Construction Cost Evidence: Bids, Published Indexes and the Contingency a Lender Will Accept

A construction budget is the one number in a feasibility study that every other test borrows. The borrower's equity, the size of each lien and the interest carried through the build are all computed from it, so an error in cost travels through the whole credit file before anyone sees it. This article sets out how a study should evidence that number: which kinds of cost evidence carry weight and in what order, how a published index can date and escalate an older figure without pretending to price a new scope, where the line between hard and soft cost falls in each program, and how large a contingency should be and what becomes of it when the work is finished. The site's archived post on construction cost benchmarks for SBA-financed projects gathers per-square-foot ranges by property type. The question here is a different one: what sort of proof a lender can rely on once a range has been replaced by a figure for one project. The 504 cost rules are read in full in the 504 method article and only summarised below.

Why the Cost Budget Decides More Than Its Own Line

Under 504, the borrower's contribution is a percentage of Project cost, excluding administrative costs ^[1], and the Debenture is capped against the same base. The regulation's general limit for the 504 loan reads: "may not exceed 40 percent of total Project cost plus 100 percent of eligible administrative costs" ^[2]. Understate the cost and every share shrinks with it, so the borrower reaches closing with a contribution sized for a cheaper building. Overstate it and the file carries debt for work that will never be bought. Either way the mistake is invisible in the ratios, because the ratios are computed on the wrong base. The sources and uses article follows the same base into the equity test.

Bank supervision reaches the budget from another side. The interagency real estate lending guidelines published with 12 CFR 365 say that a bank's policy for development and construction projects should spell out what the bank requires of feasibility studies and of sensitivity and risk analyses, "commensurate with the size and type of the project or property" ^[3]. The same list asks for standards on whether interest reserves are acceptable and how far they may go ^[4], and the guidelines expect loan administration procedures that cover documentation, disbursement and collateral inspection ^[5]. Disbursement is where a budget stops being a page in a report and becomes a schedule of checks, so the lines a study prints should be the lines the draws will follow. The conventional bank method article reads those guidelines at length.

The appraisal answers a different question. When a loan finances construction or renovation, the 2010 interagency appraisal guidelines say an institution would generally ask the appraiser for the property's as is value and, as applicable, its prospective value upon completion and upon stabilization ^[6]. Those are opinions of what the finished property will be worth, not of what it will cost to build. When the same guidelines describe the criteria a bank's portfolio monitoring should use to decide when a new appraisal or evaluation is needed, overruns and delays in construction costs appear among the changes in market conditions those criteria should address ^[7]. A feasibility study does not value the property. Its task on cost is narrower and earlier: to show that the budget the lender is about to fund is complete, current and priced by a party who will be held to the price. The article on studies and appraisals sets out how the two documents divide the work.

The site's own methodology page states the standard this series applies. It says that every material assumption must trace to a source a reviewer can verify, and it names a construction bid among those sources ^[8]. In its worked thread, the site's methodology page states that the construction budget would rest on current local bids rather than on figures from an earlier year or another market ^[9]. The rest of this article is about what that standard means in practice when a bid is not yet in hand.

The Hierarchy of Cost Evidence, From a Signed Contract to the Sponsor's Estimate

Cost evidence comes in grades, and a study should say which grade stands behind each line. From strongest to weakest, the grades are a signed construction contract, a bid that has not yet been signed, an estimate by a professional estimator or engineer, a published unit cost or index applied to the project, and the sponsor's own number. The grades differ in one respect above all: whether anyone outside the borrower has agreed to deliver the work at the stated price.

A signed contract. The USDA rule shows what the top grade looks like. A B&I lender can request that the Agency issue its loan note guarantee while the building is still under construction, unless the project relies on non-proven technology ^[10]. To do so it must show, among other things, that the borrower has signed a firm, fixed-price construction contract with an independent general contractor, with costs set out in detail and terms for change order approvals, the agreed retainage percentage and the disbursement schedule ^[11]. The lender must also show that it has checked the contractor's financial capacity and record, or that it has covered the risk of an unfinished project, for instance by requiring a 100 percent performance and payment bond on the contractor ^[12]. Where 7(a) money pays for construction, the regulation makes the borrower supply, unless SBA grants a waiver, "a 100 percent payment and performance bond and builder's risk insurance" ^[13]. A bonded fixed-price contract is strong evidence because it moves pricing risk from the borrower to a party that has priced the drawings and has a surety standing behind its number.

