation:

"In the baseline treatment [of windfall earnings] the theoretically predicted outcome of "zero offered" occured in 19% of the low-stakes bargains and 15% of the high-stakes bargains. In contrast, [in the earnings treatment] zero offers increase to 79% and 70% of the low- and high-stakes earnings treatments." (Cherry et al., 2002, 1219)

Moreover, in the double blind treatment the zero offer became even more prevalent occuring in 95% in the low-stakes and 97% of the high stakes treatments. What can we deduce from these results? First, 'legitimizing wealth', that is getting subjects to work for their payoffs before making decisions, seems to affect how subjects behave in experiments. Second, combining earning and isolation (the double blind treatment) brought the results of the experiment much closer to the theoretically predicted outcome than in almost all other experiments.

So is all lost for theories of other-regarding behavior? Where do we go from here? Replication and variation. I'll report soon on some similar papers that show similar results with earning in dictator games with some variation (the 'deservingness' of the receiver plays a role here as does culture), and other papers show how earning does not discourage subjects from cooperating in public goods games, in fact it may even encourage them to cooperate.  See you soon for more payoff and earning legitimacy in experimental economics!