In recent years, real estate prices in Seoul have risen sharply,and existing theories have not adequately explained this phenomenon. Akerlof and Shiller pointed out the sentiment factor as crucial in affecting price changes. One of the most critical issues emerging in this area is the detection of people’s sentiments. The survey method is traditionally used to understand people’s sentiments. However, some researchers have insisted that big data indexes, such as online search data, detect people's sentiments more accurately than survey data. There is also a lack of research comparing the usefulness of the survey data and internet search data, especially in the housing prediction model. Therefore, we analyzed and compared the relationship between the real estate prices and people’s sentiment using survey data and internet search data with the Vector Error Correction model (VECM) and the post-sample forecasting errors model. The models used the apartment sale price, stock price, interest rates, and sentiment index. The real estate market sentiment index was provided by the Korea Research Institute for Human Settlement and the survey data and search volume data was provided by Naver, the dominant search engine service company in Korea. The result confirmed that both indexed sentiments positively affect the apartment sale prices. Moreover, post-sample predictive capability evaluation showed that the model with online search data performed better prediction than the one with the survey data. The result means that big data indexes such as online search data detect people's sentiments with more correction than survey data. As a result, this study allows the researchers to explore more areas in the housing prediction model.
본 연구에서는 소비자 심리를 파악할 수 있는 설문 조사와 검색어 데이터를 이용하여 부동산의 대표 매물인 아파트 매매가격 간의 관계를 실증적으로 분석하고 이 둘의 결과를 비교 분석한다. 또한 사후 예측 잔차 모형(Post-sample forecasting errors model) 을 사용해 각 모형의 예측력을 비교 분석한다. 이를 통해 아파트매매가격지수를 더 잘 예측하는 심리변수를 분석을 선정하고자한다. 연구 결과는 부동산 가격 예측 요인에 대한 이해를 제고하고 향후 보다 정교화된 예측모형의 개발을 위한 자료로 활용될 수있을 것으로 기대된다. 나아가 설문에 기반을 둔 심리지수와 빅데이터에 기반을 둔 심리지수를 비교함으로써 부동산 시장에 대한 사람들의 심리를 좀 더 정확히 파악하는 데 빅데이터의 활용가능성을 제시하고자 한다.