[Students who have not studied economics may be unfamiliar with the core idea that Knetsch challenges in this selection, an idea that Coase implicitly embraced in The Problem of Social Cost. The MIT Dictionary of Modern Economics (4th ed. 1992) defines an indifference curve as "A curve showing the locus of combinations of the amounts of two goods, say X and Y. such that the individual is indifferent between any combination on that curve."]

Indifference curves are normally taken to indicate corresponding tradeoffs of goods /l for 13 or A for A over the same interval: "the rate of commodity substitution at a point on an indifference curve is the same for movements in either direction" (James Henderson and Richard Quandt, 1971, p.12). However, recent empirical findings of asymmetric evaluations of gains and losses imply that the presumed reversibility may not accurately reflect preferences, and that people commonly make choices that differ depending on the direction of proposed trades.

The evidence from a wide variety of tests is consistent with the suggestions of Daniel Kahneman and Amos Tversky (1979), and Richard Thaler (1980), that losses from a reference position are systematically valued far more than commensurate gains. The minimum compensation people demand to give up a good has been found to be several times larger than the maximum amount they are willing to pay for a commensurate entitlement. For example, when questioned about the possible destruction of a duck habitat, hunters responded that they would be willing to pay an average of $247 to prevent its loss but would demand $1044 to accept it (Judd Hammack and Gardner Brown, 1974). Respondents in another study demanded payments to accept various levels of visibility degradation that were from 5 to over 16 times higher than their valuations based on payment measures. Similar large differences in valuations of a wide variety of goods and entitlements have been reported in numerous survey studies.

These valuation differences based on responses to hypothetical survey questions prompted further, more severe, tests of the disparity using a variety of real exchange experimental designs. The results have been much the same. Even when exchanges of real goods and actual cash payments motivated the evaluations, the compensation demanded to give up an entitlement has been reported to far exceed the comparable payment measure of value. . .

. . . Other more recent tests of the evaluation disparities, or endowment effects, have shown that the observed disparities between evaluations of entitlements are not the result of wealth effects or income constraints, strategic behavior, or transactions costs, and that the differences persist over repeated binding iterations of market trials (Knetsch, Thaler, and Kahneman, 1988).The purpose here is to report the results of more direct tests of the reversibility of indifference curves. A variety of test designs were used in these exercises but in each case one-half of a sample of participants were offered one good or money for another good, and the other half were offered a similar trade but in the opposite direction.

Test 1: Exchanges of Two Goods

A straightforward preference exercise involving a choice between two goods was conducted with three comparable groups of University of Victoria students.

In one class, 76 participants were given a coffee mug and were then asked to complete a short questionnaire. The students had the mug in their possession while they answered the written questions. After the questionnaires were completed, the experimenter showed the participants a 400-gram Swiss chocolate bar and told them they could have one in exchange for their mug.3 Participants were informed that they could either keep their mugs, or have a chocolate bar. They were instructed to hold up a colored paper with the word "trade" marked on it if they preferred the candy to their mug and wished to make an exchange. All desired trades were made immediately by one of four experimenters present, there was no uncertainty of receiving the other good and no effort was required beyond raising the paper to indicate a willingness to make an exchange.

The 87 participants in the second group were offered an opportunity to make the opposite trade of giving up a candy bar, which had been given them initially, for a mug. All conditions were the same except for the direction of the exchange offer.

The 55 people in the third group were simply offered a choice between receiving a candy bar or mug.

The choices presented to people in each of the first two groups differed only in terms of which good they had to give up to obtain the other—incentives were compatible and there were no income, or wealth, effect possibilities, nor any wealth constraints. However, contrary to the expectation of an equal proportion favoring one good over the other in each group, based on the conventional assertion of economic theory and practice, the different initial entitlements and subsequent direction of potential trades heavily influenced the participants' valuations of the two goods.

3. Participants were familiar with both the mugs and the candy bars. The mugs were sold in the University bookstore for $4.95 and the candy bars were available in local shops for around $6.

The preferences indicated by the choices were:

  When given a simple choice without a prior entitlement or reference position, 56 percent of the participants (Group 3) selected a mug in preference to a candy bar. But only 10 percent valued a mug more than a candy bar when they had to give up the candy to obtain a mug (Group 2), while 89 percent of those who initially received a mug (Group 1) declined to give it up for a candy bar. The preferences for a mug over a candy bar varied from 10 to 89 percent depending entirely on the reference position of the endowment at the time of the valuations.

In this simple experiment, participants' preferences were not independent of the direction of the exchanges, as is commonly assumed. They expressed a dramatic asymmetry in valuations by weighing the loss of giving up their initial or reference entitlement far more heavily than the foregone gains of not obtaining the alternative entitlement II.

Test 2: Minimum Acceptable Exchanges

The symmetry of exchange preferences was further tested in a second experiment involving real goods and actual cash payments. As in the first test, the design of this experiment also precluded the influence of a wealth constraint and the possibility of an income, or wealth, effect. The participants were 80 students in five tutorial groups of an introductory pre-principles economics class at Simon Fraser University. The exercise was repeated in the five tutorial meetings in a single afternoon.

Forty-one individuals in three of the groups were first given two 100-gram Swiss chocolate candy bars.4 They were then asked to state the smallest amount of money they would accept to give up both of their candy bars.

The 39 members of the other two tutorial groups were each given two single-dollar bills. They were then asked the minimum number of chocolate bars they would require to give Up their two dollars. The initial offering of money or candy bars was alternated over the five participating groups.

The method of obtaining the declarations of minimum acceptable trades was designed so that stating the smallest compensation each participant would agree to was in the person's best interest. To ensure that the individuals understood that this was the case, a preliminary instruction and screening test was conducted so that they could demonstrate their understanding before becoming participants in the real exchange exercise. This preliminary test involved a hypothetical, induced, or prescribed, value and the identical bidding mechanism that was used to reveal preferences in the subsequent real good exchange.

