Unraveling endogeneity in seat capacity and Fares: Time series econometric models for airline origin-destination passengers forecasting

Abstract

Accurate prediction of origin-destination (O-D) air travel passengers is critical for airline schedule profitability analysis, as it enables airlines to align capacity with demand, optimize fare structures, and minimize operational inefficiencies. Reliable forecasts also support strategic decision-making by identifying profitable routes, reducing overcapacity risks, and enhancing network connectivity. This study explores the role of seat capacity and fares in forecasting O-D passengers through the development of three experimental models: the baseline Seasonal Autoregressive Integrated Moving Average (SARIMA), Vector Autoregression (VAR) models with endogenous variables, and SARIMAX models with exogenous variables. The models are applied to two O-D pairs, LAX-JFK and IST-LHR, and extended to a larger sample of 2000 O-D pairs for more comprehensive analysis. Results reveal that treating seat capacity and fares as exogenous variables significantly improves forecasting accuracy of passengers, with the SARIMAX models outperforming the VAR models, which incorporate these variables as endogenous factors. The findings suggest that seat capacity is best modeled as an exogenous variable, consistent with airlines’ scheduling practices, where seat capacity may vary across different scheduling periods. This study contributes to the literature by providing insights into the complex relationships between seat capacity, fares, and passengers, while offering a scalable approach for forecasting across large airline networks.