How ambiguity affects workers’ payment scheme preferences and work performance

Xueting Yang and John Hey

Abstract: This study is designed to understand workers’ understanding of, and preference for, payment schemes which are ambiguous, and to determine whether such schemes affect the work that workers do. Such schemes do not specify the payment that the worker will receive, either deterministically or even probabilistically, but rather specify it in ambiguous terms – for example, in terms of the work that the worker does. The study, therefore, is designed to elicit workers’ preferences over ambiguous payment schemes. To this end, we need to elicit preferences over ambiguity – assessing aversion to, or loving of, ambiguity. To do this, we first need to elicit preferences over risk - assessing aversion to, or loving of, risk. We therefore implemented an experiment with three basic parts: a Risk part, an Ambiguity part and a Work part. In each part, to elicit the relevant preferences, we used the Allocation elicitation method. Working was a crucial part of the experiment; this was implemented by asking subjects to solve a number of puzzles. The number that they solved influenced their payment for the Work part of the experiment. We fitted to the data four models of behaviour under ambiguity: αMaxMin, MaxMin, EU and RDEU, and ranked them in explanatory order.  EU turned out to be the best, indicating that subjects simplified the decision problems and treated ambiguity as risk. Other data confirmed this conclusion.

Parameter estimates: Yang and Hey (MS Excel , 78kb)