Andreas Fuster, Caitlin Gorback, and Paul Willen
Since the onset of the subprime crisis, many places across the United States have been affected by high levels of negative equity (meaning that borrowers owe more on their mortgages than their homes are worth), an associated flood of foreclosures, and loss of local wealth. In mid-2012, a community advisory firm, Mortgage Resolution Partners (MRP) approached the government of San Bernardino County, California (a region with particularly high levels of negative equity) and pitched the idea of using eminent domain to seize privately securitized mortgage loans in order to restructure or refinance them. The MRP proposal was largely based on a plan by Cornell University law professor Robert Hockett. In late January, this controversial plan
was abandoned by San Bernardino County, yet it remains under consideration in other counties. While a lot of the debate surrounding the plan has centered
on value judgments and legal issues, in this post we look at available data in
order to get an idea of the landscape of loans that could have been affected by
such a program in San Bernardino County.
There are a few key takeaways from our analysis. First, the share of privately securitized mortgages that are still active five years after the beginning of the subprime crisis is relatively small. Second, while a vast majority of these loans in San Bernardino County is severely underwater, the required payment (at least on the first lien) has decreased significantly for many of them. And though these loans continue to enter serious delinquency at relatively elevated rates, things look much better than they did three to four years ago, in part because house prices have increased quite rapidly over the past year. When evaluating the costs and benefits of policy options such as the use of eminent domain, these facts should be kept in mind.
We use the CoreLogic Loan Performance data set, which contains loan-level information on nearly all privately issued (non-agency) mortgage securitizations. For San Bernardino County, we find about 456,000 first-lien mortgages, comprising roughly 280,000 in the subprime category, 142,000 Alt-A (also known as “near-prime”) loans, and 34,00 prime jumbo loans. Eighty-five percent of these loans were originated between 2003 and 2007. About 70 percent are adjustable-rate mortgages (ARMs), while the others are fixed-rate mortgages (FRMs).
The chart below shows what has happened to the loans in our sample as of August 2012. At first glance, it may be surprising to see how few loans were still open (the yellow and green portions of the bars). Of all the loans, only about 68,000 (15 percent) haven’t yet ended in voluntary prepayment or entered foreclosure. About 275,000 (60 percent) have prepaid voluntarily, and 112,000 (25 percent) have foreclosed or are currently in foreclosure proceedings. The incidence of foreclosures has been particularly high for Alt-A loans, while jumbos have fared the best.
Among the loans that were still open in August 2012, 51,500 (76 percent) were current, 5,000 (7 percent) were thirty days delinquent (an early-stage delinquency often reversed in the subsequent month), and 11,500 (17 percent) were sixty or more days delinquent. We focus now on the loans in these three groups, as they would be the ones potentially affected by the eminent domain proposal.
Let’s start with the bad news. The collapse in house prices that started in 2006 has hit our sample of borrowers in San Bernardino County hard. As of August 2012, only 11 percent of remaining borrowers are estimated not to be underwater, and the median borrower has a mortgage that exceeds the value of the collateral by a factor of about 1.5. (These estimates come from the “TrueLTV” variable in CoreLogic’s data set, which tracks all liens on a property and estimates the current value of the property using an automated valuation model.) In terms of dollars, the median estimated equity position of a borrower current on his or her mortgage is −$107,000, while the median ninety-plus-days delinquent borrower is more than $150,000 underwater (see chart below).
There is some good news, however. The payments due on these mortgages have fallen dramatically for many borrowers, at least on the first lien. (We don’t have any information on the required payment on additional liens.) Nearly 70 percent of ARMs and 30 percent of FRMs now have lower payments than they did at the end of 2007, and, as the next chart shows, the declines have often been substantial, with about one-third of ARM borrowers now having a scheduled payment at least 40 percent lower than they had five years ago.
