Impact of Stock Market Indices and Other Regional Exogenous Factors on Predictive Modeling of Border Traffic with Neural Network Models

Abstract

This paper analyzes the impact of stock market indices, as indicators of political and economic stability, and other regional exogenous factors on the performance of predictive modeling of border traffic using neural network models. To prove the concept, the Saudi–Bahrain corridor through King Fahd causeway is selected as our area of study. These two countries have strong cultural ties and a wide variety of variables affects the incoming and outgoing traffic flows. Various models of artificial neural networks are constructed for different prediction horizons and look-back periods using a dataset prepared for the period from 2003 till 2013. In our study, stock market indices are proposed, for the first time, to be used in border traffic forecasting. These indices are added as a surrogate measure of the political and economic conditions of the countries which are under study. Their effects on models with varying ranges of time-series inputs and different prediction horizons are studied in detail. It is found that including stock market indices and other most relevant local factors has generally improved the prediction performance of the neural network models in all cases. Additional reduction in the prediction error is achieved by the proposed ensemble model trained with different time lags. Yet, the degree of improvement depends on the look-ahead horizon for prediction.