This paper presents the results of the only publicly available empirical study of what agreements governing subprime securitized mortgages say about mortgage modification.
The mortgage and foreclosure crisis continues to transfix the nation, even as housing markets across the country show signs of improvement.1 The government has tried to promote mortgage modifications to allow homeowners to stay in their homes.2 One particular focus of these efforts has been subprime loans that are securitized, that is, transferred into pools held by trusts and administered by servicers on behalf of investors who buy certificates. For securitized mortgages, a contract called the pooling and servicing agreement governs what the servicer may do to modify the mortgages in the pool.3
Throughout the crisis, an important factor in policy analysis has been what the pooling and servicing agreements that govern securitized subprime mortgages say about mortgage modification. Do these agreements forbid mortgage modifications, so that the most effective modification programs have to trump these agreements, raising all the issues that attend government modification of private contracts?4 Or do the agreements by and large permit mortgage modifications, so that policymakers designing modification programs should concentrate on other possible rigidities that frustrate modification?
Whether and how pooling and servicing agreements constrain mortgage modification is relevant to current policy debates. It is still an open question whether the federal government should take action to trump anti-modification provisions in private securitization agreements, for example by trying to establish the supremacy of federal servicing standards over these agreements. Another question currently under debate is whether local governments should exercise the power of eminent domain to take securitized mortgages and modify them. Municipalities across the country have considered or are currently considering this idea.5 Proponents of such a strategy argue that securitization agreements limit the modification of securitized mortgages, so local governments must step in.6 Another open question is whether differences in terms across securitization agreements impede an effective policy response to problems in the mortgage market, so that the agreements should be standardized.7 The findings reported here suggest that subprime securitization agreements are in fact heterogeneous, so the results bear on this question as well.
Subprime securitization agreements are worth studying because subprime mortgages are commonly understood as lying at the heart of the mortgage and foreclosure crisis. The Financial Crisis Inquiry Report cites “an explosion in risky subprime lending” as the first factual basis for its lead conclusion that the financial crisis was avoidable.8 The Report contains two dissents. One identifies as the lead cause of the crisis a credit bubble “the most notable manifestation of which was increased investment in high-risk mortgages.”9 The other states that “the losses associated with the delinquency and default” of “27 million subprime and Alt-A mortgages” in themselves “fully account for the weakness and disruption of the financial system that has become known as the financial crisis.”10 Although the majority and dissenters disagreed on many topics, they concurred that risky (i.e. subprime) lending was a key part of the crisis.
Subprime mortgages originated in the mid-2000s continue to be important today even though conditions in the mortgage markets have been improving by many measures. For example, as of 2012, subprime delinquencies in mortgage insurer MGIC’s portfolio were at over 60% of their 2006 level.11 Mortgage insurer Radian’s 2012 exposure to mortgages rated A- or below was more than 55% of its 2006 exposure.12
Moreover, even after every subprime loan in existence today is paid off or foreclosed on, subprime loans will remain an important case study of private-label mortgage securitization. The subprime market apparently has started to revive,13 and it will be important to understand how this market has functioned in the past as we discuss how best to regulate it in the future – including whether documentation should be standardized.
Given the importance of what securitization agreements say, it is surprising that no systematic study of the contents of these agreements appears to exist in the academic literature. This paper is a first step toward filling that gap. It presents the results of a review of transaction documents from the sixty-five largest subprime securitization programs from 2006, the last full year of the subprime securit- ization boom. These programs accounted for approximately 75% of the dollar volume of subprime mortgage-backed securities issued in 2006 about which the Bloomberg Financial Information Service has information.14 Although this paper certainly is not intended to be the last word on the contents of securitization agreements, it is the most comprehensive review of subprime securitization contract terms undertaken to date.
One notable finding is that only about 8% of the agreements by dollar volume contain outright bans on mortgage modification. Most agreements (60% by dollar volume) permit material modification subject to conditions. Among the most common conditions are that default must be reasonably foreseeable or imminent, that the servicer must follow “normal and usual” servicing practices, and that the servicer must act in the interests of certificate-holders, the securitization trust, and/or the trustee. Another relatively common condition requires insurer approval for modification of more than 5% of loans in a securitization pool.
The Bloomberg Financial Information service lists 614 deals from 2006 in the “Res B/C” category.15 Bloomberg has further in- formation on 482 of the 614 deals. These 482 deals cover $435 billion of volume, which is similar to the $449 billion volume for 2006 subprime securitization reported by Inside Mortgage Finance.16 Deals are grouped by “program.” A program is defined as a series of related deals with the same “motive force.”17 The 482 deals fall into 152 “programs” identified with different names in the database.
