mortgagetape^™

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

Rental-index data in this article is provided by Beekin, an Official AI Partner of Mortgage Tape. Beekin’s MSA-level single-family and multi-family rent indices are the input for the rent-YoY reads underlying all three findings below. Rate-lock and DSCR data is from Optimal Blue, also an Official AI Partner, via the aggregated CDL+ industry feed.

📌 Quick takeaways by role

• DSCR aggregators & non-QM originators: Note-rate is essentially flat across MSA rent-risk quintiles — 7.01% to 7.10%, an 8 bps spread across a rent-YoY range of −5.8% to +1.1%. The market is not charging borrowers for the rent shock. Simplest fix: overlay a MSA-rent-risk band on your rate sheet before your book concentrates in Miami, Cape Coral, Detroit and Phoenix by accident. Jump to Finding 1.
• Capital markets & aggregator conduits: Roughly 40% of the DSCR loans you originated in Miami / Cape Coral / Detroit at ≥1.25× coverage are now sitting in a covenant-breach band (below 1.25× but above the 1.0× hard-default line). That’s $233M UPB in Miami alone. Not a default event yet — but a re-underwrite event on any refinance, and a valuation event on any hold-to-maturity discount analysis. Jump to Finding 2.
• LGD modelers, MSR desks: In the worst rent-decline MSAs, adjusted current LTV is running 4-12 percentage points ABOVE what an appraisal-anchored LGD model reads. If your LGD anchor is a 12-24 month old appraisal — as most are — you are systematically under-reserving on the SFR investor book in Hagerstown, Canton, Cape Coral, Detroit, and Miami. Jump to Finding 3.

Two weeks ago we published a piece showing that the single-family rental market and the multi-family rental market — which had moved in near-lockstep from 2016 through 2024 — decoupled in early 2025 and have stayed apart. Median MSA-level SFR YoY is now −0.95% while MF is +1.32%, a 2.3 percentage point gap sustained for eighteen months.

That article documented the divergence. This one asks a follow-up question: has the mortgage side of the market priced it in?

The answer, across three cuts of Optimal Blue’s DSCR-flagged lock feed joined to Beekin’s MSA-level rent index, is essentially no.

Finding 1 — Note rate is flat across rent-risk quintiles

We took 66,423 DSCR-flagged CDL+ purchase locks from the trailing 12 months, joined each lock to the latest Beekin single-family rent YoY reading for its MSA (via ZIP → derived MSA/MD), and split them into five equal-sized quintiles ranked by rent-YoY. The five quintiles span the full MSA rent-risk distribution.

If DSCR aggregators are pricing rent-risk into the offered rate — the way a rational originator would — the note rate should tilt higher on loans in falling-rent MSAs.

Here is what the data actually shows:

Quintile	Avg SFR YoY	Avg DSCR (orig)	Avg note rate	n locks
Q1 (worst rent)	−5.78%	1.174	7.034%	13,285
Q2	−2.28%	1.172	7.023%	13,285
Q3	−1.51%	1.150	7.013%	13,285
Q4	−0.58%	1.183	7.019%	13,284
Q5 (best rent)	+1.14%	1.198	7.096%	13,284

Read the rate column carefully. The worst rent-risk quintile — MSAs averaging −5.78% YoY SFR rent decline — pays 7.034%. The best rent-risk quintile — MSAs averaging +1.14% YoY — pays 7.096%. That’s an eight basis point range across a seven percentage point spread in the underlying rent trajectory. The mispricing has actually inverted at the top: MSAs with the strongest rent tailwind are paying the highest rates, because higher-tier MSAs on the DSCR side skew to higher-FICO / higher-LTV loans that hit different LLPA-style stack overlays.

The origination-time DSCR is also nearly identical across quintiles (1.15 to 1.20). Aggregators are underwriting these loans on nearly identical rent assumptions and offering nearly identical rates — despite the underlying rent trajectories in the two extremes now looking nothing alike.

