Unraveling systematic and unobserved heterogeneity in willingness-to-pay for traffic safety improvement: A hybrid decision tree and latent class Tobit approach with evidence from Pakistan

Evaluating traffic safety initiatives through cost-benefit analysis is essential for policymakers seeking effective resource allocation, and understanding individuals’ willingness-to-pay (WTP) for improved traffic safety has therefore garnered increasing attention in recent years. However, important challenges remain in modeling WTP, namely, (i) the absence of systematic higher-order interactions in models, resulting in omitted variable bias and inaccurate model inferences, and (ii) the use of conventional models that presume a constant impact of explanatory variables on WTP, overlooking unobserved heterogeneity. This study addresses these gaps by integrating a decision tree with a latent class Tobit (LCT) model. The decision tree identifies potential systematic higher-order interactions that serve as inputs to the LCT model. The LCT model captures unobserved heterogeneity by partitioning individuals into distinct classes and identifying class-specific factors. The proposed modeling framework is tested on a sample of 653 car drivers from Peshawar, Pakistan. Results reveal two distinct classes in the population: “high” and “low” financial contributors. They further show that variables not significant in a conventional Tobit model become significant in different classes of the LCT model. Higher-order interactions identified through the decision tree are also incorporated in the model, capturing systematic heterogeneity in preferences. In addition, WTP for fatal and severe injury risk reductions is estimated and used to calculate the corresponding risk-reduction values. Overall, the proposed modeling framework enhances understanding of individuals’ sensitivities toward WTP for improved traffic safety and provides actionable insights for policymakers seeking to target distinct population groups.