Navigating the mortgage debt markets can be tricky in good times, but market participants may find today's market conditions more akin to walking through a minefield.
Home prices continue to decline nationwide and various reports suggest that the number of problem loans has grown, expanding beyond just subprime mortgages.
Regulations on state and federal levels are subject to changes when it comes to the loan modification process and finding out information about who has had their loan modified, when that mortgage was modified and what the loan's new terms are can be pretty tough.
Last month, 1010data sponsored a panel discussion that brought together investment professionals who offered their thoughts on various loan modification programs and what they mean for investors. The panel was presented before an audience of investment pros that included hedge funds.
Some themes touched on included what second-lien mortgages mean for the workout of first-lien home loans and how trusts record mortgage loan modifications. Also, panelists considered whether modifications were being made to a wider variety of subprime home loans and if servicers should reach out to borrowers earlier in an effort to curtail defaults and delinquencies.
The discussion was moderated by ASR special contributor Aleksandrs Rozens.
ASR: When we look back at 2008, many modifications for loans were done for subprime, two-year ARMs. Are modifications being made in other types of mortgage loans?
YALE: Absolutely. The new HAMP program, Housing Assistance Modification Program, does not distinguish among mortgage types, unlike prior programs that were specifically designed to modify short-reset ARM products. While there is a certain self-selection process in that riskier ARM programs are considerably more likely to be modified, any mortgage type is now eligible for modification.
D'ANNA: Based on the data that's available for loan performance and modification tables, two-year ARM products are the majority of what was modified in the past and they still were [the majority] in the first half of 2009. There's a deceleration in that as you're starting to see more of other loan types [being modified].
SABAPATHI: In the prime space and, to some extent, in the Alt-A space, the deterioration of the economy and jobless growth is actually what's driving the need for more loan modifications. As we continue to see more unemployment, the pace of modifications in the prime space will continue to increase and stay at a high rate for a while.
AGARWAL: I think the sheer number of modifications in subprime and the percentage of what's outstanding is pretty high. It's close to almost 2% of what's outstanding. If you look at the recent history between Alt-A, option arms, prime or prime jumbo in mortgages, and you look at what's being modified as a percentage of what's delinquent, then they're much more comparable.
In prime and Alt-A you have roughly about 2% of what's delinquent being modified. In subprime you have 4% to 5% [of what is delinquent being modified]. Given the amount of delinquencies in subprime, that 4% to 5% looks like almost two percent of the outstanding pool, whereas in prime agencies that looks like only ten basis points from the outstanding pool. The pace of modifications is picking up in non-agency sectors and in subprime, but the delinquency levels are not that high so the absolute numbers look pretty low.
GREEN: I would sort of echo what Rajesh is saying, but I think it's important to note that if you look at new modifications in the non-agency space, whether it's ARM or fixed, there are still much more subprime loans getting modified each month than Alt-A or Jumbo. There is starting to be some pickup in Alt-A and Jumbo loans [modifications] as they see an increase in delinquencies.
D'ANNA: I did some analysis with 1010data and found that over the last three to six months, the fraction of loans that are current or current-but-formerly late that are getting modified has been increasing. I noticed a significant portion of loans that are only recently becoming late for the first time. It's conceivable that these are folks who are just coming up on their first reset and they can't a rate changes.
ASR: What vintage is it? 2007?
D'ANNA: Just looking at the whole universe of first-lien, modified loans, the weighted average vintage is about second half of 2006 or so, for subprime.
GREEN: I would just add that in the 2007, the subprime origination machine started to slow down in the second half. Originators were starting to put the brakes on [subprime lending], maybe a little later than the world, in retrospect, would have wanted them to. Because Alt-A started to make up a bigger portion of the non-agency origination universe at that time (2007), you would naturally expect the Alt-A share of modifications from the 2007 vintage to be higher.
ASR: When we started looking at loan modifications, a lot of folks were considering the idea of tweaking the interest rate somehow on the loans. Now some people are saying that perhaps we should look at the principal and tweak that somehow. Is this becoming more common and how does it impact the pool?
SABAPATHI: We're seeing more principal modifications. It's a function of servicers trying to do what they can to get the borrowers' payments in line with what the borrower can afford. You'll start with dropping the rates down. A lot of the servicers we've spoken to follow a protocol where they'll drop the rates. They'll think about extending the term. If those measures don't get the payment to a point where the borrower can afford it, they'll have to consider some level of principal modification.