A bid not yet signed. A competitive or negotiated bid on complete drawings is nearly as good, provided the study reads it rather than just quoting its total. A study should record the bid date and how long the price is held, list the exclusions and allowances, check the scope against the plan set, and say who carries escalation between the bid date and the start of work. An allowance is a placeholder inside a bid, not a price, and a study should treat a large allowance as estimate grade evidence even though it sits in a contractor's letter.

A professional estimate. Before drawings are complete, the strongest evidence on offer is an estimate by a cost estimator, an architect or a civil engineer. Its weight depends on the stage of design it was prepared from and on its date. An estimate is priced at a moment, and a study that uses one should state that moment and bring the figure forward with an index, as the fourth section of this article shows. The B&I rule puts an independent professional's review of the budget on the record in one case: when the guarantee is sought before completion, the project timetable and the matching budget, with the parties responsible for paying each cost, are to be confirmed as adequate by a qualified independent consultant ^[14], for example the project architect or engineer, with experience in the project's industry ^[15]. In meeting the B&I monitoring rules the lender may rely on written reports from an independent engineer and other qualified consultants ^[16].

A published unit cost or index. Per-square-foot data and cost indexes describe averages across many projects. They are good for testing whether a bid is in a plausible range and for dating an older figure, and they are poor evidence of what one scope on one site will cost. A study should use them as a check and an escalator, never as the price.

The sponsor's number. A budget the borrower wrote is a hypothesis. It may be right, but until a third party prices the work a study should label each such line an assumption and say what evidence would replace it.

The program texts point the same way even where they do not rank evidence. Under 504, Debenture proceeds may be held in escrow to finish a minor Project component, such as landscaping or a parking lot, if that item is a minor portion of the Project and has been contracted for completion at a specified price and a specific future date ^[17]. Before the Debenture is issued, the CDC must certify that the Project was "completed in accordance with the final plans and specifications" ^[18], escrowed minor items aside, which means the finished work is checked against the same drawings the bid priced. Table 1 gathers the provisions that bear on cost evidence in each program.

Hard Costs, Soft Costs and the Lines Each Program Will Pay For

Before a budget can be evidenced it has to be divided, and the division matters twice: once for the evidence each line needs, and again for whether the program will finance it. Hard costs are the work installed on the site. Soft costs are the professional services, fees, financing charges and reserves that surround it. A public definition helps keep the boundary honest. In its construction spending statistics, the Census Bureau measures the value of construction installed or erected at the site during a period ^[21]. For a single project that value includes the materials installed, the labor of contractors and of the owner's own crews with a share of equipment rental, and the contractor's profit ^[22]. It also includes architectural and engineering work, overhead charged to the project on the owner's books, and interest and taxes paid during construction, with an exception for state and locally owned projects ^[23].

The Census definition is as useful for what it leaves out. Production machinery and special purpose equipment that fits a building to one use, such as steam tables in a restaurant or display cases and shelving in a store, are outside it ^[24], and so is land acquisition ^[25]. A study need not adopt the Census boundary, but it should draw some boundary and keep to it, because a contingency, a fee percentage or an escalation factor applied to the wrong base changes the total without anyone deciding it should.

The programs then decide which of those lines they will pay for. Under 504, eligible Project costs include costs directly attributable to the Project, including the borrower's spending on land used in the Project before it applied to SBA ^[26], and professional fees that are directly attributable and essential, such as title work, appraisals, environmental studies, design fees and the legal costs of zoning, permits or platting ^[27]. The payoff of interim financing counts too, interest, points and fees included ^[28], and so, where the Project involves construction, does a contingency reserve for overruns of up to 10 percent of construction cost ^[29]. The list of ineligible costs names, among other things, fees on the permanent Third-Party Loan, working capital, advertising, franchise fees, organization costs and counseling or management fees ^[30]. Those excluded items still have to be paid, and a study should show where the money comes from.

