Co-integration, tourism

The inclusion of nonstationary variables in a regression model can lead to spurious regression. The co-integration technique developed by Engle and Granger (1987), along with the error correction models, can solve the spurious regression problem in econometric modeling. They suggest that if a pair of nonstationary variables (e.g., arrivals and income) belongs to the same economic system (Figure 1), there should be an attractor that pushes the economy back toward equilibrium whenever it moves away. A stationary series is denoted I(0), and a nonstationary series is known as I(d), suggesting that the latter will become stationary after being differenced d times. If the linear combination of two nonstationary series y _t and x _t is integrated with order (d-b), where b > 0, then y _t and x _t are said to be co-integrated. This co-integration relationship can be denoted as (y _t, x _t) ~ CI (d, b). This two-variable case definition can be extended to a k-variable co-integration system.

Figure 1

Co...