The real challenge is if you start doing principal modifications en masse, you open up the moral hazard issues. How much do borrowers who really can afford the mortgage want to misrepresent in such a way as to re-mark their home to market? If you can do that, it's a great trade. So, you want to try to do that very selectively where it's necessary.
It's a real challenge because there's a lot of evidence that seems to suggest that negative equity is what's driving performance. If you don't want to address that negative equity point, you're not going to get better performance.
In some of the work we're doing and continue to evaluate, you also have to look at what is your payment and how does it compare to the alternative rentals in the area. It's difficult to qualify and if you don't want to address negative equity, you have to do principal forgiveness in order to get the modifications to work.
The redefault rates that we look at are very challenging because I don't think we have a sufficient map of redefault rates to decide what was the modification and how was it done. There's more work to be done to really appreciate the need for principal modifications.
AGARWAL: The modifications that happened in the, say, eight- to twelve-month time frame, looking back, if you look at the performance of the principal modifications versus the performance of the rate modifications, they're almost on top of each other. It's very hard to distinguish that principal modifications have done much better. Now, the kinds of principal modifications that were done then versus what might be currently being done may be a little different. I think some principal modifications now are more aggressive.
I think it's too early for us to know how these [2009 modifications] are going to play out because we're still only in May and June. We need to give about a good six to nine months to see how these modifications play out on the principal side. But from what we've seen in the data, I think it's really hard to make the case that the modification philosophies that were followed nine to twelve months ago were really effective.
ASR: There is also the jobless rate and the overall economy.
AGARWAL: That's absolutely true. That's a big factor.
D'ANNA: What we focus on is the change in payment. Obviously, having negative equity is going to affect someone's decision process, but if they just can't make the payments or they're just too high relative to rents, it just is unattractive. A lot of modifications done in the past didn't have sufficient impact on the actual payments that the people had to make in some types of modifications.
Just drilling into individual loans and comparing back to the remits we found cases where someone has been late for several months and everything they owed gets capitalized and they continue with the same terms [but] it doesn't solve their problem.
There are these issues of moral hazard and legal issues. The environment in which servicers have to operate has been changing with every change in law and program. The capabilities they need in order to effectively re-underwrite each of these loans they have to develop because, traditionally, servicers do not underwrite loans. They were not simply reluctant to take the risk. They didn't have the prudent capability to do it and they're building that now.
The best we can do is try to look at the data we have and learn about the effectiveness of different types of modifications. The game is changing and we have not solved it yet.
GREEN: One of the problems, [is that] you hear the term recidivism rate and re-default rate as if there is some universal definition of "redefault rate". If a loan is modified and then misses two payments, is that a re-default? Three payments? If it was modified two months ago, is it part of your re-default category today? It is not as straightforward as one might think to just define re-default matter of factly when you analyze modified loans.
One of the nice things that we do at 1010data is help people to figure out how to develop and constantly change their own unique default and redefault definitions when looking at distressed loans. 'Okay, this is a plain default, [or this is] re-default, etc.'
A lot of loans are already delinquent when they're modified. Generally, the re-default rate is very high and everyone kind of knows that. But a lot of delinquent loans in this environment are going to default anyway. So it's kind of hard to say a modification didn't help necessarily, even though the redefault rate might be sixty or seventy percent on modified loans.
D'ANNA: Right, and the conclusion that the programs were a failure and that mods are not working is relative to an expectation that we know was very high, that this was going to solve our problems [and] possibly forget the fact that these are folks who are in a cohort of loans that already have high default rates.
I looked at delinquency distribution per month after modification for different vintages and then computed a delinquency probability transition matrix. What I found was we didn't solve our problem and 70% of the people are delinquent or defaulted by a year. But, you know, for the cohort they came out of, that's actually better, right?
From a very crude standpoint you can say the modifications didn't hurt. But, from an investor's standpoint, they could have hurt by prolonging the time to recognize what was going to be a loss anyway and, possibly, change the terms in a way that were damaging to the economics of a deal.
