Quantitative Research Method
It means that the income level of the original countries has positive impact on the international tourism demand in Hong Kong. The lagged-dependent variable is significant in all countries researched except China This suggests t information from people who travelled to Hong Kong past and/or the consumer persistence features vitally in the international tourism demand in Hong Kong (Song, H. et al, 2003). The deep implication of this finding is that the more comfortable the past traveller felt in Hong Kong, the more travellers that will visit to Hong Kong in the future.
Because of this, enhancing the quality of service for travellers is important for Hong Kong to attract more tourists in the future. The P variables measures the costs of tourism in Hong Kong relative to that in the original countries. According to the Table 1, the price level of the source market is significant in eight countries. The exceptional cases are Japan and Korea. This suggests that travellers always care bout the relative tourism price in Hong Kong to that in their own countries.
The higher the price ratio, the less likely the tourists travels to Hong Kong, because the minor sign in coefficient of P. The substitute price variables is only significant in three out of ten countries, which suggests that the competitive advantage of price in other tourism destination is less crucial for tourists to choose Hong Kong as a tourism destination. It seems that the financial crisis in Asia takes a negative influence on the international tourism demand in Hong Kong, because the relative dummy variables has the minor sign in China, Japan and Thailand.
In 1997, the hand over of Hong Kong to China might take negative impact on tourism from the I-JK choosing Hong Kong as the destination, because the D97 was included in the I-JK model (Song, H. et al, 2003). Oil crisis seems have less impact on the international tourism demand in Hong Kong. The diagnostics statistics shows that most of the model passes all five diagnostics tests, except Korea which fail only one of five tests. The plot about normality test and heteroskedasticity test are listed in the Appendix I and Appendix II.
Korea demand model cannot pass the normality test, because the histogram ploy is more peaked han a normal curve. The eigen value are all less than 30. This means that all models pass the collinearity test. D-W number close to 2 indicates no serial correlation, suggesting all models pass the serial correlation test. The sample size is Just 27 which is not so much in a statistics model. But I also got the successful estimated demand model for these ten countries. Thus, the results in Table 1 are quite reasonable and convincible.