Debt Service Coverage Ratio (DSCR): Sizing Constraints, Amortizing vs IO, and the 2026 Institutional Discipline

Key Takeaways

Institutional CRE DSCR vs SFR DSCR Loans

Before going further: this article is about DSCR as the institutional sizing constraint on commercial real estate loans: CMBS conduit, bank, life insurance company, agency (Fannie Mae and Freddie Mac multifamily), and bridge debt. The math, the minimums, and the underwriting discipline all assume an income-producing commercial asset with institutional-grade financing.

If you arrived here looking for a "DSCR loan" as a single-family-rental investor mortgage product (the 1-to-4 unit residential cash-flow loans marketed to landlords without W-2 income verification), this is not the right article. a residential DSCR-loan lender is the better resource for that product category. The residential DSCR-loan market uses similar vocabulary but a different underwriting box (FICO, DTI, per-asset pricing, residential appraisal). The math below applies to the institutional CRE world only.

What DSCR Actually Measures

The debt service coverage ratio is the lender's coverage discipline. It answers a single question: if I lend against this property's cash flow, how much cushion exists between the income the asset throws off and the contractual payment I require? At a 1.25x DSCR, the property generates $1.25 of NOI for every $1.00 of debt service. At 1.50x, $1.50 of NOI for every $1.00 of debt service. The number is a buffer: the room the property has to absorb vacancy, expense pressure, or NOI compression before the loan goes into payment shortfall.

DSCR is a lender's metric, not a borrower's. Borrowers don't have a DSCR target the way they have an IRR target; what they have is a sizing constraint. The lender publishes a minimum DSCR, and the borrower's loan proceeds are capped at whatever balance satisfies that minimum on the property's underwritten NOI. Push the minimum from 1.20x to 1.30x and proceeds compress by roughly 8% on the same deal. Push it from 1.25x to 1.40x and proceeds compress by roughly 11%. The minimum is not a goalpost. It is a ceiling.

The metric matters most when NOI is the binding constraint: when the property's income is low relative to its value, low relative to its cost basis, or low relative to the loan balance. That happens in three common situations: (1) cap rates are compressed (a high-priced, low-yielding asset), which loosens LTV relative to income so that the income-based tests bind first, (2) the rate environment moved against the deal (in-place NOI was sized when coupons were 200 bps lower), or (3) the asset class itself runs thin coverage even at stabilization (urban office, full-service hospitality). In those cases, an income-based test (DSCR or debt yield) picks the loan balance; which of the two wins depends on the coupon, the amortization, and each minimum, not on the cap rate. In other cases, particularly value-add or development with high-yielding stabilized economics, LTC or LTV typically binds first.

The Formula and Its Definitions

The formula is short and the definitions are where the work lives:

DSCR = Net Operating Income ÷ Annual Debt Service

The numerator is NOI. Institutional NOI is gross revenue, less vacancy and credit loss, less operating expenses, with capital reserves typically held below the line. Some lenders size on net cash flow (NCF) instead. The Fannie Mae Multifamily Guidedefines Underwritten NCF as underwritten NOI minus a replacement reserve (a minimum of $200 per unit per year, included whether or not the escrow is funded) and computes Underwritten DSCR on NCF, so dividing NOI by debt service overstates agency DSCR. The lender's underwriting NOI is almost never the broker's pro-forma NOI. It is normalized: in-place rent roll (not market-rate projections), vacancy and credit loss at the higher of in-place or submarket average, OpEx benchmarked to comparable properties in the asset class (an expense ratio well below comparables is a red flag), management fees at-market (not in-house), and one-time items stripped out.

The denominator is annual debt service, and this is where practitioners get tripped up. Annual debt service is principal plus interest (not interest only) unless the loan is on an IO period. The mortgage constant captures this: debt service divided by loan balance, expressed as a percentage. A fully-amortizing 25-year loan at 6.5% has a constant of roughly 8.10%; the same loan on a 30-year schedule has a constant of roughly 7.59%. A pure interest-only loan at 6.5% has a constant equal to the coupon: 6.50%. The lower the constant, the higher the debt the same NOI supports at the same DSCR.