YALE: I have an example that I have worked through. Let's assume that we have a borrower who purchased a home for $200,000 three years ago. Now let's say that home is worth $150,000. Let's say that the home across the street, the same exact home, was bought for $200,000 and the owner defaulted on it, it was foreclosed and an investor bought that property.
Now, that investor is renting that property across the street and he's offering a rental payment which is equivalent to the $150,000 valuation of the home.
So the guy who is in the original house has the option of handing back the keys to his home and going across the street and renting that home for something less than what he's paying on his home today.
Let's say that this guy was delinquent or in default, and he went through one of the loan modifications program. If he goes through the loan modification program, what is his new payment going to be? Well, it's not going to be based on the rent of the house across the street. It's going to be based on 31% of his income, right? So there's a possibility he's still going to be underwater. The probability of re-default has been reduced but certainly not eliminated. The rental equivalent across the street is an important concept, because it determines the borrower incentive between staying current on his mortgage or re-defaulting on his home.
Now, add on top of that maybe the homes have dropped another $50,000 in value, and the homes are now worth $100,000. So, he's really upside down and he can go across to a different subdivision and rent for even cheaper.
The loan modification programs are based not on what the comparable rents are or what the comparable values are in that area, but they're based on this concept of 31% of what the borrower can afford. Ultimately it is the combination of what the borrower can afford to pay and the market-equivalent rents that will determine the outcome on the modified loans.
ASR: So when we look at the servicers and how they look at loan pools, at what point do you think they should jump in and start that loan modification process? Do you wait until it becomes delinquent or do you jump in ahead of time?
AGARWAL: What we've seen is that modifications of loans which were current at the time of modification has led to much better performance - as much as 50% better - than loans which were delinquent at the time of modification. So from that point of view, it seems to make sense that you catch the borrower earlier on.
There are obviously secondary benefits as well. For example, if you catch someone at the time of their reset, it's clearly an advantage, because when you look at loan data on the mortgage side and you match that up with the non-mortgage credit data that people at Equifax and Trans Union have, you'll see that a lot of borrowers in the lower parts of mortgage credit are pretty stretched. If they have that one reset up, that's enough to throw them off balance.
Clearly, there's reason to believe, by looking at the data, that over the term for which we've observed this data - roughly twelve-ish months - it makes a lot of sense to catch the current borrower.
There are other issues around that - moral hazard and so forth - that have to be addressed. Also, I think the frame work of the DTI [debt-to-income] is probably not as strong as it could be, in that it looks at DTI at the front end but really DTI at the back end is what the borrower really has to work through. If you strengthen that framework and if we address the equity question, then I think there's some hope in catching these borrowers, particularly before they become delinquent.
SABAPATHI: There's a lot of data available and there are a lot of early indicators of default probabilities and risk scoring of borrowers that servicers can use. There are going to be loans that fall through the cracks.
Ideally, you don't want to modify folks that are not going to be at risk of default. You want to try to catch only those that are most affected in preventing future defaults. So, it's going to be a difficult balance of trying to avoid that moral hazard. With some of the tools and technology we have, servicers can do a pretty decent job of trying to figure out a borrower's debt load and income stream, and what payment they can afford. We have seen evidence that current borrowers, when modified, perform a lot better.
I'm still not sure exactly what that means because I'm not sure if that's largely engendered by moral hazard. If many of these borrowers are reaching out to the servicers, they would have paid anyway. So are you really just giving up something there and not really gaining anything? I think it's still inconclusive. Conceptually, it makes sense but, it's a very difficult balance.
D'ANNA: I think one important aspect that has to be considered is what are the incentives and the risk to the servicer? I agree wholeheartedly - and we all have seen the kind of tools that we have on hand from vendors and from 1010 - it is possible to make a reasonable prediction of who is going to have problems and who will not and what the probability of default is.
But like past government-sponsored programs and efforts where there were wrong or non-existent incentives for the servicer to do what we considered to be the most beneficial thing, if those incentives aren't correct and the protections aren't there, they're not going to do it. They are not going to because there's an element of risk in estimation and for all we know they may have never had a problem anyway. I don't know if that's a risk that a servicer wants to take unless they are somehow specifically granted the approach or the direction to do so.
