Construction-to-permanent loan model: a $5.8 million takeout gap

An analyst at an institutional real estate investor asks Compound whether a permanent loan will repay the construction loan on a warehouse development. Compound builds the model from a project in a REIT's 10-Q, Maryland CMBS cap rates, the SOFR forward curve and the Treasury yield of November 13, 2026.

Compound Research · Updated Nov 15, 2026 · 13 min read

Short answer

A construction-to-permanent loan model tests whether the permanent loan sized at stabilization repays the construction loan. In Compound's model of a Maryland warehouse project disclosed in a REIT's 10-Q, the permanent loan is $30.3 million against a $35.8 million payoff, and the sponsor adds $5.8 million of equity at the refinance.

Try it now
On this page (9)
  1. 1. What goes into a construction-to-permanent loan model
  2. 2. The as-stabilized cap rate from Maryland CMBS industrial loans
  3. 3. Construction loan draw schedule: equity first, an S-curve and an interest reserve
  4. 4. How to size the permanent loan: LTV and DSCR tests side by side
  5. 5. Refinance sources and uses: the equity paydown
  6. 6. What moves the shortfall: rate, NOI, cap rate and timing
  7. 7. What public data cannot tell you about a construction loan takeout
  8. 8. Our view: size the construction loan from the takeout backward
  9. 9. Frequently asked questions

A permanent loan of $30,318,852 does not repay a construction loan that has grown to $35,799,601. That is the base case of this model for a 180,300 square foot warehouse project in Maryland. At the refinance on January 18, 2028 the sponsor adds $5,783,937 of new equity, and every one of the 50 cells in the model’s two sensitivity tables shows a shortfall (our count).

You work in acquisitions or asset management at an institutional real estate investor, and a development is on your desk. The construction lender sizes its loan on cost. The permanent lender sizes the takeout on value and on income. A construction-to-permanent loan model puts both loans on one timeline and answers one question: on the day of the refinance, does the second loan pay off the first? We asked Compound to build that model on a real project from public data. It ran on November 15, 2026 and returned one Excel workbook and a summary.

The project’s cost, size, yield and stabilization date come from a public filing. Both loans are assumptions: the request set the structure, and the workbook takes no loan from the filing.

Two facts frame every figure below. Compound’s summary reports that $53.7M of the $57.6M had already been spent, so a new construction loan from November 17, 2026 is a hypothetical structure on a project that is mostly paid for. The filing’s total also includes capitalized interest, so the model’s $1,750,000 interest reserve and $358,249 loan fee count some financing cost a second time, as Compound’s summary notes.

The shortfall is the output of those assumptions. It is neither a forecast for this property nor advice to buy, sell or lend on it.

What the workbook shows:

  • The debt service coverage test sets the permanent loan. At a 6.90% rate and a 1.25x minimum, stabilized net operating income of $2,995,200 supports $30,318,852. The 65% loan-to-value test would allow $32,260,893.
  • The construction loan is larger than either test allows. A 60% loan-to-cost commitment of $35,824,950 is drawn to $35,799,601 by month 14, with $1,724,651 of interest capitalized.
  • The as-stabilized value is 83.1% of total cost. Net operating income over a 6.03% cap rate, the median of nine Maryland industrial properties in CMBS filings, gives $49,632,142 against a cost of $59,708,249 with financing.
  • The permanent rate is 170 basis points above the project’s yield on cost. The 10-Q discloses a 5.2% expected stabilized yield. The 10-year Treasury alone closed at 5.15% on November 13, 2026.
  • The interest reserve lasts with $25,349 to spare. A $1,750,000 reserve covers interest through month 14 and runs out in month 15.

Each input on the Assumptions sheet carries a citation, the word “Request”, or the words “Analyst assumption” with a reason. The project comes from a 10-Q on SEC EDGAR. The cap rate comes from CMBS loan-level filings. The construction loan floats over Compound’s SOFR forward curve, and the permanent rate starts from the Daily Treasury Par Yield Curve.

