ANALYSIS AND FORECAST OF CROATIAN TOURISM DEMAND SEASONALITY

Purpose - Investigate the issue of seasonality of Croatian tourism demand. Design - We have established the seasonality of Croatian tourism as a whole by measuring monthly overnight stays ("overnights") using advanced automated as well as original methods. We also investigated land vs. coast contributions to Croatian hospitality industry. Methodology - Seasonal ARIMA (SARIMA) was utilized to analyze above data and forecast future demand in terms of the "overnights" time series. Approach - Having proven the standard SARIMA automated methods deficient, we developed an alternative approach and proved its reliability by a comparison to actual data. Findings - The smoothing nature of ARIMA forecasting algorithm leads to a mitigation of seasonal effects in its forecast. Therefore, an alternative approach is called for. Originality of the research - Due to the apparent shortcomings of ARIMA forecasting we developed an alternative model that pays due attention to the strong seasonal effects of tourism demand common to Mediterranean countries. Nonetheless, we anchored our approach in some existing concepts (e.g. random walk). Still, our concept of "tiers" that mimic standard quarters but preserve the information on crucial differences between peak and shoulder months is, for all our research, original. Our results demonstrate clearly that we have found a fine balance between automated methods and expert judgment.