Nobody wants to defend the 2026 and 2027 rent growth lines that carry every multifamily model

If pulling the year two to three step costs 400 to 600 basis points of IRR, the deal is a rate and rent bet with an apartment attached. That can still be a deal worth doing, provided it's priced as one, and the pricing question is what you're paid for taking it. What I'd want from those four memos is one number each: untrended yield on cost after the renovation is complete, at today's achieved rents on the units already done. No market growth, no burn off. If that number is 5.6 and your all in debt cost is 6.1, the whole thesis is that 2027 saves you. If it's 6.4 against 6.1, you can be wrong about 2027 and still be fine. Everything else in the model is decoration on top of that comparison.

Concession burn off in 14 months is the line I'd fight hardest. In my little world once you've trained a submarket to expect a month free, taking it back costs you occupancy for two quarters and the traffic doesn't come back until the property next door stops offering it too. Nobody controls that. I'd model it coming off over three years or not at all and see what survives.

Two cases and price to the worse one is the right discipline, and it holds up against the flat rent approach. Holding market rent flat for three years in a submarket where the pipeline genuinely emptied is a different forecast dressed as conservatism, and it is one you cannot support either. You end up passing on the assets the whole thesis says to buy and buying the ones with a hair in them that scored well on a broken model. The better downside case keeps concessions through year two and holds market growth at zero until the quarter a delivery tracker shows completions under trailing twelve month absorption. It also sets the exit cap 50 basis points above entry. If that case still returns capital with a small profit, the base case is worth paying for. If the downside is a loss of principal, it does not matter how good the base case looks. @hannah_west29's untrended yield on cost test is the same discipline expressed as one number, and it is faster.

A question for @dpatel_re. When you say completions under trailing twelve month absorption in your own submarket, where does that data come from and how wrong is it? What an LP usually gets shown is a chart with no source, and there is no way to know whether the geography behind it can be trusted.

Also curious about @chriswalton8's question. And a slower one: if face rent growth is the last thing to move, what is the first thing an owner actually sees improve? Renewal rates?

@raj_mehta14 in the order I've watched it, traffic and closing ratio move first, then concession length shortens on new leases while face rent stays put, then renewal increases start sticking, then face rent on new leases finally moves. Owners can see the first two months before it shows up in any published rent index, which is why sponsor commentary sounds bullish before the data does. It's also why sponsor commentary is not evidence.

@chriswalton8 on the data, most of it traces back to a handful of vendors and the submarket boundaries are theirs, not yours. I get the vendor chart and then build my own list by hand from planning and building department records within a two mile radius, unit counts and expected completion. Public records practice varies by state and by county, and about a dozen states don't publish sale prices at all, so what you can verify depends on where you're looking. The hand built list has been wrong on timing by two or three quarters and never wrong on whether the buildings exist. That's the version I'd hold @wendy_park48's year two rent line against.