by Tanta on 2/29/2008 07:18:00 AM

There has been some concern in our comments and on other blogs about this bit in Fannie Mae's 10-K published Wednesday:

Beginning in November 2007, we decreased the number of optional delinquent loan purchases from our single-family MBS trusts in order to preserve capital in compliance with our regulatory capital requirements. Although this change in practice may affect our cure rates, it has had no effect on our loss mitigation efforts and, based on current market conditions, is not expected to materially affect the “Reserve for guaranty losses.” We continue to purchase delinquent loans from MBS trusts primarily to modify these loans as part of our strategy to mitigate credit losses and in circumstances in which we are required to do so under our single-family MBS trust documents. Because we are continuing our loss mitigation efforts for delinquent loans, with a primary goal of permitting borrowers to avoid foreclosure, we do not intend to defer purchases of delinquent loans until we are required by our MBS trust documents to purchase the delinquent loans from our MBS trusts. Although we have decreased the number of our optional loan purchases, the total number of loans purchased from MBS trusts may increase in the future, which would result in an increase our SOP 03-3 fair value losses. The total number of loans we purchase from MBS trusts is dependent on a number of factors, including management decisions about appropriate loss mitigation efforts, the expected increase in loan delinquencies within our MBS trusts resulting from the current adverse conditions in the housing market and our need to preserve capital to meet our regulatory capital requirements. For example, we recently introduced a new HomeSaver Advance(tm) initiative, which is a loss mitigation tool that we began implementing in the first quarter of 2008. HomeSaver Advance provides qualified borrowers with an unsecured personal loan in an amount equal to all past due payments relating to their mortgage loan, allowing borrowers to cure their payment defaults under mortgage loans without requiring modification of their mortgage loans. By permitting qualified borrowers to cure their payment defaults without requiring that we purchase the loans from the MBS trusts in order to modify the loans, this loss mitigation tool may reduce the number of delinquent mortgage loans that we purchase from MBS trusts in the future and the fair value losses we record in connection with those purchases. The credit environment remains fluid, and the number of loans that we purchase from our MBS trusts will continue to be affected by events and conditions that occur nationally and in regional markets, as well as changes in our business practices to respond to the current adverse market conditions.

This seems to have a bunch of folks concerned that what Fannie Mae is doing is making unsecured loans to borrowers just so that Fannie doesn't have to buy the loan out of the pool and take a fair value write-down. What it says, of course, is that Fannie Mae certainly intends this program, if it is successful, to reduce the number of mortgages that have to be bought out of the MBS, because the point of the program is to avoid having to do a formal modification of mortgage. If that is successful, it should, in fact, reduce the number of loans Fannie buys out of pools for loss-mitigation purposes.

It does not say that the sole intent of the program is to avoid fair-market write-downs on loans bought out of pools. This mention of this program just appears in the part of the 10-K that deals with pooled loan repurchases. I see nothing here that says that the HomeSaver Advance loans do not involve an increase in reserves for guaranty losses or an increase in the fair value of guaranty obligations (see page 54ff for an explanation of how accounting for credit losses, actual and expected, on MBS loans is handled), instead of an FMV adjustment to an owned loan. Perhaps the question came up in the conference call; I didn't listen to it. I would certainly expect that these loans can be expected to result in higher guaranty costs in the future, and should be accounted for accordingly. The point is simply that as the loans remain in the MBS, they do not appear in the category of loans bought out of MBS and therefore requiring an FMV adjustment under the infamous SOP-03. The reduction to income (increase in reserves) would happen elsewhere in the financials, under guaranty costs.

All that may be too geeky for you; if so I congratulate you on being a normal human being. Getting beyond the GAAP issues here, I think people just really want to know what this HomeSaver Advance thingy is. According to Fannie Mae's website,

HomeSaver Advance, an unsecured personal loan, is a new loss mitigation alternative available to approved Fannie Mae servicers for eligible borrowers designed to bring a delinquent loan current without a formal loan modification. It provides funds to cure arrearages of principal, interest, taxes, and insurance (PITI), as well as other advances and fees as listed in the Highlights section below. HomeSaver Advance is documented by a borrower-signed promissory note, payable over 15 years at a fixed rate of 5% with no payments or interest accrual for the first six months.

HomeSaver Advance is designed for qualified borrowers who have fallen behind on their mortgage, but are able to resume timely payments once their loan is brought current by the advance. It helps simplify and streamline the workout process for applicable loans, as it provides an option for earlier resolution of delinquent loans.

