Short-term forecasting of Japanese tourist inflow to South Korea using Google trends data

It is found that Google-augmented models perform much better than the standard time-series models in terms of short-term forecasting accuracy, and in particular, Google models show better out-of-sample forecasting performance than in- sample forecasting.

Abstract

We utilize the Internet search data from Google Trends to provide short-term forecasts for the inflow of Japanese tourists to South Korea. We construct the Google variable in a systematic way by combining keywords to minimize mean squared or mean absolute forecasting errors. We augment the Google variable to the standard time-series forecasting models and compare their forecasting accuracies. We find that Google-augmented models perform much better than the standard time-series models in terms of short-term forecasting accuracy. In particular, Google models show better out-of-sample forecasting performance than in-sample forecasting.