사전확률분포를 활용한 주택경매낙찰가의 동태적 예측 분석

An Empirical Study on Dynamic Prediction of Housing Auction Prices Using Prior Probability Distribution

In order to forecast time series by analyzing housing auction prices, this study analyzed and predicted dynamic relations among the variables of the BVAR model presented to remedy the shortcomings of the unconstrained VAR model. This study also compared forecast errors between the BVAR model and the VECM model to verify the predictability of the analysis model. The analysis findings and implications are as follows. First, among the variables included in the BVAR model, Mortgage Rate and Default Rate had highly negative influence on housing auction prices. Second, the BVAR model had a greater predictability than the VECM model, when comparing forecast errors between the BVAR model and the VECM model. This demonstrates the constrained prediction model can predict better than the unconstrained model in analysis of housing auction prices. Third, this study verified the applicability of Bayesian inference, which is mostly used in macroeconomics or in finance to redeem over-parameter estimation of the VAR model that is mainly used to forecast the housing market.