Several studies have suggested that experience causes phenomena that are anoma- lies for NT (such as the endowment effect) to disappear, however, it is not clear
from these studies what aspects of experience matter. The experiment reported in this chapter explored how different types of experience might influence the endow- ment effect. It had four treatments. The behaviour of subjects in a control group was compared to that of subjects with experience of consuming, owning, and choo- sing goods. An endowment effect was observed but it was not as strong as those reported by earlier studies and its strength differed between treatments. Notably, the effect was significantly stronger in the two treatments in which the endowment was acquired in two steps (the choosers and owners) rather than one step (the tasters and control group). The strength of the effect differed little between the tasters and control group. It was weaker among the choosers than the owners.
There are a number of possible explanations for why the strength of the en- dowment effect varied between treatments and differed compared to earlier studies. First, when people are faced with an option to swap, they can make their decision in different ways. They can (a) either think about the problem in terms of (i) outcomes or (ii) losses and gains. They can (b) either use (i) intuition or (ii) reasoning to make the decision. How they make decisions determines whether an endowment effect occurs, and how they make their decision is influenced by experimental procedures. Second, loss aversion is a factor in all decisions but the degree to which losses weigh heavier than gains varies between decision problems and can be influenced by expe- rimental procedure. The high overall swap rate in the experiment could be explained by loss aversion being weaker when a person is not giving up their last unit of a good or when they already own a good like the one being gained. Third, people do not al- ways immediately update their reference point when their endowment changes. The differences in swap rates between treatments could be caused by differences bet- ween treatments changing the likelihood that subjects updated their reference point. Fourth, other effects in addition to loss aversion are involved. The higher swap rate of the choosers relative to the owners could be explained in terms of the choosers
being committed to the choice they made before the endowment occurred.
Survey data was collected in addition to the data on choices. The responses to the attitude to risk surveys suggest that subjects had reference dependent preferences and were loss averse. Swap decisions could be predicted to a degree using the survey responses. Subjects who swapped also gave responses to the risk survey that showed less loss aversion. Furthermore, the swap rate varied with the type of reason subjects gave to explain their decision.
This study found significant variation in the strength of the endowment effect. This has several broader implications. One should be cautious not to prematurely generalise the predictions of NT or RDT since this study suggests human behaviour is richer than simple forms of both these theories imply. Furthermore, the results cast doubt on some other explanations of why the endowment effect occurs. Plott and Zeiler (2005) argue that the endowment effect in the form of a gap between willingness to pay and willingness to accept is a symptom of subject’s misconcep- tions about experimental tasks. They support their claim by presenting experiments with and without controls for misconceptions. They find that the WTA/WTP gap is not observed when a certain set of controls for misconceptions is used. In the ex- periment presented in this chapter, the set of controls for misconceptions is constant across treatments but the degree to which the endowment effect occurs varies bet- ween treatments. Therefore, subject’s misconceptions cannot account for whether or not the endowment effect is observed.
The following are possible extensions. Exploring whether using bundles of goods rather than single items weakens the endowment effect. Testing whether ac- quiring an endowment in several steps rather than one step strengthens the endow- ment effect. Using an incentivised design to explore the link between risk aversion over modest stakes and the endowment effect (loss aversion has been cited as an explanation for both). Further tests of whether experience of choosing and tasting
influence the endowment effect. A sharper test of whether the TSH can explain differing swap rates under different experimental settings.
Appendix B
Experimenter Scripts
B.1
First Experimenter
The text in italics was read out by the experimenter. Some parts of the script differed between treatments. The labels in bold indicate which treatments each part of the script applied to.
All: The subject was invited to take a seat opposite the experimenter. This is
an experiment looking at how people make decisions. During the experiment you will be asked to make decisions and answer some questions on paper forms. De-
pending on which of the four treatments the subject was assigned to, the experiment proceeded as follows.
Control Group: To start with, please could you complete this form. The expe-
rimenter handed the subject the demographic survey and a pen.
Taster:To start with, please could you taste these two items. The experimenter
poured the subject a small cup of lemonade and offered them a small bowl containing some vegetable crisps. The packaging of the two items were in front of the subject so they could examine them if they wished. When the subject had tasted both items, the experimenter handed the subject the demographic survey and a pen, and said
Then could you complete this form
Owner: To start with, please could you pick an envelope and look inside. The
experimenter offered the subject a basket containing a number of envelopes. The subject picked an envelope and looked inside. It contained a paper token for either one bottle of lemonade or one packet of crisps. Depending on what the token was for, the experimenter handed the subject either a bottle of lemonade or packet of crisps and said This is yours to keep. Then the experimenter handed the subject the demographic survey and a pen, and said Then could you complete this form
Chooser: The experimenter placed a bottle of lemonade and packet of crisps in front of the subject and said Please could you look at these two items and choose
one of them: you get to keep the one you choose. The subject chooses one of the
items. Then the experimenter handed the subject the demographic survey and a pen, and said Then could you complete this form
All:The experimenter then presented the subject with a basket of envelopes and said please could you pick an envelope and look inside. Inside the envelope was a token for several units of the two goods. The experimenter handed the subject the goods specified by the token.
The experimenter then asked the subject to complete the two attitude to risk sur- veys. The order in which the two surveys were given to the subject was randomised. Once the subject had completed the two surveys the experimenter said The first part
of the experiment is complete. Drawing the subject’s attention to the items they have
been given: These are yours to keep. Please could you take them around the corner
and [Name of second experimenter] will complete the second part of the experiment with you.