Servicer transfer at closing keeps eating my first-year yield

Your cost stack is in the normal range and the fix is mostly negotiation and sequencing rather than a different structure.

On the fees, boarding and transfer pricing is volume-sensitive. Servicers quote a retail number to someone bringing one loan and a lower one to someone committing to a flow of files, and some will waive boarding if you agree to keep the loan with them for a minimum period. Ask for the fee schedule as a document and ask what it looks like at five loans and at twenty. Custody of the collateral file is separately negotiable and some buyers keep the originals themselves rather than paying a custodian, which is cheaper and worse if you ever need to prove chain of custody in a contested foreclosure.

The transfer gap is the more expensive item and it's largely controllable. Federal servicing transfer rules generally require notice to the borrower from both the outgoing and incoming servicer with a window during which payments sent to the old servicer can't be treated as late, and the practical version of that is you coordinate the goodbye letter and the hello letter so they arrive close together and name the same date. Get the outgoing servicer's transfer date in writing before you close and make your servicer's board date match it. Confirm the current notice requirements with your servicer, since they own the compliance.

What you haven't priced is the seller's incentive at the seasoning boundary. A note that has been performing for exactly the minimum number of months a buyer wants often goes to market precisely then, because the seller knows something about the borrower that the pay string doesn't show yet. Escrow shortfall, a hazard policy that lapsed, a property tax bill that just tripled after a reassessment. On a $50,000 note, a $3,000 escrow advance you inherit costs you more than the entire fee stack you're worried about.

The escrow point deserves more weight than it usually gets. I read the servicing comments on these files and the number of loans where the escrow analysis is stale, or the taxes were paid by the borrower directly outside escrow and nobody reconciled it, is high. Ask for the escrow ledger, not the escrow balance. A balance is a number; a ledger shows you whether anyone has actually been paying the county.

$1,400 all-in per note is roughly what I've been penciling too, and it's the reason I stopped modeling anything under about $60k. The fees don't scale down.

Isn't the simple answer here to buy into a fund that holds a pool instead of assembling one note at a time? You give up control and a management fee, but the per-note friction is somebody else's problem and it's spread over hundreds of loans. Genuinely asking what the buyers of single notes get that's worth $1,400 a file.

@verity you get to see the collateral file and say no. In a fund you're buying the manager's diligence and their incentive to deploy. Whether that's worth the friction depends on whether you'd actually do better diligence than they would, and most people assume they would.

@ember the seasoning-boundary point from the official reply is the part I'd chase. If sellers systematically bring paper to market at the moment it becomes marketable, then the average note offered at exactly twelve months of clean pays is worse than a random twelve-month note. Is there any way to see how long the seller has held it? Or does the assignment chain tell you that?