Public resource allocation, strategic behavior, and status quo bias in choice experiments.

Abstract

Choice experiments, a survey methodology in which consumers face a series of choice tasks requiring them to indicate their most preferred option from a choice set containing two or more options, are used to generate estimates of consumer preferences to determine the appropriate allocation of public resources to competing projects or programs. The analysis of choice experiment data typically relies on the assumptions that choices of the non-status quo option are demand-revealing and choices of the status quo option are not demand-revealing, but rather, reflect an underlying behavioral bias in favor of the status quo. This paper reports the results of an experiment which demonstrates that both of these assumptions are likely to be invalid. We demonstrates that choice experiments for a public good are vulnerable to the same types of strategic voting that affect other types of multiple-choice voting mechanisms. We show that due to the mathematics of choice set design, what is actually strategic voting is often interpreted as a behavioral bias for the status quo option. Therefore, we caution against using current choice experiment methodologies to inform policy making about public goods.