Why don't Lenders renegotiate more home mortgages? Redefaults, self-cures and securitization

Highlights

• From 2005 to 2011 lenders renegotiated a small fraction of delinquent mortgages.

• Securitization cannot explain the paucity of renegotiation.

• Asymmetric information between borrowers and lenders is a more likely explanation.

Abstract

A leading explanation for the lack of widespread mortgage renegotiation is the existence of frictions in the mortgage securitization process. This paper finds similarly small renegotiation rates for securitized loans and loans held on banks' balance sheets that become seriously delinquent, in particular during the early part of the financial crisis. We argue that information issues endemic to home mortgages, where lenders negotiate with large numbers of borrowers, lead to barriers in renegotiation. Consistent with the theory, renegotiation rates are strongly negatively correlated with the degree of informational asymmetries between borrowers and lenders over the course of the crisis.

Introduction

Many commentators have attributed the severity of the foreclosure crisis in the United States to the unwillingness of lenders to renegotiate mortgages. Almost every major policy action to date in the housing market has involved encouraging lenders to renegotiate loan terms in order to reduce borrower debt loads. The appeal of renegotiation is simple to understand. If a lender makes a concession to a borrower, it may prevent a foreclosure. This is a good outcome for the borrower, and possibly good for the lender as well: the lender loses money only if the reduction in the value of the loan exceeds the loss the lender would otherwise sustain in a foreclosure. According to its proponents, modification of home mortgages is a win–win–win scenario, in that it helps both borrowers and lenders at little or no cost to the government.¹

This paper adds three insights to the existing literature on mortgage renegotiation. First, despite the intuitive appeal of renegotiation described above, lenders renegotiate a small fraction of delinquent mortgages. Over the period 2005–2011, in the year after the first 60-day delinquency, lenders reduced monthly payments for about 10% of borrowers. In the early stages of the financial crisis they modified an even smaller fraction of delinquent loans. Second, securitization cannot explain the paucity of renegotiation, as the differences in modification rates between securitized and non-securitized loans are economically small and often statistically indistinguishable from zero. Third, a simple model of a lender's decision to renegotiate loans shows that information asymmetries between lenders and borrowers are a more likely explanation than securitization for both the level of modifications over time and for the modification behavior of servicers of different types of loans.

By far the most popular hypothesis in the literature and in policy circles to explain the absence of renegotiation of mortgages by lenders is the “institutional theory,” which poses that frictions in the mortgage market prevent lenders from renegotiating loans even when it is in their interest to do so. Securitization plays a central role in this narrative because the renegotiation decision is not made by the owner of the loan but by an independent agent called the servicer. As Eggert (2007) puts it, “with the loan sliced and tranched into so many separate interests, the different claimants with their antagonistic rights may find it difficult to provide borrowers with the necessary loan modifications, whether they want to or not.” Proponents of the institutional theory have maintained that because servicers do not internalize the losses on a securitized loan, they may not behave optimally. Policymakers and researchers have also argued that the pooling and servicing agreements (PSAs), which govern the conduct of servicers when loans are securitized, place limits on the number and type of modifications a servicer can perform, and that the rules by which servicers are reimbursed for expenses may provide a perverse incentive to foreclose rather than modify.²

The data cast significant doubt on the institutional theory. The dashed line in Fig. 1 labeled “1-year mod rate (Portfolio loans)” plots the time-series evolution of the modification rate associated with mortgages held in the portfolio of the lender, which is almost indistinguishable from the corresponding modification rate for the total population of loans. Indeed, at the peak of the crisis, portfolio loans were slightly less likely to receive modifications. Of course, not all securitization is the same, and the principal focus of the institutional theory has been the private or non-agency securitization market, which consists of mortgages securitized by private sector institutions rather than the Government Sponsored Enterprises (GSEs), Fannie Mae and Freddie Mac. Thus, we distinguish between whether the loan was held in portfolio, securitized by the GSEs, or securitized by a private firm (hereafter referred to as “Private Label Securities” or PLS loans). The bottom panel of Fig. 1 shows the modification rates of the two types of securitized loans relative to portfolio loans and while there are differences, no clear pattern emerges. Early in the crisis, PLS loans were more likely to receive modifications while GSE loans were less likely to receive them, and later in the crisis, their respective positions switched. The “institutional theory” cannot explain these patterns, given that the GSEs hold all credit risk, much like banks in the case of portfolio loans.³

