A Neural Network model of the impact of political instability on tourism

This paper presents an empirical integration of the dimensions of political instability with traditional exogenous variables, which are usually employed in econometric tourism demand forecasting, within a tourism demand model in order to investigate causal relationships between political instability and tourism. The work uses the POLINST Database, which contains events of political instability from 1977 to 1997 that took place in the Middle East - Mediterranean region. The model is based on a Focused Tapped Delay Line Neural Network (FTDNN) with a sliding time window of 12 months. The evaluation results show that our model can be used to achieve a good estimation of the effects of political instability on tourism. In an extended set of experiments we were able to show the relative importance of the political instability factors on tourism. Finally, our model also allowed to estimated the time lag between a political instability/terrorist event and the reduction of tourist number to the destination.