Two different research teams look at the question from two different sides.
Let’s start with the dark side: the borrowers who ruthlessly put the house back to the servicer. Then we’ll move on to some researchers we don’t hear from often enough, who start by trying to quantify the bright side: the borrowers who won’t drive away from the house.
Ruthless Default
Or, as it is now termed, strategic default (is this a PC-oriented language scrubbing?) is a hot topic. Of course, my colleague Austin Kilgore has a piece, “Next Move: Strategic Defaults,” in last month’s HousingWire Magazine (taste-test here [1] ).
The topic long been a serious concern among MBS analysts as well. Last week, for instance, analysts at J.P. Morgan Securities “put pencil to paper” to get beneath vague generalizations that suffice for most discussions of walk away risk, to compute strategic default incentive for ever agency and non-agency securitized loan.
In total, they estimated that, out of a universe of 57 million loans, roughly 27% or 15.7 million are underwater. If home prices stay flat, roughly 20% or 13 million borrowers have incentive to walk away. (Their methodology assumes borrowers evaluate the choice over a five year horizon.) A modest 3% rate of national home price appreciation shaves that down to just over 10% or 6.5 million borrowers. Average incentives range from $40,511 for borrowers on agency fixed loans, to $105,608 on Option ARMs.
The good news is that “the walk away option will not be exercised with 100% efficiency.” The JPM loan levels models estimates roughly 50% to 60% of prime borrowers with positive incentive to walk away and as high as 80% to 90% of option ARM and subprime borrowers will actually default, but the market is already pricing in default expectations of this magnitude.
Healthy Borrowers to the Other Side of the Room
One of the finest broker/dealer MBS research groups, at Bank of America/Merrill Lynch, is rarely permitted to share its research with media. (This is lamentable, because their findings are hugely relevant to the ongoing policy dialogue on Main Street and in D.C.). Last week was a happy exception (one I would have shared sooner had not 2 feet of ice and snow left me without power).
The usual approach to credit modeling is to estimate the amount, timing and severity of loan losses. Instead, BofA/ML Analysts Vipul Jain and Tim Isgro turned the problem upside down and tried to isolate “a relatively healthy group with a very low probability of default.”
To do this, they use new data provided by credit bureau Equifax. (Note, both data sets do not identify borrowers or addresses, so complex matching algorithms must be used.) By matching original loan amount, zip code and other data items in the LoanPerformance (LP) loan level security databases to those Equifax data, the BofA/ML researchers are able to ascertain if the borrower has other first-lien or second lien-mortgages (both closed end and HELOC), extent of other credit lines and utilization of revolving debt and current delinquent status on other debt.
In brief, in the universe of securitized non-agency loans, they find that almost 60% of prime borrowers have a second lien, over 50% of Alt-A borrowers, over 46% of Option ARM and not quite 25% of subprime. Prime borrowers’ seconds are much more likely to be HELOCs, and “the propensity to hold a closed-end second as opposed to a HELOC increases as one moves down in credit class.” Subprime borrowers’ second liens tend to be larger relative to the first lien – 21.4%, compared to 14.6% for prime borrowers, 19.3% for Alt-A, 14% for Option ARM.
More than Half of 2006 Securitized Non-Agency Loans Underwater
Using this data, Jain and Isgro calculate the percentages of first lien borrowers that are underwater when both first and second lien debt are considered (in the report they show results for 2006 vintages only). In ever loan type, a minority of borrowers still have equity in their homes: 45.0% of prime borrowers, 38.8% of Alt-A, 19% of Option ARM, and 41.9% of subprime.
(Note that BofA/ML’s reported results relate only to the 2006 vintage and non-agency loans. It stands to reason that a much higher percentage of loans are underwater in this sample, originated when home prices were at their peak and had the farthest to fall, than in the JPM study.)
It should come as no surprise that, when actual loan performance is in conjunction with CLTV, that the more underwater borrowers are, the more likely they are to default.
Jain and Isgro also look at the relationship between mortgage performance in context of borrower’s other debt service. Of course the propensity to be delinquent on other debt, even among borrowers current on their mortgage, rises the lower the mortgage credit class. But they tease this interesting observation out of the data: delinquency on other debts is a “much more powerful indicator of mortgage health in the better credit sectors. For poorer credit borrowers, going delinquent on other debts may be a way of life, but for prime borrowers, it is more indicative of distress.”
And they cite another recent study (one I would have loved to transmit to HousingWire readers): “We showed that an auto loan delinquency serves as greater form of financial relief for borrowers lower down in the credit spectrum. After going delinquent on auto debts, subprime borrowers showed lower mortgage delinquency roll rates.” The analysts surmised that, because auto payments make up a larger share of total debt, halting the car payment provided greater relief.
Bottom Line
Low risk borrowers are those who have CLTVs less than 100% (not underwater), have always been current on their mortgage and are current now on other debts. Looking at 2006 first lien vintages, 39.7% of prime, 27.6% of Alt-A, 12.7% of Option ARM and 16.7% of subprime borrowers satisfy these criteria. These are adjusted by observed 6-month roll rates from current to 90+ delinquent (that is, the rate at which loans move from current to 90+ delinquent over a six-month period). Their conclusion: of the outstanding balance of prime, Alt-A, Option ARM and subprime borrowers, 6%, 30%, 17%, and 11%, respectively will not default over 5 years. Interestingly, this is more or less what their credit models had been suggesting all along.
Reprinted from housingwire.com
NOTE: Linda Lowell writes a regular column, called Kitchen Sink, for HousingWire magazine [2].