거래빈도가 낮은 시장에서의 실거래 부동산 가격지수 작성에 관한 연구 - 강남구를 대상으로 -

In this article, I examine both the hedonic and time-series approaches for constructing real estate price indexes for markets that have few transactions. The thinness of the market does have a marked effect on the precision of the index estimate. The index from the hedonic model by using time dummy variables loses precision rapidly when the number of transactions drops. The imprecision is manifest in a much more volatile index. At the ten transactions per period level of market thinness, the imprecision is manifest. At two transactions per period, the index becomes wholly unreliable. On the other hand, the time-series price index shows much better in thin markets. It is more parsimonious than the hedonic price index and also potentially more accurate and less prone to outliers. It remains close to the true index when there are small transactions per period, and it does not have the volatility of the hedonic index. When there are two transactions per period, the index itself is stable, although it is prone to sample selection bias. Nevertheless, the index retains reasonable precision, and the estimate of the fundamental volatility of real estate prices is good. It achieves this by linking current transactions to preceding transactions, thereby increasing the set of comparable transactions on which to base the index.