B&I draws its own boundary. Guaranteed loan funds may pay professional service fees and charges if the Agency approves the amounts as reasonable and customary in the area, and may pay fees for construction permits and licenses ^[31]. They may also pay interest, including interest on interim financing, until the first principal payment falls due or the facility starts producing income, whichever comes first ^[32]. That last condition matters for a budget: an interest reserve sized for the whole lease-up can exceed what the rule will fund if the property starts earning before the reserve runs out. The B&I method article covers the rest of the eligible uses.

Under 7(a), the share of a loan that buys or improves real property may have a term of 25 years plus whatever period the construction or improvements need ^[33], so the build does not eat into the amortization. That changes the debt service a study models, not the cost, but it is one more reason to state the construction schedule in the cost section.

Some rules raise cost rather than define it, and a study should name the ones that apply. When SBA loan proceeds build a new building or an addition, the work must conform with the NEHRP recommended seismic provisions for new buildings or a code SBA identifies as substantially equivalent ^[34]. Under B&I, the lender must ensure that facilities built with guaranteed loan funds comply with applicable domestic procurement preference requirements, including section 70914 of the Build America, Buy America Act ^[35]; the word applicable carries weight there, and the study should say whether and how the preference reaches the project's materials. Prevailing wage rules are the opposite case. USAspending's data dictionary records whether a federal contract transaction is subject to the Construction Wage Rate Requirements, clause 52.222-6, which it ties to the wage rate requirements formerly called the Davis-Bacon Act ^[36]. The 2026 editions of 13 CFR 120 and 7 CFR 5001 read for this series do not name the Act. A study therefore has no program rule to cite for a prevailing wage premium in a 7(a), 504 or B&I budget and should not add one on the program's account; it should ask instead whether any other public money in the capital stack brings a wage rule of its own.

Dating and Escalating a Cost With a Published Index

Every cost figure has a price date, and most budgets mix several. A contractor's bid is priced this month; a civil engineer's site estimate may come from a design round two years ago; an equipment quote may be valid for ninety days. Before the lines can be added, each has to be brought to a common date, and then the unpriced part of the work has to be carried forward to the time it will actually be bought. The tool for both steps is a published price index, and the first job is to choose one that measures price rather than activity.

Construction spending is not a price index. The Census construction spending series is the wrong tool for escalation. It measures the value of work put in place ^[21], summed across all projects underway in a period, whenever they started and whenever the contractors were paid ^[37], and for some categories the estimates represent payments made rather than work done ^[38]. A value measure moves with the amount of work as well as with its price, so a rise in spending is not by itself evidence that costs rose. It is also revised: Census publishes an annual analysis of how far the monthly seasonally adjusted estimates move between the preliminary release and the second revision ^[39]. A study that cites spending to show a market is active should record the release it read. It should not use the growth rate to inflate a budget.

The Census price indexes cover housing. Census does publish construction price indexes, but for single-family houses sold and single-family houses under construction, and the houses sold index includes the value of the land ^[40]. The current series adds multifamily units under construction ^[41]. For an apartment project these are relevant, provided the index chosen excludes land. For a car wash, a hotel or a warehouse they measure a different product. Building permit data is a related source with its own limit: in the Building Permits Survey, valuation is the estimated value of the residential structure as shown on the permit, or the permit official's estimate where the permit shows none ^[42]. The Nebraska multifamily worked example uses permit valuations as a cross-check on an assumed building cost, which is the right role for them: a declared estimate filed with a permit is a reasonableness test, not a price.

The highway index prices site work. The Federal Highway Administration publishes the National Highway Construction Cost Index, which it describes as a price index for tracking highway construction costs and for converting current dollar highway spending to constant dollars ^[43]. It is quarterly and is built from state postings of winning bids on highway contracts ^[44]. In those bids each item's price includes the material, the cost of moving, placing and installing it, and profit and overhead ^[45], and FHWA describes the result as an output cost index rather than an input price index ^[46]. The index draws on 29 major components ^[47], among them grading and excavation, asphalt, base stone, drainage pipe and inlets and concrete ^[48], and water, gas and sewer utility work and concrete pavement ^[49]. Those are the items in a commercial site package: earthwork, paving, storm drainage and utility connections. Its basket is highway work, and although one component covers buildings and miscellaneous structures let on highway contracts ^[49], it does not track what a commercial building's shell or mechanical systems cost; a study that applies it to those lines is using it outside its basket.

