Wim Roach & Brian Lorenz — HUD/FHA Practitioner Series

How HUD Sizes a 223(f) Multifamily Mortgage

A Former Senior HUD Underwriter Explains the Methodology

When borrowers ask us how large a HUD 223(f) loan can get, the honest answer is: it depends on which of four independent tests is binding. HUD doesn't size a mortgage the way a conventional lender does — it runs multiple constraints simultaneously, and the loan is the lowest result across all of them. This article walks through each test, explains the math behind it, and covers why the binding constraint shifts as interest rates move. There's also a sizing tool at the bottom where you can play around with the math.

B
Brian Lorenz
Vice President — Former Senior HUD Underwriter
W
Wim Roach
Vice President
Centennial Mortgage, Inc.
Published April 2026
Last updated May 2026

I spent years on the underwriting side of HUD before moving into originations, and I've built the Excel underwriting models for nearly every lender I've worked for — including one of the top 20 largest banks in the country — which means I've run these calculations several hundred times from the inside. One of the most persistent misconceptions I encounter is that borrowers — and sometimes brokers — treat HUD loan sizing as a single-variable problem. They'll say "the property is worth $20 million, so at 80% we should get $16 million." That may or may not be right. The 80% LTV figure is one ceiling. There are three others that can cut below it, and in a high-rate environment the one that actually binds often has nothing to do with value.

01
Program Parameters

The 223(f) at a Glance — and How It Compares

The HUD 223(f) program provides FHA mortgage insurance for the acquisition or refinance of existing multifamily properties — no construction involved. It's a 35-year fully amortizing loan at a fixed rate, non-recourse to the sponsors.

The headline parameters on the loan itself compare well to FNMA and Freddie. For market-rate acquisitions and refinances, HUD allows up to 87% LTV against a 1.15x minimum DSCR. For cash-out refinances the LTV ceiling drops to 80%. Fannie and Freddie are typically capped at 75% LTV for comparable deals — so at the LTV level alone, HUD is structurally more aggressive.

Parameter HUD 223(f) FNMA / Freddie
Max LTV — Standard87%75%
Max LTV — Cash-Out80%70–75%
Min DSCR1.15x1.25x
Loan Term35 years5–12 years
Amortization35 years30 years (up to 35)
RecourseNon-recourseNon-recourse (with carveouts)
Rate StructureFixedFixed or floating

While HUD's longer timeline earns its reputation, these parameters explain why sponsors who can absorb that timeline are often better served by the 223(f) than by agency alternatives. The 35-year amortization also has a direct effect on sizing — we'll cover that in the DSCR section.

02
A Critical Distinction

Market NOI vs. Debt Service NOI

Before walking through the four sizing tests, it's worth explaining a distinction that catches borrowers off guard: HUD uses two separate NOI figures, and they feed different parts of the analysis.

Market NOI is what the property stabilizes to at market occupancy with normalized operating expenses. This is the NOI the third-party appraiser develops alongside the appraiser's sales comp valuation, and it drives the LTV calculation. That is, Market NOI divided by the cap rate equals the appraised value, and the LTV tests apply to that value.

Debt Service NOI is generally a more conservative figure that HUD derives from analyzing the property's actual operating history and the appraiser's NOI to determine what the project's actual expected NOI will be. In addition, HUD requires that the vacancy be 7% for market-rate deals (can be lower if affordable), even if the actual vacancy is lower. It also applies the actual annual replacement reserve escrow amount as determined by the Property Capital Needs Assessment (as opposed to the $250/unit escrow the appraiser generally puts in). While the Market NOI feeds the LTV loan amounts, this Debt Service NOI feeds into the DSCR test directly.

The sizing tool later in this article has separate input fields for Market NOI and Debt Service NOI. If you're working from preliminary estimates and haven't separated these yet, use the same figure in both fields to start — but understand that HUD's Debt Service NOI will almost always be lower than your underwritten market NOI once third-party reports are in.