HomeSaver Advance Highlights

*Loan amount up to the lesser of $15,000 or 15% of the original UPB for delinquent PITI, escrow advances, and advances for attorney fees and costs and up to 6 months of unpaid HOA fees (12 months, where the HOA fee is paid once per year)
*Advances may not include late charges or other ancillary fees and costs
*The full loan amount is applied directly to arrearage (borrower never receives funds in hand)
*Truth in Lending Statement and unsecured promissory note are executed at time of agreement with borrower
*Note rate at a fixed rate of 5% with 6-month no-interest/no-payment period
*Amortization period of 14.5 years after the conclusion of the 6-month no-interest/no-payment period
*Workout fee paid to servicer is $600
*Fannie Mae will contract with a third party to service HomeSaver Advance promissory notes

The general rules for borrower eligibility:

• The mortgage is delinquent in an amount equal to or greater than two full payments of principal, interest, taxes and insurance;
• The mortgage must be seasoned with a minimum of six monthly payments made since the date of loan closing;
The mortgage may secure a principal residence, second home, or investment property—owner occupancy is not required; and
• The mortgage may generally be any type of loan (i.e., fixed-rate, adjustable-rate, interest-only, bi-weekly or daily simple interest).

HomeSaver Advance does not have a loan-to-value restriction or property valuation requirement.

Borrower Eligibility

Servicers must also ensure their borrower meets the following qualifications:
• The borrower has successfully resolved the reason for delinquency;
• The borrower demonstrates a long -term financial capacity to resume making the payments on the first mortgage loan and all other debts, including any subordinate mortgage loans (verbal confirmation of financial capacity is acceptable);
• The borrower has surplus income to support an additional monthly payment of at least $200 but does not have the ability to cure the arrearage using a repayment plan within a period of not more than nine months;
• The borrower is willing to participate in HomeSaver Advance; and
• The borrower does not currently have an outstanding HomeSaver Advance note; the HomeSaver Advance option may only be used once during the life of the particular first mortgage loan.

Borrowers involved in an active bankruptcy proceeding or who have had the debt previously discharged in a bankruptcy action are not eligible for this loss mitigation option.

So what does all that mean? First, it doesn't mean that this is how Fannie will handle all troubled loan workouts; it is one possibility. The rationale for this kind of thing--which isn't unheard of, by the way, for banks and other portfolio lenders, although it's new as far as I know for Fannie Mae--is that if you have a borrower with a fairly modest past-due amount ($15,000 or less) and you have determined that the cause of the delinquency was short-term and is now fixed (like temporary job loss), you could find that the effort and expense of buying a loan out of a pool and doing a formal modification of mortgage to add this modest amount can cost you almost more than it's worth. An alternative is just to make an unsecured loan for the past-due amount, while leaving the existing loan's terms unchanged.

It's risky, of course, because you aren't securing the make-up loan amount; that means that you can't take that amount out of foreclosure recoveries, and if the borrower declares bankruptcy your make-up loan just gets tossed into the unsecured bucket with the credit cards and such. The idea is that you would only do this if the amount in question were modest enough that it's not worth the expense to secure it. It seems completely obvious to me that it's not always worth rewriting a $200,000 loan to secure another $1,000. Another $15,000? That seems rather high to me as a ceiling for this program. But cost-effectiveness is the idea here.

I would also have tightened up the verification requirements for this deal. I realize that Fannie Mae clearly has some ability to collect from servicers if they misrepresent on the borrower's situation; I am, however, getting more and more jaded by the minute about conducting these things on a rep-and-warranty basis. I'd want written documentation of the borrower's cause of delinquency and financial condition, not verbal.

However, I don't necessarily think this is a terrible idea, with the above caveats in mind. Of course it all depends on how good you are at targeting it to borrowers whom it will truly benefit. (Do remember that these are Fannie Mae loans, not those horrible subprime exploding ARMs and stuff. They aren't perfect, but they're not the worst loans in the bunch to start with.) A couple of our commenters have suggested that it seems like no more than a way to throw away $15,000 on every delinquent loan you have. I don't think it's that bad, so I thought maybe we could go through an example--and it's just an example, not a prediction--of how the math would work in deciding to offer a program like this.

Let's start by assuming we have a pool of 100 loans that are eligible for this treatment. We'll assume the average loan amount is $200,000, the average interest rate is 6.50%, and the loans are all interest only (mostly so I don't have to keep amortizing balances, but also because that gives you a slightly worse case in recovery from future foreclosure, and I'm not trying to build an optimistic scenario). I'll assume that the servicer waives all attorney's fees and there's no HOA looking for money, so all we have to work with is past-due interest payments and escrow account contributions. The average monthly interest payment on these loans is $1083, and we'll say the average monthly escrow payment is $417, since that gives us a nice round $1,500 monthly payment to play with.

If all the borrowers are six months past due--which is a lot--then the unsecured loan amount for each loan would be $9,000. Plus Fannie Mae pays $600 a pop to the servicer for the workout fee, so if we did this on all 100 loans, we'd end up with $900,000 unsecured money at risk on a $20,000,000 pool at a cost of $60,000. You can, if you want, assume that all of these loans are underwater--the math works the same way--but for convenience I will assume that the $200,000 balance represents 100% LTV (the property is valued today at $200,000).

Therefore, if we had added the $9,000 to the loan amount via a formal modification, we'd have ended up with an LTV of 104.5%. Because we didn't secure the loan, our LTV stays 100%. Remember that since the additional money is unsecured, there is less disincentive for the borrower to sell the property at break-even; the lien can be released without the additional loan amount being paid in cash at the closing table. So from a voluntary prepayment perspective, these loans should perform just like any other once-delinquent loan at 100% LTV.