Some lenders include reserves and escrows in the debt service calculation. CMBS conduit and many bank construction loans require tax and insurance escrows funded monthly, plus FF&E reserves on hospitality and capital reserves on multifamily. These flow through a "lockbox" or "cash management" structure that, in distress, captures cash before equity distributions. They don't typically show up in the debt service side of the formula, which is computed on principal and interest only, but they materially affect the property's free cash flow available to equity. The distinction matters most in stressed scenarios where the lender pulls cash that the sponsor was relying on for distributions.

Institutional Minimums by Lender Type

"What is a good DSCR?" depends entirely on the lender. The matrix below is an illustrative baseline by lender category and property type, the author's market read rather than published lender requirements. These are minimums, not targets, and they tier up for higher LTV, weaker property types, weaker markets, and looser sponsor profiles. Lenders may also apply a stressed-rate overlay (covered in the next section) on top of the headline minimum.

Illustrative institutional DSCR minimums by lender type. These are floor numbers; actual minimums tier up for weaker markets, weaker property types, and higher LTV. Agency DSCR is computed on net cash flow per the Fannie Mae Multifamily Guide, Part II, Chapter 2.

A few cluster-specific calibrations behind the matrix:

• CMBS conduit (1.20–1.25x). CMBS sizing leans on debt yield (NOI ÷ loan) alongside DSCR and LTV. A debt yield minimum binds ahead of DSCR whenever it exceeds DSCR × mortgage constant, which is about 8.6–9.5% at 1.20–1.25x and 6.00–6.50% on a 30-year schedule. Partial-IO is common on CMBS (illustratively 2-5 years IO, then amortizing on a 30-year schedule); where the DSCR test runs at the amortizing payment, the IO period relieves cash flow but adds no proceeds.

• Commercial banks (1.25–1.40x). Bank DSCR is the most variable. A regional bank lending to a long-term relationship at 60% LTV runs 1.25x; the same bank lending to a new sponsor on a non-relationship deal runs 1.35–1.40x. Bank loan docs typically include a DSCR covenant tested annually, which is operationally consequential: a covenant breach triggers cash management, not just default. The stressed DSCR overlay (covered next) may also apply on bank refinancings.

• Life insurance companies (1.30–1.50x). Life co lending tends to be conservative. Life co underwriting often layers a stressed-rate test (in-place rate + 100–200 bps at a 1.0x DSCR floor) on top of the headline minimum. On a 30-year amortizing loan at a 1.30x headline, that stress still leaves coverage above 1.0x (see the next section); the stressed test matters for IO structures and thinner headline coverage. The life co quote also tends to assume a longer amortization (25-year or 30-year) than CMBS or bank debt, which sizes proceeds higher on the same DSCR.

• Agency multifamily (1.20–1.55x). Fannie Mae sizes on Underwritten DSCR, the ratio of underwritten net cash flow (NOI less a replacement reserve) to a level debt service payment with the applicable amortization, using the greater of the actual note rate or an Underwriting Interest Rate Floor. Minimum DSCR and maximum LTV are set by tier in the Multifamily Underwriting Standards (Form 4660); the 1.20–1.55x range above is an illustrative assumption, not a reproduction of that matrix.

• Bridge / private debt (1.10–1.20x). Bridge lenders accept thin DSCR at origination because the deal is transitional: the property is being repositioned, leased up, or stabilized. The lender sizes to a stabilized DSCR at conversion, with the acquisition-date DSCR effectively a placeholder. Most bridge structures include cash management, an interest reserve, and a mandatory conversion test before the loan can be extended or refinanced into permanent financing.

Amortizing vs IO: The Proceeds Lift

The single most consequential practitioner insight in DSCR sizing is that where a lender tests DSCR on the actual interest-only payment, interest-only debt sizes materially more proceeds at the same DSCR: about 17% over a 30-year schedule and about 25% over a 25-year schedule at a 6.5% coupon. Many bridge loans and some bank or life company loans are sized this way. Fannie Mae is not: its Guide requires a level debt service payment including amortization regardless of the length of the interest-only period, so IO adds no DSCR-constrained proceeds on Fannie Mae loans. The reason is the mortgage constant. On an amortizing loan, debt service includes both interest and principal repayment. On a pure IO loan, debt service is only interest. Same NOI, same DSCR minimum, smaller denominator: bigger loan balance.

Work the math on a $5.0M NOI multifamily asset at a 1.25x DSCR minimum and a 6.5% all-in coupon. Maximum debt service at the constraint is $5.0M ÷ 1.25x = $4.0M. Now the question is: how big a loan does $4.0M of annual debt service support, and the answer depends entirely on the amortization schedule.

