Operating Expense Evidence: Building an Expense Budget a Credit Officer Can Trace

A projection for a new business has two halves, and they rarely get equal care. Revenue gets the market chapters: the trade area, the competitors, the capture rate, the ramp. Expenses often get a page of round numbers. A careful reader of the file can start at the bottom of the operating statement instead, because a line of cost that cannot be explained is easier to challenge than a demand forecast. The question for every line is the same: where did this number come from, and could someone else rebuild it?

The sections below show how a study builds an expense budget that survives that question: the order of evidence, how each major line is sourced, why the split between fixed and variable cost decides what a shortfall does, why an industry ratio is a check rather than evidence, and why a line lifted from a brochure fails. One illustrative budget runs through the whole article: a new franchised limited-service hotel of 80 rooms in Windsor Locks, Connecticut, in which every quantity is a labelled assumption and every rate is read from a public source.

Two archived posts sit close to this subject. Case Study: Operating Expenses in Orlando's Full-Service Hotels and Multi-Family and Apartments Feasibility Study: Operating Expense Analysis in Miami, Florida describe what expenses run to in two markets. This article asks a different question: how a budget for a business with no operating history earns a lender's belief, one line at a time.

What the Program Texts Ask of the Expense Side

None of the program texts prescribes an expense budget, but each sets a standard that an unsupported budget cannot meet. SBA's lending criteria begin with soundness: the applicant must be creditworthy, and each loan must be, in the rule's phrase, "so sound as to reasonably assure repayment" ^[1]. That section also binds lenders and CDCs to prudent, generally acceptable commercial credit analysis of the kind they apply to their similarly sized commercial loans that carry no SBA guarantee ^[2]. What SBA asks to see in most business loan applications includes, among other things, three years of past financial statements or tax returns beside the current ones, with a business plan where one applies ^[3]. A business that has not opened has no history to submit. Its projected operating statement stands where the historical statements would stand, and the expense half of that statement carries the weight the tax returns would otherwise carry.

The USDA text is more explicit. Under 7 CFR 5001.303, the projections in a guarantee application "must be supported by a list of assumptions showing the basis for the projections" ^[4]. Where a projection departs from the borrower's record, the credit evaluation rule demands substantiation and documentation, and it says the projections should be consistent with what the borrower has done before ^[5]. Growth in revenue or in margins should, in the rule's words, be "reasonable and substantiated in the analysis" ^[6]. For a start-up the consistency test has no record to measure against, so the whole weight falls on substantiation. The rule's definition of financial feasibility asks, in so many words, for "sufficient income, credit, and cash flow to financially sustain the project over the long term", with enough left to cover every debt the project carries ^[7]; that is the test the budget serves.

Bank supervision points the same way for real estate loans. The interagency guidelines that the FDIC publishes as an appendix to 12 CFR 365 say an institution should monitor real estate market conditions in its lending area ^[8], and among the supply and demand factors to consider they list current and projected operating expenses for different types of projects ^[9]. Where the loan finances development, construction or a completed commercial property, the guidelines go further: an institution's lending policy should, they say, call for feasibility studies and for testing how income responds to change, scaled to the size and type of the project or property, and their example of that testing reads "sensitivity of income projections to changes in economic variables such as interest rates, vacancy rates, or operating expenses" ^[10]. Every expense line is therefore both an input to the base case and one of the variables a lender is told to stress.

The expense budget ends in the numerator of the coverage ratio, and USDA's definition of that ratio reaches into the budget itself: it begins with EBITDA, takes away what the borrower can reasonably expect to spend replacing capital items, and divides the remainder by one year's principal and interest ^[11]. A study that omits a replacement reserve overstates coverage under that definition before any market assumption is tested.

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.

Turning the budget into a coverage ratio, and choosing between floors, are covered in Debt Service Coverage in a Feasibility Study: How the Ratio Is Built, and Which Floor Binds. The site's SBA program page states a 1.15x floor for standard 7(a) loans, which a start-up must reach within two years of funding ^[12], and the same multiple for 504 loans, whether measured on the borrower's history or on a projection reaching two years ahead ^[13]. The worked budget below is a stabilized year, tested against the 1.25x lender assumption, which sits above both floors.

An Order of Evidence: From a Signed Contract Down to a Sponsor's Estimate

Not all support is equal, and a study should say which kind stands behind each line. The order below is the method this series uses; it ranks evidence by how little judgment stands between the source and the number.

