Reading a non-performing note tape: are you buying the borrower or the collateral?

I'd write the foreclosure number, because it's the only one that doesn't require another human being to cooperate. A modification is a forecast about behavior. A collateral value is a forecast about a house, and houses are easier to forecast than people. I'm still preparing, though, so treat that as a preference rather than a track record.

You're framing it as two businesses. In practice the file tells you which one you're in about six weeks after you buy it, and your job before that is to make sure both endings are survivable at your price.

I hold land and long positions, so I default to the collateral side. The word I'd push back on in your post is bonus. If you priced to liquidation and the borrower reinstates, you didn't get a bonus, you got a shorter timeline, which is worth more than most people put on the sheet.

Dumb question but I want to understand the mechanics. When people say reperform, does the borrower have to agree to sign something new, or do they just start paying again? And can they stop again a month later?

@megan_doyle15 both happen. Reinstatement is the borrower catching up the arrears and going back to the original terms, no new paper needed. A modification is a signed change to the terms, usually a lower payment, sometimes arrears rolled into the balance or deferred to the end. And yes, they can stop again a month later. Re-default on modified loans is common enough that anybody buying seasoned reperformers will price your file on payment history and not on your story about it.

Which is my answer to @raj_mehta14. I don't underwrite to a mod. I underwrite to the collateral and treat a mod as the fast exit if it happens.

Nobody has mentioned what the two paths cost to run. Boarding a file with a licensed servicer plus monthly fees, forced-place insurance if the borrower's policy lapsed, delinquent property taxes, and the legal spend if you file. The legal spend is where the state matters, and it matters more than most spreadsheets admit. A judicial state can be a multiple of a non-judicial one on both time and fees, so check with counsel licensed where the collateral is before you assume a number.

I budget the foreclosure cost on every file I look at even when I'm hoping for a mod, because I've never seen a hope reduce an invoice.

Your intent lasts exactly as long as the collateral file allows it. Before deciding which business you are in, find out whether you can even enforce. Start with the original note and an unbroken endorsement chain to your seller. Then check that every assignment of mortgage is recorded in the correct order, and that the last recorded assignment matches the entity that is actually selling you the paper. Recording practice and what a county will accept differ by state, and gaps get expensive to cure when you need them cured on a deadline. Plenty of files deserve a pass where the collateral math was fine and the paper was not. A discount does not help you if you cannot foreclose and the borrower's attorney figures that out before you do.

Consider the numbers, and what might be wrong with them. $92k unpaid balance, house worth about $115k in a normal midwest suburb, ask $44k. Mod path: payment resettles at $650, taxes and insurance escrowed, collecting maybe $7,800 a year on $44k plus whatever was advanced. Sell at 12 clean payments to a reperformer buyer somewhere in the high 50s to low 60s percent of balance, for an exit around $53k to $56k. Foreclosure path: 10 to 20 months depending on the state, $6k to $12k legal, arrears taxes, a house that needs $15k, sell at $105k, net maybe $75k on $44k plus carry. The foreclosure path pays more in dollars and worse in IRR if it drags. The mod path pays less and pays sooner. So the answer swings entirely on the timeline assumption, which is the number worth trusting least.

@hannah_west29 the piece that moves the mod path is what document actually gets signed. A forbearance agreement, a repayment plan, and a full modification are three different instruments with different consequences for lien priority and for whether anything needs to be recorded, and recording requirements vary by state. It should be drafted by an attorney licensed where the property is, not from a template a note seller emailed over. This matters because a sloppy mod can hurt twice. It can fail to bind the borrower the way it was meant to, and it can complicate the sale of the paper later when a buyer's diligence reads it.

Someone underwriting from the lending side, who doesn't want to own houses, is going to lean hard toward the reperform answer. If the workout ends with holding a vacant property in a market you've never visited, you've accidentally become an operator with a rehab budget. So the discipline is to only buy files where the collateral cushion is deep enough that you could dump the REO wholesale at a loss to the appraised value and still be whole. Which is basically @marcus_hale41's underwriting with a different motive.

Is there really a live market for seasoned reperformers, or is that something people say to make the exit sound easy? @hannah_west29 quoted high 50s to low 60s like it's a screen price.

@chriswalton8 there are buyers, and pricing moves with payment history, escrow status, the state, and whatever yield the buyer needs that quarter. It isn't a screen price. It's a bid you go get, and thin files with no verified income and a hand-written payment log get bid worse than clean ones. @raj_mehta14, on the sheet the right approach is liquidation value net of every cost that can be named, then the price that leaves a margin on that number alone. Every other path is faster money on the same collateral. Paying for outcomes that require the borrower's cooperation is a mistake, since cooperation rarely survives contact with a real file.

All of this assumes the collateral floor is real. In many rural markets, a $38k house in a town of 2,600 can show three comps in eighteen months, and one of them can be a family transfer. The liquidation number in that setting is a guess with a decimal point on it. In a market like that, the reperform path isn't the softer option, it's the one that actually works, because the borrower's willingness to stay is worth more than a resale that can't be priced. Different answer from @dpatel_re's, same reasoning applied to a worse market.