03
Constraint One

The LTV Test — Value-Driven Ceiling

The LTV test is the most intuitive of the four constraints. HUD commissions an independent appraisal, which establishes market value. The appraiser applies a market cap rate to the stabilized Market NOI to help derive value, and the 87% LTV ceiling is applied to that appraised value.

Illustrative Example — LTV Test

Market NOI: $1,200,000  |  Cap Rate: 5.50%

Value = $1,200,000 / 0.055 = $21,818,182
LTV Maximum (87%) = $21,818,182 × 0.87 = $18,981,818

The LTV test produces a maximum mortgage of approximately $18.98 million. We'll carry this figure into the DSCR example below.

Because Market NOI drives value, cap rate assumption has a significant effect on the LTV-constrained loan amount. A 25-basis-point difference in cap rate on a $1.2 million NOI deal moves value by roughly $1 million — and the LTV ceiling translates that directly into proceeds. This is one reason HUD's third-party appraisal matters so much and why we spend time discussing cap rate selection with borrowers before their appraisal engagement.

04
Constraint Two

The DSCR Test — Where Rate Sensitivity Lives

The DSCR test is where interest rate assumptions have their entire effect on sizing. The mechanics are a bit different from how most lenders describe them, and there's a quirk worth understanding.

The way HUD actually sizes the DSCR-constrained loan is to multiply the Debt Service NOI by 87%. That result is the maximum allowable annual debt service — covering P&I, MIP, and all. The industry shorthand is "1.15x DSCR," which is how most originators and borrowers describe the requirement. But 1 / 0.87 equals approximately 1.149, which rounds to 1.15 — so the 1.15x figure is a convenient approximation of the same underlying calculation.

To convert that debt service budget into a loan amount, you divide by the annual loan constant. The loan constant is the sum of three components: the interest rate, the annual MIP rate, and the amortization rate. The amortization rate is the annual principal repayment per dollar of loan balance at a given rate and term — it's what's left of the annual P&I payment after stripping out the interest component. At 5.00% on a 35-year term, the amortization rate is approximately 1.056%. Add the 0.25% MIP and the 5.00% interest rate and the full loan constant is approximately 6.306%. Divide the DS NOI × 0.87 figure by that constant and you have the DSCR-constrained loan amount.

Illustrative Example — DSCR Test

Debt Service NOI: $1,080,000  |  Rate: 5.00%  |  Amortization: 35 years  |  MIP: 0.25%/yr

Max Annual Debt Service (P&I + MIP) = $1,080,000 × 0.87 = $939,600
Amortization rate (at 5.00%, 35yr) = annual P&I per $1 − 5.00% = 1.056%
Loan Constant = 5.00% (interest) + 0.25% (MIP) + 1.056% (amortization) = 6.306%
DSCR Loan Amount = $939,600 / 0.06306 = $14,900,000

The DSCR test at 5.00% produces a maximum mortgage of approximately $14.9 million — well below the $18.98 million LTV ceiling from the prior example. LTV is not the binding constraint here.

At 3.00%, the amortization rate drops to roughly 0.762%, producing a loan constant of approximately 4.012%. The same $939,600 debt service budget supports a loan of roughly $19.3 million — which exceeds the LTV ceiling and makes LTV the binding constraint instead. The crossover point where DSCR and LTV produce the same loan is around 3.12% at these assumptions. Below that rate, further rate decreases stop moving proceeds.

The 35-year amortization term is part of why HUD's DSCR test can produce more proceeds than agency lenders at equivalent rates. A 30-year amortization on the same loan amount produces a higher monthly P&I, which consumes more of the debt service budget. The longer HUD amortization schedules a lower monthly constant, stretching a given NOI to support a larger balance. Combined with the 1.15x minimum versus the 1.25x Fannie/Freddie minimum, the cumulative effect on proceeds can be substantial — particularly on lower-NOI deals.

