Using Discrete Choice Models to Estimate Non-market Values: Effects of Choice Set Formation and Social Networks

Abstract

Discrete choice models are often used to estimate non-market values. In standard models, individuals make choices considering all possible alternatives. However, in reality, the set of alternatives individuals consider may differ. Moreover, these choice sets may be influenced by the individual's social networks. For instance, Romeo was going to go to the beach; however after talking to Juliet he is also considering the mountains. Recent research has demonstrated that ignoring the choice set formation (CSF) process leads to biased estimates of non-market values. This paper develops a discrete choice model in which the choice set faced by a decision-maker is influenced by her social network. In the model, a network parameter denominated by social propensity determines the weight a decision-maker places on her network when determining what alternatives to consider. We use Monte-Carlo experiments to investigate the effects of ignoring social networks when modeling CSF. We find that when social propensity is relatively low, CSF models that ignore social networks do not lead to significant bias in welfare estimates. However, as social propensity increases and reaches a certain threshold, welfare estimates that ignore networks are significantly biased and estimates of welfare change are significantly higher than the true welfare change.