Housing price indices for small spatial units

Median-based Housing Price Indices (HPIs) generate potentially misleading indicators especially when applied to Small Spatial Units (SSUs) such as small Metropolitan Statistical Areas (MSAs) or neighborhoods within larger metropolitan regions. Given the small total number of sales in SSUs and the even smaller number of repeat sales, traditional alternatives are few. A matching-based Fisher HPI is proposed as an alternative that provides more accurate estimations for SSUs given that it both controls for housing characteristics and is not restricted just to repeated sales. This paper makes use of housing sales data from the state of Illinois, as an example of the application of the methodology to small MSAs and at the neighborhood level within larger MSAs. The results highlight some of the biases that have arisen from the use of median price indicators.