We investigate experimentally the effect of a negative endowment shock in a trust game to assess whether different causes of inequality have different effects on trust and trustworthiness. In our trust game there may be inequality in favor of the second mover and this may (or may not) be the result of a negative random shock (i.e., the outcome of a die roll) that decreases the endowment of the first-mover. Our findings suggest that inequality leads to differences in behavior. First-movers send more of their endowment and second-movers return more when there is inequality. However, we do not find support for the hypothesis that the cause of the inequality matters. Behavior after the occurrence of a random shock is not significantly different from the behavior when the inequality exists from the outset. Our results highlight that we have to be cautious when interpreting the effects on trust and trustworthiness of negative random shocks that occur in the field (e.g., natural disasters). Our results suggest that these effects are largely driven by the inequality caused by the shock and not by any of the additional characteristics of the shock like saliency or uncertainty.

We study behavior in a trust game where first-movers initially have a higher endowment than second-movers but the occurrence of a positive random shock can eliminate this inequality by increasing the endowment of the second-mover before the decision of the first-mover. We find that second-movers return less (i.e., they are less trustworthy) when they have a lower endowment than first-movers, compared with the case in which first and second-movers have the same endowment. Second-movers who received the positive shock return more than those who did not; in fact, second-movers who received the positive shock return more than second-movers who had the same endowment as the first-mover from the outset. First-movers do not seem to anticipate this behavior from second-movers. They send less to second-movers who benefited from a shock. These findings suggest that in addition to the distribution of the endowments the source of this distribution plays an important role in determining the levels of trust and trustworthiness. This, in turn, implies that current models of inequality aversion should be extended to accommodate for reference points if random positive shocks are possible in the trust game.