Producer price indexes measure the seller's side. FHWA's explanation, written about the producer price index for streets and highways, draws the distinction a study needs: a producer price index measures the selling prices received by producers, the seller's view, while the highway index measures from the buyer's side ^[50], and the highway index includes the contractor's markup over the producer's price ^[51]. An input index for steel or lumber is therefore a good way to explain why a bid moved and a poor way to escalate a whole budget, because it leaves out labor, installation and margin. Series also end: FHWA notes that the Bureau of Labor Statistics had discontinued that producer price index as of July 2010 ^[52]. A study should confirm that the index it relies on is still being published before it builds an escalation on it.

The method. Escalating with an index takes five steps, and a study should print each one. First, record the price date of the figure being escalated. Second, read the index level for that quarter and for the latest quarter published, from the same series and on the same basis; if the series is not seasonally adjusted, compare the same quarter of different years. Third, take the change as a ratio of the two levels: FHWA warns that index levels cannot be compared directly across indexes with different methods or bases ^[53], and says the accurate way to compare them is to convert each index to a percent change ^[54]. Fourth, carry the figure from the latest published quarter to the midpoint of construction at a stated annual rate, labelled as an assumption, because no index publishes the future. Fifth, write down the series, the quarters and the date the series was read, so a reviewer can rebuild the factor.

Table 2 lists every input for an illustrative owner-occupied building, and Table 3 applies the method. The NHCCI starts at 1 in 2003 Q1 ^[55]; the model multiplies the published levels for 2023 Q1 ^[56], 2025 Q1 ^[57] and 2026 Q1 ^[58] by 100 and rounds them, which leaves the ratios unchanged to the precision printed.

Rounded as shown; lines may not foot by one unit in the last place.

The site work estimate of $620,000 was priced in 2023 Q1. Between that quarter and 2026 Q1 the index rose 10.28 percent, which brings the estimate to $683,732 at 2026 Q1 prices. Carried forward at an assumed 3.0 percent a year for 1.5 years to the middle of the build, it becomes $714,729. Had the study simply carried the older estimate unchanged, that one line would have sat $94,729 below the escalated figure, before any overrun at all.

The second index row is the caution. The index level in 2026 Q1 stood 0.89 percent below its 2025 Q1 level, although some of the quarters in between rose, so a study that set its forward rate from the most recent year would have escalated the site work downward. The 3.0 percent rate in Table 2 is therefore labelled a judgment, not a reading of the index. A forward rate should be argued, with the recent path of the index, the bid climate for the trades involved and the length of the build all stated, and it belongs in the sensitivity tests described in the stress testing article. The building line, by contrast, is not escalated at all: it is a fixed-price bid on complete drawings with its price held through signing, so the contractor carries the escalation. The Ohio glamping worked example shows the same discipline applied to two older third-party budgets, escalated by a national index and labelled as assumptions where they are applied to a different project.

Contingency: How Large, Measured Against What, and What Happens to the Unspent Part

A contingency pays for things inside the agreed scope that no one could price exactly when the budget was set: ground conditions found during excavation, coordination problems between drawings, quantities that run over, a trade that rebids higher. It is not a fund for changes the owner decides to make later, and it is not a disguised reserve for operating losses. A study should state what its contingency is meant to cover, what base it is measured against, and why it is the size it is.

Size follows the evidence. The right contingency shrinks as the cost evidence improves. A conceptual estimate made before design calls for a larger allowance than a bid on construction drawings, and a bonded fixed-price contract has already transferred much of the pricing risk to the contractor, so the remaining contingency is there mainly for site conditions and owner-side items. A study should therefore set the contingency line by line or by package: smaller on the contracted building, larger on the site work that is still at estimate grade, and separate for any renovation of an existing structure, where hidden conditions are the main risk. A single percentage at the program ceiling, with no explanation, tells a reviewer that nobody asked the question.