Prompt to try

“I need to know whether the permanent loan will pay off the construction loan on a warehouse development, and how much cash the sponsor must add if it does not, so I want a construction-to-permanent loan model anchored on public facts. From the latest 10-K or 10-Q of a public industrial REIT, pick one development project, or one market's group of projects, that the REIT discloses with its square feet, total expected investment, percent leased, expected stabilized yield and stabilization date. Use those figures as the project: total cost, building size, and stabilized NOI as cost times the disclosed yield. Set the as-stabilized value from that NOI over a cap rate taken from the appraised values and NOI of stabilized industrial properties in the same state in the CMBS loan-level filings. Set the construction loan at a stated loan-to-cost, floating over the SOFR forward curve plus a spread, with an S-curve monthly draw, equity funded first, and an interest reserve. Set the permanent loan rate from the 10-year point of today's Treasury par yield curve plus a spread, with a 30-year amortization, a maximum loan-to-value of 65% and a minimum DSCR of 1.25x. Spreads, loan-to-cost and the draw curve are analyst assumptions with a note on each. Build three sections: construction, with the running balance, interest capitalized each month and the payoff at a stabilization month I can change; permanent sizing, with the loan-to-value test and the DSCR test side by side and the binding one picked by formula; and the transition, as sources and uses at the refinance. Report the construction payoff, permanent proceeds and which test set them, the equity paydown or cash back, the blended cost of capital across both loans, stabilized debt service, and cash-on-cash before and after the transition, with any paydown counted as new equity. Add a sensitivity table of the shortfall to the permanent loan rate and stabilized NOI, and a second one to the stabilization month and the cap rate. Flag a payoff shortfall, an interest reserve that runs out before stabilization, and a permanent loan rate above the project's yield on cost. Use the stored public data wherever it has the figure, and web search only for what it lacks. Put every input on an Assumptions sheet with one column that says where it came from: a cited source, or "analyst assumption" for a figure I would set myself, with the reason for the value you chose. Every output must be a live formula that traces back to the Assumptions sheet. Add a Sources sheet listing each data source, what it gave and its date. Deliver one Excel workbook and a short summary of the result, the flags, and what I would need from the seller to firm it up. Cite every sourced figure.”

Try this prompt free →

That is the request as we sent it, without its closing line, which asks that no name or email be put in the workbook.

What goes into a construction-to-permanent loan model

Five project figures come from one table in a 10-Q, and four inputs come from the request. The loan terms that decide the answer are analyst assumptions, and the sheet labels each one.

Compound picked Craftsman Circle in Hyattsville, Maryland, a row in the Development and Redevelopment Activity table of an industrial REIT’s 10-Q for the quarter ended September 30, 2026, filed November 4, 2026. The row gives a total expected investment of $57,600,000, an estimated post-development size of 180,300 square feet, an estimated stabilized cap rate of 5.2%, a stabilization quarter of Q4 2027 and 0.0% leased.

Craftsman_Circle_Construction_to_Perm_Model.xlsx
C27fx=COUNTIF(E45:E80,"<"&C11)+1
BCDE
5InputValueUnitSource / basis
7Total expected investment (total project cost)57,600,000$Terreno Realty Corp 10-Q, quarter ended 9/30/2026 (filed 11/4/2026), Development and Redevelopment Activity table, Craftsman Circle row. Includes purchase price, closing costs, redevelopment spend, capitalized interest and leasing costs.
8Building size (estimated post-development)180,300SFTerreno Realty Corp 10-Q, quarter ended 9/30/2026 (filed 11/4/2026), Development and Redevelopment Activity table, Craftsman Circle row
9Expected stabilized yield (estimated stabilized cap rate)5.2%%Terreno Realty Corp 10-Q, quarter ended 9/30/2026 (filed 11/4/2026), Development and Redevelopment Activity table, Craftsman Circle row. Terreno defines it as annualized cash NOI at market occupancy (about 95%) divided by total cost.
10Percent pre-leased at 9/30/2026-%Terreno Realty Corp 10-Q, quarter ended 9/30/2026 (filed 11/4/2026), Development and Redevelopment Activity table, Craftsman Circle row (shown as —%).
11Disclosed stabilization quarter (Q4 2027), quarter-end date12-31-27dateTerreno Realty Corp 10-Q, quarter ended 9/30/2026 (filed 11/4/2026), Development and Redevelopment Activity table, Craftsman Circle row: estimated stabilization quarter Q4 2027; quarter-end date used.
14As-stabilized cap rate (median of comps below)6.03%%Formula: median of the 9 Maryland industrial/warehouse comps in the CMBS block below (appraised value and underwritten NOI at securitization, SEC Form ABS-EE).
19Land and site cost funded at closing (% of total cost)25%%Analyst assumption: 25% for infill Washington DC-area industrial land and closing costs; the seller's budget should replace it.
20Construction period (months)10monthsAnalyst assumption: 10 months, typical for a 180,000 SF tilt-up warehouse shell; leaves 4 months of lease-up before the disclosed stabilization quarter.
21Construction loan-to-cost60%%Analyst assumption: 60%, the middle of the 55% to 65% range lenders offer on speculative industrial with no pre-leasing.
22Construction spread over 1M SOFR3.00%%Analyst assumption: 300 bps, in line with bank pricing on unleased industrial construction loans.
23Construction origination fee (% of commitment)1.00%%Analyst assumption: 1.00%, the market norm for construction loans; paid at closing out of equity.
24Interest reserve (funded inside the loan)1,750,000$Analyst assumption: sized to cover projected interest on the forward curve through the base stabilization month (about $1.73M), rounded up to the nearest $250K.
26Stabilization and payoff month (model month)14monthRequest: changeable. Base value 14 = first model month ending on or after the disclosed Q4 2027 stabilization (row 27).
3010-year Treasury par yield (close 11/13/2026)5.15%%U.S. Department of the Treasury daily par yield curve, 10-year, as of 11/13/2026 (latest close before 11/15/2026), Compound stored rates data.
31Permanent spread over 10-year Treasury1.75%%Analyst assumption: 175 bps, typical life-company or CMBS spread for a newly leased suburban Maryland warehouse at 65% LTV or less.
32Amortization (years)30yearsRequest
33Maximum loan-to-value65%%Request
34Minimum debt service coverage ratio1.25xxRequest
35Permanent loan closing costs (% of loan)1.00%%Analyst assumption: 1.00% for fees, legal and third-party reports at the refinance.
SummaryAssumptionsConstructionPermanentTransitionSensitivitySources
The Assumptions sheet, columns B to E, as Compound wrote them. We show 19 of its input rows and leave out the sensitivity steps, the 36-month forward curve and the comps block. The formula bar shows the cell that turns the disclosed stabilization quarter into a model month.

