I tested whether cheap markets really do erode on cash flow. The data said no — but n

[David Walters]

A couple of weeks ago I posted here arguing that metro size barely predicts rent yield. Dan pushed back with something sharper than what I'd written: the misconception that actually costs investors money isn't about where yield is highest — it's that the cash flow you underwrite on closing day is rarely the cash flow you keep. Cheap, high-yield markets, he argued, tend to cash-flow worse over a long hold than the spreadsheet promised. Initial cash flow flatters.

It's a widely held view and a reasonable one. So I tested it.

The setup. Initial yield is rent over price on the day you buy — clean and knowable. Realized yield is what you actually earn over the hold, and it rides on the trajectory of rent, not its starting level. The mortgage is fixed; rent is the variable. The testable version of Dan's point: the highest-yield markets should show the slowest rent growth afterward.

The test. 171 metros with at least 1,000 active listings (a noise floor so thin rural markets don't dominate the extremes). I computed each metro's entry yield as of April 2021, sorted them into five tiers, and measured how rents actually grew over the next five years — then re-ran it at four- and seven-year windows.

What came back. The opposite of erosion. Rent growth rose from the lowest-yield tier through the middle and plateaued at the top — it didn't collapse. Lowest tier (median entry yield ~4%) grew rents a median 4.7%/yr; highest tier (~7.2%) grew them 5.5%/yr. The correlation between entry yield and rent growth was positive (about +0.24), and it held in all three windows. The most recent four years were the most lopsided — top-tier metros grew rents 4.1%/yr against the bottom tier's 2.3%, when coastal demand was supposedly reasserting.

(For anyone about to raise the obvious objection — that current yield and recent rent growth share a term — I tiered strictly by 2021 entry yield, independent of later growth. Same answer. 77% of metros land in the same tier both ways.)

Here's the part I'd actually underwrite on. The effect is real but small — yield tier explains maybe 6% of the variation in rent growth. The real spread is within tiers, not between them. In the top yield tier, the Texas/oil metros did stall, exactly as Dan would predict — Lubbock 3.2%/yr, Corpus Christi 2.4%, McAllen 3.6%. But Rust Belt and Deep South metros in the same yield band did the opposite — Youngstown 6.7%, Mobile 6.7%, Montgomery 6.3%, Rochester 6.2%. Lubbock and Youngstown started at the same yield and finished 3.5 points of annual rent growth apart. Yield tier didn't sort winners from losers. The local economy did.

And the appreciation angle cuts the other way too. The usual defense of low-yield markets is "you give up cash flow for appreciation and eventual cap-rate compression." Over this hold it lost on every leg: cap rates didn't compress on either end (exit yields drifted slightly up in both the top and bottom tiers), and the high-yield tier appreciated faster — about 1.7 points/yr more home-value growth than the low-yield tier. Denver grew rents 0.6%/yr over the last four and appreciated 2.4%/yr over five; Mobile, three points higher on entry yield, grew rents 6.7% and appreciated 4.7%.

The honest limits. This is one specific 2021–2026 hold — a big Sun Belt/Rust Belt in-migration wave — and it contains no national rent recession. Run the same test across 2006–2011 and it'd look different. Which is Dan's deepest point, and the one the data can't touch: he's right that cash flow is never certain and rents can fall outright, like Detroit and Phoenix and Vegas after '08. The "erosion as a rule" claim doesn't survive the data. The "rents can fall" claim was just never in the sample.

The takeaway I'm keeping: the day-one yield barely predicts the yield you keep — in either direction. What predicts it is the economy underneath the number. Buy the yield and ignore that, and you might catch Lubbock thinking you bought Youngstown.

Credit to Dan for the pushback — it produced a better answer than either of us expected.

For those of you holding in high-yield metros: what's the local signal that actually told you the rent would hold — jobs, population, something else? Curious where people look once they get past the cap rate.

— David

[Dan H.]

Interesting, but I have some observations:

- 2021 to present is not a long hold period.

- There is a black Swann event that started in 2020 (Covid) that would have large impact on data that starts at 2021.   This black Swann event had an impact in many markets that was unusual.   I use San Francisco as a prime example.   From turn of century to near 2020, it had incredible appreciation and rent growth.   Since then, a different story.   Cities like Boise caught fire in terms of appreciation but it has a fairly small economy.   This works great when WFH is accepted virtually every where.   Can it support a go to work environment?

- I would be interested in comparing long term appreciation and rent growth. Maybe use this century for the period.   I believe appreciation will have a tighter coupling to rent growth than any other item.   Note using the year 2000 has 2 RE black Swann events (I do not consider the dot.com bomb as an RE black Swann event), the GFC and Covid.

- the GFC negatively impacted income in some markets significantly, but many RE markets had no significant impact.  My market, San Diego, RE income was flat for a handful of years at the GFC but did not fall appreciably.   

Best wishes