How Bad Was the Treasury's HAMP Report Mistake?

Yesterday, we learned that the Treasury made a mistake in reporting the number of delinquent borrowers who have permanent modifications under their HAMP mortgage modification program. The June report (.pdf) contained a new chart detailing the performance of these reworked mortgages. It looked good, a little too good. Analysts began asking questions, and Treasury soon realized that there a data discrepancy. How big of a mistake was it?

We still don't know. At this time, Treasury will only say that the mistake was due to "inconsistent reporting." It's currently working to get the numbers right. You may recall the chart in question from this post on the HAMP's June performance:

perm mod delinquency 2010-06.PNG

It looks pretty good, right? 90-day delinquencies never exceed 5%, which is extremely impressive given the likely credit quality of these individuals. Yet, part of the reason for this could be that cancelled modifications are missing from the balance. That's one explanation provided by Shahien Nasiripour of the Huffington Post. He explains what some Barclays analysts found:

The problem they identified had to do with how Treasury was calculating the rate. In the report, Treasury stated that a "HAMP permanent modification is canceled for nonpayment if it is more than 90 days delinquent." To the Barclays Capital analysts, it appeared that Treasury was thus not including those homeowners with five-year modifications who were kicked out of the program. More than 8,600 homeowners have been bounced from HAMP.

The 8,823 cancellations might sound like a lot, but really it's a not much, considering it's out of 398,021 modifications made permanent. This means just 2.2% of permanent modifications have been cancelled so far. Of course, that number will grow. But if these analysts are complaining about these not being shown as a part of the 90+ delinquency bin, then they might not have had much of an effect. Unless these cancellations are highly correlated with the earlier vintages, they would likely have increased the delinquency rates by only few percentage points. It could still be below 10% for most categories shown in the chart.

Of course, the data discrepancy could be more significant. Oddly, the quarterly vintages shown above don't appear to account for all of the modifications made permanent in each time period. For example, the chart shows 126,527 loans in the Q1-2010 population. But the report elsewhere states that 163,863 loans were made permanent during that quarter. That leaves 37,336 loans unaccounted for by the new chart. If all of those turned out to be 90+ days delinquent, then that would raise vintage's the two delinquency measures shown above from 4.1% and 1.3% to 26.3% and 23.9%, for 60+ and 90+, respectively. Of course, that would also mean the Treasury exactly excluded the bad loans only, which seems rather unlikely and particularly devious. Instead, the actual delinquency rates are probably somewhere within that bounds. That is, unless there's a broader systematic problem with the data.

Considering the high re-default rates predicted for these loan modifications exceeding 50%, even sub 25% rates of delinquency sound pretty low. How could this be? There are a few reasons why it makes sense for these loans to be performing relatively well.

First, think about how many filters these borrowers have already been through. First, they had to want a modification instead of opting for strategic default, which signals their willingness to try to keep their home. Next, they had to get approved for a temporary modification. A huge portion of borrowers don't get past that threshold. Then, they had to successfully complete the trial to be made permanent. Again, the drop off rate here is quite high. You're probably left with relatively pristine borrowers, compared to the average modification applicant.

Moreover, remember how easy it should be for these borrowers to pay for the modified mortgage. The payment is set to be just 31% of their income. That's a pretty conservative underwriting standard. Unless these borrowers have a great deal of other outstanding debt or expenses, then they should generally be able to afford this low payment.

Finally, all of these permanent modifications are less than one year old. That means the interest rate hasn't reset to a higher rate yet, which would increase the payment. That will occur with most of these loans over the first several years. But for now, they only have to satisfy the low payment amount where the modification begins.

Hopefully, the Treasury will clear up this discrepancy and provide this data going forward. It's fairly plausible that the performance of these loans is relatively decent at this time. Of course, that could still change as time passes and changing modification terms put stress on some borrowers.