Let's further assume that our estimated losses on this pool (net of mortgage insurance and including FC expenses), if we foreclosed today instead of doing this workout, are 30%. That means that foreclosing them all right now would cost us $6,000,000. If you assumed that 100% of these loans would be permanently cured and all those borrowers would pay back all of the unsecured as well as secured money, it's obviously a deal to do this.

Of course we won't assume for a moment that these will be 100% successful. Fannie Mae reported in the 10-K that of the loans it has done modifications on that have a 2-year history since the modification was put in place, going back to 2001, 60% were performing or paid in full 24 months later, and 9% had been foreclosed 24 months later (the rest were still on the books but had become delinquent again, just apparently not delinquent enough to mean foreclosure yet). We are going to assume that that is much better performance than our workouts are likely to get, even though, if the program requirements are truly fulfilled, these HomeSaver Advance deals ought to perform better than your average workout (because they're supposed to involve true "temporary" situations and clear financial capacity to carry the payments).

We will assume that after two years, only 20% of our loans are either still cured and paying as agreed, or paid in full including the unsecured amount. We'll assume another 10% of the loans paid in full (the borrower sold the home or refinanced), but the borrower stiffed us on the unsecured amount and we have to write it off.

Of the remaining 70%, we will assume that 50% end up in foreclosure after 12 months, and the other 20% end up in foreclosure after 24 months. We do that because we want to take into account the possible costs of delaying foreclosure. That is always the problem with workout calculations. It's one thing to compare the cost of a workout to the cost of foreclosure today, but if the loan re-defaults, it may end up costing you more, because foreclosure losses next year might well be worse than they are this year. For our example, we'll assume that losses in a foreclosure would be 30% today, 40% in 12 months, or 50% in 24 months. (This is an example, not a prediction, remember. I'm not building in any positive effect of my workout efforts, although logically I should; the more loans I can permanently cure, the better the recovery should be on the ones I do foreclose because it means less REO inventory.)

That gives us 20 loans with no losses except our $600 fee to the servicer or $12,000. The 10 loans that paid in full on the mortgage but stiffed us on the unsecured loans generated a $96,000 loss. The 50 loans we had to foreclose after a year generated a $4,480,000 loss ($200,000 times 50 times 40% plus $9,600 times 50). The 20 loans we had to foreclose after two years generated a $2,192,000 loss ($200,000 times 20 times 50% plus $9,600 times 20). Total losses: $6,780,000, or 34% loss severity after two years instead of 30% loss severity by foreclosing them all today.

I think it's fair to say that it doesn't take much to think we could break even here. For one thing, it's probably not likely that all loans would be six months past due; if you figured only 4 months past due, which is still severely delinquent, you get 33% loss severity after two years (because the unsecured loan amount is smaller). Add 10 loans to the mortgage paid but unsecured loan written off group and take 10 out of the foreclosure after 12 months group, and you've actually got losses at 24 months at 29%, or just slightly better than foreclosing today.

My point is that you have to remember with workout calculations that while delaying foreclosure might increase the severity of loss on the foreclosures, you do save a lot of money on the ones you cure, and that offsets the calculations on an aggregate level. In practical terms, a program like this is just a lot faster and easier than the buy-loan-out-of-pool-modify-mortgage thing, and it's fair to count in the plus column the extent to which that frees up resources to work with the loans that don't qualify for this kind of deal (the ones that you'll have to do a full-blown mod on). Of course, you have to add back the fact that anything that's faster and easier is going to suffer from adverse selection problems--it does tend to be the lazy, cheapskate servicer's first choice of loss mit options even if it shouldn't be.

At the end of it, I think I'd say this isn't a terrible way to handle the workouts for those borrowers who were tight on mortgage affordability but not impossibly over their heads--say originally qualified at 42% DTI--and who therefore ended up several months past due after an incident (cut back on work hours, unforeseen medical expenses, that kind of thing) that would, in a less expensive relative to income housing cost environment, probably have been tolerable. As long as those borrowers are working with the servicer and want to hang onto the home, this gets them over the rough patch without raiding their retirement accounts or borrowing from the local loan shark, and I can get behind the wisdom of that, at least.

I don't think most troubled borrowers are necessarily in that situation--too many, sadly, are really in over their heads. But if you limit the HomeSaver Advance thing to the salvageable borrowers, you salvage those borrowers quickly and fairly cheaply, freeing up your real effort and expense for dealing with the much more troubled cases. So if that's what they're up to, I guess it's OK with me--not like they asked my opinion--but I do hope OFHEO is keeping an eye on this kind of thing for us, and that we all get to see some periodic reporting about how well this initiative is working out.

As far as the cynical view that this is just a way for Fannie Mae to avoid buying mortgages out of MBS? Doesn't everyone want them to limit their portfolio expansion? As long as they're accounting properly for the increased guaranty obligation here, why not leave the loans on the MBS's books?