본 연구는 칼만필터 앨고리즘의 상태공간모형을 이용하여 1995년부터 2011년까지 미국, 일본, 중국의 월별 관광객수, 소비자물가지수, 국내총생산, 환율을 통해 관광객의 수요탄력성을 파악하고자 한다. 본 연구의 결과를 요약하면, 첫째, 관광수요모형의 칼만필터 추정결과는 3개의 소득탄력성이 통계적으로 유의하나, 자체가격과 대체가격탄력성은 모든 수요모형에서 유의하지 않았다. 수요탄력성의 기대부호는 3개의 소득탄력성의 경우만이 양(+)의 부호로 일치하여 소득탄력성 변수만이 한국관광수요의 결정요인이었다. 관광수요는 3개의 자체가격탄력성이 모두 탄력적이었고 양(+)의 값을 보여주어 대체재의 관계가 존재하여 한국관광가격은 비경쟁적이며 가격경쟁에 목표를 둔 전략들로 추가적인 수익을 창출할 것이다. 미국의 가격탄력성은 매우 크고 일본의 소득탄력성은 탄력적으로 사치재의 관계를 나타내고 대체가격탄력성은 모두 비탄력적이고 미국은 대체재, 일본과 중국은 보완재의 관계가 존재하였다. 둘째, 상태공간을 이용한 미국, 일본 및 중국의 수요탄력성의 칼만필터 추정결과는 소득탄력성에 있어서 글로벌 금융위기기간인 2007년과 2008년에 미국은 이전과 거의 동일하고, 일본은 약간 높고 중국은 조금 낮아졌다. 2011년의 소득탄력성이 미국은 증가, 일본은 대폭 증가, 중국은 이전과 거의 동일하여 한국으로의 여행은 미국과 중국의 관광객들에게는 필수재, 일본의 관광객들에게 사치재로 인식되었다. 가격탄력성에 있어서 글로벌 금융위기기간인 2007년과 2008년에 미국은 이전보다 조금 높고, 일본은 대폭 높고 중국은 약간 높아졌다. 2011년의 가격탄력성이 미국은 대폭 증가, 일본은 거의 동일, 중국은 대폭 감소하였다. 따라서 본 연구에서 이용된 상태공간모형이 일정한 모수를 가정하는 전통적인 회귀분석방법보다 모수의 불안정을 추정하는 관광수요모형의 칼만필터 앨고리즘과 더불어 관광객들의 시간가변 수요탄력성을 더욱 정확히 파악하는데 적절하다는 것을 확인하였고, 또한 관광수요모형의 수요탄력성은 관광정책과 관광제도의 변경, 소비자의 취향변화, 기대감 등과 같은 비관측요인들로 인하여 시간가변적일 가능성이 존재하였다.
The purpose of this study was to examine the elasticity in the demand for tourists through the relation between tourists’s arrivals of source country and consumer price index, gross domestic products, the exchange rates in Korea, the United States, Japan, China by using a state space model with Kalman filter algorism from January 1995 to December 2011.
First, in the Kalman filter estimates of tourism demand models, three income elasticities were statistically significant, while the own price elasticities and cross price elasticities were not significant in all demand models. The expected sign of demand elasticities were consistently positive with three income elasticities as the determining factor of Korea tourism demand. According to the tourism demand, three own price elasticities were positively elastic and related to substitutes and the tourism price was non-competitive in Korea. This indicated that the elastic tourism price will create additional revenues in strategy based on the goal to price competition. The price elasticity was very large in the U.S. and the income elasticity was elastic as the luxury goods in Japan, but the cross price elasticity was inelastic. Thus they showed that there were the flexible relations to the United States as substitutes, Japan and China as complements.
Second, by utilizing the State Space, the results of the Kalman filter estimates of demand elasticities showed that three income elasticities increased from 1995 to 2000 and the United States and China were slightly higher in 2003, but Japan was much lower. And the United States were about the same as before in 2007 and 2008 (during the Global financial crisis) and Japan increased slightly, but China was a little lower. In terms of the income elasticity in 2011, it increased in the United States and China, but Japan increased significantly. Thus the travel to Korea was recognized as necessities for the tourists in China and the United States, but as luxuries for tourists in Japan. In the price elasticity, the United States and Japan decreased from 1995 to 2000, but China increased, following the United States and Japan were slightly higher in 2003, but China was almost the same. Thus the United States was a little higher than before and Japan was significantly higher, but China rose slightly in 2007 and 2008. In the price elasticity in 2011, the United States increased significantly and Japan was about the same, but China decreased significantly.
Third, the research potential applications need to be further investigated. First, the equation of state of demand models in this study was based on empirical study of the demand for consumer goods such as home, food and general including durable or not durable goods (Song, Romilly and Liu, 1998; Brown, Song and McGillivray, 1997), which were assumed to follow the walk process. However, the decision-making process of tourists was not necessarily the same as consumers and the tourism products and services consumer demand in general can not relate at all. Therefore, in a great deal of research, the equation of state, setting such as autoregressive (AR), moving average (MA) or autoregressive moving average (ARMA), needs to be done in order to grasp whether it is more appropriate than the course of random walk. This can be done by examining the predictability of the state space model and the state equation of another. Second, this study was focused on investigating whether that could be used to analyze the elasticity of tourist demand related to long term determinants using the state space model. If this results could be compared with the alternative tourism demand model through the prediction performance of the state space model, they were possible to become more help with practitioners and policy makers of tourism. Third, the legitimate process theoretically should be developed for the instability structure and the state space model need to determine more desirable models than the fixed parameter models. The aspects should be detailed in analysis of future research.
Therefore, this study confirmed that the state space models used were appropriate than the traditional regression analysis approaches to examine the Kalman filter estimates of tourism demand models and the time-varying demand elasticities. And in the future, this models will be considered to be particularly useful in simulating the structural changes of tourism demand models that have been frequently altered by unobserved factors such as consumer taste, expectations and policy and regime changes.