Forecasting national port cargo throughput movement using autoregressive models

Abstract

Port services demand planning plays an important role in port capacity planning and management. It enables ports to anticipate, prepare for, and respond to changes in demand, fostering operational excellence and customer satisfaction in the port and maritime industry. This article explores the use of a multivariate forecasting model to predict port cargo throughput movement at a national level considering macroeconomic indicators. The statistical model is used to analyze how the port cargo throughput movement in Mexico is affected by changes in the level of industrial activities in both Mexico and the United States, and to generate a projection of the national port cargo throughput movement for the upcoming years. To achieve this, a multivariate time series analysis with vector autoregressive models was constructed using monthly frequency data from 2010 to 2022. The results of the autoregressive model indicate that the proposed macroeconomic variables have a Granger-causal effect on port cargo throughput movement. It was also found that an incremental shock from the U.S. economy has a positive effect that is transmitted temporarily during the first six immediate months, while changes in the national economic activity also have a temporary positive effect, but only during the first immediate period. Traditional forecasting performance metrics are used to evaluate the effectiveness of the proposed model.