WP313_BG_exp2_2021

Abstract

In our experiment, participants engage in three subsequent one-shot investment games. Investors initially choose between keeping or investing 10 monetary units (MU) with recipients, who, in the case of investment, decide how much of the tripled amount to repay to investors. The three investment games differ solely with a view to whether (i) no one, (ii) only the investor, or (iii) both the investor and recipient have access to a machine prediction about whether the recipient is most likely to pay back more than 10 MU (repayment) or not (no repayment). We employ the strategy method, allowing us to observe recipients’ behavior for both actual and counterfactual predictions.