The yield needs a careful read. The filing defines its estimated stabilized cap rate as annualized cash net operating income at market occupancy, generally 95%, divided by total cost. That is a yield on cost. The request told Compound to use it that way: stabilized net operating income is $57,600,000 times 5.2%, or $2,995,200, which is $16.61 per square foot.

Four inputs carry the label “Request”: the 30-year amortization, the 65% maximum loan-to-value, the 1.25x minimum coverage, and the stabilization month, which you can change. The base value of 14 is a formula result. Cell C27 counts the forward curve periods that end before the disclosed quarter end of December 31, 2027 and adds one.

The analyst assumptions are the ones a lender’s term sheet would replace. The construction loan is 60% of cost at 300 basis points over 1-month SOFR with a 1.00% origination fee. Land and site cost is 25% of total cost, funded at closing, and construction takes 10 months. The permanent loan prices at 175 basis points over the 10-year Treasury with 1.00% of closing costs. Each has a one-line reason in column E. The loan-to-cost note, for example, says 60% is the middle of the range lenders offer on speculative industrial with no pre-leasing.

The as-stabilized cap rate from Maryland CMBS industrial loans

The value at stabilization is net operating income over a cap rate, and the request asked for that cap rate from CMBS filings in the same state. Compound found nine comps, took the median of 6.03%, and the value came out below cost.

Every CMBS deal files loan-level data with the SEC on Form ABS-EE, with each property’s appraised value and underwritten net operating income. Compound kept Maryland industrial and warehouse properties with an MAI appraisal dated 2023 or later and occupancy of 90% or more at securitization.

Craftsman_Circle_Construction_to_Perm_Model.xlsx
C14fx=MEDIAN(C85:C93)
BCDEFGHI
84PropertyCap rateCityAppraisal dateAppraised valueUnderwritten NOIOccupancyCMBS deal
8511900 Trolley Lane5.54%Beltsville04-07-2525,300,0001,400,518100.0%Benchmark 2025-V16 Mortgage Trust
863300 75th Avenue6.35%Landover05-02-2367,800,0004,305,412100.0%Benchmark 2023-B39 Mortgage Trust
8740-52 Derwood Circle6.66%Rockville08-29-2515,800,0001,052,500100.0%BANK 2025-BNK51
88Brandywine Regency Warehouse & Distribution Center5.78%Brandywine11-15-25138,300,0007,996,872100.0%BBCMS Mortgage Trust 2025-C39
89Clinton6.42%Clinton08-13-247,900,000507,120100.0%BANK 2024-BNK48
90Dunlop Protective Footwear - Havre De Grace, MD6.03%Havre De Grace03-01-2411,300,000681,932100.0%BMO 2024-5C4 Mortgage Trust
91Fedex Ground - Cumberland5.07%Cumberland02-28-2346,300,0002,346,515100.0%BANK5 2023-5YR1
92McGaw Commerce Center5.51%Columbia02-14-2450,100,0002,760,44994.6%BMO 2024-5C4 Mortgage Trust
93Middle River Distribution Center6.16%Middle River01-12-2622,400,0001,378,733100.0%Benchmark 2026-B43 Mortgage Trust
SummaryAssumptionsConstructionPermanentTransitionSensitivitySources
The CMBS comps block of the Assumptions sheet, columns B to I, as Compound wrote them. Property names are as the filings give them. Column J of the sheet, left out here, holds the SEC accession number of each ABS-EE filing.