At the top are amounts fixed by law or by a contract the borrower will sign. A statutory tax rate is the clearest case: every employer owes social security tax of 6.2 percent of wages as section 3121(a) defines them ^[14] and hospital insurance tax of 1.45 percent ^[15], with no estimate involved. Minimum wage law, a franchise agreement's fee schedule, a management contract and a signed ground or equipment lease belong in the same tier. Such amounts can still be wrong in a budget, but only through arithmetic or a misread document, both checkable.

Second come published rates applied to a stated quantity. A town's mill rate times the property's assessment, or a utility price times the building's load, is only as good as the quantity, and the rate itself moves from one period to the next. The study's job is to name the rate, its period and its source, and to defend the quantity separately.

Third are quotations, such as an insurance proposal or a service contract bid: specific to the property, not yet binding, and liable to expire.

Fourth are benchmarks: industry ratios from government surveys, trade association studies, a franchisor's own disclosures. They show whether a budget is in a plausible range. They cannot show that it is right for this building.

At the bottom sits the number with no stated basis: the sponsor's estimate, a template's percentage, a line copied from a marketing brochure. It may be a reasonable placeholder while evidence is gathered, provided it is labelled as an assumption and replaced before the file goes to credit.

The Worked Budget: One Hotel, Every Line Labelled

The hotel is budgeted for its stabilized year. It is not a real project and none of its quantities are measured: the occupancy, rate, staffing hours, energy load, assessment and insurance allowance are assumptions chosen to show the method. Wherever a public rate exists, the illustration uses it, and Table 2 prints each input with its basis, so a reader can see which numbers are sourced and which are judgment.

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

Table 3 builds the operating statement from those inputs. Each line is a quantity times a rate, a share of a named revenue line, or a stated allowance. At 72 percent occupancy and an average daily rate of $150.00 the hotel sells 21,024 room nights and earns total revenue of $3,216,672. Operating expenses come to $2,346,185, an operating expense ratio of 72.9 percent, leaving EBITDA of $870,487 and, after a replacement reserve of $128,667, net operating income of $741,820.

The bottom three lines of Table 3 matter as much as the total. Variable expenses, the lines that move with rooms sold or with revenue, come to $1,084,825; fixed expenses, which do not move with volume in the stabilized year, come to $1,261,360, or 53.8 percent of operating expenses. That split is a product of the line-by-line build: a budget expressed as one percentage of revenue has no split at all, and the section on fixed and variable cost below shows why that matters.

Payroll: A Staffing Plan, a Wage Floor and the Burden on Top

Payroll is the largest line in the worked budget, and a study should build it from hours and rates by position, because hours and rates are what can be checked. The worked budget separates two kinds of hours. Some positions are staffed on a schedule regardless of occupancy: a hotel's front desk is open through the night whether ten rooms are sold or eighty, and the budget carries an assumed 40 staffed hours a day for the front desk, night audit, breakfast and maintenance. Other hours follow volume: housekeeping is budgeted at an assumed 30 minutes per occupied room, which at the stabilized occupancy gives 10,512 hours a year. Salaried managers are a third block, fixed and paid whether or not the hotel fills.

The rate side starts with a floor that is a matter of law, not judgment. The federal minimum wage is $7.25 an hour ^[18], and the federal statute expressly does not excuse compliance with a state law or municipal ordinance that sets a higher minimum ^[40]. In Connecticut that higher minimum rose to $16.94 an hour on January 1, 2026 ^[16]. The state floor is not fixed: Connecticut law adjusts it each year by the change in the federal employment cost index over the twelve months ending the previous June 30 ^[17], and the state's Department of Labor already lists $17.48 from January 1, 2027 ^[25].

Table 4 sets the worked budget's housekeeping wage against those floors. The assumed $17.50 an hour clears the 2026 floor by $0.56 and the 2027 floor by $0.02. From January 2027, in other words, the statute all but sets the housekeeping wage, and every later year's floor will be set by an index the sponsor does not control. A multi-year projection that holds wages flat while the legal floor is indexed upward understates payroll in every year after the first. The fix is mechanical: tie each wage at or near the floor to the floor's own schedule, and escalate the floor by a stated assumption for the years not yet announced.

