The Loan Analytics Methodology Series: Data, Underwriting, Asset-Class and Program Methods

This series sets down, article by article, how Loan Analytics builds and reads the evidence behind a lender-grade feasibility study. The site's methodology page describes the study itself: the trade area, demand, competition, the site, the financial model and the coverage test. The series goes one level down. It explains where each number in a study comes from, what a public loan record can and cannot prove, what each lending program actually tests, and how a study answers each of those tests with evidence a credit officer can check.

The articles are written for three readers at once. A borrower can use them to see what a lender will ask before the application goes in. A loan officer or a Certified Development Company can use them to judge whether a study in the file answers the program's own questions or only repeats the sponsor's hopes. A reviewer can use them to trace any figure back to the page or the loan file it came from. Every article follows the same rules, so a reader who learns how one is built can audit all of them.

How Every Article in the Series Is Built

Four rules hold across the series, and the gates the articles pass before publication enforce them.

First, every outside fact is quoted from a saved copy of a public primary source, with the address beside it in the Sources list: the Code of Federal Regulations, the United States Code, the Federal Register, SBA's program and data pages, USAspending, Census, BEA, the Federal Reserve, the FDIC, NCHS and USCIS. Where an article says what a regulation requires, the words come from the regulation. SBA's business loan rules, for instance, allow the agency to ask for a feasibility study among the conditions it attaches to a loan ^[1].

Second, every loan figure is printed from the Loan Analytics loan dataset and carries its key. The dataset holds SBA 7(a) and 504 approvals from the agency's public loan files; the slice the articles use most covers fiscal years 2017 to 2026, which holds 647,329 ^[2] approvals, of which 182,784 ^[3] fall inside the asset classes the site covers. Each figure carries either a FIGURES id (a national figure) or a cell key (a state, asset class or program slice), and the Sources entry names the dataset release it came from, so a reader can find the exact row.

Third, loss and cancellation rates are measured by approval cohort and printed as floors, because loans still open can still be charged off, and a rate resting on fewer than 30 loans is withheld rather than shown. The articles in the first group explain why in detail.

Fourth, where an article discusses debt service coverage, the equity a borrower brings, or how an acquisition is tested, it uses the same conventions as the site's worked examples, and says so in one fixed line:

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.

The loan figures this page leans on are collected below with their keys.

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.

Group A: How the Loan Data Is Built and Read

The first group explains the loan dataset itself: where it comes from, how it is checked, how loans are sorted into asset classes, what its rates mean and where it falls silent. For scale, SBA's files for fiscal year 2025 alone carry 84,840 ^[4] approvals across both programs.

• How the Loan Analytics SBA Loan Dataset Is Built: Six FOIA Files Reconciled to SBA's Year-End Reports: the public loan files, the data dictionary and the independent check against SBA's own annual totals.
• From NAICS Code to Asset Class: How SBA Loans Are Assigned, and Where Two Industries Share One Code: how an industry code on a loan becomes an asset class, and why some classes must disclose a neighbour that shares their code.
• The Approval-Cohort Charge-Off Floor: Why a Published SBA Loss Rate Can Only Understate: cohorts, seasoning and the reason a loss rate taken today can only rise.
• Why a Rate on Fewer Than 30 Loans Is Withheld: Small Samples and the Suppression Rule: what happens to a rate when only a handful of loans stand behind it.
• Cancellation Is Not Charge-Off: The Two Ways an SBA Approval Can End Without Repayment: approvals that never fund against loans that fund and fail.
• What the SBA Loan File Cannot Tell You: The Limits of FOIA Data in Underwriting: the fields that are missing or withheld, and what a study supplies instead.
• USDA Business and Industry Loans in Public Data: What USAspending Shows and What It Hides: the award records the federal spending site publishes and how far they reach.
• How to Read a Loan-Outcome Page: Counts, Floors, Ranks and the Keys Behind Every Figure: a walk through every block of a state and asset class page.

Group B: Underwriting Method

The second group covers the analytical core of a study, the tests a lender applies when it decides whether the cash flow can carry the debt. SBA's own rules treat a feasibility study as something the agency may require as a condition of a loan ^[1], and these articles set out what such a study has to contain to be worth requiring.

• Debt Service Coverage in a Feasibility Study: How the Ratio Is Built, and Which Floor Binds: what goes into the numerator and the denominator, and which floor sizes the loan.
• Global Cash Flow: Testing the Borrower, Its Owners and Its Affiliates as One Repayment Source: combining the operating company, the owners and their affiliates without counting a dollar twice.
• Equity Injection and Sources and Uses: How a Study Shows Where Every Dollar Comes From: a sources and uses table that foots, and the equity a borrower brings at the loan the cash flow supports.
• Buying a Business With a Loan: Why Coverage Is Tested on Historical Earnings, Not the Buyer's Plan: normalising the seller's history, and why a buyer's plan is a second test, never a substitute.
• Sensitivity and Stress Testing: Break-Even Points, Shock Sizes and What a Lender Should See: how large a shock to test, taken from the published record, and how far the base case sits from the point where coverage fails.
• A Feasibility Study Is Not an Appraisal: What Each Document Answers, and What Each Cannot: the question an appraisal answers against the question a study answers, and why the study never states a value.
• Trade Area and Demand Methods: Capture Rates, Penetration Rates and Gravity Models: drawing a trade area from public data and the arithmetic of capture, penetration and gravity.
• Supply and Comparable Analysis: Counting Competitors, the Pipeline and the Properties a Study Leans On: counting competitors and the pipeline, and choosing and adjusting the properties a study leans on.
• Ramp-Up and Absorption: How Long a New Project Takes to Stabilise, and How a Study Proves It: the curve from opening to stabilisation, the evidence for it, and the reserve that carries the gap.
• Construction Cost Evidence: Bids, Published Indexes and the Contingency a Lender Will Accept: bids, published indexes, escalation and the contingency, ranked by how much weight each can bear.
• Operating Expense Evidence: Building an Expense Budget a Credit Officer Can Trace: an expense budget built line by line from rates a credit officer can look up.
• Reading Rejection From the Loan Outcomes: Where SBA Approvals Cancel, Where Loans Fail, and What the File Cannot Say: what the cancellation and charge-off floors say by asset class, and what they cannot say about why.

