earticle

영어

This study predicted crises expressed as a surge in housing price in Chinese housing market using Probit Model among EWS methods, and analyzed influence on crises of policy variables through the prediction. Prior to full-blown empirical analysis, it reviewed over housing policies announced and being still supplemented by considering the process of the formation of Chinese housing market. Through this, it found out that the housing market in China is greatly influenced by government policies rather than moved by the economic mechanism. Partially, it considered variables that would affect crises of Chinese housing market based upon Probit Model, one of EWS methods and built an early warning system for the housing market. And then, through in-sample & out-of sample prediction, predictability on possible crisis occurrence within a year was verified. As a result, in-sample predictions showed 6 prediction fallacies out of a total of 43 predictions, whose valid predictability reached 86% while out-of-sample prediction showed 2 prediction fallacies out of a total of 6 predictions, whose valid predictability reached 80%. After that, it built an early warning system using only policy variables based upon Probit Model and conducted in-sample & out-of-sample prediction experiments. As a result, in-sample predictions showed 10 prediction fallacies out of a total of 43 predictions, whose valid predictability reached 76.7% while out-of-sample prediction showed only 1 prediction fallacy out of a total of 10 predictions, whose valid predictability reached 90%. Both of the two early warning systems showed considerably high predictability in in-sample & out-of-sample predictions. In-sample predictions showed higher predictability in a model considering variables other than policy variables while out-of-sample predictions showed valid predictability, 80% & 90% in both models respectively, but it’s hard to say that statistically significant difference was found. It’s hard, however, to confirm that effective early prediction is possible using Chinese housing market early warning systems built on the basis of only the results verified by this thesis or that price surges in Chinese housing market can be predicted.

한국어

본고에서는 EWS기법 중 프로빗 모형을 사용하여 중국 주택시장에서 주택가 격 급등으로 표현되는 위기를 예측하여 그 예측력을 통해서 정책변수의 위기에 대한 영향력을 분석하였다 EWS의 주요 기법 중 하나인 프로빗 모형에 기반하여 주택시장 위기에 영향 을 줄 수 있는 변수들을 고려하여 중국 주택시장에 대한 조기경보체제를 구축 하였다. 그 후 샘플 내 예측과 샘플 외 예측을 통해서 1년 이내 위기 발생에 대한 예측력을 검정하였다. 그 결과 샘플 내 예측의 경우 총 43번의 실험 중 6 번의 예측오류가 발생하여 유효예측력은 86%에 달하였다. 그리고 샘플 외 예 측의 경우 총 10번의 실험 중 2번의 예측오류가 발생하여 유효예측력은 80%에 달하였다. 그 후 역시 프로빗 모형에 기반하여 정책변수만을 가지고 조기경보체제를 구 축하고 역시 표본 내 예측과 표본 외 예측실험을 진행하였다. 그 결과 표본 내 예측에서는 총 43번의 실험 중 10번의 예측오류가 발생하여 유효예측력이 76.7%에 달하였다. 표본 외 예측에서는 총 10번의 실험 중 1번의 예측오류만이 있어 유효예측력이 90%에 달하는 결과가 나타났다.