The floor only bounds the rate from below. The rate a hotel must actually pay to staff a front desk overnight in a given labor market is a market question, and the evidence a study should cite for it is an occupational wage survey for that market, the operator's own payroll records at comparable properties, or current job postings, with the source named. Where the only support is that a wage clears the minimum, the study should say so, because a wage set at the floor in a tight labor market is a staffing risk the lender should see.

On top of wages sit the employer's taxes and benefits. The employer's social security and Medicare taxes are statutory, at 6.2 percent and 1.45 percent of wages ^{[14] [15]}. The federal unemployment tax is imposed at 6 percent of total wages as section 3306(b) defines them ^[41], and a study should work out the net federal and state unemployment cost from those definitions and the employer's state rate notice rather than adopt a round figure. Workers' compensation and health benefits are quotations and plan terms. The worked budget carries all of these other items as one assumed 18.0 percent of wages, labelled as an assumption, which with the federal taxes brings the burden to 25.7 percent of wages in Table 4.

One more payroll item belongs in every budget for an owner-operated business: the owner's own labor. If the owner will manage the hotel, the hours that a paid general manager would work still have to be paid for, either as a salary in the budget or as an explicit statement that the owner's return includes that work. A budget that leaves the owner's labor out shows income that a buyer, a receiver or a hired manager could not earn without paying for that work. Global Cash Flow: Testing the Borrower, Its Owners and Its Affiliates as One Repayment Source takes up the same point from the owners' side.

Utilities: A Published Price Times a Load Someone Has to Defend

A utilities line is a price multiplied by a quantity, and the two halves need separate evidence. For price, the Energy Information Administration publishes average electricity prices by state and customer sector every month. In August 2026 the average price to commercial customers in Connecticut was 19.29 cents per kilowatthour, against 23.47 cents in August 2025 ^[26]. For the first eight months of 2026 the average was 21.75 cents, against 23.57 cents a year earlier ^[19], and the national commercial average for the same months was 14.04 cents ^[42].

Three features of those figures shape how a study should use them. The 2026 values are preliminary estimates from a cutoff model sample, while the 2025 values are final ^[43], so a price read today may be revised. A single month can sit well away from the year, as the August 2026 figure does beside the year to date. And the state average blends customers of very different sizes, since utilities may classify commercial customers by NAICS code or by demand or usage limits in a rate schedule ^[44]. The average is a reasonableness check on the hotel's cost, not its tariff. The evidence that belongs in the file is the serving utility's rate schedule for the class the hotel will be billed under, applied to a twelve-month load profile.

USDA's rules for one of its energy programs show the standard well. For energy efficiency improvement projects, the energy audit described in appendix C to subpart D of 7 CFR 5001 reports the average price per unit of energy the customer paid over the most recent 12 months, or an average of two to five years ^[45]. The appendix applies to that program, not to B&I budgets, but the principle carries: a utilities line should rest on a year of prices, never on the cheapest month available.

The worked budget follows that principle by using the year-to-date price, which on an assumed 950,000 kilowatthours a year gives electricity of $206,625. Table 8 shows what the opposite choice would do: priced at the August figure, the same load costs $183,255, and coverage rises from 1.29x to 1.33x with nothing about the hotel changed. A study that picks its period should say which period it picked and why.

Natural gas shows a different hazard. EIA's annual series puts the average commercial price in Connecticut at $10.71 per thousand cubic feet in 2025, after $10.39 in 2024 and $12.85 in 2022 ^[20], and the worked budget applies the 2025 figure to an assumed 3,600 thousand cubic feet for a cost of $38,556. The monthly series for 2026, however, shows a Connecticut value for only two of the six months published ^[46]; the rest are marked NA, which the table's key defines as not available ^[47]. A public series can simply run out, and when it does the gas utility's tariff is the only evidence left. The worked budget has no published price for water, sewer and waste removal and carries an assumed $48,000, to be replaced by the serving utilities' rate schedules.

The quantity half is the harder one to support. Kilowatthours and gas volumes for a building that does not exist come from an engineer's load estimate or from metered consumption at comparable properties the operator runs; neither is in the public record. The study should print the load it used and its source, so that a reviewer can tell a price problem from a consumption problem.