YALE: Under the new rules, the borrower has to certify that they are in financial hardship. With the official federal program, they have to sign an affidavit that says: 'I can't make my mortgage payments.' That is one of the first steps. I would think with that legal document, someone is signing on the bottom line and if, subsequently, it's found out that they do not have financial hardship, then they may lose their house.
GREEN: When you sign that affidavit, you're swearing to the U.S. government that this is what you make because no one can prove what they weren't making two years ago. At least today if you say, 'This is what I make,' and if you're lying about that, you're lying to the FBI. You are lying to the U.S. Congress. I suppose the idea is that that's supposed to potentially scare people.
ASR: When trustees record all the information on the loan modifications, is this being done in a uniform fashion or is it subject to each trustee's methods?
AGARWAL: I think there's a lot of discrepancy. If you just look at the ABX deals, for example, I think almost thirty of them don't have loan level/loan modification information. They have loan modification information but in many cases it's aggregated.
I don't think it's necessarily the trustee that's doing this on their own accord because with the same trustee you'll find different servicers or originators who have different levels of information. Tools like loan performance let you sort of have another way of verifying for yourself essentially how much modification might be happening and tying it out to those aggregate numbers, if you will. They may not match exactly, but can get pretty close.
SABAPATHI: Part of the issue we have is that a lot of the data fields in non-performance do not contemplate the level and the type of modifications that we're seeing today. Until that gets expanded to capture that information we're all facing a real challenge.
Some of the work that's being done in Project Restart - if that gets exported to existing deals - is going to provide a lot of value in terms of the information. But, I don't think that's going to happen anyway.
The unique loan I.D. should really help borrowers who have other issues. That debt load [information] is valuable, so there's information that can come out in the near future that can help out. But, in the near term there's a real information gap and a lot of work to be done.
With 1010, the platform helps and we have done our own modification queries to help figure out what type of mods are being done. [However], it's very crude and it's somewhat speculative.
GREEN: One of the things we [1010data] do is give the customers an ability to calculate when a probable modification occurred. For example, flagging a subprime ARM the date it went from a nine percent coupon to a two percent coupon.
There are a lot of obvious cases not always reported by servicers. But again, there are periodic caps, life caps, periodic floors, life time floors which make the task of determining rate modifications more complicated. Still, you can get pretty close to identifying most rate mods on 1010data's platform but not all modification types are as easy to identify. For example, if a loan is current and then its principal is reduced dramatically, there's no loss recorded. Was it curtailment or a principal reduction? Hard to say for sure without the servicer telling you explicitly.
If you examine remittance reports from different deals, you're going to see different trustees and servicers reporting varying degrees of detail on the loans modifications, some doing a really good job, some providing limited or no detail.
Investors are just trying to piece together all the information about a loan from the data they have access to, but certain modifcation types, like forbearance, are very difficult to identify from raw data alone. In addition, some modification may change the nature of loan completely, i.e from ARM to fixed, fixed to ARM. These types of modifications tend to be rare, but do occur and are hard to detect.
YALE: There's a lot of good current information about what is going on with mods as far as what's the change in interest rate or the change in the principal balance. The problem we're going to see going forward is the loan terms have changed.
So it's not going to be a 3-1/2% loan for the remaining life of the deal. It's going to be three and a half for a couple of years, and then it's supposed to ratchet up over time. We're not going to know what that ratcheting up is going to be. We don't know what that is at this point. So the terms have changed. We know what the current terms are but we don't know what the forward looking terms are. As far as projecting cash flows it's going to be difficult to project what the cash flows are likely to be on the loan.
D'ANNA: And hopefully if the mods have been effectively applied, we'll have reason to project them out.
ASR: So when we modify the principal, is that recorded as a loss to the trust?
AGARWAL: Yes, it's recorded as a loss to the trust. In the case of rate modifications - getting away from principal modifications for second - you usually see the negative curtailment in the trust. On the principal side you are seeing losses accorded to the trust.
SABAPATHI: On the other hand, one of the provisions is for principal forbearance which is really not an immediate principal reduction. It's forbearance which contemplates a reduction in principal that would not be supported by interest from the borrower.
There is a fair amount of ambiguity and uncertainty as to how that should be treated. Should that principal, which is no longer interest bearing, be considered a write down or not? I think that's going to be evolving and it's going to be important for investors to continue to think through that and provide feedback to the servicers, to the industry and help form a view as to how that should be treated.