Proceeds at 1.25x DSCR minimum, $5.0M NOI, 6.5% coupon, varying amortization structure. The full-IO loan sizes 24.5% more proceeds than the 25-year amortizing equivalent (16.7% more than the 30-year), on identical NOI, identical coupon, identical DSCR, where DSCR is tested on the actual IO payment. Mortgage constants assume monthly payments.

The proceeds lift is rate-sensitive. At a 4.0% coupon (the 2021 vintage), the constant on a 30-year amortizing loan is roughly 5.73% versus 4.00% IO, a 43% lift in proceeds at the same DSCR. At a 6.5% coupon (the 2026 vintage), the lift compresses to roughly 17% versus 30-year amortizing. The IO proceeds advantage widens at lower rates, which made IO sizing especially attractive to borrowers in the 2020-2021 low-rate vintage.

Lenders restrict or neutralize IO for exactly this reason. Typical patterns, as the author's market read:

• CMBS partial-IO. 2-5 years of IO followed by 25- or 30-year amortization is a common conduit structure. Where the DSCR test runs at the post-IO amortizing payment, the IO period gives the borrower cash flow flexibility without sizing the loan against the lower constant. Whether a full-term IO loan is sized on the IO payment or an amortizing payment depends on the lender and the deal; confirm the basis in the term sheet.

• Agency multifamily IO. Fannie Mae's Guide (Part II, Chapter 2, Section 203.02) requires Underwritten DSCR to use the same level debt service payment, including amortization, regardless of the length of the interest-only period. An IO period therefore changes cash flow and pricing on agency loans but does not raise DSCR-constrained proceeds.

• Bank IO. Construction loans are IO during construction (no amortization on undrawn principal anyway). Conversion to permanent typically requires conversion to a fixed-rate amortizing structure. Permanent bank loans vary; partial IO is sometimes negotiated for relationship borrowers.

• Life co IO. Life companies often limit IO to lower-LTV loans on the strongest property types (trophy multifamily, industrial), and many life co loans amortize from day one on a 25- or 30-year schedule.

• Bridge IO. Bridge debt is almost universally IO during the transitional period. The loan sizes to the stabilized DSCR at conversion, not the in-place DSCR.

Stressed DSCR: The Post-2022 Discipline

Many lenders add a stressed DSCR test on top of the headline minimum. The underwritten coupon at origination is one rate; the stressed coupon is that rate plus an increment (illustratively 100–200 basis points), and the lender requires the deal to clear a 1.0x DSCR (or sometimes 1.10x) at the stressed level. It is most associated with life company and bank lending. The arithmetic is simple: stressed DSCR = headline DSCR × (mortgage constant at the base rate ÷ mortgage constant at the stressed rate).

The rationale: when rates rose sharply from 2022, the in-place DSCR cushion compressed on floating-rate paper and on fixed-rate paper at refinancing. Stressed DSCR builds the rate-volatility shock into origination: if the loan is to be extended or refinanced into a higher-rate environment, the property's cash flow must support that scenario at a thin (1.0x) but positive coverage.

Work an example. A trophy multifamily asset throws off $5.0M of NOI. The life co quotes 1.30x DSCR on a 5.75% coupon. Headline DSCR test:

• Max DS at 1.30x = $5.0M ÷ 1.30x = $3.85M
• At 5.75% coupon, 30-year amortization, monthly pay, constant = 7.00%
• Max loan = $3.85M ÷ 7.00% = $55.0M

Now apply the stressed test: in-place 5.75% + 150 bps stress = 7.25%, 1.0x minimum coverage at stress:

• Max DS at 1.0x stressed = $5.0M ÷ 1.0x = $5.0M (the full NOI is the ceiling)
• At 7.25% stressed coupon, same 30-year amortization, constant = 8.19%
• Max loan at stressed test = $5.0M ÷ 8.19% = $61.1M

The headline 1.30x test sizes to $55.0M; the stressed 1.0x test sizes to $61.1M. On this deal, the headline test binds and the stressed test is not the active constraint: at $55.0M the stressed DSCR is 1.11x. Push the headline DSCR to 1.25x and the headline sizes to $57.1M with a stressed DSCR of 1.07x; the headline still binds. On a 30-year amortizing loan the stressed test becomes active only when headline coverage is thin or the shock is large. From a 6.5% base, a 1.25x headline needs about +230 bps and a 1.20x headline about +185 bps before stressed DSCR falls to 1.0x; from a 5.75% base, 1.30x needs about +260 bps and 1.25x about +219 bps. Interest-only debt reaches the floor sooner: a 1.30x IO loan at a 6.5% coupon falls to 1.0x at +195 bps. The stressed test is most likely to matter on:

• Floating-rate deals where some lenders use the SOFR cap strike (the cap rate at the cap purchase) as the stressed rate, effectively pricing the stressed scenario into the cap economics
• Interest-only or partial-IO loans where the stressed test is run on an interest-only payment (a 1.20x IO headline at a 6.5% coupon falls to 0.92x at an 8.5% stressed rate)
• Hospitality and select-service lodging where DSCR cushions are thinner to begin with

The Federal Reserve's2026 stress test scenariosare a regulatory analogue at the bank-balance-sheet level: they include a 39 percent decline in commercial real estate prices. That is a price and macro shock rather than a rate shock, so it is an analogy to, not a version of, the deal-level stressed-DSCR test.

The Constraint Hierarchy: DSCR vs LTV vs Debt Yield vs LTC

DSCR is one of four sizing constraints, and the lender sizes to the lowest of the four. The full constraint stack is:

• DSCR (cash-flow test). NOI ÷ annual debt service. Governs when income is the binding variable: when NOI is low relative to value, cost, or loan balance. Covered in this article.

• LTV (value test). Loan balance ÷ appraised value. Governs when value is the binding variable, typically on properties trading at premium multiples or in markets where appraisals are compressed. Covered inLTV vs LTC: When Each Governs.

• Debt yield (yield test). NOI ÷ loan balance. A central CMBS conduit sizing test, it binds when the asset trades at a premium and the loan-to-value optically clears but the actual debt yield is thin. CMBS pivoted to debt yield post-GFC because DSCR is gameable (lower the coupon, lengthen the amortization, the DSCR rises) but the asset's yield to the lender doesn't change. Covered inDebt Yield: Why Institutional Lenders Prefer It.

• LTC (cost test). Loan balance ÷ total project cost. Governs on construction, value-add, and development: deals where the appraised value at exit is speculative and the cost basis is the only verifiable anchor.

On any given deal, all four tests run in parallel and the lender takes the minimum. The institutional discipline is to compute all four explicitly on every term sheet: sponsors who only model DSCR consistently overestimate proceeds because they miss the deals where debt yield or LTC binds tighter. Conversely, sponsors who only model LTV miss the DSCR ceiling on income-thin assets.

A practitioner heuristic: Compressed cap rates loosen LTV, so DSCR or debt yield tends to bind on stabilized core/core-plus; which of the two wins depends on the coupon, the amortization, and the two minimums. Rate-stressed refinancings tend to be DSCR-bound. Debt yield binds on premium-priced CMBS conduit deals. LTV binds on trophy/gateway-market deals where lenders cap absolute leverage despite ample coverage. LTC binds on value-add, construction, and ground-up development. Knowing which constraint is likely to bind beforerunning the math saves iteration time on the deal model.

2026 Rate Environment and Sizing

The 2026 rate environment is the central variable in DSCR sizing today. On May 29, 2026 theUS Treasury par yield curveshowed 3.69% at three months and 4.45% at ten years, an upward-sloping curve, andSOFRwas 3.63%. By October 7, 2026 the ten-year yield had risen to 5.28% per theFederal Reserve H.15. Illustrative all-in coupons by lender type, rounded and struck off the late-May curve (re-price against the current curve; the ten-year yield was 83 bps higher on October 7 than on May 29):

• CMBS conduit: 6.00–6.50% all-in (roughly T+150 to T+200)
• Commercial banks: 6.50–7.00% all-in (often floating-rate over SOFR with cap requirements; illustrative)
• Life insurance companies: 5.75–6.25% all-in (roughly T+125 to T+175, fixed-rate)
• Agency multifamily (Fannie/Freddie): 5.50–6.00% all-in (roughly T+100 to T+150)
• Bridge / private debt: 6.63–8.63% all-in (SOFR of 3.63% + 300-500 bps floating, often with floor)

These rates are 200-300 bps above the 2021 vintage and the central problem in 2026 refinancing. A 2021-vintage multifamily loan originated at a 3.50% fixed coupon, 30-year amortization, and roughly a 1.25x DSCR, with the lender's underwritten NOI then sufficient for the size. Suppose the same property's NOI today is up 15% on rent growth since origination; it faces a refinancing market with coupons at 6.00% or higher and similar 1.25x DSCR minimums. The mortgage constant moved from roughly 5.39% (3.50% / 30-year amortizing) to roughly 7.19% (6.00% / 30-year amortizing), a 33% increase in required debt service per dollar of loan balance.