The results reported here are based on a review of transaction documents from deals in the sixty-five largest subprime securitization programs. The sixty-five largest programs include securities with $323 billion in aggregate principal at issuance, or about 75% of the dollar volume for subprime securitizations in 2006 about which Bloomberg has some information.
The review covered transaction documents from one randomly selected deal in each program. The numerical results reported be- low are based on aggregating the results of these reviews of sixty-five deals.
A key assumption of the approach is that deals within pro- grams are likely to be similar. This assumption is based in part on guidance from a major New York law firm with extensive experience in securitization and in part on input from the American Securitization Forum. To check the assumption that modification provisions are likely to be the same across programs, the author reviewed half the deals in the two largest programs, a Countrywide program and a First Franklin program. This supplemental review turned up no significant differences in the material-modification provisions among deals in either program.
1. Subprime securitization contracts may expressly bar, expressly authorize, or remain silent on material modification. Express authorization is the most common arrangement (60% of contract volume), followed by silence (32% of volume), and express bar (8%).
The chart below shows the relative prevalence of the three types of contract term. (The chart is omitted. Please refer to the original pdf document.)
Outright bans on material modification are relatively rare, but the situation where there is no express authorization to make material modifications is fairly common. “Material” modifications are defined here to include long extensions of loan maturity (more than a year or so) and reductions of principal or interest. Material modifications, as defined here, do not include short extensions of time to pay or waivers of late fees or penalty interest. Many contracts provide for such minor modifications without authorizing more significant ones.
The paper now turns to a discussion of the two largest categories: the one in which material modification is expressly authorized and the one in which material modification is neither expressly authorized nor expressly barred, in turn.
2. When material modification is expressly authorized, it is subject to conditions.
The review did not identify any contracts that simply authorized the servicer to modify contracts without conditions on the exercise of this authority.18 The chart below illustrates the proportion of the dollar volume of 2006 subprime mortgage-backed securities subject to various conditions. The percentages are relative to the total volume of securities that have express modification provisions (that is, relative to the green 60% slice in the chart above). Totals add up to much more than 100% because more than one constraint applies to the typical loan.
Certain general standards are extremely common: Servicers typically must follow normal and usual servicing practices, act in the interests of investors, and service loans in the same manner as they service their own loans. It is also common for securitization documents to require the servicer to service the loans that are the subject of the transaction in the same manner as other loans that it services for third parties, although this type of provision is less prevalent (29.3% of principal volume). Clearer standards could promote modification by reducing litigation risk.
Provisions that require reasonably foreseeable or imminent default before material modifications are allowed are also extremely common (83.4% of principal volume is subject to such provisions), and provisions requiring actual default as a prerequisite for material modification exist, though they are not common. Presumably, standards based on loan-to-value and debt-to-income ratios could supply an objective means of meeting a “reasonably foreseeable or imminent default” test.
It is also common to require permission from third parties involved with the transaction for material modifications. Such parties include the rating agency, the credit insurer, or the trustee. More than half the total principal volume covered by this study is subject to such a requirement. (The figures reported here include provisions that re- quire permission only when 5% or more of the total volume or number of loans is modified).
3. Even when material modification is expressly authorized, not all types of material modification are necessarily permitted.
The most common form of express authorization to make material modifications takes the form of allowing the servicer to modify “any” term of the mortgage loan as long as specified conditions are met. However, not all authorizations take this form; some authorize only a subset of material modifications. Among 2006 subprime securitizations for which some material modification is permitted, 23% of the dollar volume of principal is governed by provisions that do not expressly authorize material term extensions, that do not expressly authorize principal reductions, or that expressly limit or do not expressly authorize interest rate reductions. Examples of the latter class of provision include those that prohibit loan modification to reduce interest below 5% or below half the rate otherwise applicable under the contract.
4. When material modification is not expressly authorized, the contract typically contains a broad provision empowering the servicer “to do any and all things that may be necessary or desirable in servicing the loan,” or words to that effect. Even when such language is absent, the grant of power to service is a basis for arguing that the servicer may modify the loans.
Turning to the smaller group of securitized subprime mortgage loans for which modification is not expressly authorized (the burgundy slice of the pie graph above), approximately two thirds of the dollar volume of principal in this class is covered by the broad, catch-all grant of power above, which appears to provide a contractual basis for making modifications that satisfy the other standards in the contract.
5. In cases where material modification is not expressly authorized, there are contractual constraints on the power to modify, frequently arising from the agreements’ general provisions.
Where power to make material modifications is not express, if it is assumed that the power to make material modifications may be inferred from the general grant of power to the servicer to service the loans (and possibly to “do all things necessary or desirable” to do so), this implied power will be limited by the general servicing standards in the agreement and, frequently, by specific modification constraints as well.19 The chart below (omitted, please refer to the original pdf document) illustrates the relative importance of various constraints on any implied power to modify. The percentages are expressed relative to the aggregate principal volume of securities that contain neither express authorization nor express prohibition of material loan modifications.