The 504 ceiling and the base. In a 504 Project involving construction, the contingency reserve for overruns is an eligible cost only up to 10 percent of construction cost ^[29]. The passage does not define construction cost further, so a study should say which lines it used. The example measures the contingency against site work and building only, leaving out land, design fees, permits and interest; applying the percentage to a base that includes land would inflate the reserve with money that buys no construction.

What the 504 rule does with the unspent part. The less obvious 504 rule concerns the end of the job, and it is short: "the Debenture amount will be reduced by the amount that the unused contingency reserve exceeds 2 percent of the anticipated Debenture" once the Project is complete ^[59]. An oversized contingency therefore does not become cash for the borrower or a cushion for the operating years. Beyond a small threshold it simply shrinks the Debenture, and because the Debenture was sized as a share of Project cost, the financing that was arranged around it changes shape at closing. Table 4 shows the arithmetic at three sizes.

At the 7.5 percent base the contingency is $232,105, and the anticipated Debenture at the 40 percent general ceiling is $1,811,386 ^[2]. The rule lets $36,228 of unused contingency stand, 2 percent of that Debenture ^[59]. If none of the contingency is spent, the Debenture falls by $195,877; if half is spent, by $79,825. At the 10 percent ceiling the reserve is $309,473 and the cut on a job that finishes on budget would be $272,626. The example treats the Debenture as 40 percent of Project cost and leaves out administrative costs, which the regulation adds on top of that share ^[2]; a real 504 structure should be run with the actual split and fees, as the 504 method article does. The point survives the simplification: a contingency set higher than the evidence supports is not free, because the excess comes back out of the Debenture at completion.

What USDA asks of a contingency. B&I sets no percentage in the provisions read here, but it asks a harder question when the lender wants the guarantee before construction is finished. The lender must show that there is enough cash flow to complete construction, including contingencies for cost overruns, plus working capital during start-up ^[15], and must show to the Agency's satisfaction that contingency funding is in place to meet unforeseen overruns without asking for more guaranteed assistance ^[60]. That is a test of the funding behind the contingency, not of its size. A study supports it by showing which source pays the contingency line and that the source is committed.

What the bank texts say. The interagency lending and appraisal guidelines quoted in this article do not use the word contingency at all. They ask for feasibility studies and sensitivity analyses on construction projects ^[3] and for interest reserve standards ^[4], and they treat overruns as a market condition to monitor ^[7]. A conventional lender's contingency is therefore a matter of its own policy, and the study's task is to show the contingency's basis clearly enough that the lender can apply that policy to it.

When the Budget Is Wrong: Overruns, Draws and Who Pays

Conventions. This article follows the conventions of the Loan Analytics worked examples. Coverage is tested against the program floor the site's program page states, where the program has one, and against a stated 1.25x lender assumption, and the higher of the two binds. Equity is measured at the loan the cash flow supports, with fees recomputed on that loan. An acquisition is tested on the seller's historical earnings, not on projections.

A contingency absorbs an overrun only up to its own size, and the loan does not grow to cover the rest. Under the conventions of this series, the loan is the one the project's cash flow supports, so cost above the budget lands on the borrower's equity unless another committed source is named. Table 5 runs three overruns through the example's 7.5 percent contingency.

A 5.0 percent overrun on construction cost, $154,736, fits inside the reserve. A 10.0 percent overrun, $309,473, exhausts the contingency and leaves $77,368 for the borrower to fund, 1.7 percent of Project cost. At 15.0 percent the uncovered amount is $232,105, 5.1 percent of Project cost. Those figures are the reason a study should state, in the cost section, which party funds an overrun and from what: a cash reserve held back at closing, a guarantor with liquid assets, or a contractor bound by a fixed-price contract. Where the answer is that nobody has committed, the study should say so, because the sensitivity of coverage to cost overruns is then really a test of the borrower's liquidity.

