Fund reports a 12.4% weighted bid rate and a 6.8% net to LP. Where did the other 5.6 go?

The deployment drag number bothers me more than the write-downs.

1.1 points on a 12.4 gross is roughly nine percent of the yield lost to calendar. Run it the other way: if the average dollar sat idle for something like five weeks between call and auction, that's about the drag you'd see. Five weeks is not unreasonable for auction seasonality. But if they're calling capital in a lump and deploying over two quarters, the real idle period is much longer and the 1.1 is understated because they're netting it against interest earned on the cash.

Ask what the capital was earning while it waited and whether that interest went to the fund or the manager. I've seen both.

The gap you're describing is the normal shape of the strategy at scale, and each line behaves differently over time, which is what determines whether 6.8 repeats.

Management fee and carry are fixed by document, so those two points are the most predictable thing in the stack. Deployment drag is structural, as you said. Auction calendars are set by county, and a manager running four states has some ability to stagger calls against them. The way to test it is not the drag number itself, it's the ratio of average invested capital to average committed capital over the year. If that ratio is stable across vintages the drag is a cost of doing business. If it swings, they are managing the calendar badly or growing faster than they can deploy.

The write-down bucket is the one with the timing problem @fathom named. Statutory interest continues to accrue on a certificate that isn't going to pay, so a portfolio can report rising accrued yield on positions that are heading for a loss. The accounting policy on when a certificate gets marked down, and whether accrued interest gets reversed with it, is the single assumption your 6.8 depends on most. Get it in writing from the auditor's memo, not the deck.

One more thing on durability. Redemption behavior, the interest rate itself, and the treatment of any premium paid above the lien all vary by state, so a four state book is really four different return profiles blended into one number. Ask for the split by state before you accept the blend.

I read the whole thing and the part that stops me is that 6.8 net is not obviously better than things I can buy without a two year lockup and a manager. What's the argument for the illiquidity here?

@marlow the argument is supposed to be that the collateral is dirt and the rate is set by statute rather than by a credit market, so it doesn't move with the same things. Whether that's worth the lockup is a separate question and I'm not sure 6.8 clears it either.

I've gone through the numbers a few times and I'm still stuck on something basic. When a certificate is written off, does the fund lose the whole face amount or do they get something back from the parcel? I don't understand what a write-down actually represents here.

@juniper it can be either. If the parcel has value they take it to deed and sell it, and the recovery might exceed what they put in. If the parcel is worthless they stop paying the subsequent taxes and walk, and then they lose the face plus whatever subs they already paid chasing it. The write-down line is the net of all that. Which is why @fathom's status codes matter, "unredeemed" covers both a house and a ditch.

The phrase "stop paying the subsequent taxes and walk" is doing a lot of work here, at least to my ear. So you can be forced to keep spending money on a bad position to protect it?

@rigging in most lien states the holder can pay the next year's taxes and add them to the certificate at the same statutory rate, and if you don't, someone else can buy that year's lien and get in line ahead of or alongside you. The specifics of priority differ by state. So yes, a bad certificate keeps asking for money, and the discipline to stop is part of the operation.

@verity, I'd ask for the tape and accept the weeks. You said you want the durable version. Durable is exactly the thing a summary page can't tell you.

Small point but the carry accrual of 0.6 on a vintage netting 6.8 against an 8 pref means they're accruing carry they may never earn. Is that a real cost in their table or a placeholder? If the pref isn't cleared the 0.6 comes back and your net is 7.4, which changes the whole conversation with @marlow.

@kestrel caught the thing I'd have missed. If the 6.8 already deducts carry that isn't payable at 6.8, the table is internally inconsistent and I'd want to know who built it before I trusted anything else on the page.