The nine cap rates run from 5.07% for Fedex Ground - Cumberland to 6.66% for 40-52 Derwood Circle in Rockville. The median is the Dunlop Protective Footwear property in Havre De Grace at 6.03%. Eight of the nine were 100.0% occupied at securitization, and McGaw Commerce Center in Columbia was 94.6% occupied.

Cap rates of nine Maryland industrial properties in CMBS filings

Percent. Underwritten net operating income over appraised value at securitization, lowest to highest. The median comp is highlighted

Fedex Ground - Cumberland (Cumberland)5.07McGaw Commerce Center (Columbia)5.5111900 Trolley Lane (Beltsville)5.54Brandywine Regency Warehouse andDistribution Center (Brandywine)5.78Dunlop Protective Footwear - HavreDe Grace, MD (Havre De Grace)6.03Middle River Distribution Center (Middle River)6.163300 75th Avenue (Landover)6.35Clinton (Clinton)6.4240-52 Derwood Circle (Rockville)6.66Value-weighted cap rate of the nine comps5.82

Compound workbook, Assumptions sheet, CMBS comps block, from SEC Form ABS-EE asset-level data. Appraisals are dated from February 2023 to January 2026.

Compound also computed a value-weighted rate of 5.82% and marked it “reference only”. It is lower than the median because the largest comp, Brandywine Regency Warehouse & Distribution Center, carries a $138,300,000 appraisal at 5.78%. We agree with using the median. One large property should not set the cap rate for a 180,300 square foot building.

At 6.03%, $2,995,200 of net operating income is worth $49,632,142, or $275.28 per square foot. Total project cost with the interest reserve and the loan fee is $59,708,249. The value is 83.1% of that cost, about $10.1 million less (our arithmetic). The reason is a comparison of two inputs: the 5.2% yield on cost from the filing is 83 basis points below the 6.03% median cap rate from the comps (our arithmetic). The comps are appraisals of other, leased properties across Maryland. The result describes how the model’s value input sits against its cost input, and it is no estimate of what this building will sell for.

Construction loan draw schedule: equity first, an S-curve and an interest reserve

The construction loan is sized on cost, funded after the equity, and charged interest that it lends to itself. By the payoff month it has grown to $35,799,601, of which $1,724,651 is capitalized interest.

The commitment is a single formula, =C8*(C6+C7)/(1-C8*C9). The origination fee is 1.00% of the commitment and is itself a project cost, so the commitment depends on a total that depends on the commitment. The formula solves that loop in one cell: 60% of the $57,600,000 investment plus the $1,750,000 interest reserve, grossed up for the fee, is $35,824,950. The fee is $358,249, total cost is $59,708,249, and required sponsor equity is $23,883,300, or 40.0%.

Craftsman_Circle_Construction_to_Perm_Model.xlsx
G43fx=MIN(G41,$C$13-SUM($C43:F43))
BDEFGHIJKLMNOPQRS
29Model month0123456789101112131415
31Period end11/17/2612/17/261/19/272/17/273/17/274/19/275/17/276/17/277/19/278/17/279/17/2710/18/2711/17/2712/17/271/18/282/17/28
33PhaseClosingConstructionConstructionConstructionConstructionConstructionConstructionConstructionConstructionConstructionConstructionLease-upLease-upLease-upLease-upAfter payoff
36S-curve cumulative % of construction complete-2.8%10.4%21.6%35.2%50.0%64.8%78.4%89.6%97.2%100.0%100.0%100.0%100.0%100.0%100.0%
41Total uses before interest ($)14,758,2491,209,6003,283,2004,838,4005,875,2006,393,6006,393,6005,875,2004,838,4003,283,2001,209,600-----
43Sponsor equity14,758,2491,209,6003,283,2004,632,250------------
44Construction loan draw for costs---206,1505,875,2006,393,6006,393,6005,875,2004,838,4003,283,2001,209,600-----
50All-in rate-6.97%7.19%7.31%7.37%7.51%7.57%7.63%7.69%7.70%7.70%7.58%7.53%7.47%7.47%7.66%
51Interest accrued (actual/360)----1,18141,88073,746124,677170,713186,011221,934227,886220,524219,999236,100228,374
52Interest funded from reserve (capitalized)----1,18141,88073,746124,677170,713186,011221,934227,886220,524219,999236,10025,349
54Closing balance ($)---206,1506,082,53112,518,01118,985,35724,985,23429,994,34733,463,55834,895,09235,122,97835,343,50235,563,50135,799,60135,824,950
59Reserve closing1,750,0001,750,0001,750,0001,750,0001,748,8191,706,9381,633,1921,508,5161,337,8031,151,792929,858701,972481,448261,44925,349-
SummaryAssumptionsConstructionPermanentTransitionSensitivitySources
The monthly schedule on the Construction sheet, model months 0 to 15 of 36, as Compound wrote it. We show 12 of its rows. The formula bar shows the equity-first rule in month 3: equity funds the month's costs until the required equity is used up, and the loan funds the rest.