USDA's rule lays out the sequence plainly. Under 5001.205(g)(1) the borrower's own money is spent first and guaranteed loan money afterwards ^[61], and the early guarantee provision adds that in all cases the borrower's equity is injected ahead of any guaranteed loan funds ^[11]. The lender or its agent monitors construction and makes the reviews and inspections needed to keep the work in line with codes, plans, specifications and contract documents ^[62], and whenever the guarantee is issued the lender must report any problem in project development to the Agency within 15 calendar days of finding it ^[63]. When the guarantee is sought before completion, the lender and borrower must also provide a construction schedule, and the projected cash flow needs should mirror the quarterly construction costs as the work proceeds ^[64]. The rule backs this with a sanction on the lender: failure to provide adequate construction planning and monitoring so that a guaranteed project is completed with the funds available is among the listed causes for which the Agency can revoke a lender's approved status ^[65].

The 504 rules put the weight on the interim lender, because the Debenture is issued only once the CDC has certified completion ^[18]. Any source of interim financing must have the experience and qualifications to monitor all Project construction and progress payments, and if it lacks them SBA may require the interim loan to be managed by a bank or a professional construction manager ^[19]. Two certifications come first: the interim lender states how much it has disbursed, and the CDC states that the Project matches its final plans and specifications ^[18]. The interim loan's interest, points and fees are eligible Project costs ^[28], so the interest line in the budget should come from a draw schedule rather than from a round number.

A study supports that work in three ways. It should present the budget in the same packages the contract and the draw requests will use, so a monitoring report can compare spent against budget line by line. It should build interest during construction from a monthly draw curve, which also gives a bank the basis to apply its interest reserve standards ^[4]. And it should run the operating projection with the overrun funded the way the study says it will be funded, so the coverage result is not quietly computed on a budget that has already been exceeded. The Maryland car wash worked example shows a cost basis split between the 504 Project and the items paid in cash outside it, with its contingency stated against the ceiling.

What the SBA and USDA Loan Records Can Say About Construction Risk

The public loan files cannot test a construction budget, and it is worth being exact about why. SBA's 504 data dictionary defines ThirdPartyDollars as the third party loan amount and GrossApproval as the SBA/CDC loan amount ^[66]; among its fields there is none for total Project cost, the construction budget, the contingency or an overrun. The file shows how much was lent, not what the project cost or whether it came in on budget. The article on the limits of the SBA file covers the other gaps.

What the record does show is scale and timing. In FY2025, 504 approvals numbered 6,762 ^[67], a gross figure that still includes approvals cancelled afterwards, and they carried $7.80 billion ^[68] of debentures, which measures the SBA/CDC loans alone and not the Projects behind them. Median 504 debentures over FY2017 to FY2026 were $2,048,000 ^[69] for hotels and motels, $1,094,000 ^[70] for car washes, $890,000 ^[71] for self storage and $860,000 ^[72] for gas stations and truck stops. Those medians describe the size of the loans, not of the budgets or of how much of each Project was construction, and the asset class data pages give the full figures by class.

Timing is the second reading. A Debenture is issued only after the CDC certifies that the Project was built to its final plans and specifications, escrowed minor items aside ^[18], and at September 30, 2026, two to three years on, 17.30 percent ^[73] of the approvals made in FY2024 showed no disbursement. The record does not say which of those approvals finance construction, or why any of them has not funded, and a study should not read a cause into the share. The 504 cancellation rate for the FY2017 to FY2024 approval cohort is 13.60 percent ^[74], an approval-cohort floor that can rise but cannot fall while approvals remain undisbursed; the cancellation article explains why a cancellation is not a loss and what the rate can and cannot tell a reader.

USDA's public record is thinner still. The dataset's USDA figures are USAspending award aggregates: 256 ^[75] core B&I new awards in FY2025, with a gross face value of new awards of $2,382.6 million ^[75], and nothing about the construction budgets behind them; the USDA public data article explains what those aggregates hold. Neither record can tell a lender whether a contingency was large enough. That evidence lives in the contract, the draw reports and the completion certificates, and the cost section of a study is where it should first be assembled.

