Co-integration in Tourism

The concept of co-integration posits that the linear combination of two or more nonstationary time series can be stationary if they are co-integrated (Granger 1981). Co-integration and error correction models are largely applied to solve the spurious regression problem resulting from the inclusion of nonstationary variables in a regression model.

Two main approaches to modeling co-integration are single-equation residual-based and system-based analyses. One precondition for the above tests is that all variables in the co-integration regression should be integrated with the same order. When different orders of integration are identified, alternative tests such as the autoregressive distributed lag bounds test can be applied (Pesaran et al. 2001).

Since the mid-1990s, a considerable number of tourism researchers have adopted the co-integration method to address a variety of topics related to tourism demand modeling and forecasting. Numerous efforts have been undertaken to empirically...