Reflecting economic activity through ports: The case of Australia

Abstract

With approximately 85% of global trade moved by sea, the relationship between ports and the economy has become symbiotic. Identifying and tracking this relationship is sought by both port economics and port forecasting literature. Tackling both challenges -i.e., the ports-economy relationship and forecasting- at once, can only be pursued by data-driven factor models, through their ability to reduce the dimensionality of large cross sections of time series. We find fertile ground in applying, for the first time, a factor modeling approach to the Australian port sector by utilizing a disaggregate dataset of 2765 series representing national and regional port activity for 20 years. Through our model, we establish a quantifiable connection between ports and the economy and demonstrate their capacity in reflecting economic activity. We assess a rich lead-lag structure in our dataset and trace its cyclical properties. Using the same method, we compare the Australian and U.S port sectors, revealing insights on their structural differences. Finally, utilizing our model as a forecasting device, we report favorable short and mid-term forecasting performance against benchmarks.