Finding 2 — Covenant-breach at-risk population by MSA

DSCR loans typically carry a 1.25× maintenance covenant. A borrower originated at 1.25× DSCR in a market where market rent has fallen 8% will, mechanically, now be sitting at approximately 1.16× — still above the 1.0× hard-default line, but below the covenant threshold that gives the servicer or the aggregator a re-underwriting handle.

We took all DSCR originations from the last 24 months in Optimal Blue CDL+ where the origination DSCR was ≥ 1.25×, modeled current DSCR as orig × (1 + MSA_sfr_yoy), and counted per MSA the share of loans now sitting between 1.0× and 1.25× — that is, in covenant breach.

Top MSAs by covenant-breach share (min 100 origination loans):

MSA	SFR YoY	Total orig ≥1.25	Breach n	Breach %	Breach UPB ($M)
Cape Coral–Fort Myers, FL	−8.1%	247	124	50.2%	$39M
Grand Rapids–Kentwood, MI	−5.6%	113	54	47.8%	$16M
Miami–Fort Lauderdale, FL	−6.3%	1,126	481	42.7%	$233M
San Antonio–New Braunfels, TX	−3.4%	216	87	40.3%	$17M
Port St. Lucie, FL	−6.5%	101	40	39.6%	$10M
Dallas–Fort Worth, TX	−1.7%	466	173	37.1%	$51M
Detroit–Warren–Dearborn, MI	−6.9%	488	180	36.9%	$27M
Las Vegas–Henderson, NV	−2.5%	208	76	36.5%	$26M
Phoenix–Mesa–Chandler, AZ	−3.4%	434	141	32.5%	$53M
Memphis, TN-MS-AR	−4.9%	468	164	35.0%	$24M

Miami is the number that matters most. Roughly $233M of the 24-month DSCR originations in Miami-Fort Lauderdale sit in the covenant-breach band today. None of these loans has “defaulted” in any traditional sense. Every one of them is presumably current on scheduled debt service. But every one is a re-underwrite event the moment the borrower needs to refinance out of a bridge, extend a balloon, or move to a longer-term takeout at prevailing rates. The workout economics on a covenant-breach loan are always worse than on a loan re-underwritten inside the covenant.

The concentration in Florida is especially notable because Florida is where the 2020-2022 SFR boom pulled the most out-of-state institutional capital in. The same MSAs that saw the strongest rent tailwind three years ago now show the deepest rent correction, and the DSCR aggregator book acquired at the peak of that tailwind is where the covenant-breach exposure is concentrated.

Finding 3 — LGD anchor lag by MSA

The third cut is the most consequential for a portfolio manager and the least visible in normal LGD reporting.

Standard loss-given-default modeling anchors on the origination appraisal. That appraisal is by construction 12-24 months old by the time any loss event might crystallize, and even a Broker Price Opinion refresh is anchored on comps that lag the actual rent-signal turn by another quarter or two.

Income property valuation, in economic reality, tracks market rent via a fairly stable cap-rate multiple. If market rent in an MSA falls 7%, income-approach valuation falls roughly 7% too, over a period much shorter than a typical appraisal refresh cycle. That means in a falling-rent MSA, the actual current LTV on the outstanding investor book is materially higher than the appraisal-anchored reading a LGD model produces.

We modeled adjusted current LTV as base_loan_amount / (appraisal × (1 + MSA_sfr_yoy)) for DSCR CDL+ originations in the last 24 months. The formula assumes MSA cap rates hold roughly steady over the window, so income-approach valuation moves roughly 1-for-1 with market rent; where cap rates are also expanding (rising exit yields on a falling NOI), the LTV lag understates the true drift, and where they’re compressing it overstates it. Cap-rate stability is a decent baseline for the trailing 24 months but is not a permanent assumption. The top MSAs by LTV lag:

