Kumar, Zilvinas Does Market Experience Eliminate Market Anomalies? John A List Summary List hypothesizes that the market experience can eliminate the anomaly known as the endowment effect.1 While endowment effect may appear in experiments or situations in which agents are inexperienced, as agents gain market experience they overcome this “flaw” in their thinking and behave “rationally.” Methods Overview List collected his data by examining trading patterns of sports memorabilia at a sports card show in Orlando, FL, and trading patterns of collector pins in a market constructed by Walt Disney World at Epcot Center, also in Orlando. List claims that such a method of data collection is advantageous because it is realistic and unobtrusive. One problem cited by List is determining which direction the path of causation follows. The correlation between reduced endowment effect and market experience may be exhibited because the subjects learn from their experience to behave more rationally (treatment effect). However, the cause chain may also flow in reverse: Certain subjects are simply more rational by nature, they naturally do not exhibit the endowment anomaly, and as a result of this fact they are more willing to buy and sell goods at the market prices, thus gaining “experience.” Whereas other subjects may be more irrational by nature, naturally exhibiting the endowment effect, and as a result demand anomalously high prices to induce them to sell. Thus they sell less and become less “experienced” (selection effect). To eliminate this ambiguity, List reexamined the same subjects a year later, to see if their behavior was different as a result of their increased experience during that year. Phase I, The Sports Card Market At the sports memorabilia market, subjects were randomly given one of two more or less equally valuable sports cards (we will label these good A and good B). They filled out a survey, and were then asked whether or not they wanted to exchange their good for the other good. Given that the two goods were initially assigned randomly, we expect that exactly fifty percent 1 The endowment effect is a name given to a certain seeming contradiction to the standard assumption that individuals have stable, well defined preferences, and make rational decisions in accordance with those preferences. The endowment effect, first discovered and labeled by Thaler (1980), refers to the fact that individuals demand much more to give a good up than they would pay to acquire it. This, of course, seems to contradict the assumption that individuals have stable valuations of goods. Although it can be argued that the endowment effect is not inconsistent with stable preferences, but simply that a stable feature of humans’ preferences is that the disutility of having to give something up is greater than the utility of acquiring it. Kahneman and Tversky labeled this hypothesized property loss aversion. Another related concept is what Samuelson and Zeckhauser (1988) called status quo bias – namely, they hypothesized that people simply have a preference for maintaining the status quo, thus once endowed with an object they are resistant to both buying more of it or selling it at the current market price. of people were assigned the good they preferred less (each person has a fifty percent chance of receiving their favored good, whichever of the two that is), thus we expect a trading volume of fifty percent. A volume of less than this would constitute evidence for the existence of an endowment effect – people value goods more when when they have them. The closer the trading volume is to fifty percent, the lesser the magnitude of the endowment effect. List thus tried to estimate the magnitude of the endowment effect within two subject subsets: i) Card dealers with booths at the convention (whose average level of experience was relatively high, as measured by a question on the survey regarding average number of trades per month) and ii) Non-Dealers (whose average level of experience was relatively low). List also calculated the magnitude of the endowment effect within subsets of the non-dealers – list split the non-dealers up into “experienced” and “inexperienced” categories based on whether they made more or fewer than 6 trades per month (the median number of trades per month among non-dealers). Phase II, The Pin Market The second phase of the study, conducted at the pin market, was conducted so as to avoid any possible biases specific to the sports memorabilia market. The methods used in this phase were similar, and so I won’t go into detail about them. Phase III, Return to the Sports Card Market In the third phase, List returned to the sports card market a year later and conducted the same experiment (with two new goods, of course) on the subjects he’d tested earlier. The purpose of this, recall, was to eliminate the possibility of a selection effect. Results Overall, the results confirmed the hypothesis that market experience reduces the magnitude of the endowment effect. In phase I, percentage of trades conducted by dealers numbered 43.6% and 45.7%, both close to the theoretically predicted 50%, whereas percentage of trades among non-dealers numbered 20% and 25%. Alternately, the null hypothesis had a pvalue of .32 among dealers (32% probability that there is no endowment effect among dealers) and a p-value of .001 among non-dealers (.1% probability that there is no endowment effect among non-dealers). Finally, regression models estimated a statistically significant positive effect of trading experience on number of trades among non-dealers. The results in phase II were similarly supportive of the hypothesis, suggesting that the hypothesis is robust across different types of markets. Results in phase III, again, were supportive, however List mentions the possibility of a selection bias – namely, only those subjects who remained interested in sports card trading returned to participate in the later experiment. However, statistical tests conducted by List indicated that this was not a major problem. Relevance to Markets To investigate the potential impact of an endowment effect on equilibrium social welfare, List constructed a theoretical partial equilibrium model with two goods in which some consumers were “experienced” (exhibiting no endowment effect) and some were “inexperienced” (exhibiting an extreme form endowment effect for good 1 – namely, they refused to ever lower their consumption of good 1). The result was an equilibrium that was considerably less efficient than the normal competitive case without endowment effects; furthermore, most of the efficiency loss was suffered by the inexperienced consumers. However, the situation may not be so severe, for if even some economic agents are rational, strategic interaction may yield a form of Coase’s invariance theorem. List gives the following example. Suppose Gary and Milton’s valuations of a lamp are given by the following table: Now if the lamp is initially given to Milton, no exchange will occur, for even though Gary’s WTA is higher than Milton’s, Gary does not value the lamp at 200 until it is already in his possession. Thus, an inefficiency results – the lamp is not being distributed to the person who values it most. However, suppose there is a third party, John, who is profit maximizing. John can purchase the lamp from Milton at 160 and “lend” it to Gary. Now that Gary has the lamp, its value to him will shoot up to 200, and John can threaten to take it away unless Gary pays him the 200. Thus the lamp will end up being distributed to the person who values it most, and John will make a profit for his unselfish altruism. Conclusion and Alternate Theory In conclusion, Lists hypothesis is confirmed. I propose, however, a potential rational explanation of the endowment effect. The endowment effect may be viewed as rational, if one takes it to be simply a form of risk aversion: You may not know how valuable something is to you until you have it, thus you undervalue goods when purchasing them, because you are discounting for the possibility that the good may not be as valuable as anticipated.