Controlling for Specific Years and Tourist-origin Countries

In this subsection, a modified tourism model of gravity-type is used to allow for the heterogeneity of tourist-origin countries and specific years. The estimation results are demonstrated in Tables 6-8. As shown in column (1) of Table 6, the coefficient of per capita GDP is 0.253, being positive and statistically significant, which means that holding other constant, a 10% rise in per capita GDP would be associated with a 2.53% increase in tourist arrivals. The coefficient of distance is -1.958, being negative and statistically significant, indicating that tourist arrivals should decrease 19.58% if the distance between two countries increases by 10%. The coefficient of percentage of industry added-value to total GDP is 0.911, being positive and statistically significant, predicting that holding other tings constant, a 10% rise in the industry value in percent of GDP would be associated with a 9.11% increase in tourist arrivals.

The population variable is employed into the model to test if the estimates of the threshold variables are sensitive. In column (2), the coefficient of per capita GDP is 0.755, being positive and statistically significant, which means that holding other constant, a 10% rise in per capita GDP would be associated with a 7.55% increase in tourist arrivals. The coefficient of distance is -2.224, being negative and statistically significant, indicating that tourist arrivals should decrease 22.24% if the distance between two countries increases by 10%. The coefficient of percentage of industry added-value to total GDP is 1.172, being positive and statistically significant, predicting that holding other tings constant, a 10% rise in the industry value in percent of GDP would be associated with a 11.72% increase in tourist arrivals. The coefficient of population is 0.507, being positive and statistically significant, which indicates that tourist arrivals should increase 5.07% if the population of country i rises by 10%.

22

Next, the variables of bilateral trade flows and the exchange rate in tourist-origin country are also added as an estimator in the model. In column (3), the coefficient of per capita GDP is 0.153, being positive and statistically significant, which means that holding other constant, a 10% rise in per capita GDP would be associated with a 1.53% increase in tourist arrivals. The coefficient of distance is -1.027, being negative and statistically significant, indicating that tourist arrivals should decrease 10.27% if the distance between two countries increases by 10%. The coefficient of percentage of industry value-added to total GDP is 0.540, being positive and statistically significant, predicting that holding other things constant, a 10% rise in the industry value in percent of GDP would be associated with a 5.4% increase in tourist arrivals. The coefficient of population is 0.06, being positive but statistically insignificant, which indicates that tourist arrivals should increase 0.6% if the population of country i rises by 10%. The coefficient of total trade is 0.789, being positive and statistically significant, which means that holding other constant, a 10% rise in total trade should be associated with a 7.89% increase in tourist arrivals.

In column (4), the coefficient of per capita GDP is 0.143, being positive and statistically significant, which means that holding other constant, a 10% rise in per capita GDP would be associated with a 1.43% increase in tourist arrivals. The coefficient of distance is -1.092, being negative and statistically significant, indicating that tourist arrivals should decrease 10.92% if the distance between two countries increases by 10%. The coefficient of percentage of industry value-added to total GDP is 0.649, being positive and statistically significant, predicting that holding other tings constant, a 10% rise in the industry value in percent of GDP would be associated with a 6.49% increase in tourist arrivals. The coefficient of population is 0.072, being positive but statistically insignificant, which indicates that tourist arrivals should increase 0.72% if the population of country i rises by 10%. The coefficient of total

23

trade is 0.781, positive and statistically significant, which means that holding other constant, a 10% rise in total trade should be associated with a 7.81% increase in tourist arrivals. The coefficient of exchange rate is -0.028, being negative and statistically insignificant, indicating that tourist arrivals should decrease 0.28% if the exchange rate falls by 10%.

To examine whether other social-economic factors drive tourism flows, the variables of urban population ratio and price level are also introduced into the model.

In column (5), the coefficient of per capita GDP is 0.069, being positive and statistically significant, which means that holding other constant, a 10% rise in per capita GDP would be associated with a 0.69% increase in tourist arrivals. The coefficient of distance is -0.991, being negative and statistically significant, indicating that tourist arrivals should decrease 9.91% if the distance between two countries increases by 10%. The coefficient of percentage of industry added-value to total GDP is 0.568, being positive and statistically significant, predicting that holding other tings constant, a 10% rise in the industry value in percent of GDP would be associated with a 5.68% increase in tourist arrivals. The coefficient of total trade is 0.824, being positive and statistically significant, which means that holding other constant, a 10%

rise in bilateral total trade should be associated with a 8.24% increase in tourist arrivals. The coefficient of exchange rate is -0.022, being negative but statistically insignificant, indicating that tourist arrivals should decrease 0.22% if the exchange rate falls by 10%.