Rate environment and the binding constraint: In a low-rate environment, the DSCR test frequently produces a higher loan amount than the LTV ceiling, making LTV the binding constraint and making rate changes below a threshold mostly irrelevant to proceeds. In a higher-rate environment, DSCR bites first. If you're sizing a deal at 6.00% and re-running it at 5.00%, the proceeds difference often runs $1–2 million on a mid-market property — the entire difference lives in the DSCR test.

05
Constraint Three

The HUD Statutory Mortgage Limit — Per-Unit Dollar Caps

The third constraint is the one borrowers are least familiar with: HUD publishes statutory per-unit mortgage limits by bedroom count that establish an absolute dollar ceiling on the loan, independent of both value and income. These limits exist because HUD mortgage insurance is a federal program with congressional authorization tied to specific loan ceiling amounts.

The base limits are set by statute and adjusted upward via a High Cost Percentage (HCP) factor — currently 270% uniformly across all markets. HUD adds 100% of the appraised land value to the statutory ceiling and 100% of the cost of building space that is not attributable to residential use — referred to as "Cost Not Attributable." This includes commercial square footage, standalone community buildings, hallways, and other non-residential components.

Illustrative Example — Statutory Test (Elevator Building)

Using elevator per-unit base limits (approximate; confirm current figures with HUD) and the same property from the LTV example (appraised value $21,818,182):

1-BR: ~$88,832  |  2-BR: ~$108,925  |  3-BR: ~$136,424  |  4+BR: ~$154,259  |  0-BR: ~$80,170

A 100-unit elevator property with 60 one-bedroom and 40 two-bedroom units would calculate as:

(60 × $88,832 × 2.70) + (40 × $108,925 × 2.70) = $14,390,784 + $11,763,900 = $26,154,684

Adding Cost Not Attributable (10% of appraised value) and 100% of land value:

Cost Not Attributable = $21,818,182 × 10% = $2,181,818
Statutory Maximum = $26,154,684 + $2,181,818 + $2,000,000 (land) = $30,336,502

In practice, the statutory limit rarely constrains a 223(f). The deals where it bites are not large-unit properties in average markets — they are properties in very high-rent metros like Manhattan or San Francisco, where both the LTV and DSCR tests would otherwise produce a very large loan. In those markets the income and value support an outsized mortgage, and the statutory per-unit cap is what ultimately cuts the number. The reason this happens infrequently even in high-cost markets is that land values in those same metros tend to be substantial, and adding 100% of land value to the statutory ceiling pushes the limit high enough that it doesn't bind. It is worth running the test, but it should not drive early-stage deal decisions in most markets.

06
Constraint Four

The Cost-of-Refinance Criterion — How HUD Handles Cash-Out

The fourth criterion works differently from the first three, and it's frequently misunderstood. It isn't simply a floor that the loan has to clear — it's the mechanism HUD uses to govern whether a refinance produces cash-out, and if it does, it caps the LTV at 80%.

Here's how it works: this criterion sizes the loan at the greater of 80% LTV or the total cost of the refinance transaction. The cost of refinance includes the existing debt payoff, any prepayment penalty, required repairs, replacement reserve escrow deposits, third-party report costs, lender fees, HUD fees, and legal costs.

If the cost of refinancing is less than 80% LTV — meaning the property has enough equity that even at 80% the proceeds exceed what it costs to close — the loan produces cash-out. The borrower walks away from closing with funds in excess of the transaction costs. This is the normal refinance outcome on a well-valued, low-debt property.

If the cost of refinancing exceeds 80% LTV, HUD will size the loan to exactly cover those costs — but only up to the 87% LTV ceiling from Criterion One. A refinance where the gross loan lands between the cost of refinance and 87% LTV produces no cash-out but also doesn't require the borrower to bring cash to closing. A refinance where even 87% LTV isn't enough to cover the cost of refinancing is called a cash-in refinance — the borrower has to contribute funds at closing to make the deal work, which is uncommon but possible on heavily leveraged assets with significant repair needs.