Loan Data Behind This Article

Every figure in this table comes from the Loan Analytics loan dataset, analytics-loan-data freeze-3 (commit 360c856), and carries its key. SBA rates are approval-cohort floors and are withheld when fewer than 30 loans stand behind them. USDA figures are USAspending award aggregates; a USDA rate in the dataset is a credit subsidy rate, not a loss rate.

A Checklist for the Cost Section of a Study

The questions below let a lender or borrower test the cost section of a study; every one rests on a rule or method set out earlier in this article.

Is every line graded? Each budget line should say whether it rests on a signed contract, an unsigned bid, a professional estimate, a published benchmark or the sponsor's figure, and each assumption should say what evidence would replace it. The site's methodology page states that material assumptions must trace to a verifiable source, a construction bid among them ^[8].

Is every price dated? A figure without a price date cannot be escalated or checked. Where an older estimate is used, the study should show the index, the two quarters, the factor and the date the series was read, and it should confirm that the series is still published ^[52].

Is the index the right one? Construction spending measures activity, not price ^[21]. The Census price indexes describe housing, and one of them includes land ^[40]. The highway index prices horizontal site work from the buyer's side, including the contractor's markup ^[51], and should not be stretched to the building.

Is the forward rate argued? Escalation beyond the latest published quarter is an assumption, and the example's index level was 0.89 percent lower in 2026 Q1 than a year earlier, after rising 10.28 percent over three years. The study should say why it chose its rate and test the budget at other rates.

Is the contingency sized to the evidence and measured on a stated base? Under 504 the reserve is eligible only up to 10 percent of construction cost ^[29], and the unspent part above 2 percent of the anticipated Debenture comes back out of the Debenture at completion ^[59]. Under B&I, when the guarantee is issued before completion, the contingency has to be funded, not just budgeted ^[60].

Are the program's own cost rules in the budget? Seismic provisions under 13 CFR 120.174 ^[34], applicable domestic procurement preferences under 7 CFR 5001.205(a)(4) ^[35], bonding and builder's risk insurance on a 7(a) construction loan unless SBA waives them ^[13], and the eligible and ineligible lines under 13 CFR 120.882 and 120.884 ^{[29] [30]} should each appear where they apply, and a prevailing wage premium should appear only where some funding source actually imposes one.

Who pays an overrun? The study should name the source, show that it is committed, and run the projection with the overrun funded that way. Under B&I, borrower funds go in before guaranteed funds ^[61], and a failure of construction planning and monitoring is a listed cause for revoking the lender's approved status ^[65].

Does the budget match the draws? The lines in the cost section should be the packages the contract, the draw requests and the monitoring reports will use, and interest during construction should come from the draw curve, so that the bank's interest reserve standards ^[4] and the lender's inspections can be checked against the same numbers. Operating costs after opening are a separate evidence problem, taken up in the operating expense article.

Related Pages on Loan Analytics

• The Loan Analytics methodology series (all articles)
• How a feasibility study is done: the Loan Analytics methodology
• SBA feasibility studies (7(a) and 504)
• USDA feasibility studies
• Bankable feasibility studies for conventional loans
• Loan data by state
• Loan data by asset class
• Worked examples (case studies)
• Worked Example: SBA 504 Express Car Wash Feasibility Study in Maryland
• Worked Example: Conventional Bank Construction-to-Permanent Multifamily Feasibility Study in Nebraska
• Worked Example: USDA Business and Industry Loan Glamping Resort Feasibility Study in Ohio
• Construction cost benchmarks for SBA-financed projects: per-square-foot data across 9 property types and 4 U.S. Census regions
• U.S. 75+ Population by State: The Senior Housing Demand Screen, Rebuilt for Underwriting
• Operating Expense Evidence: Building an Expense Budget a Credit Officer Can Trace
• Equity Injection and Sources and Uses: How a Study Shows Where Every Dollar Comes From

Commission a Feasibility Study

Loan Analytics prepares lender-grade feasibility studies for SBA 7(a), SBA 504, USDA Business and Industry, conventional and EB-5 financing, built on the methods this series sets out. A study is not an appraisal: it tests whether a project's market, costs and cash flow can carry the debt a lender is asked to approve. To discuss a study for your project, contact Loan Analytics.

Daniel Smith, MAI, Loan Analytics