legal environment and the choice of default resolution alternatives: An empirical analysis, The

Richard A. Phillips* Eric Rosenblatt**

Abstract. In addition to standard foreclosure, three other methods of resolution for mortgage defaults are available: bankruptcy protection, surrender of deed to the lender, and pre-foreclosure sale. This paper develops a model that specifies the choice of resolution method as a function of the state-specific legal environment and local area economic conditions. A large national data set is used to estimate a multinomial logit choice model for the 1987 to 1991 period. The results indicate that the choice of default resolution alternative is sensitive to the legal environment. The results imply that selected legal reforms will tend to improve the efficiency of the default resolution process.

Introduction

When mortgage loan defaults occur, lenders bear legal, administrative and opportunity costs that accrue in direct proportion to the total time required to resolve the default (Clauretie, 1987). For this reason, recent research has focused on methods to expedite the default resolution process. In general, the focus of this research is the effect of various statutes and regulations on resolution time. For example, Clauretie (1989) finds that private mortgage insurers (conventional loans) have higher aggregate loss rates in states that use the judicial method of foreclosure, have right of redemption laws and prohibitions against deficiency judgment liens (these terms are explained below). Clauretie (1990) also analyzes a large number of FHA defaults and finds that losses per default are higher in judicial foreclosure and right of redemption states. Bible (1988) analyzes the average time for a sample of Louisiana (a judicial state) foreclosures. He argues that judicial states, other factors being equal, are likely to be associated with a longer average time to foreclosure.

The results of these studies imply that lender costs (losses) can be reduced by specific legal reforms that will reduce the average time required to foreclose on a delinquent loan. The present research contributes to the analysis of default resolution by recognizing that borrowers face a "menu" of choices for delinquency resolution. In addition to the standard foreclosure procedure other alternatives, such as voluntary surrender of deed to the lender (friendly foreclosure), sale of the property prior to foreclosure, and filing bankruptcy are available. Our approach presumes that these choices will be influenced by the legal environment and that the mix of statutes available in a given state will effect the resolution choice. We employ a large national data set to generate multinomial logit estimates of the effects of the legal environment on the default resolution choice. The results indicate that the non-foreclosure resolution choices, which are generally more efficient, will tend to be promoted by elimination of redemption and deficiency prohibition statutes. Because these outcomes usually involve cooperation among lenders and borrowers, the results imply that specific steps to encourage such outcomes will tend to reduce default losses. In addition, the results allow inference on the legal environment that is most conducive to reduced default resolution time. The paper is organized as follows: section one details the empirical model; section two presents a description of the data used in the analysis; section three summarizes the results.

The Model

The foreclosure and liquidation of residential properties is initiated by the decision of borrowers to default. A large literature has analyzed the determinants of both the probability and timing of defaults.1 However, as noted, given that default occurs, the borrower and lender face additional choices regarding the resolution of the defaulted loan. In addition to standard judicial or power of sale foreclosures (FORECLOSE), defaults may also be resolved by: filing of bankruptcy protection (BANKRUPT), surrender of deed to the lender (DEED), and sale of the property prior to foreclosure (PRESALE). Two of the choices are unilateral by the borrower: BANKRUPT and FORECLOSE. While foreclosure is technically initiated by the lender, the borrower's choice not to seek other methods of resolution initiates the process. The remaining two choices are bilateral in that they indicate cooperation among the parties to resolve the delinquency. Note, however, that the borrower makes the choice to pursue these options by initiating contact with the lender.

The BANKRUPT choice is a delaying action: it does not resolve the default. It is treated as a resolution alternative in our model due to the extended delays associated with the bankruptcy process. Because of these delays, the BANKRUPT choice imposes additional costs on the lender. These costs include losses on the nonperforming loan as well as administrative and legal expenses. Since these costs are borne exclusively by the lender, they are not relevant for the borrower. However the BANKRUPT choice is not costless for the borrower: costs include legal fees, reduced access to credit and intangible costs such as damage to personal reputation. The BANKRUPT choice is selected when the benefits to the borrower (relative to all other choices) exceed the costs associated with bankruptcy.2 With the DEED choice, title to the property is voluntarily surrendered to the lender (also known as "friendly" foreclosure). In the usual case, the borrower will be subject to deficiency claims because the loan balance plus costs borne by the lender is likely to exceed the sale proceeds. However the lender often writes off the deficiency in exchange for a relatively small fee. The advantage for the lender is reduced losses due to a reduced resolution time. The benefit for the borrower is a reduction in the costs of default: the credit history will not show a foreclosure.