MSA	SFR YoY	Loans	Orig LTV	Adjusted current LTV	LTV lag
Hagerstown-Martinsburg, MD-WV	−14.4%	235	72.9%	85.1%	+12.2 pp
Canton-Massillon, OH	−9.6%	121	73.6%	81.4%	+7.8 pp
Cape Coral-Fort Myers, FL	−8.1%	1,270	69.8%	75.9%	+6.1 pp
Lynchburg, VA	−6.8%	150	73.2%	78.6%	+5.4 pp
Detroit-Warren-Dearborn, MI	−6.9%	1,212	71.5%	76.8%	+5.3 pp
Milwaukee-Waukesha, WI	−6.5%	924	73.6%	78.7%	+5.1 pp
Miami-Fort Lauderdale, FL	−6.3%	7,156	65.9%	70.3%	+4.4 pp
Port St. Lucie, FL	−6.5%	505	67.8%	72.5%	+4.7 pp
Grand Rapids, MI	−5.6%	279	70.6%	74.9%	+4.2 pp
Panama City, FL	−5.5%	429	70.6%	74.7%	+4.1 pp

Hagerstown is the outlier — SFR down 14.4% in a market that had a genuine bubble in 2021-2022 and is now working through the reversion. But the Miami row is where the aggregate exposure sits: 7,156 DSCR loans in the CDL+ 24-month window, and each one carries an adjusted current LTV that’s on average 4.4 pp higher than the appraisal your LGD model is running against.

Four percentage points of LTV translates to material LGD divergence. On a 70-LTV book, a 4-point shift moves you from ~70% LTV to ~74% LTV — for a DSCR investor loan in a stress scenario, that pushes the modeled loss on default from ~15% severity into the ~22% range depending on your severity curve.

What to do about this

If you are on any of the following seats, there is a specific, actionable read on these three findings:

DSCR / non-QM aggregators: You have latitude to overlay MSA rent-risk on your rate sheet before the market forces you to. Beekin’s MSA-level SFR YoY panel is the input; a −4% or worse MSA reasonably justifies 25-50 bps of additional pricing. The Q1 quintile in this analysis pays 7.03%; on rate-parity math, that should probably be closer to 7.30-7.50% given the tail risk. Doing this before your competitors gives you an incremental margin sleeve and it steers your pipeline away from the concentration risk documented in Finding 2.

Aggregator conduits & MSR desks: The DSCR loan book acquired in the 2023-2024 window in Florida, Michigan, Texas Sunbelt, and Arizona is materially different in current-DSCR terms than what its origination-DSCR profile suggests. Any prepay-and-default cash flow re-projection anchored on origination-DSCR is now overstating the covenant-safe fraction of the book. Beekin’s MSA panel lets you compute the current-DSCR-adjusted view — as demonstrated in Finding 2 — and rerun your projections against it.

LGD modelers & portfolio risk teams: LGD models anchored on origination appraisal are systematically under-reading LTV — and therefore under-reserving expected loss — in the top-15 MSAs listed in Finding 3. The fix is not to wait for the next appraisal refresh cycle. A MSA-rent-YoY adjustment to the appraisal input is the simplest available upgrade; the more sophisticated version replaces the appraisal anchor entirely with a rolling cap-rate-based income-approach current-value estimate.

A note on scope. The DSCR cohort here is Optimal Blue’s CDL+ industry-aggregate lock feed with dscr_ratio populated, over the last 12-24 months. This captures a large share of the non-QM DSCR origination universe but not all of it — DSCR originations that never touched the OB pipeline (some smaller correspondent conduits, some captive-lender flows) are outside this sample. The findings should be read as directionally representative rather than as a full-book census.

The larger point stands regardless: an MSA-level rent signal is the single most useful piece of extra data for pricing, monitoring, and reserving on the DSCR investor book. It’s been available since 2015. Until this month, it wasn’t joined to loan-level lock data in a way that lets you see the pattern. Now it is.

Interested in getting these numbers refreshed for your portfolio, or building the covenant-breach + LGD-adjustment queries against your own book? Contact us or try the analysis in chat directly.