How should modification re-default rates be built into non-performing note exit assumptions?

Question

Take a small pool of nine non-performing firsts, average UPB 118k, average as-is value about 141k, all in judicial states, with the seller asking 62 percent of UPB. A typical model assumes six of the nine can be modified back to performing and sold as seasoned re-performers at 78 to 85 cents on modified UPB after 12 payments. The part that model cannot defend is re-default. Published figures run anywhere from 20 to 40 percent within two years on modified loans, and if a third of the re-performers go back down after 14 months of work getting them there, the exit multiple being underwritten is fiction. On top of that, re-performer buyers want 12 clean payments minimum and some want 18, so the capital sits longer than modeled. How do people actually underwrite the modification path? Is the right answer to price every loan as if it goes to foreclosure and treat a successful mod as upside?

Rena Expert · Accepted Answer

Pricing every loan to the collateral outcome and treating the modification as upside is the discipline that survives a bad year. Not because modifications don't work, they do, but because the mod path has two variables you don't control and the foreclosure path has one. On foreclosure you control timeline poorly and value moderately. On a mod you're also depending on a borrower's income staying intact for the seasoning period plus however long the note buyer holds it after you sell.

On your specific pool: 62 percent of 118k is about 73k per loan against 141k as-is. That's roughly 52 percent of value, which is not a deep entry for judicial states where a contested foreclosure can run 18 to 30 months depending on the state and the court. Stack on it. Advanced taxes and forced-place insurance during the hold, legal at maybe 3 to 6k uncontested and more if anyone answers, and rehab on whatever you take back, because a borrower who stopped paying two years ago has usually stopped maintaining too. Deferred maintenance on a 141k house eating 15 to 25k is common. Run that and your 52 percent of value gets thin fast.

On re-default, the number that matters more than the published range is the borrower's payment-to-income after the mod. Mods that capitalize arrears and stretch term without changing the borrower's actual monthly burden re-default at the high end of your range. Mods that cut the payment materially hold better. So underwrite each borrower's capacity rather than applying a pool-level percentage, and if you can't get income documentation, you don't have a mod model, you have hope.

The other thing your model probably doesn't carry: nine loans in judicial states means nine separate legal engagements, and your attorney's capacity becomes the bottleneck. Ask what their current docket looks like before you assume parallel timelines.

gregpowell66 · Answer

The rehab number is where to push hardest. A house that has sat in default two years in a cold climate has a real chance of a freeze burst, and that runs well past a 15k problem once you add plumbing, drywall, flooring and mold remediation. Renovation crews see it regularly. You cannot underwrite it from a drive-by BPO.

wendy_park48 · Answer

Before the modification math, check whether any of the nine have been previously modified. Prior mod paperwork that was executed sloppily, say a missing borrower signature or an unrecorded modification that affects lien priority, gives a defense attorney something to work with and turns an uncontested timeline into a contested one. It is not unusual for the servicer's own mod agreement to be the weakest document in the stack.

tonyv88 · Answer

On the ones you take back, occupancy status drives everything after the sale. Post-foreclosure eviction timelines and whether cash for keys is worth offering vary state to state, and a cooperative borrower moving out on a scheduled date is worth paying for when the alternative is a sheriff lockout three months later.

paulgrant17 · Answer

Do you know if any of the nine borrowers are actually living in the houses? A rented-out house with a tenant paying someone else changes the picture and tenant protections after foreclosure differ by state.

bmoreno23 · Answer

Reading this to learn, and the thing I hadn't understood is that the borrower's income is the asset in a modification. Everyone talks about the collateral. But if you can't document income you're guessing at the one input that decides whether the mod holds.

samantha_ruiz · Answer

@brandonlee36 the 12 to 18 payment seasoning requirement is the quiet killer in your model, not the re-default rate. You're not modeling 14 months, you're modeling 14 plus 18, and your IRR on a 73k check tied up for nearly three years is a different animal even if every single borrower performs. Have you run it as a straight multiple instead of annualized? Sometimes that's the more honest way to look at a workout book.

luis_carrera · Answer

Second the multiple framing. The other question is what happens if the re-performer bid market softens during seasoning. That model is underwriting an exit price 30 months out on paper nobody has bid on yet.

How should modification re-default rates be built into non-performing note exit assumptions?

8 replies