PRESALE indicates a resolution of the delinquency by sale of the property prior to foreclosure. The PRESALE choice involves cooperation between the borrower and lender to expedite sale of the property. In many cases, borrowers in default having a positive equity position will unilaterally initiate sale on their own prior to contact with the lender. This situation is precluded in our analysis since the PRESALE cases included in our data are limited to situations where the borrower has zero or negative equity: the lender's losses are minimized but not eliminated.

The choice of resolution method is modeled in a utility maximizing framework: we assume that the chosen resolution alternative is that one that maximizes borrower utility. The probability of resolution of the delinquent mortgage for each choice is given by:

The predicted relationships of the explanatory variables are as follows: CURRENTLTV raises the likelihood of cooperative resolution (DEED or PRESALE). As CURRENTLTV is higher, the ratio of the loan balance to market value is larger. There is a greater incentive for both parties to cooperate to reduce losses. Because of the incentives to cooperate, there is a reduced likelihood of resolution by BANKRUPT or FORECLOSE. JUDICIAL indicates states that use the judicial method of foreclosure. In the judicial foreclosure method, the lender must initiate a costly court action to acquire title to the property. The alternative is the power of sale method which expedites the process because a trustee may initiate foreclosure without a court order. Because JUDICIAL states are associated with a longer resolution time there is an increased likelihood of resolution by standard foreclosure (FORECLOSE). JUDICIAL is predicted to negatively effect the probability of the BANKRUPT choice because of the increased incentive to file bankruptcy in power of sale states. Due to the shorter duration of the foreclosure process in such states, gains from costless shelter are increased with the BANKRUPT choice. JUDICIAL is also predicted to lower the probability of the cooperative choices. Because power of sale states hasten the foreclosure process and impose higher costs on defaulters, there is a higher probability of resolution by PRESALE and DEED.

Higher APPREC values indicate a stronger regional housing market. Such markets are generally associated with a reduction in market time and a higher contract to ask price ratio. The relative ease of liquidation raises the costs of the unilateral choices (BANKRUPT and FORECLOSE) and lowers the probability that they will be chosen. Both parties are predicted to benefit from a cooperative resolution. With resolution by PRESALE, lender losses are reduced due to the expected higher percentage of principal recovered. Because deficiencies are lower, lenders are more inclined toward discharge which clearly benefits borrowers.

REDEMPTION indicates delinquencies that occur in states having redemption statutes. These laws provide the option to recover a foreclosed property for some period after the fact by payment of principal balance and all costs. From the lenders' perspective redemption rights are equivalent to a conditional lien which raises liquidation costs because marketing is more difficult.3 Thus longer REDEMPTION periods impose additional costs on lenders which increases the incentive for a cooperative resolution. With resolution by PRESALE or DEED, redemption rights are suspended.

DEFICIENCY identifies defaults that occur in states having laws that permit judgment liens to satisfy foreclosure sale deficiencies. Because lenders have the right to recover all administrative, legal, marketing, and other costs, there is an incentive for the borrower to cooperate with the lender. Thus there is a greater likelihood of resolution by PRESALE or DEED. However states having deficiency provisions are predicted to increase the probability that the borrower will seek bankruptcy protection. This is because bankruptcy settlements usually result in the discharge of deficiencies. In effect, the discharge feature of the bankruptcy code has the effect of nullifying deficiency recovery provisions.

The variable BALANCE indicates the amount of the unpaid principal balance. A larger BALANCE increases the incentive for lenders to seek a cooperative resolution because of the increased losses on the nonperforming loan. PER CA PINC proxies for the overall economic strength of the region and the unknown income of the borrower at the time of default. Higher PERCAPINC values correlate with increased housing demand and the likelihood of the PRESALE choice is higher. In addition, insofar as PERCAPINC proxies for individual earnings, the probability of BANKRUPT is predicted to be lower. The rationale is that a stronger economy and increased individual earning potential raises the costs of bankruptcy, which may include reduced employment opportunities. FLORIDA and TEXAS are included in the model to control for housing market conditions that were specific to these states. The economic conditions and the magnitude of defaults in these states increases the likelihood of resolution by BANKRUPT and FORECLOSE relative to the entire sample.

TIME is included in the model to control for the effect of the length of time since default: a longer TIME increases the likelihood of resolution by standard foreclosure (FORECLOSE). Exhibit 1 lists the predicted relationship for each of the variables included in the model.

