engleski

Models Of Overnights From Tourism Demand In Croatia 1960-98

Some of the EU countries as a Austria, Italy, Germany, Netherlands, and UK are the most important and traditionally fastest growing markets for tourists to Croatia. The purpose of this paper is to investigate movements in the long-run demand for tourist travel by these fife origin countries for Croatia. The cointegration and error correction models are very useful in tourism demand analysis, where the long-run equilibrium behaviour is expected to be a major concern of policy-makers whilst the short-run dynamics provide information for short-term forecasting. In this paper we use the general-to-specific modelling approach of outbound tourism demand from the main set of countries (Austria, Italy, Germany, Netherlands, UK), which generate tourism overnights stays in Croatia. We start with the construction of a general demand model using OLS. Various restrictions on the parameters are tested to determine which type of model (static, autoregressive, growth rate, leading indicator, partial adjustment, finite distributed lag, or dead start model) is appropriate, and diagnostic checking (sum of squared residuals, Breusch - Godfrey LM test for autocorrelation, Jarque-Bera normality test, Engle test, White test, RESET test for model mis-specification) is then carried out in the model selection process. The data are annual and cover the period 1960-98. The following variables are included in the model: the number of overnight stays in Croatia taken by particular country residents (ONS), real personal disposable income of particular country (PDI), and a relative cost-of-tourism variable of particular country (RC) and dummy variable which indicate war conditions from 1991 to 1995. The procedure for doing the error correction model (ECM) is not as simple as in employing the OLS method since it is quite complicated and tedious. Moreover, a set of cointegrating variables is necessary. In proceedings we use the same data to testing for cointegration, we find that there is no cointegration relationship between chosen variables, so modelling error correction procedure is not required because it could result spurious regression. This fact is attested, by our research, which showed, at first hint, that the Engle and Granger two-stage error correction model performs much poorer results than the OLS estimation method.

overnights; demand; tourism; general-to-specific modelling approach; cointegration; ECM; Croatia

nije evidentirano