In column (6), the coefficient of per capita GDP is 0.078, being positive and statistically significant, which means that holding other constant, a 10% rise in per capita GDP would be associated with a 0.78% increase in tourist arrivals. The coefficient of distance is -1.253, being negative and statistically significant, indicating that tourist arrivals should decrease 12.53% if the distance between two countries

24

increases by 10%. The coefficient of percentage of industry value-added to total GDP is -0.071, being negative and statistically insignificant, predicting that holding other tings constant, a 10% rise in the industry value in percent of GDP would be associated with a 0.71% decrease in tourist arrivals. The coefficient of total trade is 0.844, being positive and statistically significant, which means that holding other constant, a 10%

rise in total trade should be associated with a 8.44% increase in tourist arrivals. The coefficient of price level is -0.003, being negative but statistically insignificant, indicating that tourist arrivals should decrease 0.03% if the price level falls by 10%.

In column (7), the coefficient of per capita GDP is 0.155, being positive and statistically significant, which means that holding other constant, a 10% rise in per capita GDP would be associated with a 1.55% increase in tourist arrivals. The coefficient of distance is -0.96, being negative and statistically significant, indicating that tourist arrivals should decrease 9.6% if the distance between two countries increases by 10%. The coefficient of percentage of industry value-added to total GDP is 0.53, being positive and statistically significant, predicting that holding other tings constant, a 10% rise in the industry value in percent of GDP would be associated with a 5.3% increase in tourist arrivals. The coefficient of total trade is 0.842, being positive and statistically significant, which means that holding other constant, a 10%

rise in total trade should be associated with a 8.42% increase in tourist arrivals. The coefficient of exchange rate is -0.027, being negative but statistically insignificant, which means that tourist arrivals should decrease 0.27% if the exchange rate falls by 10% The coefficient of urban population ratio is -0.526, being negative and statistically significant, predicting that tourist arrivals should decrease 5.26% if the urban population ratio falls by 10%.

In column (8), the coefficient of per capita GDP is 0.09, being positive and statistically significant, which means that holding other constant, a 10% rise in per

25

capita GDP would be associated with a 0.9% increase in tourist arrivals. The coefficient of distance is -1.255, being negative and statistically significant, indicating that tourist arrivals should decrease 12.55% if the distance between two countries increases by 10%. The coefficient of percentage of industry value-added to total GDP is 0.058, being positive but statistically insignificant, predicting that holding other tings constant, a 10% rise in the industry value in percent of GDP would be associated with a 0.58% increase in tourist arrivals. The coefficient of total trade is 0.846, being positive and statistically significant, which means that holding other constant, a 10%

rise in total trade should be associated with an 8.46% increase in tourist arrivals. The coefficient of exchange rate is -0.018, being negative but statistically insignificant, which means that tourist arrivals should decrease 0.18% if the exchange rate falls by 10% The coefficient of urban population ratio is -0.131, being negative but statistically insignificant, predicting that tourist arrivals should decrease 1.31% if the urban population ratio falls by 10%. The coefficient of price level is -0.002, being negative but statistically insignificant, indicating that tourist arrivals should decrease 0.02% if the price level falls by 10%.

The estimation results of total tourism flows have been shown above. In next subsection, this thesis turns to make sensitivity analyses and discuss how robust the above results are.

26 Note: Standard errors in parentheses, ^* p < 0.05, ^** p < 0.01, ^*** p < 0.001.

 Table 6. The Estimation Results of Total Tourism Flows

(1) (2) (3) (4) (5) (6) (7) (8)

27

 Table 7. The Coefficient Estimates of Year Dummies

(1) (2) (3) (4) (5) (6) (7) (8)

lnVTT lnVTT lnVTT lnVTT lnVTT lnVTT lnVTT lnVTT

Y1997 -0.021 0.021 -0.024 -0.014 -0.022 -0.032 -0.020 -0.024

(0.155) (0.136) (0.102) (0.102) (0.102) (0.107) (0.101) (0.107)

Y1998 -0.016 0.069 0.024 0.038 0.024 -0.024 0.039 -0.006

(0.155) (0.137) (0.102) (0.103) (0.103) (0.113) (0.101) (0.114)

