The structure of stated preference questions to value consumption from public infrastructure can vary depending on the conditions of consumption facing the household. Specifically, a good could be offered as a quasi-public or quasi-private good. This paper demonstrates how consumption from two alternative electricity allocation options can be valued using two types of stated preference questions. Since surveyed households were asked two types of questions, the authors develop a joint model of a contingent valuation question and a contingent quantity behavior response that allows for correlation in error terms across models. In their application to two villages in Rwanda, the authors find higher WTP for electricity consumed as a quasi-private good rather than a quasi-public good, with four hours of electricity per day, only in the evening. They also find correlation in the error terms across the two models, suggesting that their joint estimator is more efficient than estimating each model individually.