Modelling dependence between tourism demand and exchange rate using the copula-based GARCH model

Abstract

This paper investigates dependence between tourism demand and exchange rate, using the case of China, and from a new perspective by using copula–GARCH models. The empirical results show that the volatility of exchange rate is not a determinant factor in fluctuation of China's inbound tourism demand from the countries being studied. Furthermore, only Russia exhibits risk-adverse behaviour with extreme SUR depreciation, or CNY appreciation associated with an extreme decline in arrivals. Third, introducing the tail dependence and dynamic dependence between growth rates of tourism demand and exchange rate add much to the explanatory ability of the model. The findings of this study have important implications for destination manager and travel agent as it helps to understand the impact of exchange rates on China inbound tourism demand and provide a complementary academic approach on evaluating the role of exchange rates in the international tourism demand model.

Keywords:

• tourism demand
• exchange rate
• GARCH model
• copula
• dependence

Acknowledgements

The authors thank the Editors, Professors C. Michael Hall and Chris Cooper and anonymous referees for their helpful comments which help to improve the manuscript significantly.

Notes

1. World Tourism Barometer-Advance Release, January 2013.

2. If both and are either smaller or larger than 0.5, this infers that the dependence at time t is higher than previously (Wu, Chung, & Chang, 2012). If that is not the case, then the dependence at time t is lower than previously.

3. China inbound tourism means non-residents travel to China for leisure, business and other purposes.

4. According to the National Bureau of Statistic of China, top six source countries in 2010 are South Korea, Japan, Russia, the USA, Malaysia and Singapore.

5. The X-12ARMA method proposed by the US Census Bureau is the most commonly used method for seasonal adjustment. Readers may refer to Findley, Monsell, Bell, Otto, and Chen (1998) for more methodological details about X-12 ARIMA.

6. The ZA unit root test considers structural break(s) in the intercept, linear trend or both (Zivot & Andrews, 1992).

7. Testing no correlation between u and v: . With α = 0.10, we get the critical value of 10% . If the dynamic conditional dependence lies beyond the critical value of 10%, the dynamic conditional dependence is significant at the 10% level.