Preserving Homeownership: Progress Needed to Prevent Foreclosures
Testimony to The U.S. Senate Committee on Banking, Housing, & Urban Affairs Hearing
Chairman Dodd, Ranking Member Shelby, and distinguished members of the Committee, I thank you for your invitation to testify today. My name is Paul Willen, and I am one of the Senior Economists and Policy Advisors at the Federal Reserve Bank of Boston, which as you know is one of the twelve regional Reserve Banks in the Federal Reserve System. I would like to stress that the views I share with you today are mine, not necessarily those of the Federal Reserve Bank of Boston, the other Reserve Banks, or the Federal Reserve’s Board of Governors.
In the time allotted today I plan to briefly summarize some key findings in the research that I and several talented co-authors have done over the last two years – findings that I think are particularly relevant to the issue of foreclosure prevention. I have also submitted a written statement to the committee, which contains more detail on our research, and which I respectfully request be accepted for the record.
I hope that my comments today and our broader research will be helpful to the Committee, as you consider the important issues that are the focus of this hearing.
The limited success of foreclosure prevention strategies undertaken to date results, at least partly, from reliance on theories about the causes of the crisis that – while intuitively appealing – are at odds with the data. In my remarks today I will focus on four facts from the data which contradict widely held beliefs about the causes of the crisis:
- Resets of adjustable rate mortgages have not been the main driver of borrower payment problems.
- Household life events like job loss and illness played a central role in the surge in foreclosures that started in 2007, even prior to the start of the recession.
- Most borrowers who got subprime mortgages would not have qualified for a prime mortgage for that transaction.
- The practice of securitization is not the main reason that lenders have failed large numbers of home mortgages. A more plausible explanation is that it is simply unprofitable for them to do so.1
I would respectfully submit that policies that ignore these facts - however well intentioned - will address some smaller problems while regrettably ignoring much more serious ones. According to the conventional wisdom, large payment increases associated with the first reset of subprime adjustable rate mortgages led to large numbers of foreclosures. To test this in the data, researchers – including but not only my co-authors and me – have looked at a large sample of individual loan histories which provide information about both the expected payments owed by borrowers and whether borrowers made those payments. If resets were truly important, we would expect to see a dramatic increase in the likelihood that a borrower has trouble with his or her payment to coincide with the first reset of an adjustable-rate-mortgage. But we see no such relationship in the data and, in fact, the majority of borrowers who default on subprime adjustable rate mortgages start missing payments long before the rate increases with a reset.2
Part of the reason for the confusion about the resets is the widespread and, we have found, incorrect belief that rates on subprime ARMs spike dramatically at the reset. Our research reveals that in fact the so-called "teaser" rates on subprime mortgages were very high to begin with. Indeed the phrase "teaser rate" is something a misnomer as it was typically 3 percentage points higher than the rate on an equivalent prime mortgage. The bump in rates at the reset, which is typically tied to six month London Inter Bank Offered Rate (LIBOR), was only about 3 percentage points when LIBOR peaked in 2007, and the Fed Fund rate cuts in the fall of 2007 largely eliminated the reset as an issue entirely. Starting in 2008, most subprime mortgages saw no change in the rate at the reset. The fact that there was no improvement in loan performance corresponding to interest rate cuts suggests the limited scope of resets as a problem. Allow me a point of clarification that is more than mere semantics. Some commentators have erroneously equated subprime mortgages with alternative-mortgage products like so-called Option-ARMs. Option-ARMs, which allow borrowers to pay less than the interest on the loan, and make up for it by adding to the principal balance, were not generally marketed to subprime borrowers, and our investigation of the data suggests that the typical pool of subprime loans had no Option ARMs at all. In fact, the majority of problem subprime loans were fully-amortized loans and many of them were, in fact, fixed rate mortgages. Option ARMs have been and will continue to be a problem but they are not, nor have they ever been, the main source of problems in the mortgage market.
A second point. The conventional wisdom until very recently minimized the role of so-called "life events" like unemployment and illness in generating defaults on subprime mortgages. People argued that life-events could not explain the surge in defaults in 2007, because there was no underlying surge in unemployment or illness that year. But I believe that view reflects a misunderstanding of the interaction of house price depreciation and life events in causing default. Foreclosures rarely occur when borrowers have positive equity, for the simple reason that a borrower is almost always better off selling if they have to leave the house anyway. Thus, detrimental life events have no effect on foreclosures when prices are rising. Consider that in 2001, Massachusetts suffered a fairly severe recession which led to a big increase in delinquencies, but the number of foreclosures actually fell to a record low, as shown in the chart I have included with my testimony (Figure 1). But when home prices fall, some borrowers can no longer profitably sell, and then the income-disrupting life-events really take a toll. Thus we did not need to see a surge in life-events to get a surge in foreclosures, but rather a fall in house prices – which is exactly, and unfortunately, what we saw.
In understanding the role of unemployment in foreclosures, for example, one has to understand that large numbers of households suffer job losses – “separations,” in the lingo of labor economics – even when the economy is doing well. Even in the summer of 1999, in the best labor market in a generation, 300,000 individuals filed new claims for unemployment insurance every week. Because house prices were rising rapidly, few of these job losses ended in foreclosure. But the recession that started at the end of 2007 and worsened dramatically in the fall of 2008 has aggravated the problem. The separation rate has increased and importantly, the finding rate – the rate at which unemployed worked get new jobs – has fallen to record lows. While a recession certainly makes the foreclosure problem worse, it is not necessary to generate large numbers of employment-related foreclosures.
