Econometric Forecasting of Tourist Arrivals Using Bayesian Structural Time‐Series*

Abstract

This article introduces the Bayesian structural time series (BSTS) as a potential tool for forecasting in the tourism literature. Using data on Australian tourist arrivals in New Zealand, the forecasting accuracy of the estimated model is evaluated using a fixed partitioning approach. The MAPE of the fitted model is 3.11 per cent for the validation stage and 2.75 per cent for the test stage. The BSTS outperforms two other competing models both in the validation and test stage. In addition to forecasting, BSTS also estimates the trend, trend slope, and seasonality that change over time.