Mortgages, once defaulted, will rarely pay

22 Jun 2010 by Jim Fickett.

Once a loan defaults, that is, becomes at least 60 days delinquent, it will rarely recover and pay as agreed. The effect of modifications is mostly to delay the problem – loans that have defaulted and then recovered to current status go on to re-default at the rate of about 55%/year. This is not stopping the administration from trying – as HAMP fails, Fannie and Freddie are being used to provide further modifications. It is possible that this could change the outlook in home prices from “another leg down” to “stagnant for several years”.

Laurie Goodman and her team at Amherst Securities have released another invaluable study, entitled Liquidation Lags & Modifications—Impact On Housing Data, Non-Agency Cash Flows (FT Long Room membership required).

Recovery from default is rare

The most important point of this paper is that once a borrower defaults (defined as becoming at least 60 days delinquent), he or she will rarely end up paying as agreed.

In the graph below (based on the table “Exhibit 2” in the paper), the date on the x-axis is the time at which a loan became 60+ days delinquent for the first time, and the y-axis is the percent of loans, from a particular default date, that were current or paid off one year later (click for larger image).

The black line is for all loans and the blue line is for modified loans. Three things are clear:

  • The fundamental trend for recovery from default has been downward over the last few years
  • Modifications have at least temporarily helped (all of the recent upturn, in the rate at which defaulted loans become current, is due to modifications)
  • Even with the help of modifications, the vast majority of defaulted loans fail to recover

The help from modifications may or may not last.

The final way we make the case that once a loan is 60+ delinquent it has a low chance of recovery is via the behavior of loans that used to be 60+ and are now re-performing. We define a re-default as a loan returning to the 60+ delinquency bucket. … loans are now re-defaulting at a rate of 6.5%/month or 55%/year, regardless of whether they are sub-prime, option ARM, Alt-A or prime. This represents a real improvement from the 10- 12%/month or 70%/year default rates of late 2009, but is still an astronomically high number. Moreover, the re-performing loans are prepaying at a rate of <2%/year. If re-performing loans are re-defaulting at an annual rate of 55%, and prepaying at close to zero, it suggests the loans do not re-default this year, they stay in the pool and may re-default next year. …

So here’s the bitter truth. Once a loan transitions into the 60+ bucket for the first time, it has a low chance of eventual recovery. Less than 7% of the borrowers catch themselves up on their own; another 20% get modified and re-default at over 50% per annum, at least in the early years of a 30-year contract. Thus, in analyzing the market, we need to bear in mind that most of these loans will eventually liquidate.

Housing statistics are currently badly distorted

the typical path from delinquency to foreclosure to REO sale has become more complex (as modifications become more commonplace) and has greatly lengthened. This, of course, challenges the integrity of traditional housing statistics, as seriously delinquent loans become “current” at an ever-increasing pace (courtesy of loan modifications), and borrowers who stop paying their mortgage can live in their home payment free for a much longer period before receiving foreclosure notices. It appears the market is much more responsive to liquidation rates than any other statistic. This is a lagging indication of defaults, and is biased more by servicer behavior than actual borrower financial health. …

current via modification. In other words, they are not delinquent because the delinquent payments were either forgiven or added to their loan balance.

Washington will keep trying to delay the problem

I made the point yesterday that HAMP is probably having a negative net effect currently on liquidations, and will soon also have a negative effect on delinquencies. The HAMP document makes the point that about half of the loans with modifications canceled by HAMP receive another modification elsewhere. Here is Goodman's take on that:

What happens to the loans that fall out of the HAMP modifications? …

The question is: will these borrowers liquidate, putting further pressure on home prices? Servicers are supposed to test to see if these investors can qualify for an alternative, proprietary modification program. It is unclear what percent of them will do so, and no numbers are being tabulated. Thus far, in the private label securitized universe, we have seen some upturn in the movement from 90+ to foreclosure, but it is very muted.

Investors should be aware that the fallout may be less than anticipated. Fannie Mae and Freddie Mac have each introduced alternative modification programs (“Alt Mod” for Fannie, “HAMP Backup Mod” for Freddie) for borrowers who have made all three payments during the trial modification period, but have failed to provide the documentation (mostly because their income was too high to qualify for the reduced payment). Under both programs, one of the clear requirements is that a borrower’s front end DTI must be greater than 20% for all loans with LTV < 80%. It appears that any loan with LTV ≥80% is eligible for either an “Alt Mod” or “HAMP Backup Mod,” regardless of DTI. (Yes, you are reading this correctly. The borrower initially overstated income on his original application to qualify for a more expensive home than he could otherwise afford, he then understated income to qualify for a loan modification, but did not provide the required documentation because it did not support the stated number. But the GSEs are giving the borrower a permanent non-HAMP modification anyway.)

In other words, although liquidations are rising, Fannie and Freddie will be used to delay as many of them as possible, for as long as possible. Thus, although I still expect liquidations to rise, they may rise only slowly, and it may take many years to clear the backlog. If that is correct, housing prices might not fall as much as some are currently expecting, but would be stagnant for a long time.