Research Note: World Tourism Forecasting – Keep It Quick, Simple and Dirty

Abstract

Building multivariate causal models for world tourism or world subregions could be very difficult as the necessary short-term information, such as monthly data on the independent variables determining tourism demand, is either not available or very labour and cost intensive to obtain. In order to solve the short-term forecast problem on a global scale, a quasi-causal model was constructed to explain international arrivals. This model was based on a REGARIMA approach, which used as its exogenous variable the flexible trend of the arrivals being explained through the model. The flexible trend was identified by the HP-filter method and indicated in the model important exogenous aggregated information. ARIMA models were developed and ‘absolute no change’ forecast values computed in order to benchmark the forecast accuracy of the REGARIMA model. The study demonstrated that the simple ARIMA approach could not, in general, be outperformed by the more complex quasi-causal REGARIMA approach. Both time series approaches outperformed the ‘absolute no change model’.