단변량 시계열모형과 계량경제모형의 정확도 비교

A Comparison of Accuracy between Univariate and Econometric Models

The purpose of this paper is to compare accuracy among tourism forecasting models to predict tourism demand to South Korea from Japan and U.S.A. For this end, annual data from 1970 to 2000 were used for within sample accuracy, whereas data from 2001 to 2004 were employed for testing period in order to test post sample accuracy. The criterion used for testing accuracy is value of MAPE. The empirical results indicate that econometric model for Japan and exponential smoothing for USA performed best in terms of within sample accuracy, while naive 1 model for USA and ARIMA model for Japan outperformed others on the basis of ex post forecasting accuracy. The results suggest that no single forecasting method performs consistently best across within and post sample accuracies.