Y1999 0.023 0.093 0.050 0.042 0.035 0.045 0.045 0.044

(0.155) (0.136) (0.102) (0.102) (0.102) (0.116) (0.101) (0.117)

Y2003 -0.219 -0.225 -0.222^* -0.223^* -0.222^* -0.221^* -0.215^* -0.218^*

(0.155) (0.136) (0.102) (0.102) (0.102) (0.106) (0.101) (0.106)

Y2008 0.075 -0.178 -0.230^* -0.226^* -0.207^* -0.283^** -0.245^* -0.288^**

(0.151) (0.134) (0.101) (0.101) (0.100) (0.104) (0.100) (0.105)

Y2009 0.044 -0.158 -0.033 -0.025 -0.002 -0.112 -0.026 -0.104

(0.154) (0.136) (0.102) (0.102) (0.102) (0.109) (0.101) (0.109)

Y2010 0.108 -0.197 -0.244^* -0.235^* -0.212^* -0.317^** -0.246^* -0.313^**

(0.153) (0.137) (0.103) (0.103) (0.102) (0.104) (0.101) (0.104)

Y2011 0.153 -0.208 -0.295^** -0.283^** -0.258^* -0.379^*** -0.302^** -0.376^***

(0.154) (0.138) (0.104) (0.104) (0.103) (0.104) (0.102) (0.105)

Y2012 0.203 -0.167 -0.208 -0.195 -0.167 -0.298^** -0.210^* -0.292^**

(0.158) (0.142) (0.107) (0.107) (0.106) (0.110) (0.105) (0.110)

Y2013 0.248 -0.146 -0.171 -0.158 -0.127 -0.284^* -0.162 -0.271^*

(0.164) (0.148) (0.111) (0.111) (0.110) (0.117) (0.109) (0.118)

Note: Standard errors in parentheses, ^* p < 0.05, ^** p < 0.01, ^*** p < 0.001.

28

 Table 8. The Coefficient Estimates of Country Dummies

(1) (2) (3) (4) (5) (6) (7) (8)

lnVTT lnVTT lnVTT lnVTT lnVTT lnVTT lnVTT lnVTT

CHN 1.468^*** -0.180 -0.013 -0.186 0.001 -0.083 -0.067 -0.183

(0.361) (0.343) (0.257) (0.278) (0.259) (0.260) (0.257) (0.278)

DEU 1.083^*** 0.217 -0.127 -0.163 -0.095 -0.072 -0.193 -0.115

(0.176) (0.169) (0.128) (0.130) (0.125) (0.127) (0.126) (0.132)

JPN 1.502^*** -0.0637 0.378^* 0.393^* 0.553^*** 0.063 0.546^*** 0.105

(0.212) (0.224) (0.169) (0.169) (0.147) (0.308) (0.145) (0.310)

KOR -1.107^*** -1.835^*** -0.784^*** -0.768^*** -0.652^*** -1.059^*** -0.533^*** -0.990^***

(0.230) (0.211) (0.166) (0.166) (0.155) (0.168) (0.155) (0.181)

MYS 0.189 0.454^* 0.116 0.003 -0.013 -0.090 0.027 -0.159

(0.207) (0.184) (0.139) (0.155) (0.156) (0.145) (0.154) (0.169)

PHL -1.284^*** -1.116^*** -0.311^* -0.423^* -0.370^* -0.876^*** -0.315 -0.882^***

(0.229) (0.202) (0.156) (0.171) (0.169) (0.173) (0.167) (0.195)

THA -0.025 0.014 0.300^* 0.219 0.249 0.089 0.008 -0.014

(0.212) (0.187) (0.140) (0.149) (0.149) (0.144) (0.158) (0.167)

USA 4.226^*** 2.771^*** 1.265^*** 1.285^*** 1.332^*** 1.213^*** 1.208^*** 1.214^***

(0.188) (0.202) (0.168) (0.168) (0.166) (0.165) (0.167) (0.168)

VNM -0.860 -0.540 -0.436 -0.367 -0.406 -0.754^* -0.551 -0.733

(0.575) (0.507) (0.380) (0.382) (0.382) (0.375) (0.379) (0.378)

Note: Standard errors in parentheses, ^* p < 0.05, ^** p < 0.01, ^*** p < 0.001.

29