Equity goes in first. At closing on November 17, 2026 the sponsor funds $14,758,249: the land and site cost of $14,400,000 and the loan fee. It funds all of months 1 and 2 and most of month 3. The loan’s first draw is $206,150 in month 3.

The rest of the cost follows an S-curve over 10 months. It runs at 2.8% of construction spend in month 1, 14.8% in each of months 5 and 6, and 2.8% in month 10. The two peak months draw $6,393,600 each. The draw curve is an analyst assumption that a contractor’s schedule of values would replace.

Interest accrues on the opening balance each month at the 1-month forward rate plus 300 basis points, on an actual/360 basis. The forward curve starts at 3.97% for the first period and reaches 4.70% in month 10, so the all-in rate rises from 6.97% to 7.70% during construction. The reserve pays the interest, which means the loan funds it and the balance grows by it.

Construction loan balance at each month end

Closing balance, months 0 to 14. Costs are drawn through month 10. In the four lease-up months the balance grows only by capitalized interest

Month 0, closing$0.00MMonth 1, construction$0.00MMonth 2, construction$0.00MMonth 3, construction$0.21MMonth 4, construction$6.08MMonth 5, construction$12.5MMonth 6, construction$19.0MMonth 7, construction$25.0MMonth 8, construction$30.0MMonth 9, construction$33.5MMonth 10, construction$34.9MMonth 11, lease-up$35.1MMonth 12, lease-up$35.3MMonth 13, lease-up$35.6MMonth 14, lease-up$35.8M

Compound workbook, Construction sheet, closing balance row. Interest at 1-month forward SOFR as of November 13, 2026 plus an assumed 300 basis points.

The balance is $34,895,092 when construction ends in month 10. Four months of lease-up add only interest, and the payoff on January 18, 2028 is $35,799,601. The balance-weighted all-in rate through payoff is 7.58%. The schedule counts no lease-up income before stabilization, a choice stated at the top of the sheet.

The reserve deserves a second look. Compound sized it at $1,750,000 against projected interest of $1,724,651, so $25,349 is left at payoff and month 15 is the first month it cannot cover. The flag for a reserve that runs out before stabilization reads “OK”, with one month of room. The reserve was sized to the projected interest, so the pass follows from the sizing.

How to size the permanent loan: LTV and DSCR tests side by side

The permanent loan is the smaller of two numbers: a share of value and the loan whose payments the income covers 1.25 times. Here the coverage test gives $30,318,852 and the value test gives $32,260,893, so coverage sets the loan.

Craftsman_Circle_Construction_to_Perm_Model.xlsx
D27fx=D26/D25/$C$21
BCD
24TestLoan-to-valueDebt service coverage
25Constraint (max LTV / min DSCR)65%1.25x
26Sized off (value / NOI)49,632,1422,995,200
27Maximum loan ($)32,260,89330,318,852
28Annual debt service at maximum loan2,549,6432,396,160
29Implied loan-to-value65.0%61.1%
30Implied DSCR1.17x1.25x
31ResultNot bindingBinding
33PERMANENT LOAN RESULT
34Permanent loan proceeds ($)30,318,852
35Test that sets proceedsDSCR
36Stabilized annual debt service ($)2,396,160
37Monthly debt service199,680
38DSCR at proceeds1.25x
39Loan-to-value at proceeds61.1%
40Debt yield9.88%
41Yield on cost (disclosed)5.20%
42Permanent rate less yield on cost (bps)170
SummaryAssumptionsConstructionPermanentTransitionSensitivitySources
The sizing tests and the result on the Permanent sheet, as Compound wrote them. The formula bar shows the DSCR test: net operating income, divided by the minimum coverage, divided by the annual debt constant.

The rate is two cells. The 10-year Treasury par yield closed at 5.15% on November 13, 2026, a sourced figure. The 175 basis point spread is an analyst assumption, described on the sheet as typical of a life company or CMBS loan on a newly leased suburban Maryland warehouse. Together they make 6.90%. With 30-year amortization, a year of principal and interest is 7.90% of the loan amount. That figure is the annual debt constant.

The loan-to-value test is 65% of $49,632,142, or $32,260,893. Payments on that loan would be $2,549,643 a year, and net operating income would cover them 1.17x, below the 1.25x minimum. The coverage test works backward from income: $2,995,200 divided by 1.25 and by the debt constant is $30,318,852. That loan is 61.1% of value, inside the 65% limit. The model picks the smaller with =MIN(C27,D27) and names the binding test in the next cell, so the label changes when the inputs do.