Section 3 systematically analyzes the differences in modification rates across portfolio, PLS, and GSE loans. Here, the role of government policy potentially clouds our findings. After the Lehman Brothers bankruptcy in September of 2008, the federal government took a much more active role in the financial system in general and in the mortgage market in particular. Through the Troubled Asset Relief Program (TARP), the government took equity stakes in most of the large mortgage servicers, and then with the Home Affordable Modification Program (HAMP), it directly intervened in the modification process. For that reason, our focus is on loans that became delinquent prior to the end of 2008, as loss mitigation efforts can be observed in the absence of government intervention. In that period, a multivariate regression analysis that controls for most of the observable underwriting characteristics of loans confirms the findings in the bottom panel of Fig. 1: prior to the government interventions in the mortgage market, servicers were no more likely to modify portfolio loans than PLS loans. This finding is quite robust, and holds in a number of alternative specifications and subsamples.

An alternative explanation, the “information theory,” which is modeled explicitly in Section 4, is that, in the presence of uncertainty, foreclosure can trump renegotiation even when the losses from foreclosure exceed the direct cost of renegotiation to the lender. As Wang et al. (2001) point out, most models of debt renegotiation address the case of a distressed firm facing multiple debt holders, whereas the relevant setup for understanding mortgage renegotiation is one where debt holders face a large group of potential defaulters.⁴ The underlying issue is simple: the mortgage borrower has private information about his or her current financial state and willingness to repay the mortgage, which the lender and servicer do not observe.⁵ For example, researchers and policy makers have pointed to the moral hazard problem whereby borrowers may deliberately default to obtain a lower mortgage payment despite being able to afford their stipulated payment.⁶ We show robust examples to illustrate that even when the borrower truly cannot afford the stipulated mortgage payment but could afford a smaller mortgage with higher net present value to the lender than the recovery from foreclosure, foreclosure can still be the profit maximizing course of action for the lender.

The data are generally consistent with the information theory. According to the theory, lenders weigh the benefits of preventing a particular foreclosure with the cost of providing that same concession to all observationally equivalent borrowers. These costs include the possibility that borrowers will “self-cure” without a modification (i.e. that they exit delinquent status without assistance) or that borrowers will redefault after receiving a modification. According to our data, at the beginning of the foreclosure crisis in 2007 over 50% of seriously delinquent borrowers cured their delinquency without receiving a modification. This means that if lenders could not screen borrowers well, a large percentage of the money spent on loan modifications would have been wasted because most borrowers would have become current on their loans without assistance. High self-cure rates in the initial stages of the crisis are thus consistent with the low modification rates during this time period. However, with the onset of the recession and the corresponding increase in the unemployment rate, self-cure rates fell dramatically. As Fig. 1 shows, self-cure rates fell from almost 70% in 2006 to 25% in 2009. Consistent with the information theory, there is a substantial negative correlation between self-cure rates and modification rates in the aggregate time-series, so that as self-cure rates dropped over the course of the crisis, modification rates increased. In addition, this negative correlation also holds for the differences in self-cure rates between PLS and portfolio loans, as these are highly negatively correlated with differences in modification rates between the two loan types.

Section 5 discusses the interpretation of our findings. The empirical evidence suggests that information issues between borrowers and lenders are first-order in understanding the willingness of lenders to renegotiate loans (both the aggregate trends and the relative differences between investors holding the loans, as shown in Fig. 2, Fig. 3, Fig. 4). As for the importance of the institutional theory, the weight of the empirical evidence in this paper as well as the previous literature suggests that it played a minor role at best. First, in the pre-intervention sample, securitization simply has no economically meaningful effect on the probability of a modification. Second, to the extent that differences emerge in modification rates between PLS and portfolio loans after 2009, it is not clear how they should be interpreted. Relative to the large upward trend in the time-series of modification rates for all types of loans over the sample period, the differences in modification rates between PLS and portfolio loans are small. In addition, there is no direct evidence from the previous literature that those differences are due to frictions inherent in the securitization process as opposed to simply unobserved differences in the original quality of loans and/or unobserved differences in the financial situation of delinquent borrowers across the loan types. Furthermore, the model in Section 4 shows that depending on the self-cure and redefault risk, foreclosure can be more profitable for a lender than modification. As a result, one cannot interpret higher modification rates as evidence of higher efficiency or can one interpret lower modification rates as evidence of frictions. Finally, we review institutional evidence on the servicing agreements and SEC filings of servicers of PLS loans and argue that contracts did not give incentives to servicers to foreclose when modification was in the interest of investors.