October 1, 2009
(Note: This interview originally appeared in the Financial Update Focus podcast series.)
Moderator: Welcome to the Federal Reserve Bank of Atlanta's Financial Update Focus podcast. Today, we're joined by Kris Gerardi, an economist in the research division here at the Atlanta Fed. He will be speaking about loan modifications for borrowers in danger of foreclosure. Thank you for joining us today, Kris.
Kris Gerardi: Thank you, Tom.
Moderator: Kris, it seems that lenders should find loan modification attractive; after all, something is better than nothing. Why do you think lenders have not modified more loans in danger of foreclosure?
Gerardi: Well, Tom, I think there are numerous possible reasons for the low level of modifications that we have seen in the mortgage market so far. In the paper we discuss two specific reasons: redefault risk and what we term as self-cure risk. These risks are not analyzed or discussed by market observers, and I think they could be very important. Redefault risk is pretty well known. It corresponds to the possibility that a loan modification will be unsuccessful, in which case the borrower simply defaults at some point after the modification is performed. In this case the lender basically goes through the foreclosure process, but the key point is that the process will be delayed, and the lender will end up receiving payment from an eventual foreclosure sale later down the line than it would have otherwise if it had not provided modification in the first place. If house prices during this time fall further, and we've seen this happen so far, then the lender will lose more in this scenario. Self-cure risk corresponds to the case in which a lender provides a modification to a borrower who does not really need one. In this case the borrower would have repaid the full amount of his mortgage, so the lender has basically given away money to the borrower without having any effect. This phenomenon is exactly analogous to the discussion of preventive health services. Performing a preventive test that successfully prevents a disease probably saves costs in those cases. The problem is that it is nearly impossible to tell beforehand who will get the disease, so preventive tests must be applied to a large sample of people, which costs a lot of money. If the disease is very common, then preventive testing may save costs overall, but if it isn't very common, then preventative testing will probably raise costs.
In terms of how quantitatively important these two types of risks are, in the data that we use for our analysis, we see about 30 percent of borrowers who are seriously delinquent on their mortgages recovering and becoming current after 12 months. We also see redefault rates somewhere between 30 and 50 percent over a six-month horizon after a modification is performed. I would argue that these are statistics and nontrivial risks.
In addition to these possibilities, there is also the risk of what I'll call moral hazard, which we don't discuss in this paper but have discussed in previous papers. The risk of moral hazard corresponds to the situation in which borrowers who are not in danger of foreclosure view modification programs as very appealing and so have an incentive to purposefully miss mortgage payments to qualify for these programs. I think this is something that lenders and servicers both think of as nontrivial factors.
Finally, there is always the possibility that lenders and servicers are simply waiting on the sidelines for a better government program to come along and provide even bigger subsidies and incentives to modify mortgages. Given the current uncertain political environment, this explanation may not be such a long stretch.
Moderator: Interesting. You point out in your research that policymakers have encouraged lenders to modify troubled mortgages. In your opinion, is there an effective incentive for loan modifications that policymakers could provide to lenders who hold troubled mortgages?
Gerardi: That's a good question, Tom. There is a proposal by one of my coauthors at the Federal Reserve Bank of Boston that advocates directly subsidizing mortgage payments for unemployed individuals in the form of a loan that must be eventually paid back once the individual finds a job. I like this plan. One of the nice things about this plan is that it would leverage off of the infrastructure of unemployment insurance, so that the setup and monitoring costs wouldn't be too prohibitive. The nice thing about this plan is that is would be directly helping borrowers instead of trying to indirectly provide incentives for services who the deal with borrowers themselves.
Moderator: Of the mortgages that have been modified, is there enough information, and has enough time passed, to tell whether these borrowers have succeeded in making their payments or whether they have slid back into default?
Gerardi: Well, many of the larger modification programs have just taken off over the past few months, so I think we need a little more time to analyze these. In our analysis, we looked at modifications going back a few years, and our data ran up through March of 2009. We looked at redefault rates for mortgages originated up until the end of 2008. We then looked at redefault rates over a six-month horizon. We found between 30 and 50 percent of modified loans redefault depending on the specific sample we looked at, which is a very high rate considering that six months is not a very long horizon. This is not inconsistent with the results from a similar analysis performed by the OTS [Office of Thrift Supervision] and OCC [Office of the Comptroller of the Currency]. Now, one of the silver linings in our analysis was that the typical modification seems to be changing from one which actually raised mortgage payments for borrowers to one that reduces them. Many of the modifications that have been performed up until now are actually not characterized by lower principal, lower interest rates, or lower payments, which may be surprising to many people. In most instances the servicers simply took the late payments with interest and rolled them into the balance of the loan. This basically serves to increase the monthly mortgage payment, and obviously this does not provide much relief to borrowers in financial distress; in fact, quite the opposite. In the data, we see that these types of modifications are decreasing over the past year or so, and modifications that involve reducing mortgage payments are increasing, which is good news. As you might expect, we find significantly lower redefault rates on these types of concessionary modifications.