Debt service on the permanent loan is $2,396,160 a year, or $199,680 a month. The debt yield is 9.88%.

The project's yield against the cost of each loan

Percent. The yield on cost is from the 10-Q and the Treasury yield is the November 13, 2026 close. Loan spreads are analyst assumptions

10-year Treasury par yield, November 13, 20265.15Yield on cost disclosed in the 10-Q5.2As-stabilized cap rate,median of nine CMBS comps6.03Permanent loan rate6.9Construction loan all-in rate, balance-weighted7.58Annual debt constant, 30-year amortization7.9

Compound workbook, Permanent sheet, cells C7, C11, C17, C19 and C21, and Construction sheet, cell C24.

The chart explains why coverage binds. The project earns 5.2% on cost. The permanent loan costs 6.90% in interest and 7.90% with amortization. When each dollar of debt costs more per year than each dollar of cost earns, a 1.25x coverage rule holds the loan to a small share of cost. The permanent proceeds are 50.8% of total project cost (our arithmetic), and the construction loan was 60%. Compound’s third flag reports the same thing in one line: the permanent rate is 170 basis points above the yield on cost.

Refinance sources and uses: the equity paydown

The transition is one small table. Uses are the construction payoff and the new loan’s closing costs, sources are the permanent proceeds, and the difference is cash from the sponsor.

Craftsman_Circle_Construction_to_Perm_Model.xlsx
C15fx=MAX(0,C8+C9-C14)
BCD
5SOURCES AND USES AT REFINANCE
6Refinance date01-18-28
8Construction loan payoff35,799,601
9Permanent loan closing costs303,189
10Cash returned to sponsor-
11Total uses ($)36,102,789
14Permanent loan proceeds30,318,852
15Sponsor equity paydown (new equity)5,783,937
16Total sources ($)36,102,789
20EQUITY AND CASH-ON-CASH
21Initial sponsor equity23,883,300
22Construction interest paid by sponsor-
23Equity invested before transition ($)23,883,300
24Plus: equity paydown at refinance5,783,937
25Less: cash returned at refinance-
26Equity invested after transition ($)29,667,237
27Stabilized NOI2,995,200
29Annualized construction interest at stabilization2,718,320
30Cash flow before transition ($)276,880
31Cash-on-cash before transition1.2%
32Stabilized permanent debt service2,396,160
33Cash flow after transition ($)599,040
34Cash-on-cash after transition2.0%
36BLENDED COST OF CAPITAL ACROSS BOTH LOANS
37Construction loan balance-weighted all-in rate7.58%
38Construction loan amount (payoff)35,799,601
39Permanent loan rate6.90%
40Permanent loan amount30,318,852
41Blended cost of debt (amount-weighted)7.27%
43FLAGSValueStatus
44Payoff shortfall (sponsor paydown, $)5,783,937FLAG
45Interest reserve runs out before stabilization (first uncovered month)15OK
46Permanent rate above yield on cost (bps)170FLAG
47Construction not finished by stabilization month (months of slack)4OK
SummaryAssumptionsConstructionPermanentTransitionSensitivitySources
The Transition sheet, columns B to D, as Compound wrote them: sources and uses at the refinance, equity and cash-on-cash, the blended cost of debt and the four flags. We leave out the two check rows and three label rows.

Uses total $36,102,789: the $35,799,601 payoff and $303,189 of closing costs at 1.00% of the new loan. The permanent loan supplies $30,318,852. The sponsor supplies the other $5,783,937, and no cash comes back. The paydown is 24.2% of the equity the sponsor had already invested (our arithmetic), and it raises that equity from $23,883,300 to $29,667,237.

What the refinance needs and what each sizing test provides

Dollars at the January 18, 2028 refinance. The paydown is the payoff plus closing costs, less the permanent proceeds

Construction loan payoff$35.8MPermanent loan closing costs$0.30MLoan-to-value test: maximum loan$32.3MDSCR test: maximum loan,the permanent proceeds$30.3MSponsor equity paydown$5.78M

Compound workbook, Transition sheet, cells C8, C9 and C15, and Permanent sheet, cells C27 and D27.

The request asked for cash-on-cash before and after, with the paydown counted as new equity. Before the transition, the model annualizes construction interest at the payoff-month rate of 7.47% on the payoff balance, which is $2,718,320. Net operating income less that interest leaves $276,880, or 1.2% on $23,883,300. After the transition, net operating income less $2,396,160 of debt service leaves $599,040, or 2.0% on $29,667,237.