Apply that to a worked refinancing. A $50M loan originated in 2021 at 3.50%/30-year required $2.69M of annual debt service against $3.35M of underwritten NOI (1.24x). Refinance today at the same property: NOI is now $3.85M (+15% on growth). The new lender quotes 6.00%/30-year/1.25x DSCR. Max DS at the new constraint = $3.85M ÷ 1.25x = $3.08M. Max loan at the new constant (7.19%) = $3.08M ÷ 7.19% = $42.8M, which is $7.2M below the original $50M. Assume the original loan was a five-year-term loan maturing now, so the remaining balance after five years of amortization is about $44.8M. The borrower then needs to bring about $2.0M of new equity to pay off the existing loan, even though NOI is up 15% over the hold period. This is the refinancing problem at the deal level: the higher constant outruns the NOI growth.

The DSCR-binding outcomes follow predictably. (1) The borrower brings new equity to plug the gap: the common resolution for sponsors with dry powder and conviction. (2) The borrower extends the maturity with the existing lender at a higher rate, often with a cash sweep until DSCR rebuilds: the common resolution for bank refinancings where the relationship matters. (3) The borrower trades the asset at a price that resolves the basis: the common outcome on weaker properties or weaker sponsors. (4) The asset defaults or is discounted-payoff'd: the outcome on properties where the gap exceeds the equity's risk appetite. All four outcomes are possible.

How to Model It

A useful DSCR sizing model needs to do four things: (1) compute the headline DSCR at in-place NOI on every term sheet, (2) compute the stressed DSCR at in-place + 100-200 bps stress, (3) compute the proceeds at constraint across amortizing and IO variants, and (4) cross-reference DSCR-binding proceeds against the LTV/LTC/debt yield minimums to identify the actual binding test. Most Excel debt-sizing tabs do (1) cleanly, (2) inconsistently, (3) rarely on the same sheet, and (4) almost never.

The institutional workflow that closes the gap:

• 1. Normalize NOI. Rebuild from rent roll and T-12. Strip one-time items. Benchmark OpEx to comparable properties for the asset class, and deduct replacement reserves where the lender sizes on net cash flow. Don't quote the broker's NOI: quote your own.

• 2. Build the four-constraint cross-check. On a single sheet, compute loan balance at the DSCR constraint, the LTV constraint, the debt yield constraint, and the LTC constraint. Output the minimum of all four as the binding proceeds figure.

• 3. Run amortizing and IO variants. Both on the same DSCR. The IO column is the ceiling on what a lender can stretch to where it tests DSCR on actual IO payments.

• 4. Apply the stressed-rate overlay. In-place coupon plus the lender's stress increment (100–200 bps), 1.0x DSCR floor at stress. Document the stressed-rate assumption inline wherever the lender applies one.

• 5. Compare term sheets on the same basis. The CMBS quote at 1.20x sizes higher than the life co quote at 1.40x at the same NOI and rate. The bank quote at 1.30x with a stressed-rate overlay may size lower than either. Normalize all four lender types on the same sheet, then compare on proceeds, blended cost of capital, prepayment flexibility, and recourse profile.

The Apers Marketplace pocket models below run this workflow as a single-sheet pattern. AQ-110 (multifamily) layers the four-constraint sizing on a 10-year monthly cash flow with IO and amortizing variants on the debt tab. AQ-141 (opportunistic with bridge) handles the bridge-to-perm conversion DSCR specifically, sized to stabilized DSCR at conversion, with the acquisition-date DSCR carried as a placeholder. AQ-301 (anchored retail) extends the same pattern to the more complex retail tenant-rollover schedule, where the underwritten DSCR runs on rolled-down rents rather than in-place.