Areas for Further Research
As noted, this paper is a first step toward understanding contractual restrictions on mortgage modification. Although a review of deals within the two largest subprime securitization programs supports the proposition that modification provisions are consistent across deals in a single program, further testing of this hypothesis may be desirable.
Assuming the technique of sampling by program is justified, it would be ideal to extend the research to years before 2006 and to non-subprime deals. Additional aspects of these contracts may also be relevant to mortgage modification. For example, prepayment penalty waivers are typically subject to standards that are different from those governing other loan modifications, and it appears based on casual observation that these standards are quite heterogeneous and often stringent. Another area for further research is contractual limits on servicer liability. A casual review suggests that these limits are both widespread and heterogeneous, suggesting that different servicers face widely varying levels of liability risk if they modify mortgage loans in a manner inconsistent with the contract documents.
Perhaps the most striking finding of the study reported here is just how heterogeneous subprime securitization agreements’ mortgage modification provisions actually are. Many different provisions appear in many – but not most – subprime mortgage agreements. It is very difficult to form a simple picture of what the typical agreement says and to make policy based on that picture, because it appears that for the most part there is no typical agreement. It is easy to imagine that policymakers will demand some kind of standardization as part of any effort to revive the moribund private mortgage securitization market.
Another noteworthy finding is that outright contractual bans on mortgage modification seem rare, appearing in only 8% of the sample. That is not to say that pooling and servicing agreements pose no significant obstacles to mortgage modifications: the restriction on modifying more than 5% of loans in a mortgage securitization pool without insurer approval, which appeared in nearly a third of agreements that impliedly authorize modification, certainly seems important.20
Informed discussion of the contents of existing pooling and servicing agreements will help advance the debate over a number of issues currently under debate, including the use of eminent domain to take mortgages, the use of federal powers to override the agreements, and whether pooling and servicing agreements should be standardized. This paper has taken a first step toward promoting such informed discussion.
- Two articles about the foreclosure crisis ran in the New York Times on July 25, 2013. See Shaila Dewan, New Defaults Trouble a Mortgage Program, N.Y. Times,, July 25, 2013, at B2 (describing level of defaults among homeowners who received mortgage modifications under TARP program and quoting Treasury source as saying “the housing market and the economy are improving”); Trip Gabriel, Welcome Mat for Crime as Neighborhoods Crumble, N.Y. Times, July 25, 2013 (explaining that in Cleveland, Ohio the foreclosure crisis’s “legacy of abandoned homes has frayed neighborhoods, leaving behind those who cannot afford to get out, while providing shelter to people on the social margins” and suggesting that this situation contributes to crime in the city). ↩
- See, e.g., Lisa Prevost, Loan Modifications, Proactively, N.Y. Times, June 28, 2013 (describing FHFA’s creation of new Streamlined Loan Initiative to promote modifications and differences between the new program and the preexist- ing Home Affordable Modification Program). ↩
- See generally Adam Levitin & Tara Twomey, Mortgage Servicing, 28 Yale J. On Reg. 1 (2011) (describing servicing of securitized mortgages). ↩
- See, e.g., Anna Gelpern & Adam Levitin, Rewriting Frankenstein Con- tracts: Workout Prohibitions in Mortgage-Backed Securities, 82 S. Cal. L. Rev. 1075, 1149 (2009) (“[I]t would be relatively simple to legislate away both the contractual and TIA barriers to amending RMBS PSAs”). ↩
- See Jody Shenn, Eminent Domain Plan Decried by DoubleLine Sees New Life, Bloomberg News, July 17, 2013 (reporting that Chicago and San Bernardino County rejected a plan to seize mortgages using eminent domain last year but that cities in California and Nevada, including North Las Vegas, El Monte, and Richmond, are proceeding with such plans). ↩
- See, e.g., Robert C. Hockett, It Takes a Village: Municipal Condemnation Proceedings and Public/Private Partnerships for Mortgage Loan Modification, Value Preservation, and Local Economic Recovery, 18 Stan. J. L. Bus. & Fin. 121, 138-43 (2012) (arguing that structural and contractual impediments to modifications are part of the justification for use of local eminent-domain powers to take mortgages). ↩
- See, e.g., Patricia A. McCoy, The Home Mortgage Foreclosure Crisis: Lessons Learned 37 (May 7, 2013) (unpublished manuscript), http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2254672. ↩
- National Commission On The Causes Of The Financial And Economic Crisis In The United States, The Financial Crisis Inquiry Report xvii (2011). ↩
