Fisher Center for Real Estate and Urban Economics

This paper considers the suitability of four widely used real estate price indices (the S&P Case-Shiller Index; the FHFA index; the NCREIF NPI; and the historical NAREIT series) for valuing and monitoring the credit risk of U.S. mortgages. Our evaluation focuses on four properties of these indexes: unbiased estimation of the drift of real estate price dynamics; unbiased estimation of the volatility of real estate price dynamics; suitability for measuring correlation between interest rates and real estate prices; suitability for measuring growth dynamics, or the effects of real options components of real estate values. We find, despite their widespread use, that these indices generate downwardly biased estimates of the idiosyncratic volatility of prices around the index and, thus systematically undervalue embedded default options in mortgage products. We also identify other specific properties, or required assumptions underlying the construction of these indices, that are likely to be problematic for mortgage pricing. These properties include the fact that none of these indices adequately address the index number problem for real estate assets. These price metrics thus mix price and quantity dynamics in non-transparent ways leading to likely biases in the measurement of the correlation between real estate "prices" and economic fundamentals, such as interest rates, and expected skewness in real estate price distributions arising from the real option components of real estate values. Overall, this analysis suggests there is a need to re-evaluate the use of these indices for mortgage risk management and to develop dynamic hedonic price indices that both address the index number problem and allow for flexible specifications for price and quantity dynamics.