The Impact of Default and Foreclosure on Housing Values: Rings Vs. Neighborhoods Approach

See Gangel et al. (2013b), Lin et al. (2009), Daneshvary and Clauretie (2012) and Biswas (2012).

Shelby County includes seven cities: Arlington, Bartlett, Collierville, Germantown, Lakeland, Millington, and Memphis.

Deficiency judgment sufficient to fully satisfy indebtedness on real property after trustee’s or foreclosure sale, Title 35, Fiduciaries And Trust Estates Chapter 5 Judicial or Trust Sales statute, Tenn. Code Ann. § 35–5-118 (2013).

In 2010, the Tennessee legislature passed House Bill 3057, effective on September 1, 2010, that provides post-foreclosure protection to borrowers regarding mortgage deficiency balances. The new statute allows debtors to question whether sale prices on foreclosure properties are truly representative of the “fair market value” of the property. Under this law, debtors have a right to attempt to prove “by a preponderance of the evidence that the property sold for an amount materially less than the fair market value…”. If successful, debtors may increase credit amounts and reduce deficiency balances. We tested whether this legislation and change in Tennessee statute has had an effect on foreclosure rates and consequently their impacts on neighborhood sales prices by assigning a dummy variable of one subsequent to the stature taking effect. We find no statistical significance for this variable, thus conclude that the statute does not materially affect our study. Results are available upon request.

See: http://​www.​foxnews.​com/​leisure/​2013/​03/​14/​realtytrac-homes-repossessed-by-banks-fall-in-february-to-lowest-level-since/​

We have no way of identifying short sales; however, since they technically do not involve a foreclosure scenario, we do not feel that a short sale situation will impact overall neighborhood property values to the same extent as foreclosures since short sales result from overall decline in property values.

According to the zoning code of City of Memphis, “R-E” means Residential-Estate, “R-15”: Residential Single-family-15, “R-10”: Residential Single-family-10, “R-8”: Residential Single-family-8, “R-6”: Residential Single-family-6, “R-3”: Residential Single-family-3, “RU-1”: Residential Urban-1, “RU-2”: Residential Urban-2.

We felt that neighborhoods which contain less than 20 parcels would be too small for any meaningful analysis.

For additional information regarding the ACS nationwide survey, see the American Community Survey Information Guide, U.S. Department of Commerce, Economics and Statistics Administration, U.S. CENSUS BUREAU, census.​gov.

P-values are all greater than 0.05 indicating the linear interpolation using 2000 and 2010 and regression estimates using more data points based on ACS 5-year estimates essential do not differ. The p value is using the z-value computed based on equation \( Z=\frac{\beta_1-{\beta}_2}{\sqrt{SE{\beta}_1^2+ SE{\beta}_2^2}} \) where β₁ and β₂ are coefficients from two regressions and SEβ₁ and SEβ₂ are the associated standard errors. The results are provided upon request.

This approach was used by Spahr and Sunderman (1998), Sunderman and Birch (2002) and Sunderman and Spahr (2006).

See Neter et al. (1983) for a discussion of this concept.

Removing the outliers resulted in a slight increase in adjusted R² from 0.95 to 0.98; however, all significant variables remained significant when outliers were removed. Removing sales outliers did not make a major change in results; however, since the objective of the model is to estimate the impact of foreclosure rates on neighborhood property values, it was our opinion that deleting the outliers improves the accuracy of the model even though coefficients may be biased relative to alternative coefficients estimated from the full sample.

In the Bryan and Colwell (1982) approach there is one variable to represent the beginning of each of the years in the analysis period. The two dummies closest to the sale date are assigned values that sum to unity, with the two values being proportionate in each case to the closeness of the sale to that year’s beginning and end. The resulting estimated path of price is a point on a log linear function that moves smoothly from the beginning of each year to the beginning of the next year. Shifts in log linear slope occur only at the beginning of each new year. The system provides more annual flexibility than linear or quadratic movements, being essentially an unconventional piecewise linear technique, with nodes at each year end within the period analyzed.

A clause commonly inserted in a mortgage and deed of trust that grants the creditor or trustee the right and authority, upon default in the payment of the debt, to advertise and sell the property at public auction, without resorting to a court for authorization to do so. Once the creditor is paid out of the net proceeds, the property is transferred by deed to the purchaser, and the surplus, if any, is returned to the debtor. The debtor is thereby completely divested of any interest in the property and has no subsequent right of redemption—recovery of property by paying the mortgage debt in full (West’s Encyclopedia of American Law, 2008).

The propensity score is the conditional probability of being a sales comparable treated based on individual covariates. Rosenbaum and Rubin demonstrated p scores can account for imbalances between foreclosed and non-foreclosed properties and reduce bias by resembling randomization of subjects into either group.

Variance inflation factors, one for each explanatory variable, measure the extent to which variances of the estimated regression coefficients are inflated as compared to the variance if explanatory variables were not linearly related. The largest factor among the variables is used as the indicator of the severity of multicollinearity. For a discussion of VIF, see Neter et al. (1983).

For brevity, we do not report results for AvgMonthlyRate(i,t = 24) in Table 4; however, all relevant variables are highly significant and Table 4 displays the relationship between neighborhood default rates and impacts on sale prices.

We also measured the impact on sales prices of average monthly foreclosure rates calculated for 12 month periods between 13 and 24 months and between 25 and 36 month prior to property closing dates. We found that none of the linear, quadratic nor cubic terms were significant. This suggests that acute neighborhood foreclosure rates occurring over a shorter time period have relatively shorter durational impacts on sales prices.

The results are provided upon request.

We use GENMOD (a procedure in SAS to control for dependence within neighborhood) and the coefficient estimates and z-statistics results were not materially different from Table 4.