Using financial incentives, we study how portfolio choice (how much to invest in a risky asset) depends on three well-known behavioral phenomena: ambiguity aversion, the illusion of control, and myopic loss aversion. We find evidence that these phenomena are present and test how the level of investment is affected by these motivations; at the same time, we investigate whether participants are willing to explicitly pay a small sum of money to indulge preferences for less ambiguity, more control, or more frequent feedback/opportunities to choose the investment level. First, the observed preference for ‘‘control’ ’ did not affect investment behavior and in fact disappeared when participants were asked to actually pay to gain more control. Second, while people were indeed willing to pay for less ambiguity, the level of ambiguity did not influence investment levels. Finally, participants were willing to pay to have more frequent feedback opportunities to change their portfolio, even though prior research has shown that people invest less in risky assets (and earn less) in this case. (JEL B49, C91, D81, G11, G19) I.

This paper focuses on egocentric biases in financial decisions. Subjects first design a portfolio, whereby each combination of assets yields the same expected return and variance of returns. They are then confronted with two alternative portfolios; the average portfolio and the portfolio of one's selected expert. Illusion of expertise prevails if one prefers nevertheless the own portfolio. Using the random price mechanism reveals that most subjects prefer their own portfolio to the average or the expert's portfolio. Illusion of expertise is shown to be stable individually, over alternatives, and for both elicitation methods, willingness to pay and to accept.

This paper investigates (i) the robustness of hindsight bias in experimental asset markets, (ii) the time invariance of the di#erent experimental risk elicitation methods of certainty equivalents and binary lottery choices, and (iii) their correspondence.