Forecasting Port Container Throughput with Web Search Indexes

The growth rate of port container throughput is a barometer of the economic development of the port located country and city. Port container throughput forecasting can support decision making and policy formulation for maritime transportation. Due to the international industrial transfer, policy changes and the COVID-19 pandemic, the global trade fluctuation is increasing, which brings challenges to port container throughput forecasting. To enhance the predictive accuracy of port container throughput, this study explores to construct web search indexes from search query data (SQD), which are collected from Baidu and Google. Stationarity, cointegration and Granger causality tests are conducted to examine whether the web search indexes can be predictors. Using data from the Port of Shanghai, this study verifies the effectiveness of web search indexes. The empirical results show that web search indexes can help to improve the forecasting performance of models. Particularly, a model with hybrid Baidu and Google indexes can achieve better forecasting performance than the model with single Baidu or Google index.