Improving Willingness-to-Pay Elicitation by Including a Benchmark Good

In this article, we propose and validate a simple way to augment the standard procedure that researchers use to elicit willingness to pay (WTP) for a good or service. The objective is to improve the accuracy and precision of measured WTP by eliminating some of the variation in raw responses that is unrelated to true WTP. The context for the validation exercise is a study in Uganda about parents’ preferences for goods for their children. To understand these preferences, we elicited WTP for various goods for children using the Becker-DeGroot-Marschak (BDM) method (Becker, DeGroot and Marschak, 1964). In the most common version of BDM, the respondent gives their take-it-or-leave-it purchase decision at several prices, knowing that one of the prices will be chosen at random and a transaction will or will not take place at that price, depending on their response. The respondent’s weakly dominant strategy is to answer truthfully. The augmentation we propose is to measure WTP for another good (“benchmark good”) and to use WTP for the benchmark good as a control variable (or, more generally, to adjust WTP for the focal goods). The benchmark good needs to be unrelated to the good the researcher is interested in and unrelated to the independent variables of interest. For example, in our context, the focal goods were human capital goods for children. We were interested in how child gender affects WTP for human capital goods, as well as how mothers’ and fathers’ WTP differ. We used household