All participants were given the same form for the preliminary demonstration and screening test:

THIS FIRST PART OF THIS EXERCISE IS HYPOTHETICAL ONLY NO ACTUAL PAYMENTS WILL BE MADE. HOWEVER, DEPENDING ON THE RESPONSE OF PARTICIPANTS IN THIS GROUP, A SECOND PART MAY BE CONDUCTED IN WHICH ACTUAL PAYMENTS WILL BE MADE.

IF THE SECOND PART IS CONDUCTED, YOU CAN ONLY WIN, YOU CANNOT LOSE.

Assume that the person conducting this exercise will pay you $2.50 for the ticket attached to this sheet. Now assume that before you redeem the ticket you are given an opportunity to sell it. The price you will be offered for it will be determined by a random draw of one of six cards with the dollar amounts: $0, $1, $2, $3, $4, or $5 on them.

What is the smallest of these offer prices you would accept to give up your ticket?

$_______ (use only whole dollars—i.e., no cents). If the offer price on the randomly drawn card is the same or higher than the price you stated above, you will receive the amount of money on the card in exchange for your ticket. If the offer price on the card is lower than your indicated minimum selling price, you will keep the ticket (and turn it in for the $2.50).After the participants read the instructions for the preliminary test, the experimenter gave a further explanation and answered

4. These candy bars sell for around $2 each in most shops.

questions. Participants then completed the forms. The experimenter checked the forms to ensure that at least 75 percent of the individuals knew how to respond in a way that served their best interest (by stating $3 as the minimum acceptable price). If more than 25 percent failed to answer in this way the experiment was terminated for that whole group.6

Of the 80 individuals in the five groups that went on to the second part involving the real exchanges, only 14 (18 percent) failed to answer the first part correctly.7 Before proceeding with the second part, the experimenter explained the bidding mechanism again, and pointed out the reasons why other answers could work to the disadvantage of the respondent. These further explanations, and the screening of participants by their demonstrated ability to reveal their true minimum values in this market, ensured that individuals participating in the real exchange experiment fully understood the value revealing incentives of the bidding procedure.

Participants in the second, real exchange, part of the experiment were given their endowment of either candy bars or dollar bills and were told that these were theirs to keep. Those with an entitlement of candy bars then received the following second form:

THIS SECOND PART WILL INVOLVE REAL PAYMENTS. THERE ARE NO RIGHT OR WRONG ANSWERS. YOUR RESPONSE SHOULD REFLECT ONLY YOUR OWN PREFERENCES.

You have been given 2 candy bars. They are yours to keep. However, if you like—and the choice is entirely yours—you may be able to trade the 2 candy bars for money. We want to know about your personal preferences in terms of the smallest number of dollars you would accept to give up the two candy bars.

The amount you will be offered to give up your 2 candy bars will be determined in the same way as it was in Part 1 of this exercise. That is, an offer price, in dollars, will be determined by a random draw of one of the six cards with amounts ranging from $0 to $5.If you are willing to give up the 2 candy bars for $5 or less, write the smallest of these offer prices that you would accept to give up both of the candy bars in the blank below. If it would take more than $5 for you to agree to give up your two candy bars, then write a 6 in the blank, and you will then keep your candy bars.$ (use only whole dollars—i.e., no cents). If the offer amount on the randomly drawn card is the same or higher than the number you have written above, you will receive the amount of

money on the card in exchange for your 2 candy bars. If the offer on the card is lower than the number you have written above you will be allowed to keep the candy bars.

6. This occurred in four other groups.

7. All but 3 of the 14 individuals answering incorrectly specified a selling price lower than the most profitable one, an error that would bias the results toward lower evaluations of the initial entitlements and symmetry and away from the much higher values of losses subsequently revealed.

Individuals receiving an initial endowment of money were given a comparable form.

The bid procedure was then explained yet again and questions answered. The random draw of the card was made after the forms were completed, and exchanges were made according to the indicated minimum acceptable trades.

The results again demonstrated a strong aversion to giving up an initial entitlement.

The participants' dollar value of the candy bars is given either by the minimum amount demanded to give them up, or by the number of candy bars required to give up dollars. The proportions of the two samples valuing a candy bar equal to or greater than $1 and $2 are given in the following tabulation.

The participants valued the candy or money significantly more when considering the prospect of giving it up than they did when they had the opportunity to acquire either in exchange for their original entitlement. The average evaluations of the candy bars, based on the responses of all of the participants in each original endowment group, were approximately $1.83 with the entitlement and $0.90 without it. That is, the people starting with the candy bars valued them at a little over twice as much as the people did who were given money.

As in the earlier test, the relative preferences of participants in this exercise depended on reference entitlements and the subsequent direction of trade offers....

The results of the exercises reported here are consistent with other empirical tests of people's evaluations. The greater value ascribed to entitlements that might be lost over ones that might be gained gives rise to the disparity between bid and offer values. This appears to be similar to the observed nonreversibility of indifference curves in the data reported here....The widespread irreversibility of indifference curves would not imply that people will not make any trades, or that consumers will not change future consumption patterns in response to changing relative prices. However, the presence of irreversibilities would imply that fewer trades will be made than predicted by standard assumptions, and they offer little assurance that the shifts in future consumption will be as complete or as prompt as would be expected if indifference curves were reversible.

In a similar way, the presence of irreversibilities also suggests that common presumptions of the potential gains from trade may often be overstated.... And the commonly observed aversion to giving up concessions in negotiations may well be due to a similar asymmetry in valuations of gains and losses.