One key policy concern I see is the likelihood that the problem of negative home equity and job loss will persist even after the economy recovers. A borrower with negative equity is, unfortunately, somewhat like a patient with a weak immune system – shocks easily absorbed by a "healthy" homeowner can prove fatal to a homeowner with negative equity. To see this depicted, please note again Figure 1. In Massachusetts, house prices stopped falling in 1992 and a vigorous economic recovery started the following year; but we saw elevated foreclosure numbers for the next five years. The reason is, I believe, rather easily determined: homeowners who bought at the peak of the market in 1988 did not have positive home equity and the protection it brings from foreclosure until house prices fully recovered the 1988 peak in 1998.
My third point relates to the oft-made claim that many borrowers who used subprime mortgages were "steered" into subprime loans and, in fact, would have qualified for prime loans. Part of the problem here relates again to a misunderstanding of what a subprime loan is. What differentiates a subprime loan from a prime loan is not the loan itself – a subprime adjustable rate mortgage is no different from a prime adjustable rate mortgage – but rather the characteristics of the transaction: the size of the down payment, the ratio of the monthly payment to income, the credit history of the borrower, the level of documentation provided by the borrower, among other things.
Careful analysis of the data shows that the vast majority of borrowers who took out subprime loans could not have qualified for prime loans. We looked at a large sample of subprime mortgages in New England in 2007 and defined a prime loan as a loan to an owner-occupant, with a loan-to-value ratio of 90% or less, full documentation of income and assets, a borrower FICO scores of 620 or higher, and a monthly payment that was less than 45% of monthly income. Only 9.6% of the mortgages identified as subprime met these criteria. Furthermore, that subset of prime-qualifying buyers got mortgages with characteristics very similar to prime mortgages available at the time – 65% had fixed interest rates and the average initial interest rate for these loans was 6.7%.3
It should be clear that borrowers may well have been steered into transactions that required subprime loans. For example, a real estate agent may have convinced them to buy an expensive house or a mortgage broker may have encouraged them to do a cash-out refinance that in either case required a loan that no prime lender would approve given their income and credit history. But conditional on the actual transaction, there is no evidence right now that borrowers who used subprime loans could have qualified for a prime loan. The evidence typically cited to make the claim that borrowers were steered is that over the period 1999 to 2006, the fraction of borrowers who used subprime loans but had FICO scores typically associated with prime mortgages increased sharply, going from about 35% to 70% in our data. What this evidence fails to take into account is that over that same period, all the other characteristics of the loans deteriorated sharply: the average LTV for a subprime borrower with 660 FICO went from 82% to 95%.4
My fourth and final comment today relates to foreclosure prevention strategies. Foreclosures are bad for homeowners, but they are also bad for lenders, which typically recover less than half the principal owed to them. So it seems natural to think that borrowers and lenders could work together to arrive at some happy medium in which the borrower gets to stay in his or her home and the lender continues to receive payments, albeit smaller ones. In our most recent paper, we find that such renegotiation is extremely rare. Through careful statistical work using a dataset with 29 million active residential loans, we were able to look at borrowers in the year after they became seriously delinquent. Our main finding is that lenders are reluctant to renegotiate loans: only about 3 percent of the seriously delinquent borrowers in our sample received payment reducing loan modifications in the year subsequent to their first 60-day delinquency.5
A leading explanation for this relative paucity of renegotiation is the view that since most loans are securitized now, the fragmented ownership and contractual complexity inherent in such transactions makes it difficult for borrower and lender to come to a mutually beneficial agreement. But our data does not support this theory. We find servicers equally reluctant to modify loans, whether they are owned in portfolio or serviced on behalf of securitization trusts.
We argue that a more plausible explanation for the unwillingness of lenders to renegotiate is that it simply isn’t profitable. The reason is that lenders face two risks that can make modification a losing proposition. The first, which has been recognized as an issue by many observers and researchers, is "redefault risk" – the possibility that the borrower who receives a modification will default again, and thus the modification will have only served to postpone foreclosure and increase the loss to the investor as house prices fall and the home itself (the collateral) quite possibly deteriorates. The second risk, which has been largely ignored but I believe is no less important, and arguably more, is "self-cure risk" – the possibility that the borrower would have repaid the loan without any assistance from the lender. About a third of the borrowers in our large sample are current on their mortgages or prepay a year after they become sixty days delinquent. An investor would view assistance given to such a borrower as "wasted" money.
Let me conclude by saying that my observation, rooted in our investigation of the data, that servicers and investors may find modification unprofitable should not be misconstrued as suggesting that modification is not desirable for society at large and the economy. The private net present value and the social net present value of a modified loan may well be very different. An investor may have an urgent need for cash that leads it to find the short-term payoff of a foreclosure far more attractive than the uncertain longer-term (but potentially larger) payoff from a modified loan.
We hope that these empirical findings about the crisis add important, and per- haps unexpected, insights to your work as policymakers. Thank you again for the opportunity to appear before you today. I would of course be happy to address any questions you might have.
- Full Banking, Housing, & Urban Affairs Senate Hearing
- Reducing Foreclosures: No Easy Answers, NBER Working Paper 15063 (2009).
- Why Don't Lenders Renegotiate More Home Mortgages? Redefaults, Self-Cures, and Securitization, FRBB Public Policy Discussion Paper 09-04 (2009).
- Just the Facts: An Initial Analysis of the Subprime Crisis, Journal of Housing Economics, 17(4):291-305 (2008).
- A Proposal to Help Distressed Homeowners: A Government Payment-Sharing Plan, FRBB Public Policy Brief 09-01 (2009) .