Tourist demand is subject to considerable variations, a fact which aggravates the development of forecast models of sufficiently adequate accuracy. This study develops models that permit including most if not all factors of influence. To this end, due consideration was given to calendar effects as well as unknown special effect in time series models and econometric approaches in an attempt to improve the quality of the forecast. Results showed that for the data set used, a combination of complex data adjustment procedures and adequat model structures substantially improved the accuracy of the forecast or simple approaches lose out to more complex ones. Accordingly, the central issue of this study, i.e. whether complexity matters, can certainly be answered quite simply by "yes, it does."
calendar effects econometric indicator approach multivariate time series model outlier detection transfer function