- Id. at 417-18 (dissenting statement of Commissioners Hennessey, Holtz-Eakin, and Thomas). ↩
- Id. at 451 (dissenting statement of Commissioner Wallison). ↩
- MGIC Investment Corp. Annual Report (Form 10-K) March 1, 2013, at 25 (2012 data); MGIC Investment Corp. Annual Report (Form 10-K) (March 1, 2011), at 22 (2006 data). ↩
- Radian Group Inc. Annual Report (Form 10-K) (February 22, 2013) at 99 (primary insurance in force data for 2012); Radian Group Inc. Annual Report (Form 10-K) (March 10, 2009) at 119 (primary insurance in force data for 2006). Moreover, Lender Processing Services reports that 35% of delinquent mortgages in July 2013 were Alt-A and subprime. Lender Processing Services, LPS Mortgage Monitor: August 2013 Mortgage Performance Observations (2013). LPS’ involvement in the robosigning scandal impairs its credibility. See David McLaughlin, LPS Reaches $120 Million Deal in “Robosigning” Probe, Bloomberg, Jan. 31, 2013; Dustin A. Zacks, Robo-Litigation, 60 Clev. St. L. Rev. 867, 902 (2013) (describing allegations against LPS and employees). However, there is no obvious motive to falsify this figure and it is broadly consistent with our other observations. ↩
- See Katherine Rushton, Fresh Fear over U.S. Subprime Lending, Telegraph, June 1, 2013 (discussing resurgence of the U.S. subprime mortgage market); Center for Public Integrity, Subprime Lending Execs Back in Business Five Years After Crash, Sept. 14, 2013, http://www.publicintegrity.org/2013/09/11/13327/subprime-lending-execs-back-%20business-five-years-after-crash(documenting that many former executives of subprime lenders are developing “new loans that target borrowers with low credit scores and small down payments, pushing the limits of the tighter lending standards that have prevailed since the crisis.”). ↩
- Bloomberg has some information on 482 subprime issues from 2006. These 482 deals cover $435 billion of volume, which is similar to the $449 billion volume for 2006 subprime securitization reported by Inside Mortgage Finance. The sixty-five programs reviewed here cover 80% of this volume, but not all documents were available for review for some programs, so the actual coverage is approximately 75%. ↩
- This list was accessed by using the “DQRP” function. The definition of “subprime” is drawn from classifications made by the Bloomberg Financial In- formation Service. The “Res B/C” loans are categorized as “subprime” in this paper. Based on interactions with Bloomberg representatives, it appears that Bloom- berg put each deal in the “Res B/C” category if the following three categories made up more than 50% of the dollar volume of principal in the deal at the time of classification: (1) “B- or C-rated loans.” Bloomberg made this determination based on the loans’ description in the prospectus; most frequently the prospectus described the loans as “subprime,” “scratch-and-dent,” “blemished-credit” or the like. (2) Home equity loans. Here the governing criterion was whether the loans’ purpose was to take equity out of the home rather than for purchase, although apparently deals might also be put in this category if the arrangers described the deal as a home-equity deal. (3) Loans that were thirty days or more delinquent at the time of classification. ↩
- Adam B. Ashcraft & Til Schuermann, Understanding the Securitization of Subprime Mortgage Credit, Fed. Res. Bank Of N.Y. Staff Reports No. 318, at 4 (citing Inside Mortgage Finance data). Because there is no single definition of “subprime,” complete agreement on dollar volumes would not be expected. ↩
- The motive force might be the mortgage originator, the underwriter, or another party. This definition of “program” was turned into a working definition by assuming that deals in the same numerical sequence are part of the same pro- gram. For example, we located 26 deals in Bloomberg with names “CWL 2006-1” through “CWL 2006-26.” We were able to locate information about the sponsor, depositor, master servicer, and underwriter for 21 of these deals, and all 21 shared a sponsor, depositor, master servicer, and underwriter. We assumed that these 26 deals were part of the same program. This pattern holds in most cases: deals in the same numerical sequence typically have the same originator or underwriter, suggesting that the same motive force is involved in each deal. ↩
- One program, covering approximately 2% of the dollar volume, authorized modification if, in the servicer’s reasonable and prudent judgment, the modification “could be in the best interest” of investors. This appears to be the most flexible standard encountered in the documents. ↩
- Agreements that do not expressly authorize material modifications often expressly authorize minor modifications, such as short extensions of time to pay or waivers of late fees or penalty interest. ↩
- This restriction is less important if rating agencies or other stakeholders do not count successful modifications against the 5% cap, as has been reported. See Diane E. Thompson, Foreclosing Modifications: How Servicer Incentives Discourage Loan Modifications, 86 Wash. L. Rev. 755, 784 (2011) (citing Monica Perelmuter & Waqas I. Shaikh, Standard & Poor’s Criteria Revised Guidelines For U.S. RMBS Loan Modification And Capitalization Reimbursement Amounts 3 (Oct. 11, 2007). ↩