Controlling for endogeneity between bus headway and bus ridership: A case study of the Orlando region

In this study, we develop an advanced econometric model that considers the potential endogeneity of stop level headway in modeling bus ridership. We recognize that bus stops with higher potential demand are also like to have higher frequency of buses (or lower headway between buses). We consider headway endogeneity by proposing a simultaneous equation system that considers headway and ridership in a joint framework. The proposed model is developed employing stop level ridership data from the Orlando region for 11 quadrimesters (four-month time periods). The presence of multiple data points for each stop allows us to develop panel models for headway, boarding, and alighting. The headway variable is modeled using a panel ordered logit model while the ridership variables are modeled using a panel grouped ordered logit models. The model estimation results justify the consideration of headway endogeneity in bus ridership analysis. To illustrate the value of the proposed model, a validation exercise and a policy analysis exercise are conducted