Moderator: Kris, I don't want to get overly technical, but in your paper you discuss the contract frictions in securitization trusts, which are a popular scapegoat in both the academic literature and media coverage of the crisis, as a stumbling block to greater numbers of mortgage loan modifications. What are these frictions, and are they really to blame?
Gerardi: There are a number of issues with respect to securitization that market observers have pointed to in potentially impeding loan modifications. First, there are legal documents called pooling and servicing agreements, or PSAs, which govern the conduct of servicers when loans are securitized. These agreements sometimes place explicit limits on the number of modifications a servicer can perform. It's unclear what percentage of securitization trusts are actually impeded in terms of their modification behavior by these PSAs, but there are a few studies that claim maybe 30 to 40 percent have strict limits.
Another popular scapegoat has been tranche warfare. In the process of securitization, a number of mortgages are pooled together, and multiple securities or bonds are then created from the cash flows that the pool of mortgages yields. Different types of investors often purchase the different securities, which results in a situation in which a pool of mortgages has numerous claimants. It is often thought that because the securities do not all receive the same timing of cash flows, the various investors may have different objectives for the treatment of delinquent mortgages. For example, investors at the very top of the capital structure, who receive cash flows from principal and interest payments first, may desire a rapid foreclosure process, as this will mean basically a quicker payout. But investors at the bottom of the capital structure, who may receive very little or nothing at all in the event of significant foreclosure filings, may desire more modifications, as this may increase their probability of receiving future payouts. These different incentives have been blamed for holding up mass modification efforts.
Finally, many have pointed to principal-agent problems between the investors in mortgage-backed-securities and the institution hired by the investors to service the mortgages. Because of the timing and structure of reimbursement payments from the investor to the servicer for servicing delinquent mortgages, many observers believe that the servicers may not be acting optimally when it comes to renegotiating with troubled borrowers.
Now, despite these and other potential frictions in the securitization process, our empirical analysis provides strong evidence against the role of securitization in preventing renegotiation. The data set that we use in the analysis includes loans that are serviced for private securitization trusts that are not sponsored by any of the government-sponsored enterprises, like Fannie Mae or Freddie Mac. These are so-called "private-label" loans, and these are subject to all the contract frictions that I just described. In addition, the data set also includes loans that are owned by servicers themselves, so-called "portfolio" loans, which are immune to such problems. In the paper, we compare renegotiation rates, or modification rates, controlling for observable characteristics of the loans across these two different types of mortgages. For our various definitions of renegotiation, including the both payment-reducing modifications as well as the payment-increasing modifications that I just previously discussed, we find that the differences in the likelihood of modification in the 12 months subsequent to the first serious delinquency between these two types of mortgages is neither economically nor statistically significant. Now, it may still be the case that the frictions I discussed above do actually exist in the securitization trusts, but I think that what our study has shown unequivocally is that these frictions are certainly not binding, if they are there.
Moderator: While we all know there is no silver bullet to solving the foreclosure crisis, based on your research what do you think needs to occur to see improvement in the situation?
Gerardi: Well, Tom, that's a great question. I think that we probably need to take action on at least two fronts. The first is preventing foreclosures from occurring, which to date has been the primary focus of policymakers in the crisis. However, as I mentioned above, I would recommend taking a more direct route to helping borrowers who have experienced temporary financial problems, such as job loss, with their mortgage payments. The second front is to help households who have already lost their homes to foreclosure, or who are going through that process right now, and also to try to reduce the stock of houses that have become vacant because of foreclosure. To the extent that we are unable to help many households avoid foreclosure in the first place, I think that it is crucial to try to provide a better safety net for them and to at least attempt to provide an affordable source of housing. In addition, I think it may be worthwhile to consider providing resources to local housing authorities in an attempt to eliminate the stock of vacant homes that is currently plaguing local housing markets. These issues have been largely ignored thus far, and I think they should be made a priority as it becomes increasingly clear that there are many families and houses that we cannot prevent from foreclosure.
Moderator: Kris, thank you very much for sharing your research with us. Again, we've been speaking today with Kris Gerardi, an economist at the Atlanta Fed. This concludes our Financial Update Focus podcast on loan modifications. For more information, please see our third quarter 2009 edition of Financial Update. On our Web site, frbatlanta.org, you can read Kris's working papers about this topic. Thanks for listening, and please return for more podcasts. If you have comments, please send us e-mail at firstname.lastname@example.org.