Mortgage payments have fallen for two reasons. First, many borrowers who took out adjustable-rate mortgages during the “subprime era” are paying very low interest rates now, much lower than what they originally paid. As the next chart shows, the median interest rate on ARMs (more than half the remaining loans in our sample) is currently 3.25 percent. The falling rates are due to a large extent to the historically low short-term interest rates to which these loans are indexed (most commonly, the six-month Libor). In fact, nearly 87 percent of ARM borrowers have seen interest rates decrease relative to what they paid five years ago. (Still, for some of them, the payment increased, because their loans have started amortizing.)
Second, while interest-rate reduction is automatic for ARM borrowers, FRM holders can only get lower rates through modification. Such adjustments have been quite common in our sample: about 40 percent of the remaining loans have been modified at least once, with many loans altered several times. The most common form of modification was interest-rate reduction, with about two-thirds having this characteristic. Other types have been much less common. Only 9 percent of modified loans saw their balance reduced, 6.5 percent received term extensions, and 5 percent switched from ARM to FRM or vice versa.
So, at what rate do the remaining loans default, and how has this rate changed over time? The next chart plots for each month the share of loans newly entering ninety-day delinquency, separately for ARMs and FRMs. The rate of delinquency has strongly decreased since 2009, especially for ARMs. (Two recent New York Fed staff reports, available here and here, document a significant decline in delinquency rates following payment reductions, even for underwater borrowers.) Nonetheless, the hazard of serious delinquency arguably remains high relative to historical standards.
Additional good news is shown in the final chart: San Bernardino house prices, after falling by more than half peak to trough and languishing for more than three years, started appreciating in February 2012, growing 9 percent in the eight months through October, or at an annualized growth rate of more than 13 percent. When we look at house prices for individual ZIP codes and weigh them by the remaining balance of non-agency mortgages, the corresponding increase for the same period was 7.5 percent (or 11 percent annualized).
Overall, the situation in San Bernardino County appears to be improving. While a large fraction of borrowers remain dramatically underwater, a number of life rafts in the form of low interest rates, loan modifications, and recently increasing house prices have kept many from drowning. Facts such as those presented here should be important considerations in the cost-benefit analysis of the eminent-domain idea or related proposals.
The views expressed in this post are those of the authors and do not necessarily reflect the position of the Federal Reserve Bank of New York or the Federal Reserve System. Any errors or omissions are the responsibility of the authors.
Andreas Fuster is an economist in the Federal Reserve Bank of New York’s Research and Statistics Group.
Caitlin Gorback is a research analyst in the Research and Statistics Group.
Paul Willen is a senior economist and policy advisor in the research department of the Federal Reserve Bank of Boston.
Thank you for your comments. To Bob Vivona: Voluntary prepays are a combination of sales and refinances. With respect to the former, it is rare that a borrower will pay off the entire mortgage if it substantially exceeds the sale price of the house. It is difficult of us to measure exactly how frequently this is done, however, because our mortgage datasets do not contain the sales price of a property when a borrower prepays and moves. As for refinances, government programs (such as HARP) have allowed borrowers to refinance and thus voluntarily prepay their loans even when they have deep negative equity, but these programs have not included the privately-securitized loans we are looking at, and as a result, refinances of underwater privately-securitized loans have been very rare. Regarding your second question: it is likely that principal reductions of privately-securitized mortgages (through eminent domain or other policies) would have had more of an effect if they had been done at the height of the crisis, in 2008/9. To Robert Hockett: We appreciate you taking the time to respond to our post. We agree that looking at defaults alone may understate the ‘true’ cost of being underwater, as consumption, home maintenance, or mobility are likely also affected (though the magnitudes of the effects are very much debated among economists). The main point of our post was simply to point out that, at least as long as house prices don’t fall again or the interest rates on the adjustable-rate loans jump upwards, the worst in terms of defaults seems to be behind us.