YALE: One thing I noticed on the Fannie Mae Web site, reading over some loan modification documents, [is that] they stated that if a loan gets curtailed - a loan that's been modified and that has a forbearance attached with it - that curtailment is to be applied to the forborne principal. It's not to be applied to the principal balance that is currently interest bearing.
So, that would imply that if you're going to pay off your loan then you would pay off the forborne principal first and the interest bearing principal after that. It remains to be seen what is really going to happen with all of these loans in four years, five years, [when] people start moving out of their homes and paying off the loan, and the proceeds from the sale of the house will not cover the entire balance, including the forborne principal of the loans.
ASR: When you try to figure out the value of some of the debt that has had loan mods, it stands to impact future cash flows and that makes it tricky to value what security that's out there. How do you establish value to securities with debt that has seen loan mods?
GREEN: Well, we actually created a new function on 1010data called "expand" that makes it easier for customers to forecast into the future at the most granular loan level (i.e future mark to market LTVs, future coupons on a loan, option arm recast etc.). I think most of our customers are always looking at the present value of the principal payments and the interest payments when valuing a bond. If it's a subprime pool with lots of loans that are expected to have their rates lowered, they're going to have to assume a much lower WAC (Weighted Average Coupon) on all of the pools future interest payments. How much lower - 1%, 2%, 3%? That's the hard part to figure out.
They're likely going to project some sort of sliding scale down, that's going to get lower and lower as more and more loans in the pool get modified. But the interest portion of the valuation of those cash flows is definitely going to go down if there continues to be a lot of rate modifications. If the principal reduction really starts becoming a factor, they'll have to factor that in too. Principal reduction might be a good thing for some investors, for some tranches, because investors will be getting some principal back sooner, that is if they're at the top of the capital structure. Modifications though definitely makes it more complicated to value securities.
SABAPATHI: You have all these ramifications across difference tranches. Given the supply/demand dynamics that we face today, where there's not a tremendous buyer base - and largely again, because of the uncertainty - when you look at these securities, you run various sensitivities through different types of principal mods and interest rate mods. It is going to extend the cash flows meaningfully because prepayments will be very low on a lot of these mods.
If you take a person down to a 3% rate, why would they ever prepay unless they have to move? If they can keep that three percent rate for the next 20, 30 years, and if it is close to a rental alternative, I would think you're going to have a very low prepayment rate.
So, given some of the extension characteristics and the uncertainty of how the different dynamics are going to play through here, you have folks running securities through very, very severe assumption sets ... you'll see very distressed pricing.
It's going to persist for the near term, until you really see more clarity as to how these mods are performing. Once there's more clarity and there's a deeper buyer base that starts emerging back into the market, then you can start thinking about what is your expected assumption are. Right now you don't really have to worry about expected assumptions, you just run it through your worse case.
D'ANNA: That's exactly what we do. We're just assuming that if these people don't get help, they'll just default.
AGARWAL: It's very hard to value these cash flows, particularly if you go out six, twelve months or beyond that because you've got the element of modifications. Let's say you're modifying roughly 2% of subprime every month these days. Last year it used to be 1% so we don't know where that's going to end up.
On top of that, you've got to layer in intervention from the government and how they might guide this whole process. That makes the whole process of evaluation extremely difficult. Then, you have things like principal modifications. There again, the loan balance is going up but you're taking losses in lieu of that up front.
Given the rate of deceleration or the rate of decline in home prices, you don't know what the new severity might be like or the new dollar value realized from the sale of the house six, twelve or eighteen months from now. That further complicates the valuation of loans under the modification regime.
ASR: So we have to see some slowdown in price drops that we see every month?
AGARWAL: That would probably help. That wouldn't solve the problem but it would certainly help.
GREEN: Just getting back to expanding on the point Mani was making. So if the interest rate is so low that the buyer has no incentive to ever refi, that loan is not worth hundred cents on the dollar. I think that's something a lot of government officials might not understand, the duration risk that all the asset managers are looking at.
If they're getting a 3% coupon and their principal back in twenty-five years, they're going to bid a lot less than 100 cents because they need to be compensated for that duration risk.
D'ANNA: Ideally, the trade off would be factored into the calculation of whether to do the modification or not - a standard that determines break even. But the likelihood that that's going to be correct is not super high. It may get close, but it's going to be flawed.