The Data and Descriptive Statistics

The primary sources of data are the servicing and foreclosure tracking systems of a large national lender. During the period analyzed, this institution serviced more than 100,000 residential and commercial mortgage loans distributed over the entire U.S. The data analyzed are the subset of defaulted (residential only) loans recorded by the tracking system. Included are a total of 2,612 defaults that occurred between August 1987 and December 1991. For all observations, the data contain detailed information on the location of the property, the original purchase price, the contract mortgage interest rate, the principal balance at default, the month of default, and the date of default resolution.

The default month is defined as the first month that the full principal and interest payment is not received. TIME is the number of days from the beginning of the default month until resolution by FORECLOSE, DEED or PRESALE.4

Exhibit 2 presents descriptive information on the data and the geographic distribution of sample observations. Possibly reflecting the effect of adverse regional economic conditions for the period analyzed, Florida, Texas and Colorado have the largest number of defaults. The states are listed in descending order of average TIME values. Note that JUDICIAL states tend to have longer average TIME values.

Supplemental data were used to construct other variables used in the empirical analysis. APPREC, the estimate of housing appreciation during the default period, is computed using the Fannie Mae-Freddie Mac repeat sales index which is computed for each state (Li, 1995). APPREC is the percentage change in the index for the relevant state from default until transfer of the title.

Exhibit 3 presents descriptive statistics for the variables included in the model and other variables of interest. The 2,612 observations were nearly equally divided among JUDICIAL (52%) and power of sale states (48%) states. Only 13% of the sample observations were not resolved by FORECLOSE: DEED (3%) and PRESALE (10%). A small percentage of the states in the sample (16%) have REDEMPTION statutes. Almost all of the states (94%) have DEFICIENCY provisions.

Exhibit 3 indicates that the average length of time from origination until default was four years and four months: the average TIME value was 342. The average original property value at origination was $91,781 and the average original mortgage loan amount was $78,767, so that the average sample loan-to-value ratio was .88.

Exhibit 4 presents the means for the same variables by the type of resolution. Exhibit 4 indicates that 218 (8%) of defaults involve the BANKRUPT choice, 82 (3%) are resolved by the DEED choice, and 271 (10%) are resolved via PRESALE. For defaults accompanied by BANKRUPT filings, the average TIME value increased to 524 days, a 58% increase over the average TIME for all defaults (342 days). DEED and PRESALE resolutions are associated with lower TIME values on average at 235 and 234 days, respectively. The mean TIME value for judicial states is 386 days and 295 days for power of sale states.

Results

Exhibit 5 presents the results of the logit derivative estimates. The derivative estimates and t-statistics are reported for each value of CHOICE. As detailed in Exhibit 5, the derivative estimates generally have the predicted signs and almost all are highly significant.

The results indicate that CURRENTLTV has a large impact on CHOICE. As CURRENTLTV increases there is, other factors being constant, a greater likelihood of the PRESALE outcome and lower probabilities of FORECLOSE and BANKRUPT. This result corresponds to expectations: as anticipated losses for the lender increase (costs to the borrower), there is a greater incentive to resolve the default in a cooperative fashion. JUDICIAL states are associated with a reduced likelihood of BANKRUPT and PRESALE choices and are more likely to be resolved via the FORECLOSE option. This is probably due to the increased time associated with judicial foreclosures which reduces the incentive for borrowers to file bankruptcy. Alternatively, in power of sale states, which are associated with a shorter foreclosure time period, there is an increased likelihood that the borrower will file bankruptcy. JUDICIAL states reduce the likelihood of a cooperative resolution: there is a reduced probability of PRESALE, perhaps because borrowers are less inclined to negotiate, and there is no effect on the likelihood of DEED surrender.

The results indicate that APPREC has the expected effects on choice: the probability of BANKRUPT and FORECLOSE is reduced and PRESALE is increased. Note however, that the magnitude of the effects is very small. REDEMPTION time has a negative effect on the BANKRUPT choice: a longer redemption time period increases the value of the option to redeem and reduces the gains associated with filing bankruptcy (which voids the redemption option). The probability of resolution by PRESALE is, however, increased: lenders have a greater incentive to negotiate and assist in a PRESALE in order to avoid the problems that redemption rights create with respect to foreclosure sales. As expected, longer REDEMPTION periods increase the likelihood of the normal FORECLOSE procedure because borrowers still retain the option to reclaim the property after the foreclosure sale. This implies that borrowers value the option even though it is seldom exercised.