The practical interaction between Criterion One (87% LTV) and this criterion is that HUD has a built-in 80% LTV cash-out cap. If a deal is producing cash-out, the loan cannot exceed 80% of appraised value — the moment the cost of refinance is fully covered and there's money left over, the applicable LTV ceiling drops from 87% to 80%. Borrowers sometimes ask why HUD caps cash-out at a lower LTV than a standard refinance — this is why. The 87% ceiling is reserved for deals where the proceeds are consumed by the transaction itself.

The sizing tool below includes fields for existing debt payoff, prepayment penalty, and estimated repair escrow so you can model where this criterion lands relative to the upper constraints.

07
Putting It Together

Which Test Bites — and When

The actual loan amount is the lowest result across all four tests. In practice, the pattern runs roughly like this:

  • High-rate environments (roughly 5.50%+): The DSCR test typically produces the lowest number. The same NOI budget supports a smaller loan at higher rates, so DSCR bites before LTV. Improving the DSCR test means increasing Debt Service NOI — rent increases, expense control, lease-up — rather than re-appraising.
  • Low-rate environments (roughly below 3.50%): DSCR often produces more proceeds than LTV, making the LTV ceiling the binding constraint. In that environment, the loan is effectively a multiple of appraised value, and rate changes below a threshold barely move the needle on sizing.
  • Very high-rent markets: The statutory limit can cut the loan when income and value would otherwise support a very large mortgage — Manhattan, San Francisco, and similar metros. In most markets it is not a factor.
  • Deals with cash-out potential: When the cost of refinancing comes in below 80% LTV, the loan produces cash-out — and HUD caps that at 80% LTV rather than the standard 87%. Understanding where the cost-of-refinance criterion lands early tells you whether you're looking at a cash-out refinance or a transaction-cost-covering refinance, and which LTV ceiling actually applies.

There's no universal answer for which test will bind on your deal — it depends on the interplay of value, NOI, rate, unit mix, and the cost of getting the transaction done. The tool below is designed to give you a quick read on all four before you engage third-party reports.

Interactive Tool

Run the Numbers on Your Deal

Enter your assumptions below. By the time you've read through the sections above, you understand what each input is doing and how to play around with the numbers. Have fun — none of the results are a commitment or underwriting from us. Call us if you want a real quote.

$
Stabilized NOI used for appraisal / LTV test
$
Conservative NOI for DSCR test — typically lower than Market NOI (see Section 02)
%
Used with Market NOI to determine appraised value
Affects LTV and DSCR multipliers (87% market / 90% affordable)

%
Contact us for current rate guidance → This is your assumption, not a quote.
$
Outstanding principal balance to be retired at closing
$
Estimated prepayment premium on existing debt
$
Cost of repairs identified by PCNA. Included in cost-of-refinance calculation (see Section 06). Enter 0 if unknown.

Third-party report costs, legal fees, and HUD/lender fees are estimated and included in the cost-of-refinance calculation automatically.


Elevator buildings have higher per-unit statutory limits
$
100% of land value is added to the statutory ceiling

HCP factor of 270% applied uniformly per current HUD guidelines. Cost Not Attributable estimated at 10% of appraised value.



Hypothetical Loan Amounts — Based on Your Numbers

Ready to run your deal against real numbers?
Talk to the Team →

We Can Size Your Deal the Right Way

The numbers above give you a reasonable sense of which constraint is likely to bite and what range to plan around. It doesn't replace us doing a proper sizing — where we look at real financials, rent rolls, market data, etc. to get a real, thorough understanding of all these inputs.

If you have a deal you'd like to run through a more detailed preliminary analysis — with current rate guidance, market-specific HCP factors, and an honest read on deal structure — we're happy to spend 30 minutes on it.

Try the Calculator