When the housing bubble began to burst in 2006-2007, many communities on the urban fringe as well as central-city neighborhoods started to experience the incidence of rising foreclosure. The general claim is that foreclosures also hurt neighbors by devaluing the nearby properties. Immergluck and Smith (2006) find that each additional conventional single-family foreclosure within one-eighth of a mile of a single-family home in Chicago is associated with a drop of approximately 0.9 percent in property value. Lin et al. (2007) finds significant negative effects of foreclosures on the nearby property value within a radius of 0.9 kilometer. The potential benefits of intervention thus depend substantially upon how strong the neighborhood effects are.

This paper offers a new way of conceptualizing and then estimating the potential neighborhood effects of foreclosures. In the context of the successful modeling of housing prices, submarket identification is very crucial since property prices in different submarkets are determined by different functional relationship. On the one hand, two unique characteristics of housing – spatial fixity and durability— suggest that a submarket definition should focus on spatial relationships. Most estimates of the impact of foreclosures have implicitly defined submarkets in this way. On the other hand, housing stock heterogeneity, particularly within center-city neighborhoods, suggests that an alternative definition may be appropriate. This approach defines submarkets as dwellings that are close substitutes for one another across a wide range of characteristics not restricted to location (Grigsby et al, 1987). It suggests that submarkets may be comprised of non-contiguous pieces of real estate. Thus, units that are not close by could be viewed as substitutes by the potential owner or renter. For example, a triple-decker is presumably of a lower quality than a single-family housing, even though they may locate next to each other and may have similar selling prices.

A more appropriate definition of submarkets suggests that foreclosure has an impact on excess supply (and reduced value) within an entire housing submarket, whether or not properties are immediately adjacent to each other. In contrast, filtering approach suggests cross-submarket effects. As foreclosed properties come on to the market by adding to the local supply of available units, there may be a longer-term housing spill over from one submarket to another as a consequence of shifts in relative prices or changes in housing quality.

This study uses a dataset that covers twenty years (1988-2008) of housing values from Worcester, Massachusetts, a mid-sized New England city with a wide variety of housing types. The study develops three alternative classifications of housing submarkets (spatial, hedonic and similarity in characteristics) and it examines how much the definition of the submarket matters for estimates of the impact of foreclosures on housing values. The hedonic approach implies that existing housing stocks are arrayed in a spectrum of submarkets according to the quality (hedonic-value) implicit in their packages of dwelling components. Clustering approaches are based on the principle of minimizing the dissimilarities in dwelling characteristics within a cluster and maximize the across cluster heterogeneity.