A hybrid econometric–machine learning framework to support market development in intercity passenger transport: the causal and predictive analytics of economic mobility features

Capturing potential travel demand is crucial for carriers to improve their market performance, especially in developing economies with an emerging middle class and increasing socioeconomic inclusion. However, the impact of upward economic mobility on deregulated transport systems and how carriers can capitalize on this trend to increase revenues remain unclear, as this phenomenon is influenced by several confounding factors. This study aims to estimate and decompose the impact of the inclusiveness boom and bust in Brazil on its domestic intercity travel industry. By utilizing Instrumental Variables Least Absolute Shrinkage and Selection Operator (IV-LASSO) and Quantile Regression, our high-dimension sparse approach intends to estimate the effects of a set of economic mobility features on travel markets. We also employ a meta-machine learning approach based on Stacking Regression to assess the contribution of these features to revenue generation. Our findings suggest that airlines are more efficient than bus carriers at implementing market development strategies to achieve effective market inclusion. The customer retention rate for bus carriers is 32% lower, indicating the need to enhance demand management. Moreover, Stacking outperforms base machine learners in predicting revenues for both transport modes. Finally, an event study carried out for the economic downturn period shows a persistent adverse effect on demand and prices and identifies the moments when the machine learning models perform most poorly.