Property Tax and Insurance: Lines Someone Else Sets

Property tax is the clearest case of a line the borrower does not control. In Connecticut the arithmetic is published by the state's Office of Policy and Management: a mill is one dollar of tax for each thousand dollars of assessment, and each fiscal year's mill rate is based on the grand list from two years before ^[21]. The bill is the assessment multiplied by the mill rate and divided by one thousand ^[48].

Windsor Locks shows why the rate is only half the line. The state's mill rate dataset gives the town a real and personal property rate of 28.14 for fiscal year 2025 ^[49], 23.99 for fiscal year 2026 ^[27] and 22.25 for fiscal year 2027 ^[50]. A falling rate is not by itself a falling tax. The bill is the product of the rate and the assessment, and if the assessments on the grand list rise at the same time, the bill can hold steady or climb while the rate falls. A study that quotes a mill rate without the matching assessment, or applies one year's rate to another year's grand list, has a number with the right name and the wrong meaning.

For a new building the harder half is the assessment, because the finished property is not yet on the grand list. The study does not set it and does not value the property; it should obtain the assessor's estimate for the completed building, or state the basis it used and flag the line as provisional. The worked budget carries an assumed assessment of $4,200,000 at the fiscal 2026 rate of 23.99 mills, a tax of $100,758. Table 8 shows that an assessment 25 percent higher raises the tax to $125,947 and takes net operating income down to $716,631, just short of the $719,361 that the 1.25x lender assumption needs; the coverage line in that column prints as 1.25x only because it is rounded.

Insurance is set by the market and by the lender's requirements together. Once a 7(a) loan or a 504 project exceeds $500,000, SBA requires hazard insurance covering all of the collateral ^[22], and where a building, machine or piece of equipment that SBA money builds, buys or improves lies in a special flood hazard area, flood coverage is required as well ^[51]. Under the B&I rules the lender must ensure the borrower obtains and keeps borrower and project insurance in substance and amount similar to what lenders in the industry ordinarily require ^[52]. Those requirements define the coverages, and the coverages define the premium. A budget line for insurance built before anyone knows what the lender will require is a guess about two things at once.

Both lines also protect the collateral. The B&I definitions list advances for property taxes, rent, hazard and flood insurance premiums and emergency repairs among the protective advances a lender may make to preserve its security ^[53]. When a borrower stops paying its taxes or its premiums, the lender may end up paying them. That is a reason for a study to support these two lines with the tax bill and a written quotation rather than an allowance, and it is why the worked budget's $1,400 per room for insurance is labelled as a placeholder. A quotation 40 percent above it costs $156,800 instead of $112,000 and pulls coverage to 1.21x in Table 8.

Franchise Fees, Management, Repairs and Reserves

Franchise fees are among the best documented expenses a new business can have, and the loan data shows how often the line arises. Of the hotel and motel approvals under SBA's 7(a) program in fiscal 2017 to 2026, 64.69 percent ^[54] carry a franchise code in the loan record; for fitness centers the share is 48.47 percent ^[55]. For borrowers in those classes, a large part of the expense budget is set by a contract with the franchisor.

Federal rules make that contract readable before it is signed. Item 6 of the franchise disclosure document must present, in a table, all other fees the franchisee must pay to the franchisor or its affiliates, or that they impose or collect for a third party ^[23]. Where a fee may increase, the franchisor discloses the formula that determines the increase or its maximum amount ^[56]. The franchisor must furnish its current disclosure document at least 14 calendar days before the prospective franchisee signs a binding agreement or makes any payment ^[57], so a sponsor who has signed, or is about to sign, will have received the document, and the study can work from the actual fee table. The study should then use it: the royalty, the marketing and reservation charges, any technology or loyalty fees, and the definition of the revenue each fee is charged on. A royalty on rooms revenue and a royalty on gross revenue are different lines, and the disclosure document says which applies.

The disclosure document can also carry performance figures, but on the franchisor's terms. Item 19 lets a franchisor give actual or potential financial performance information only if it has a reasonable basis and the information appears in the document ^[58]. Where a brand publishes operating results for its outlets, a study can use them as a benchmark, after checking which outlets they cover and what they leave out. They are a franchisor's representation, not the borrower's evidence.

The worked budget assumes a royalty of 5.5 percent and marketing and reservation fees of 4.0 percent of rooms revenue, together $299,592, both labelled as assumptions to be replaced by the chosen brand's fee table. If the actual royalty were two points higher, the line would rise to $362,664 and coverage would fall to 1.18x in Table 8, the largest single effect of any line tested there.