Thank you for the clear and concise analysis. If I understand the point correctly, most of the troubled “real estate bubble” mortgages have been closed and normal market conditions may soon prevail leaving little reason for imminent domain action. Concerning mortgages that “prepaid voluntarily”, is there any information about realized paid-in equity losses? The completed foreclosure rates, which are really sad to think about, indicate some significant realized negative wealth effects anyway. If done prior to August 2012, would the proposed imminent domain action have relieved some of these realized losses or increased them?
One of the worst ideas ever created. E.D. started out as an idea to seize underwater mortgage from capacity paying borrowers. There is a reason why it has been rejected in Chicago and in S.B. The post effect on future homeowners ( Pricing wise) outweighs any benefits this would bring. Not in America! http://loganmohtashami.com/2012/07/18/dr-housings-new-frankenstein-eminent-domain/
A few quick reactions to Andreas, Caitlin, and Paul’s helpful post: First, it’s very heartening to read the good news here – both (a) that so many in San Bernardino County have benefited by monthly payment reductions (mainly attributable to lower interest rates) and (b) that home prices in the County are showing a somewhat more resilient upward trajectory than they have experienced in previous post-bust peak periods. One cannot but hope that this continues, and that other hard-hit counties across the nation experience counterpart treds. Second, the authors’ principal claim is incontravertible – the facts that they marshall must indeed figure into any CBA to which the eminent domain plan is subjected. The same holds, of course, in respect of further facts going forward, whether conditions continue to improve, level-off, or turn south again. Finally third, notwithstanding the foregoing, a few cautionary notes bear sounding: 1) First, as documented in my forthcoming Current Issues piece to which Andreas, Caitlin, and Paul’s post will ultimately appertain, home prices nation-wide have of course cycled over the years since the bust, and some analysts are predicting that the numbers for the past several months, once available (there’s always a lag), will show another significant retreat. Sam Zell is the latest: http://video.cnbc.com/gallery/?video=3000144911&play=1. Time will tell. 2) Next, as quite a few analysts – including some who are not enamored of the eminent domain plan – often observe, principal-reduction is the most effective means of forestalling delinquency, default and foreclosure, even if other stages of the HAMP ‘waterfall’ – including rate-reduction and term-extension – are of course helpful. That too of course has to be factored in to the CBA. 3) Next, it bears noting that principal-reduction is about more than delinquency-, default-, and foreclosure-prevention, even though it’s partly about those. It’s also about at least two other related but likewise partly independent things – namely (a) market failure and efficiency, and (b) consumer spending and consequent growth and employment. As to (a), if (i) there are loans locked in PLS trusts whose EVs can actually be raised, through writing down principal, by more than the transaction costs occasioned by buying them out of the trusts and then writing them down, and (ii) only improvident PSAs, drafted when too few anticipated the prospect of market-wide crash, now prevent the write-downs, then (iii) we are currently living with avoidable deadweight loss. To recoup that lost value would be a good thing. The ‘if’s here can be and are contested, of course; and again, the EV gains would have to exceed the transaction costs to be worth the candle. But there’s reason to believe these conditions are met. Finally, as to (b), there seems to be broad – though of course not uncontested – agreement that mortgage debt overhang constitutes a significant drag on consumer spending, rendering macroeconomic recovery that much more difficult. (The ‘We Might Be Japan’ conjecture.) FRB and FRBNY white papers and policy presentations often note the connection – the wealth effect’s possible operation in reverse, so to speak. Insofar as this is the case – and again, I know it’s contestible and contested – then principal reduction would seem to be better than interest rate and term refis on this score as well. That too, then, would be well to factor in to the CBA. I hasten to add that I don’t take Andreas, Caitlin, Paul and myself to be disagreeing on anything. We’re only pointing different things out. In the proverbial ‘final’ analysis, the cost-benefit analysis will of course run on the basis of empirical data. We all doubtless agree that under different empirical circumstances, the CBA will yield different results – and that we must all accordingly continue the great work that Andreas, Caitlin and Paul here have done in assembling some of those all-important empirics. All best, more soon, and thanks again, Robert