Common Mistakes

Seven errors we see repeatedly on DSCR sizing, each of which can overstate proceeds in an IC memo:

• Using the broker's pro-forma NOI without normalizing. Broker NOIs add back T-12 vacancy below market, treat one-time expenses as non-recurring, normalize management fees to a "market" rate even when the property is owner-managed, and bake in rent growth that's already priced into the asking price. Rebuild from the rent roll and T-12 every time. Differences between broker NOI and the lender's underwritten NOI are common, and the lender's number is the one that sizes the loan.

• Using stabilized NOI when the lender sizes to in-place. Stabilized NOI is the borrower's projection; in-place NOI is the rent roll annualized. The lender sizes to in-place for stabilized core/core-plus and to stabilized-at-conversion for bridge/transitional, never to the borrower's year-3 or year-5 projection on a permanent loan. Sponsors who model DSCR off stabilized NOI on a permanent loan structurally overstate proceeds.

• Confusing amortizing and IO DSCR. The 1.45x DSCR a sponsor pitches at IC is often the IO DSCR on an IO/amortizing partial-IO structure: the amortizing-period DSCR on a 30-year schedule at a 6.5% coupon is about 1.24x, which fails a 1.25x post-IO test. Always quote both. The post-IO amortizing DSCR is the relevant constraint for the loan's long-term cash flow profile.

• Ignoring the stressed-DSCR test. A deal that clears its headline DSCR can still fail 1.0x at stressed rates when it is IO or partial-IO, thinly covered (near 1.20x), or the shock is large, and it will get re-traded or repriced. The stressed test is common at life co and banks; running only the headline DSCR can miss the binding constraint on those deals.

• Treating DSCR minimums as targets. The lender's 1.25x minimum is a floor, not a goalpost. Sizing to exactly 1.25x leaves zero cushion for NOI compression, vacancy spikes, or rate moves on floating-rate paper. Many sponsors size above the minimum (illustratively 1.30-1.35x on amortizing structures), accepting lower proceeds for coverage cushion. The proceeds-maximizing strategy of sizing exactly to the minimum is a sponsor-friendly underwriting choice that lender's covenants and bond-rating frameworks discourage.

• Modeling DSCR in isolation from the other three constraints. The most common error. Sponsors who only model DSCR systematically overestimate proceeds on premium-priced CMBS deals (where debt yield is tighter), on construction deals (where LTC is tighter), and on gateway-market trophy assets (where LTV is tighter). The four-constraint model is the only one that gives an accurate proceeds figure for any given lender.

• Quoting "DSCR" without specifying lender type and property type. "A 1.30x DSCR" means very different things at a CMBS conduit (possibly loose, with debt yield the likelier binding test), a life company (probably the headline, with a stressed test on top), a regional bank (probably with a covenant attached), and an agency deal (a tier-based minimum tested on net cash flow). Always qualify the DSCR you're quoting with the lender category and the rate basis.

Related Articles

DSCR is one of four sizing constraints in the institutional debt-analysis cluster. Sibling deep-dives:

• Debt Yield: Why Institutional Lenders Prefer It: the comparator. DSCR is gameable; debt yield isn't. The post-GFC CMBS pivot explained.
• LTV vs LTC: When Each Governs: the collateral and cost-basis constraints. Where they bind and where DSCR overrides.
• Interest Rate Risk: Caps, Swaps, and Hedging: the hedging response to stressed-rate sizing. Cap strikes and the stressed-DSCR test.
• Prepayment Analysis: Defeasance and Yield Maintenance: IO loans typically have different prepayment economics; the connection to the DSCR sizing decision.
• Refinancing Risk: Maturity Wave and Default Rate Sensitivity: the 2026 context. The deals where DSCR sizing fails are the maturity-wave deals.
• Cap Rate Calculator and Formula: compressed cap rates loosen LTV so that income-based tests such as DSCR and debt yield govern. The cross-cluster connection.

FAQ

Sources

Reference materials and authority sources behind this article:

• Fannie Mae: Multifamily Guide, Part II, Chapter 2, Section 203 (Underwritten NCF and Underwritten DSCR). mfguide.fanniemae.com.
• Federal Reserve: 2026 Stress Test Scenarios. federalreserve.gov/publications/2026-stress-test-scenarios.htm.
• Federal Reserve: H.15 Selected Interest Rates. federalreserve.gov/releases/h15.
• U.S. Department of the Treasury: Daily Treasury Par Yield Curve Rates. home.treasury.gov.
• Federal Reserve Bank of New York: Secured Overnight Financing Rate (SOFR). markets.newyorkfed.org.