Management is a contract line where a contract exists and a judgment line where it does not. Where a third-party management agreement is in hand, the budget should carry its base fee and any incentive fee as the agreement defines them. Where the owner will manage, the study should still charge a management fee or a manager's salary, for the reason given under payroll. The worked budget carries an assumed 3.5 percent of total revenue.

A repairs allowance describes a building of a particular age, and a study should say which. The worked budget carries an assumed $75,000 for a new building, and a projection that runs over the loan's term should state how that allowance changes as the building ages. Replacement spending is a separate line. Under USDA's coverage definition it is deducted before the ratio is struck ^[11], and the worked budget sets an assumed reserve at 4.0 percent of revenue, below EBITDA. A schedule of what will need replacing, from soft goods to roofs and mechanical systems, is the evidence that turns that percentage into a defensible number.

Fixed Against Variable: Why the Split Decides the Downside

A budget that names a cause for every line also shows how each line behaves when the projection is wrong, which is the purpose of the split at the foot of Table 3. Variable lines in the worked hotel follow either rooms sold (housekeeping hours, guest supplies, laundry and breakfast food) or revenue (franchise fees, channel and card fees, the management fee). Everything else stays put in the short run: the front desk is staffed, the managers are paid, the building is heated and insured and taxed whether or not the rooms fill.

Table 5 runs the stabilized year with occupancy 10 percent and 20 percent below the base, and with the average daily rate 10 percent below it. Occupancy 10 percent lower cuts total revenue to $2,895,005, but operating expenses fall only to $2,237,702, because the fixed block of $1,261,360 does not move. Net operating income falls to $541,502 and coverage to 0.94x, against the $719,361 of income the lender assumption needs. At 20 percent lower occupancy, coverage is 0.59x.

The rate case is worse than the occupancy case with the same revenue. A 10 percent lower average daily rate also brings revenue to $2,895,005, but the hotel still sells as many room nights, so the housekeeping hours, the supplies and the laundry do not fall at all; only the fees charged as a share of revenue do. Operating expenses fall only to $2,289,199, net operating income to $490,005 and coverage to 0.85x. The two shortfalls look identical on the revenue line, yet one leaves $541,502 of income and the other $490,005, and only a budget built line by line can show the difference.

The guidelines' request for sensitivity analysis ^[10] is met only when the shocks run through a budget that knows which lines move. How large the shocks should be, and where the break-even points fall, is the subject of Sensitivity and Stress Testing: Break-Even Points, Shock Sizes and What a Lender Should See.

Expense Ratios Are a Check, Not Evidence

An expense ratio compares a budget with something larger than itself, and the USDA rules expect such comparisons: the lender's credit evaluation includes spreadsheets and analysis of the financial statements with appropriate ratios and comparisons with industry standards ^[59]. The comparison is useful, but the public data itself shows why it is not proof.

The Census Bureau's Service Annual Survey publishes expenses by industry for employer firms. For accommodation, NAICS subsector 721, it estimates gross annual payroll of $62,455 million in 2022 ^[24] against revenue of $295,698 million ^[29], a ratio of 21.1 percent. Two years earlier, payroll was $42,382 million ^[24] against revenue of $161,725 million ^[29], a ratio of 26.2 percent. Between 2020 and 2022 the subsector's revenue rose much faster than its payroll, and the ratio fell with it. A ratio carries the revenue level of the year it was measured in, and applied to a projection it imports that year's conditions without saying so.

The subsector is also broad. It groups traveler accommodation, recreational accommodation such as RV parks and camps, and rooming and boarding houses, dormitories and workers' camps ^[60], and its traveler group spans hotels, motels, casino hotels and bed-and-breakfast inns ^[61]. A ratio for that population describes no limited-service hotel in particular. And the survey's categories are its own. Its total expenses of $200,622 million for 2022 ^[39] include $21,571 million of depreciation and amortization ^[36] and $8,074 million of lease and rental payments for land and buildings ^[37], items a property's operating budget either excludes or carries on a different line. Another $47,113 million sits in all other operating expenses ^[38], so that 23.5 percent of the industry's total expenses cannot be matched to any budget line. The survey also warns that its dollar figures are published in millions and its results may not be additive ^[62].