The blended cost of debt across both loans is 7.27%. Compound weights the construction loan’s 7.58% by its payoff balance and the permanent loan’s 6.90% by its proceeds. The loans run one after the other, so 7.27% is an amount-weighted average of two rates and the project never pays it in any one month.

What moves the shortfall: rate, NOI, cap rate and timing

Net operating income and the permanent rate move the shortfall by millions. The cap rate changes it only at 6.53%, the highest column in the grid, and a delay costs the sponsor interest out of pocket instead of a larger payoff.

Craftsman_Circle_Construction_to_Perm_Model.xlsx
D8fx=Construction!$C$19-MIN(Assumptions!$C$33*D$7/Assumptions!$C$14,D$7/Assumptions!$C$34/(-PMT($C8/12,Assumptions!$C$32*12,1)*12))*(1-Assumptions!$C$35)
BCDEFGHI
5SHORTFALL ($) VS PERMANENT LOAN RATE AND STABILIZED NOI
6Stabilized NOI ($)
7Permanent loan rate2,695,6802,845,4402,995,2003,144,9603,294,720
85.90%7,055,1455,458,2313,861,3172,264,403667,489
96.40%7,356,2435,776,0574,195,8702,615,6841,035,497
106.90%8,785,5037,284,7205,783,9374,283,1542,782,371
117.40%10,103,4618,675,8987,248,3345,820,7714,393,208
127.90%11,320,5759,960,6298,600,6837,240,7385,880,792
15SHORTFALL ($) VS STABILIZATION MONTH AND CAP RATE
16As-stabilized cap rateMemo
17Stabilization month5.53%5.78%6.03%6.28%6.53%Sponsor-paid interest
18104,879,4294,879,4294,879,4294,879,4295,400,517-
19125,327,8385,327,8385,327,8385,327,8385,848,926-
20145,783,9375,783,9375,783,9375,783,9376,305,025-
21165,809,2865,809,2865,809,2865,809,2866,330,374426,247
22185,809,2865,809,2865,809,2865,809,2866,330,374895,810
SummaryAssumptionsConstructionPermanentTransitionSensitivitySources
Both tables on the Sensitivity sheet, as Compound wrote them. Each cell is a formula that recomputes the payoff, less the smaller sizing test net of closing costs. A positive figure is a paydown by the sponsor. The memo column is interest the sponsor pays after the reserve runs out.

In the first table the smallest shortfall is $667,489, at a 5.90% permanent rate with net operating income 10% above the base. The largest is $11,320,575, at 7.90% with net operating income 10% below. At the base rate, the shortfall is $7,284,720 with income 5% lower and $4,283,154 with income 5% higher.

A lower rate helps less than it seems. At 5.90% and base income the shortfall is still $3,861,317, because the loan-to-value test takes over as the limit once coverage loosens.

The second table has a pattern worth noticing. Across cap rates from 5.53% to 6.28% the base-month shortfall is the same $5,783,937. The coverage test binds in all four columns, and coverage does not depend on value. At 6.53%, the highest column in the grid, the value test binds and the shortfall rises to $6,305,025. For this project a better appraisal would not produce a larger loan.

Timing works in two stages. Stabilizing at month 10 lowers the shortfall to $4,879,429, since less interest has been capitalized. Stabilizing at month 16 or month 18 gives $5,809,286 in both rows: once the reserve is spent, the balance stops at the $35,824,950 commitment. The cost of delay moves to the memo column. The sponsor pays $426,247 of interest directly through month 16 and $895,810 through month 18.

We ran one more number from the workbook’s cells. For the coverage test alone to produce enough proceeds to cover the payoff and closing costs at 6.90%, stabilized net operating income would need to be about $3.57 million, roughly 19% above the base and a 6.2% yield on the filed cost (our arithmetic).

What public data cannot tell you about a construction loan takeout

Public filings give the project’s total cost, size and expected yield. The developer’s budget, its contracts and the lender’s term sheet hold the inputs that decide the takeout, and each one replaces a labeled assumption in this workbook.

