mortgagetape^™

Written by Mortgage Tape Team—a group of industry analysts leveraging our proprietary mortgage-domain language models to synthesize and decode housing data.

This article was written in collaboration with Beekin, an Official AI Partner of Mortgage Tape, using Beekin’s rent-market analytics firm providing MSA-level single-family and multi-family data and indices.

📌 Quick takeaways by role

• DSCR / non-QM investor desks: The single-family rental market has quietly turned. Across 461 major MSAs, the median SFR rent-index YoY reading is −0.95% at Aug 2026 — down from +2.5% just eighteen months ago. Multi-family (2-5 unit apartment) rents are still growing at +1.3% in the same footprint. That 2.3 pp gap changes DSCR math on the SFR book in a way an MSA-blind aggregate rent index would obscure. The correction is concentrated in Detroit (SFR −6.9%), Miami (−6.3%), Phoenix (−3.4%) and San Antonio (−3.4%). Jump to DSCR takeaways.
• Capital markets & MSR desks: MBS / private-label pools weighted toward SFR (single-family investor loans, DSCR aggregators) face a rent-cash-flow shock that pools weighted toward small MF (2-4 unit non-QM) don’t. Prepay + default assumptions on the two sides should diverge. Aggregate rent indices (ZORI, BLS OER) don’t split by property class, so the signal is invisible from the standard data feeds. Jump to capital-markets takeaways.
• Non-QM originators & aggregators: If your book skews SFR-DSCR you’re already seeing the pressure. If you’re competing on 1-4 unit non-QM (small MF), the underlying rent trajectory is materially stronger than the SFR narrative suggests — that’s actual pipeline strength. Bay Area MF is running +7.2% YoY, Chicago MF +3.9%, NY MF +3.7%. Jump to origination takeaways.

Everyone quoting rent data uses either ZORI (Zillow) or the BLS shelter series. Both are useful. Neither splits single-family from multi-family. That matters right now because those two asset classes have decoupled — a fact that is invisible in any aggregate rent print and is only barely surfacing in the SFR trade press. The Beekin rent index does split them, at MSA level, back to 2015. This is what it shows.

What the data shows

Beekin publishes two monthly rent indices per MSA — one for single-family (SFR) and one for multi-family (MF). The MF series here is scoped to small residential multi-family: 2–5 unit properties (the mortgage-side non-QM small-MF universe), not the 50+ unit institutional apartment complexes tracked by NCREIF or CoStar’s commercial series. Methodology is same-store, log-interpolated — the same broad approach Case-Shiller uses on home prices. The index is normalized to 1.0 in January 2015 per MSA per property class. The panel we’re using is 461 MSA/MD entries with matched CBSA codes, running through August 2026.

1. 2016-2024 — lockstep. The two series moved together. YoY correlation across those nine years is 0.94. Both showed the rate-boom acceleration in 2021 (both peaked around +4.5-6% YoY through mid-2022), both decelerated in 2023-2024 as the rate-shock reset regional housing markets.
2. Early 2025 — first daylight. The two series pulled apart in Q1 2025. SFR fell through zero for the first time since 2016; MF held around +1.5-2%.
3. Mid-2026 — sustained ~2 pp gap. At August 2026, median SFR YoY is −0.95%, median MF YoY is +1.32% — a 2.27 pp gap that has now persisted for eighteen months. Not a one-print noise event.

The reasons are macro (SFR supply from build-to-rent + iBuyer + institutional owners has caught up to demand; MF absorption tightened as multifamily new-supply pipeline drained through 2024-2025) but the practical point for mortgage-side underwriting is simpler: the two loan books look identical on paper and are now trending oppositely.

Geography — nearly universal, with a DC exception

Cutting the same YoY reading across the top-35 US metros makes the pattern concrete:

34 of 35 major MSAs show MF outpacing SFR at August 2026. The only exception is DC, where SFR is +2.9% and MF is −0.6% — a reversal of the national pattern, likely reflecting the DC region’s specific commercial-real-estate stress and remote-work permanence bleeding into the apartment market. Every other major metro shows the same directional pattern with different magnitudes.

Deepest SFR corrections (top 5): Detroit (−6.9%), Miami (−6.3%), San Antonio (−3.4%), Phoenix (−3.4%), Columbus OH (−2.4%). Miami is particularly notable because the metro was one of the strongest SFR-appreciation stories of 2020-2022 — the correction is a reversion in the same market that saw the biggest run-up.

Strongest MF outperformance (top 5): San Francisco (+7.2%), San Jose (+7.1%), Chicago (+3.9%), New York (+3.7%), Baltimore (+2.8%). The Bay Area MF strength is the story of tech re-in-migration + a multi-year apartment supply drought resolving; Chicago and NY are the “gateway city return” narrative in an asset class that lags home-price signal by 18-24 months.

Notable narrow gaps: Seattle (SFR +0.7% / MF +1.1%), Riverside (+0.3% / +1.3%), Boston (+0.4% / +1.8%). These metros show the same directional pattern but the two series haven’t decoupled meaningfully — SFR-heavy investor pools in these markets are still tracking apartment fundamentals reasonably well.

