Abstract
Housing price indices (HPIs) are employed to assess the impact of the business cycle, monetary policy, housing policies, and local market dynamics. However, comparative empirical analysis of different HPI methodologies has not been conducted to measure why or when they may diverge and whether these differences are meaningful. Two leading US HPI choices, the repeat-sale transactional (S&P Case–Shiller) and characteristic-based hedonic (Zillow) indices, although highly correlated, generate different distributions and time-series properties primarily at the city level. The spread between these two HPI choices measures the difference between housing market transaction intensity and a willingness-to-pay characteristic valuation. We find that transactional indices are more volatile, with HPI spreads associated with both macro and local drivers. The transactional index will rise more rapidly in a market with increased buying (positive macro and local market conditions) and fall further in a market with increased selling (negative macro and local market conditions) relative to a hedonic index. A buyer- or seller-biased spread between a transactional and hedonic housing price index (HPI) may impact policy judgments during housing market extremes.