SABAPATHI: When you get a rate that low, part of what has to happen is you need time to see performance. If you do take the rate down that low and the payment is very low, we could conjecture today that the re-default rate should be low. These folks should want to stay in that home because it's a very attractive transaction.
YALE: I think there are a lot of challenges to creating a robust loan modification model.
First you would have to create a model that would identify which loans would be modified. That model would quite possibly have two components, a transition model that would go from current to loan mod, and then another transition model that would be from non-performing to loan mod. From there you'd need additional transition models to track the modified loans to either prepayment or liquidation.
So all of a sudden you're talking about developing a whole lot of transition models. Those transition models can be based to a large extent on your existing models but it's going to be difficult to specify all that and put that into a model and then come up with some good numbers on it.
So for many players it will be much easier to just tweak your existing models by adding a layer on top of your existing model, say by reducing prepayment assumptions over the next six to 12 months, then reducing interest rates, and increasing loss severities.
SABAPATHI: If the servicers are successful in modifying a lot of would-be troubled borrowers and keep that foreclosure supply off the market and you do actually see stabilization of housing prices, that could help.
Part of what's sad is that it's difficult for the government to come out with a program that appreciates current market dynamics and yields and the fact that some of these securities continue to be distressed despite all the efforts. I think part of that is there's just a lot of uncertainty. At the fundamental level, reducing foreclosure supply or trying to curtail the potential for foreclosure supply is an important objective that needs to be accomplished to stabilize housing. The hope is that these mods do keep foreclosures off the market.
D'ANNA: That's why we need beneficial government intervention. If everyone else does the right thing and modifies but I don't necessarily have to and I can just foreclose and get out at some lower cost, then I'm not going to participate, right?
Or, if I'm a servicer [and] if I don't have to do it or I'm not properly incented I'm not going to.
SABAPATHI: Governments have provided servicers with a lot of good incentives here, right?
D'ANNA: I think we're converging on that. Each version has gotten better.
ASR: Do you think there's enough consistency?
D'ANNA: I think there's a wide gulf between the theory and application. And, you know, the inertia that a corporation like a servicing organization has and the lack of resources they have causes change to come slowly and to be executed poorly through no fault of their own.
ASR: How do home equity loans and lines of credit impact the ability to modify the first lien?
SABAPATHI: In terms of modifications, there have been real challenges in terms of how much should a senior investor give up without a restructuring of the subordinate investor.
Part of what the revised guidelines for second liens that were announced tried to accomplish is to make sure that there is a concurrent second lien modification along with the first with lien. If there's a modification, I think the first and second will be done concurrently. Clearly, there are going to be servicers that may choose to follow a different path. In terms of modifications, of the first- and second- [lien mortgages], I don't think there's necessarily an issue to modify the first lien, as long as there's some dialogue and discussion with the second lien holder and a concurrent modification.
ASR: I'd imagine it makes it easier if your servicer is handling both the first and second lien.
AGARWAL: I'd agree with that. The second lien does complicate things. Aside from the fact of modifying the first lien and the second lien, and getting the lender and the servicer at the table at the same time, you've got reps and warrants that have to be provided by the person who's the lender for the Hope for the Homeowners. That's pretty onerous for someone to do when they see the borrower has either been delinquent in the past and or has a high chance of becoming delinquent. They don't want to take that kind of risk.
I think second liens continue to be an issue. The big issue really is that people are under water in their equity, and there's been a sea change in psychology in terms of the number of people who have chosen to exercise that mortgage put and all the programs that we've tried.
In a lot of cases, if you look at the rest of their balance sheet, they're so leveraged it's very difficult to make a solution work under the current frame work.
ASR: Sort of akin to the 125 LTVs in the late '90s?
D'ANNA: The lenders are kind of effectively the landlord now because they essentially are going to own the house. Approaching it from that perspective would take a change in view that at least is not entertained by the government or by banks. They don't want to be landlords.
Good Hill Partners
Managing Director/Head of Research
Director, Mortgage Credit Analytics
Director of Mortgage Analytics
Citi Alternative Investments
Managing Director/Senior Portfolio Manager
(c) 2009 Asset Securitization Report and SourceMedia, Inc. All Rights Reserved.