Demand forecasting within Montenegrin tourism using Box-Jenkins Methodology for seasonal ARIMA Models

The purpose of this paper is to construct adequate seasonal ARIMA models, using Box-Jenkins methodology, and to implement them in order to forecast short run flows of tourist arrivals and tourist overnight stays in Montenegro. Time scope covers ten years, from 2001/01 to 2010/12, while twelve months of 2011 are out-of-sample forecasts. Close inspection of related time series was applied which revealed no extreme and unusual specificities in the data. Therefore, only economic impacts have been affected the time series. This was important because econometric intervention analysis was excluded from models designing and building. As a result, our approach was based on time series modelling without need to take care of any structural breaks. Modified Box-Pierce and Jarque-Bera test statistics confirmed good quality of the models. Further, the results show excellent forecasting performances of specified models. According to forecasting output, Montenegro can expect upgrowth in terms of tourist arrivals as well as in terms of tourist overnight stays. The model has shown around 7,25% rise in arrivals, which is about 91 thousands tourists more in 2011 compared with the previous year. On the other hand, the calculated rise of overnight stays is around 8,42%, or about 670 thousands more than the year before.