ABSTRACT
Our main focus so far in this book has been on the single equation tourism demand model, in which an endogenous tourism demand variable is related to a number of exogenous variables. The single equation approach depends heavily on the assumption that the explanatory variables are exogenous. If this assumption is invalid, a researcher would have to model the economic relationships using a system of (or simultaneous) equations method. The popularity of simultaneous equations approaches dates back to the 1950s and 1960s within the context of structural macroeconomic models which were used for policy simulation and forecasting. In estimating these structural models, restrictions were often imposed in order to obtain identified equations. Sims (1980) argued that many of the restrictions imposed on the parameters in the structural equations were ‘incredible’ relative to the data generating process, and hence he suggested that it would be better to use models that do not depend on the imposition of incorrect prior information. Following this argument, Sims developed a vector autoregressive (VAR) model in which all the variables, apart from the deterministic variables such as trend, intercept and dummy variables, are modelled purely as dynamic processes, that is, the VAR model treats all variables as endogenous.
