Behavioural Economics

The Role of Expectations in the Decision to Trust

Because trust is so fundamental to so many transactions – and betrayal of that trust could cause psychological and financial trauma – it is important that we make sure that behaviour in the investment game does indeed reflect a willingness to trust strangers. The two studies above discussed above suggest that both trust and other-regarding preferences matter. But what if people are mostly motivated by a desire to share and to a lesser extent by trust?

Uri Gneezy, Werner Güth and Frank Verboven attempted to understand the behaviour of the senders by having people take part in a trust game where they systematically varied the amount that the receiver could return. In this study, amounts sent were only doubled rather than tripled. So, if the sender sent $10 then the receiver got $20. In one treatment, the receiver could return only up to $2, regardless of the amount that he/she received from the sender. In the second treatment,

$10, the average amount sent is $6.50, while for repayments of $18 the average amount sent is $5.63. These two amounts are not statistically different. The reason why the amount sent when the upper bound on repayment is $18 is not different from when the upper bound on repayment is $10 is probably that senders did not expect the receiver to send back much more than $10 even with a higher repayment amount and so the raising of the upper bound from $10 to $18 did not influence decisions much. Even a presentation mode which focuses on relative, rather than absolute, returns, coupled with questionnaires designed to induce strategic reasoning, does not get rid of trust among the senders. For example, in another experiment, they employ five treatments. The first is a baseline no history treatment, second one is a social history treatment. Third one shows the different amounts received by the receivers and the amounts they returned. In the fourth treatment, they use the baseline no history treatment.

but then also ask the senders to fill out a questionnaire prior to making a decision about sending money. The questionnaire is designed with two specific purposes. First, it was to ensure that senders understood the design and considered their decisions carefully before making them. Second, it was to help participants determine how much to invest by having the senders ponder the consequences of their decisions before they made them, thus reducing the potential for confusion. The senders were asked the following questions.

1. (a) How much money do you think you will send?
2. (b) How much money will your paired receiver receive if you send this much?
3. (c) How much money do you think will be returned to you?
4. (d) How much money would you return if you were the receiver?

The researchers thought that this fourth treatment (which should prompt strategic reasoning) would lead to significant drops in both the amounts sent and, consequently, the amounts returned. In a fifth and final treatment, they not only

have the senders fill out the questionnaire but also present them with a version of Figure 9.3 showing the various amounts returned by the receivers.

IS A TRUSTING DECISION ANALOGOUS TO A RISKY ONE?

Any time the sender in the investment game decides to repose trust on the receiver, that is, whenever someone decides to trust a stranger, he/she is implicitly taking a chance. There is some chance that the recipient of that trust will turn out to be trustworthy and repay that trust, making both parties to the transaction better off. But there is also a chance that the receiver will renege and take the entire amount, leaving the trustor worse off than if he/she had not trusted at all. Thus, the decision to trust may be thought of as being similar to buying a lottery ticket. With some chance, you will make a lot of money, but with some chance, you will earn nothing and lose the amount you spent buying the ticket(s). Do people who are confronted with a situation where they have to repose their trust

In a stranger behave as if they are essentially buying a lottery ticket? By and large, the answer turns out to be in the negative. The mental algorithm that is called upon when asked to repose trust in a stranger seems to be substantially different from the one that is called upon when people buy a lottery ticket.

One of the early attempts to disentangle trust from risk was undertaken by Chris Snijders and Gideon Keren. They look at a simpler version of the Berg-Dickhaut-McCabe investment game. In the Snijders-Keren version of the game, the sender has two options - (1) to send all $10 so that the receiver gets $30 or (2) to send nothing. If the sender chooses the second option (which is similar to the sender's decision to send nothing in the investment game) then both the sender and the receiver end up with some default amount. For the sake of convenience, let us say this sum is $10. So, in the absence of trust, each party gets $10. However, if the sender does

isthat they try to evaluate people’s attitudes towards risk within the context of the investment game itself, whereas a better option would be to evaluate this using a different task. Bohnet and Zeckhauser have participants take part in three different games. First, they play the binary choice version of the investment game where the sender has two options, as in Snijders and Keren. Next, they take part in a second game where the senders are essentially making a lottery choice. They are posed the following question: suppose they send all $10 and there is some chance that they will get back $20 (i.e., gain $10) and some chance that they will get back $0 (i.e., lose $10). Senders are asked to state under what circumstances they would be willing to send $10. Would they do it if the chance of getting back $20 is 50% and getting back $0 is 50% (which implies an expected return of $10)? How about if the chance of getting back $20 is 60% or 70% and so on? The researchers had already decided

The actual chance of getting back $20 prior to the beginning of the experiment. Suppose the chance of getting back $10 and thereby making an extra $10 is 50% and the chance of getting back nothing is 50%. Every participant who states that he/she is willing to send all $10 as long as there was a 50% chance of getting back $20 then plays this game. If he/she stated that he/she would not send any money unless the chance of getting back $20 was more than 50%, then he/she did not have to play the lottery game and simply kept the initial endowment of $10.

However, one issue with the lottery game is that this is an individual decision-making game where there is no receiver, while in the investment game there is a sender and a receiver, and we have seen that often the senders do care about what happens to the receiver. Thus, Bohnet and Zeckhauser have their participants take part in a third game – the risky dictator game. The risky dictator game is similar to the lottery choice game. But now,

Bohnet and Zeckhauser find that sender behaviour is indeed different in the investment game compared to the lottery choice game or the risky dictator game. People are much less willing to send money and take the chance of being exploited in the investment game while their behaviour in either the lottery choice game or the risky dictator game are not different. Bohnet and Zeckhauser comment: "Our results suggest that the decision to trust is influenced by more than just risk... They behave as though there is a betrayal cost above and beyond any dollar losses".

Altro esperimento (libro) risultati: While it