What this means for DSCR underwriting

DSCR (Debt Service Coverage Ratio) loans price off market rent. Most non-QM DSCR programs cap at 1.15-1.25× coverage on a market-rent basis; a 5% market-rent drop can drop a marginal loan below its underwriting floor. The Beekin data implies:

1. SFR-DSCR book drift toward re-underwriting. A meaningful share of the 2022-2024 vintage is quietly slipping. A borrower originated at a 1.10 DSCR on Miami single-family rent in 2023 is likely running closer to 1.02-1.05 today on the same property, purely from the rent slippage. The loan is still current — but any borrower who needed to refinance out of a bridge or into a longer-term takeout at prevailing rates faces a materially harder underwrite.
2. Small-MF DSCR holding steady. MF-DSCR (2-4 unit non-QM investor) is not seeing the same pressure. Same 2023 vintage on a 4-unit small MF in the same market has typically seen rents flat-to-up, so the current-DSCR is stable. Underwriting the two products off a common “market rent” assumption is now wrong.
3. Aggregator conduit exposure is asymmetric. DSCR aggregators concentrated on SFR (Sachem, Angel Oak’s SFR sleeve, several NRZ conduits) carry different WAC-erosion risk than aggregators with balanced SFR/small-MF mix.
4. Cash-out refi eligibility tightening in correction MSAs. Miami, Phoenix, Detroit — even at the same LTV, the property’s implied cash-flow yield has fallen. Cash-out programs pricing off a 12-month lookback DSCR will filter out loans that would have qualified against 2024 rents.

Underwriting rule of thumb. An MSA-level aggregate rent index conceals underlying performance. SFR loans are facing cash-flow compression; small multi-family properties are still holding their yields. Any DSCR risk model that doesn’t split by property class is reading the wrong number.

The mechanism is: market rent trajectory is a first-order feature of any DSCR risk model, and it needs to be split by property class. Using an MSA-level aggregate rent index (ZORI or CS-like) blends SFR and MF and hides the divergence.

What it means for MBS / MSR

Two second-order consequences worth flagging:

1. SFR-DSCR pools need a rent-shock haircut. For any deal analytics that estimate DSCR migration or forward default probability, the SFR-DSCR pools deserve a haircut relative to MF-DSCR pools in the same MSAs. The magnitude of the haircut is MSA-dependent (−7% Detroit SFR rents vs +1.8% Detroit MF rents implies a very different marginal-default assumption on the two loan sides of the same geography).
2. Prepay behavior probably diverges too. SFR investors facing declining rents have higher forced-sale prepay pressure (property valuations reset; owners under NOI stress deleverage); MF investors don’t. Standard prepay curves that pool the two are likely too optimistic on SFR and too pessimistic on MF at the current rate node.

We haven’t yet quantified the AUC lift on including MSA-level SFR-vs-MF divergence as a scoring feature — that’s a separate research pass. First-order intuition says it should matter most for DSCR default and prepay targets.

What it means for originators + aggregators

For originators competing across the DSCR / non-QM stack:

• SFR-DSCR pipeline coming in weaker. Any conversation with a repeat SFR investor client is now going to include “the rent numbers don’t work anymore” more often than they did a year ago. Position accordingly — pre-underwrite a stress scenario before quoting.
• 1-4 unit non-QM pipeline should look healthier. Same investors moving to small MF still see rent growth. If you’re originating both products, the pipeline story on the MF side is more constructive.
• Bay Area, Chicago, NY MF is a real opportunity. +3.7-7.2% rent growth on small-MF properties in those metros is a genuine cash-flow story. Broker/aggregator marketing should reflect that these are the strongest metros for MF originations today — most competitive markets right now.
• Retention risk on 2022-2024 SFR-DSCR originations. The rent trajectory implies more borrowers looking to refinance or delever over the next 12 months than the pool’s LTV history alone would suggest. Servicing books should think about retention outreach on this specific cohort.

Methodology

Data source: Beekin monthly rent index, MSA-level, single-family and multi-family series, 2015-01 through 2026-08. Same-store methodology with log-interpolation between observations — the same broad approach Case-Shiller uses on home prices. Index normalized to 1.0 in January 2015 per (MSA, property class). Cohort restricted to 461 MSA/MD entries that match Mortgage Tape’s msa_name_lookup (i.e., excludes ~535 micropolitan areas where same-store panel depth is thinner and can produce flat 0.0% YoY readings). Median YoY = 50th percentile of the (current index / index 12 months prior − 1) distribution across the cohort per month.

What we didn’t do

• We didn’t source-attribute Beekin against a Case-Shiller residual test. Case-Shiller reports home prices, not rents, so a direct level comparison isn’t meaningful — but a covariance test between Beekin’s index and CS in the metros both cover would be a useful next validation pass.
• We didn’t yet layer this signal into any Mortgage Tape scoring model. DSCR default, prepay, and repurchase models could all plausibly benefit from a msa_sfr_yoy + msa_mf_yoy pair as features. Estimated AUC lift TBD once the retrain runs.
• We didn’t split by loan vintage. A 2023 SFR-DSCR origination against 2023 rents faces different current-DSCR pressure than a 2020 origination against 2020 rents. A vintage cut would sharpen the mortgage-book risk view.

Try it yourself. Ask Mortgage Tape: “what’s the current SFR vs MF rent trajectory for Miami” — the underlying Beekin panel is queryable at MSA level.