Table 6 sets the survey's ratios beside the worked budget's. Payroll with taxes and benefits takes 24.0 percent of the budget's revenue, close to the industry's 25.9 percent for payroll and fringe benefits. Repairs at 2.3 percent of revenue sit near the industry's 1.7 percent, and property tax at 3.1 percent sits below the industry's 3.8 percent for all governmental taxes and license fees, a broader category. Energy is the outlier: electricity and gas take 7.6 percent of the budget's revenue against 2.1 percent for the industry. The difference is a prompt, not a verdict. The industry ratio is national, measured in 2022, and priced at national averages, while the budget is priced at Connecticut's 2026 commercial rate of 21.75 cents ^[19] against a national average of 14.04 cents ^[42]. The ratio cannot say whether the budget's load is right; only the engineer's estimate and the tariff can. What the gap does is tell the credit officer exactly which line to ask about.

A ratio does real damage when it replaces the budget instead of checking it. Table 7 compares the line-built budget with the same hotel budgeted at an assumed operating expense ratio of 66 percent, standing in for a ratio copied from a template or a brochure, applied to the same revenue.

At the base revenue the copied ratio gives operating expenses of $2,123,004 instead of $2,346,185, adds $223,181 of income and shows coverage of 1.68x instead of 1.29x. The larger error appears when revenue falls. A ratio applied to revenue turns every fixed cost into a variable one, so the copied budget sheds expenses as fast as revenue falls. With occupancy 20 percent lower, the copied ratio still shows coverage of 1.34x, while the line-built budget shows 0.59x; the copied ratio has invented $430,817 of income. A lender reading the copied version would see a project that survives its stress test, and the stress test would be measuring nothing.

The Brochure Line: Why a Number Without a Basis Fails

Expense figures can circulate in places that were never meant to be evidence: a broker's offering memorandum, a franchise sales presentation, a sponsor's spreadsheet built from a previous deal. The trouble with such a figure is not that it is necessarily wrong. It is that nothing about it can be checked. Often it has no stated period, no stated quantity, no stated rate, and no stated property. Measured against USDA's demand for stated assumptions behind every projection ^[4], or SBA's call for prudent commercial credit analysis ^[2], a line whose basis is unknown has no defence when a reviewer asks for one.

Several specific traps follow from where such numbers come from. A property tax figure taken from a seller's last bill reflects the assessment on the roll when that bill was issued, and a sale or new construction may change the assessment. An insurance figure taken from the seller's policy reflects the coverages the seller chose, not the ones the lender will require. A payroll figure for an owner-operated business may contain no pay for the owner. A management fee or a replacement reserve may be missing entirely where the seller managed the property and paid for replacements only when something broke. Each omission flatters income, and none is visible unless the study rebuilds the line.

For an acquisition, the remedy is to start from the seller's actual statements and adjust them openly, as Buying a Business With a Loan: Why Coverage Is Tested on Historical Earnings, Not the Buyer's Plan sets out. For a new business there are no statements, so every line has to be built. Where a study cannot yet support a line, it should keep the figure, label it as an assumption, say what evidence will replace it, and test what happens if the replacement comes in higher. The site's worked examples follow that discipline. The Worked Example: USDA Business and Industry Hotel Feasibility Study in Connecticut carries its property tax and insurance as stated assumptions because no sourced figure was available to it; the Worked Example: SBA 7(a) Gym Feasibility Study in Hawaii prices electricity from an EIA rate and sets wages against the state minimum; and the Worked Example: USDA Business and Industry Loan RV Park Feasibility Study in North Dakota carries its property tax as an assumption to be confirmed with the county assessor.

What the Unproven Lines Do Together

Each evidence gap in a budget can look small on its own, which is what makes the combination easy to miss. Table 8 takes the worked hotel's base case and changes one line at a time, then four together.

Raising every hourly wage by one dollar, the kind of move a tight labor market or the indexed floor can produce, takes net operating income from $741,820 to $710,267 and coverage to 1.23x. A higher assessment, an insurance quotation above the allowance and a higher royalty take net operating income to $716,631, $697,020 and $678,748 in turn, with coverage falling to 1.21x on the insurance case and 1.18x on the royalty case. Together, the four adverse lines raise operating expenses to $2,510,800, cut net operating income to $577,205, and take coverage to 1.00x, with nothing about demand, occupancy or rate having changed. The base case cleared the 1.25x lender assumption by a thin margin; four unproven expense lines remove it.

