Irrational Economist_ Making Decisions in a Dangerous World - Erwann Michel-Kerjan [42]
This implication raises another question about the division of labor. My view is that some degree of collaboration with neuroscientists by some economists is necessary to create good technology transfer from neuroscience to economics (much as a company importing goods from another country should have bilingual workers who actually travel back and forth). If economists simply settle for whatever neuroscientists have to offer, economists will not get the ideas and studies they most want. Through collaborations they might.
AN APPLICATION: EYE-TRACKING AND BACKWARD INDUCTION IN BARGAINING GAMES
The example I would like to discuss is based on studies of alternating-offer bargaining games ( Johnson et al., 2002). It is not an example of neuroeconomics per se, because the only variable measured is indirect—namely, which information people look for on a computer screen. However, the hope is that in future studies such measures of cognitive process will be linked directly to brain activity and other psycho-physiological correlates to form a tentative picture of the mechanisms and behavioral outputs involved.
Alternating-offer bargaining games are fashionable to study as models of how bargaining could be studied noncooperatively with mathematically precise results (and later linked to cooperative approaches). In the games we study, bargaining occurs over three stages, between two players. At each of the three stages, the first player offers a division of a known sum of money to a second player. If the offer is accepted by this second player, the game ends. If the offer is rejected, the proposed amount is divided by two and the second player who rejected the offer can make the next offer (offering rights alternate).
Thus, in the first stage, player 1 offers a division of $5 to player 2. If rejected, player 2 offers a division of $2.50 to player 1. If rejected, player 1 offers a division of $1.25 to player 1. If that final offer is rejected they earn zero. (Note that the last round is an ultimatum.)
If both players have mutual knowledge that the players are self-interested and rational, the unique sub-game perfect equilibrium is that the first player should offer around $1.25 to player 2, who should accept it because if s/he does not, s/he will have only $1.25 at the next round to divide, thus the maximum alternative amount.
In the 1980s a controversy emerged from a series of experiments using variants of this game.2 The controversy revolved around three distinct issues: What would happen to the results in subsequent rounds after the players learned by experience, and what were the “suitable learning conditions”? (2) Did the subjects have a taste for fairness or some kind of social preference involving distaste for inequality? (3) Were the subjects looking all the way ahead or not?
Eric Johnson had earlier developed a computer mouse-based technology to measure the acquisition of information. In a study published in 2002, he as well as Sankar Sen, Talia Rymon, and I used this technology to hide the dollar amounts available at the three stages in boxes that could be “opened,” revealing the dollar amount, by moving a cursor into the box. (The box amounts varied across trials to limit memory.)
The behavioral result was similar to that found in most earlier studies in this field: Players offered an average of $2.11, and offers of less than $1.80 were rejected about half the time. The question, however, is whether they were reasoning strategically (in “equilibrium”) but adjusting for fairness, or failing to fully compute an ideal strategy.
Measuring what subjects are looking at on the computer screen is potentially a way for economists to distinguish between these two different interpretations. In fact, on about 10 percent of the trials, players did not even open up the third box to see how much was available. In a boundedly rational way, their lazy pattern of looking made some sense because only 8 percent