  • The line-item budget and spend to date. The 10-Q gives one total, defined to include the purchase price, closing costs, redevelopment spending, capitalized interest and leasing costs. A budget that separates those lines would remove the overlap with the model’s reserve and fee. The developer’s draw history would replace the 25% land share and the 10-month build.
  • The construction contract and schedule. A guaranteed maximum price contract, the permits and the expected certificate of occupancy date set the draw curve and the construction period.
  • The leasing pipeline. Stabilized net operating income here is cost times a disclosed yield. The filing shows the building 0.0% leased at September 30, 2026. Letters of intent, rent per square foot, free rent, tenant improvements and commissions would turn that yield into a rent roll, and the sensitivity table shows what 5% of income is worth.
  • The construction lender’s term sheet. Loan-to-cost, spread, fee, the index the loan agreement names, any rate floor, the reserve’s size, extension options and recourse are all assumptions here. The forward curve in the workbook is market-implied compounded SOFR, and the Sources sheet says it is not Term SOFR.
  • The permanent lender’s quote. The model applies the Treasury yield of November 13, 2026 to a loan that closes in January 2028. A quote would give the spread, any interest-only period, the lender’s own coverage and loan-to-value limits, and whether the rate can be locked forward.
  • An appraisal. The as-stabilized value rests on the median of nine appraisal-based cap rates from across Maryland, dated 2023 to 2026. They are not sale prices, and none is the subject property.

Our view: size the construction loan from the takeout backward

On these inputs the gap is structural. A 60% loan-to-cost construction loan cannot be taken out by a permanent loan that coverage holds to about half of cost.

The usual order is to size the construction loan on cost and check the takeout later. This workbook is an argument for the reverse. Permanent proceeds net of closing costs are about $30.0 million, or 50.3% of total cost (our arithmetic). A construction loan near that size would be repaid in full, and it would also capitalize less interest than the one modeled. The sponsor then knows on day one that the equity requirement is closer to half the cost than to 40.0%, and the $5,783,937 is in the capital plan before the first draw.

The other side has weight. The permanent rate is today’s 10-year yield applied 14 months ahead, and a fall of 100 basis points would cut the shortfall to $3,861,317 on the first table. The spread assumptions may be wide for a fully leased building. And a yield on cost of 5.2% may be acceptable to an owner that does not plan to borrow against the building.

Three things would change our view. A signed lease at rents that lift net operating income toward $3.57 million would close the gap on the coverage test. A permanent quote with an interest-only period would loosen coverage, and the value test at $32,260,893 would then set the loan. A line-item budget would show how much of the $57,600,000 is already interest. Until then we would carry the paydown as a real use of cash.

This page isolates the takeout. A real estate development pro forma covers the whole project, from the choice of metro to yield on cost and returns. A bridge-to-perm multifamily model asks the same refinance question about an existing building in lease-up. An industrial warehouse acquisition model starts where this one ends, with a leased building and a rent roll, and a real estate pro forma sizes a single acquisition loan.

Construction-to-permanent loan model (Excel)Compound’s workbook from this run: a Summary with the key results and four flags; the Assumptions sheet with a source or a reason for every input, the 36-month SOFR forward curve and nine CMBS comps; the monthly Construction schedule; the Permanent sizing tests; the Transition sources and uses; two Sensitivity tables; and the Sources list.Download · Free

Frequently asked questions

What is a construction-to-permanent loan model?

It is a model that carries a development from the first construction draw to the refinance. It has three parts: the construction loan balance by month with capitalized interest, the permanent loan sized on loan-to-value and debt service coverage, and the sources and uses on the day one loan repays the other. Its main output is the cash the sponsor adds or receives at that refinance.

How is a permanent loan sized at takeout?

A lender computes the largest loan under each of its limits and lends the smallest. In this model the limits are 65% of as-stabilized value and 1.25x coverage of debt service by net operating income. Coverage gives $30,318,852 and value gives $32,260,893, so coverage sets the loan. The model picks the binding test by formula.

What happens if the permanent loan does not pay off the construction loan?

The sponsor pays the difference in cash at the refinance, often called a cash-in refinance or an equity paydown. In this model the difference is $5,783,937, which includes $303,189 of closing costs on the new loan. The alternatives are a loan extension, more leasing before the refinance, or a different lender, and the model’s stabilization month input shows what a delay does to the payoff and to interest paid out of pocket.

How do you size an interest reserve on a construction loan?

Project the loan balance month by month from the draw schedule, apply the forward rate plus the spread to each month’s opening balance, and add up the interest through the stabilization month. This model projects $1,724,651 and sets the reserve at $1,750,000. The schedule then reports the first month the reserve cannot cover, which is month 15 against a payoff in month 14.

Why does the cap rate not change the shortfall in this model?

The debt service coverage test sets the loan, and coverage depends on income, the rate and the amortization. Value enters only the loan-to-value test. In the second sensitivity table the shortfall is $5,783,937 at every cap rate from 5.53% to 6.28%, and it is higher at 6.53%, the highest column in the table, where the value test becomes the smaller of the two.

CoatueCentri ConsultingAvra783 Capital PartnersOctahedron CapitalArix ResearchCoatueCentri ConsultingAvra783 Capital PartnersOctahedron CapitalArix ResearchCoatueCentri ConsultingAvra783 Capital PartnersOctahedron CapitalArix Research

See this workflow on your own deal

Try it now