Portfolio Choice and Risk Attitudes: An Experiment

We study the following basic intuition: when faced with a decision how to split their investment between a risky lottery and an asset with a fixed return, people increase the proportion invested in the risky option the more they like the lottery. We find counter-examples to this, and in fact we find no simple relation between preferences between lotteries and the fraction invested in them. We use three well-documented biases (ambiguity aversion, the illusion of control and myopic loss aversion) to show this. First we replicate the previous results in a laboratory experiment with financial incentives, and then test whether participants are willing to explicitly pay a small sum of money in line with the bias (pay for less ambiguity, more perceived control, or more frequent information about portfolio performance). We then study how portfolio choice depends on these biases. With the parameters chosen, the illusion of control was eliminated when participants were asked to pay to gain more control, and the bias did not affect investment behavior (i.e., participants invested in a risky option the same fraction when faced with more or less control). In the ambiguity treatment, people were willing to pay for less ambiguity, but again the level of ambiguity did not influence investment. Finally, in the myopic loss aversion treatment participants were willing to pay money to have more freedom to choose, even though (in line with the documented bias) they invested less when having more freedom to change their investment.