The column priced at the August electricity rate runs the other way, lifting coverage to 1.33x: a convenient choice can flatter a budget as easily as an unlucky one can hurt it, and the cure for both is to state the source, the period and the quantity of every line.

Expenses and the SBA Loan Record: A Blank Page With One Useful Mark

No operating statement travels with an SBA approval into the public loan file, so the dataset is silent on the question this article asks: whether any borrower's costs came in above or below the budget in its study. What the SBA Loan File Cannot Tell You: The Limits of FOIA Data in Underwriting lists the fields the file does carry. Among them is a franchise code, and that one field touches the expense budget directly, because it marks the borrowers whose fee lines are written by a franchisor.

Counting cancelled approvals, as the dataset does, the 7(a) program logged 11,604 ^[55] gross approvals to fitness centers in fiscal 2017 to 2026, 8,942 ^[54] to hotels and motels and 645 ^[63] to RV parks and campgrounds. The franchise share for RV parks and campgrounds is 8.37 percent ^[63], far below the hotel and fitness shares given earlier: most approvals in that class carry no franchise code. Where a borrower has no franchise, the study has no disclosure document to read, and payroll, utilities and repairs carry more of the budget's weight.

Outcomes are printed here as approval-cohort floors, counting the loans approved in fiscal 2017 to 2021 that have charged off so far against those disbursed; later charge-offs would push each floor up, never down. The hotel and motel floor under 7(a) is 0.99 percent ^[54], for fitness centers 9.48 percent ^[55] and for RV parks and campgrounds 0.00 percent ^[63], a floor that says no loan in those cohorts has charged off to date, not that none will, while the floor across every 7(a) loan approved in those years is 6.33 percent ^[64]. The dataset records the outcomes, not their causes; nothing in it says whether an expense projection had any part in a charge-off, and a study should not read one in. How the floors are built and why they understate is set out in The Approval-Cohort Charge-Off Floor: Why a Published SBA Loss Rate Can Only Understate. The figures, their bases and their keys are printed in Table 9.

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.

A Credit Officer's Trace, Line by Line

The method reduces to a set of questions a reviewer can put to any expense budget, and a study prepared for a lender should answer each one before it is asked. The common thread is the USDA requirement that projections carry the list of assumptions behind them ^[4], applied to every line rather than to the total.

• Payroll. Is it built from hours by position and a rate for each? Does every wage clear the floor that binds, federal or state ^[40], and does the projection move with the state's announced schedule ^[25]? Are the employer's taxes computed at the statutory rates ^[14], and is the owner's labor paid for?
• Utilities. Does the price come from a stated source and a full year, not a single month ^[19]? Is the quantity an engineer's load or metered consumption, and is it printed?
• Property tax. Does the line show the mill rate, its fiscal year and the assessment it is applied to ^[48], and who supplied the assessment?
• Insurance. Does the line reflect the coverages the program and the lender require ^[22], and is it a quotation or an allowance?
• Franchise fees. Does the budget use the chosen brand's current fee table ^[23] and the fee base it defines?
• Management, repairs and reserves. Is there a management charge even where the owner manages, an allowance for repairs that fits the building's age, and a reserve deducted where the coverage definition requires one ^[11]?
• Ratios. Has every industry comparison been reconciled to the budget's definitions before it is read ^[62], and does the budget stand on its lines rather than on a ratio?
• Behaviour. Does the budget mark which lines are fixed and which are variable, and do the stress tests run through those lines rather than through a percentage of revenue ^[10]?

A budget that answers those questions can still be wrong, because a market can turn and a quotation can expire. But every error in it can be found, measured and corrected by the people who have to rely on it, and that is the sense in which this article calls a budget traceable.

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: USDA Business and Industry Hotel Feasibility Study in Connecticut
• Worked Example: SBA 7(a) Gym Feasibility Study in Hawaii
• Worked Example: USDA Business and Industry Loan RV Park Feasibility Study in North Dakota
• Case Study: Operating Expenses in Orlando's Full-Service Hotels
• Multi-Family and Apartments Feasibility Study: Operating Expense Analysis in Miami, Florida
• Construction Cost Evidence: Bids, Published Indexes and the Contingency a Lender Will Accept
• Debt Service Coverage in a Feasibility Study: How the Ratio Is Built, and Which Floor Binds

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