ABSTRACT
This study examines (1) whether leisure tourism can contribute to economic growth and (2) if it does, whether its growth effects are constant across levels of economic development. Leisure tourism differs from business tourism in the causal relation with economic growth. In addition, the natural and heritage-related resources for leisure tourism are limited and not reproducible. This notion has a significant implication for the relationship between the growth effects of leisure tourism and the level of economic development. Thus, the current study focuses on leisure tourism and controls for the effects of business tourism. As an economy grows, the growth effects of leisure tourism are expected to diminish due to a lack of continued productivity improvement in the tourism industry. The empirical findings in this research reveal that leisure tourism contributes to economic growth at an early stage of economic development, but its contribution becomes weaker as the economy develops.
Notes
1 Business tourism in this study includes only the travellers whose genuine purpose of trips is directly related with the destination country.
2 An exception is Yang, Lin, and Han (Citation2010) which analyses the determinants of international tourist arrivals in China, especially for World Heritage Sites. Since China is very successful in attracting foreign capital, the authors control for business travellers using the inflow of foreign capital.
3 Since the exports of goods and services include the exported tourism services, we subtract the tourism receipts from the international trade (Durbarry Citation2004; Pan, Liu, and Wu Citation2014).
4 In this case, the coefficient for , , captures the indirect effect via . However, since our causality test focuses on the coefficient for , , this indirect effect does not cause a problem.
5 Since estimation and test results can be heavily influenced by extreme values, we winsorized the observations on %∆TOUR at the 1 and 99 percentiles; extreme values are set to the 1 percentile (−45.93%) or to the 99 percentile (55.85%). The results in this study are qualitatively the same as the ones obtained without the winsorization.