Econometric Analysis of Pandemic Disruption and Recovery Trajectory in the U.S. Rail Freight Industry

Abstract

To measure the impacts on U.S. rail and intermodal freight by economic disruptions of the 2007-09 Great Recession and the COVID-19 pandemic, this paper uses time series analysis with the AutoRegressive Integrated Moving Average (ARIMA) family of models and covariates to model intermodal and commodity-specific rail freight volumes based on pre-disruption data. A framework to construct scenarios and select parameters and variables is demonstrated. By comparing actual freight volumes during the disruptions against three counterfactual scenarios, Trend Continuation, Covariate-adapted Trend Continuation, and Full Covariate-adapted Prediction, the characteristics and differences in magnitude and timing between the two disruptions and their effects across nine freight components are examined. Results show the disruption impacts differ from measurement by simple comparison with pre-disruption levels or year-on-year comparison depending on the structural trend and seasonal pattern. Recovery Pace Plots are introduced to support comparison in recovery speeds across freight components. Accounting for economic variables helps improve model fitness. It also enables evaluation of the change in association between freight volumes and covariates, where intermodal freight was found to respond more slowly during the pandemic, potentially due to supply constraint.