The results indicate that DEFICIENCY states increase the likelihood of all resolution choices. The effect of DEFICIENCY on BANKRUPT, PRESALE and DEED is as predicted: there are gains for one or both parties with each of these resolution choices. The value of the BANKRUPT choice is increased in DEFICIENCY states because deficiencies are usually discharged and borrower liability is reduced. Alternatively, there is a greater likelihood of the cooperative choices because, given that bankruptcy is not filed, borrower liability is reduced by agreement with the lender. PERCAPINC has the predicted effect and is statistically significant but the effects are very small. PERCAPINC lowers the probability of the BANKRUPT and FORECLOSE choices. Recall that these are the only outcomes having adverse effects on an individual's credit history. In a strong regional economy the costs of BANKRUPT and FORECLOSE may be higher due to the possible adverse impact on employment and investment opportunities.

Conclusion

Previous research has implied that the efficiency of the foreclosure process may be improved with specific legal reforms: First, the implementation of power of sale provisions in all states will reduce resolution time in judicial foreclosure states. Secondly, all states should allow deficiency judgments to be pursued for losses in excess of foreclosure sale recoveries. Third, statutory redemption laws should be repealed. The logit results reported here clarify that such reforms will tend to produce efficiency gains. In addition, the results clarify that the gains derive from the effect of the legal environment on default resolution choices by altering the relative costs and benefits. For example, the results imply that efficiency gains are likely when redemption statutes are eliminated because fewer delinquencies will involve the BANKRUPT and FORECLOSE outcomes: the benefits of these choices are reduced. Likewise, DEFICIENCY provisions, according to our results, will increase the benefits (to borrowers) of the DEED or PRESALE choice and raise the likelihood that they will be chosen. However, our findings also imply that the efficiency gains from certain of the policy actions may be less than anticipated due to the changes in the behavior of borrowers. For example, the results suggest that borrowers may respond to the implementation of power of sale foreclosure procedures (intended to reduce TIME) by filing bankruptcy. Power of sale provisions increase the benefits of filing bankruptcy and increase the likelihood that it will occur. Thus in some cases policies intended to reduce delinquency time may not produce the expected efficiency gains.

Notes

'See, for example, Epperson et al. (1985), Springer and Waller (1993). 2There are two benefits to the borrower for the BANKRUPT choice. First, the borrower may benefit from costless shelter service until eviction. Secondly, deficiencies (borrower liability for the deficiency of sale proceeds relative to outstanding liabilities) are frequently discharged by the court. 3For details of this procedure, see Green (1993).

4Clauretie (1989) finds that lender losses are higher in states having right of redemption laws which suggests a longer time in delinquency.

5"Workouts," defined as defaulted loans that were subsequently brought up to date or restructured by the lender, were censured from the database.

References

Bible, D. S., An Empirical Study of Residential Mortgage Foreclosures with an Emphasis on Appraised and Assessed Values, Real Estate Appraiser and Analyst, Winter 1988, 54, 30-39. Clauretie, T. M., The Impact of Interstate Foreclosure Cost Difference and the Value of Mortgages on Default Rates, AREUEA Journal, 1987, 15:3, 152-67. , State Foreclosure Laws, Risk Shifting, and the Private Mortgage Insurance Industry, Journal of Risk and Insurance, 1989, 56: 3, 544-54. , A Note on Mortgage Risk: Default v. Loss Rates, AREUEA Journal, 1990, 18:2, 202-7. and T. Herzog, The Effect of State Foreclosure Laws on Loan Losses: Evidence From the Mortgage Insurance Industry, Journal of Money, Credit & Banking, 1990, 22:2, 221-33. Clauretie, T. M. and M. Jameson, Interest Rates and the Foreclosure Process, Journal of Risk and

Insurance, December 1990, 57, 701-11.

Green, W. H., Econometric Analysis, New York: Macmillan, second edition 1993. Epperson, J. F., J. B. Kau, D. C. Kennan, and W J. Muller, Pricing Default Risk in Residential

Mortgages, AREUEA Journal, 1985, 13:3, 261-72.

Li, Y, Residential House Price Volatility and Index Value, working paper, Fannie Mae, 1995. Phillips, R. A. and E. Rosenblatt, The Timing of Residential Foreclosures, unpublished working

paper, 1995.

Springer, T. and N. G. Waller, Termination of Distressed Residential Mortgage: An Empirical Analysis, Journal of Real Estate Finance and Economics, 1993, 7:1, 43-54.

*Department of Finance, Real Estate and Insurance, Virginia Commonwealth University, Richmond, Virginia 23284-4000.

**Fannie Mae, 3900 Wisconsin Avenue, NW, Washington, DC 20016-2892. Date Revised-August 1995; Accepted-February 1996.

Copyright American Real Estate Society 1997
Provided by ProQuest Information and Learning Company. All rights Reserved