Distressed Home Sales: Temporary Discount or Permanent Devaluation

Abstract

In this paper we distinguish the two sources of price discount associated with foreclosure sales, temporary liquidity discount as a result of the transaction that typically seeks a quick resolution in a thin market, and permanent unobserved shock to property value at the property or neighborhood level. By finding the pair of regular (non-distressed) transactions before and after a given foreclosure sale for the same property, we compare the holding period returns relative to other properties that did not experience foreclosure sales. Thus, the temporary effect introduced by the foreclosure sale transaction is eliminated and we observe only the permanent return differential.We first employ the commonly used repeat-sales methodology by constructing a unique repeat-sales dataset for Connecticut from 1988 to 2011 in which those ever-foreclosed properties are identified and the corresponding pre- and post-foreclosure regular sales are paired to construct repeat sales pairs together with other repeat-sales pairs from never-foreclosed properties. Various control factors are used to identify these properties. At first glance, in general those ever-foreclosed properties (REO properties ) do not appreciate significantly differently from the non-distressed properties, seemingly suggesting the discount associated with a distressed sale is a temporary transaction-induced effect and does not persist over time. However, close examination reveals that permanent appreciation difference does exist and varies with the timing of the distressed sale relative to the housing market cycle. Compared with normal properties, properties that had an REO sale appreciate less if the distressed sale was during the pre-boom period or the most recent downturn, yet more if the distressed sale was during the boom period.In addition, we apply a matching approach to construct a control group of regular properties that are similar to the group of REO properties in terms of housing characteristics, location, and time of sales. Matching reduces the selection bias arisen from the potential systematic difference between foreclosed and non-distressed properties. Moreover, matching mitigates the concern about disparities in housing appreciation among different housing market segments (e.g. properties of different price tiers) by balancing the characteristics between control and treatment groups. The results confirm those from the repeat-sales approach: less appreciation is associated with those properties foreclosed in the pre-boom period, evidencing that there might be some unobserved property or neighborhood problem that causes the property to appreciate slower, which may have contributed to the mortgage’s default. Further, the appreciation premium associated with properties foreclosed during the housing boom period is positive but less significant after the matching; while those foreclosed post-boom show a persistent price discount, indicating these foreclosures are also likely caused by collateral risk, though credit risk dominates.Our results contribute to the understanding of the foreclosure discount. Observed price or growth differential for these properties during the foreclosure sale should be viewed as the combination of a persistent component and a temporary one. Potentially a property’s foreclosure history can be used in an automated home valuation model or default forecast model to control for the price appreciation differential that is not captured by other observables.