Group C: Asset-Class Method

The third group takes one asset class per article and applies the general methods to it. Each sets out the demand model for the class as equations with labelled inputs, the metrics a lender reads and how each is evidenced, the class's own record from the cells with every rate printed as a floor, the eligibility rules that bear on it, and a short worked model. The full worked example for a class, where the site has one, is linked from its article. The record differs widely from class to class: the car wash class alone carries 2,713 ^[6] 7(a) approvals in fiscal years 2017 to 2026, while four of the fifteen (RV and boat storage, glamping, short-term rental and retail as a whole) are not separate classes in the dataset at all, and each of those articles names the neighbouring record it prints instead.

• Car Wash Underwriting Method: Membership Demand, Tunnel Throughput and the Class's SBA Record: members against single washes, tunnel cars per hour, and the 811192 code that also counts truck and mobile washes.
• Hotel Underwriting Method: Fair Share Demand, RevPAR Evidence and the Class's SBA Record: fair share and penetration for a hotel, room nights from a state tourism series, what lodging tax collections can and cannot show, and franchise fees in the expense line.
• Gas Station Underwriting Method: Fuel Gallons, Inside Sales and the Class's SBA Record: gallons from traffic counts, inside sales and margins, and the tank rules that shape the environmental file.
• RV Park Underwriting Method: Site Nights, Seasonality and the Class's SBA Record: site nights by month, transient against long-term sites, and the passive-income question long stays raise.
• RV and Boat Storage Underwriting Method: Registrations, Space Demand and the Self-Storage Code It Shares: demand from vehicle and vessel registrations, space per unit, and a loan record that is the self-storage record.
• Assisted Living Underwriting Method: Age and Income Qualified Demand, Licensing and the Class's SBA Record: age and income screens, penetration, and the state licence that decides whether the business can open at all.
• Multifamily Underwriting Method: Rent Evidence, Absorption and an SBA Record That Almost Stops: rent and absorption evidence, the passive-business limit, and an SBA record that almost stops from fiscal year 2024.
• Owner-Occupied Retail Underwriting Method: Sales Capture, Occupancy Rules and the Retail Loan Record: why the dataset has no single retail class, sales capture from Census retail series, and the occupancy shares.
• Marina Underwriting Method: Slip Demand, Boat Registrations, Submerged Land Leases and the Class's Record: slips from vessel registrations, wet against dry storage, and leased submerged land as the collateral question.
• Gym and Fitness Center Underwriting Method: Membership Penetration, Attrition and the Class's SBA Record: members per resident, monthly attrition, dues evidence, and a 7(a) record in which franchised approvals are common.
• Glamping Underwriting Method: Unit Nights, Rate Evidence and a Loan Record That Shares the Campground Code: unit nights and rate evidence for tents and cabins, and a loan record that is the campground record.
• Short-Term Rental Underwriting Method: Nightly Rates, Local Registration Rules and Why the SBA Record Is Silent: nightly rates and local registration rules, and why SBA's rules and the dataset both leave the class out.
• Self-Storage Underwriting Method: Square Feet per Capita, Lease-Up and the Class's SBA Record: square feet per resident and its limits, lease-up, and the vehicle storage that shares the code.
• Restaurant Underwriting Method: Seats, Turns, Check Averages and the Full-Service and Limited-Service Record: seats, turns and the check average, read separately for full-service and limited-service rooms.
• Data Center Underwriting Method: Power, Tenant Commitments and a Thin SBA Record: power and tenant commitments as the demand evidence, and an SBA record kept under a code that also holds hosting and data processing firms.

Group D: Program Method

The fourth group takes the financing programs one at a time and reads each program's own rules: what the regulation or statute tests, and how a study answers each test. The rules differ more than borrowers expect. USDA's Business and Industry guarantee, for example, starts from a definition of a rural area that leaves out larger cities and towns and their urbanized fringe, subject to exclusions the rule itself lists ^[8], a test no SBA loan has to pass.

• SBA 7(a) Method: What 13 CFR 120 Tests and How a Feasibility Study Answers Each Test
• SBA 504 Method: The Job Test, the Project Split and What a Study Must Show the CDC
• USDA B&I Guarantees Under 7 CFR 5001: What the Agency Tests and How a Study Answers
• Conventional Bank Loan Method: Appraisal Rules, Supervisory LTV Limits and the Study a Bank Reads
• EB-5 Method: What USCIS Tests in the Business Plan and the Job Model, and How a Study Answers

Reading the Series With the Worked Examples

The series explains methods; the worked examples apply them to a single hypothetical project in a single state, with every assumption labelled. The two are meant to be read together. A reader who wants to see a coverage test computed end to end, on the conventions above, should open a worked example in the same program; a reader who wants to know why the test is set up that way should come back here. On the site's own methodology page, the coverage threshold is whatever the program and the lender demand, placed in a band between two ratios whose floors shift with the program and the size of the loan ^[9]; the program articles show where each of those floors comes from in the program's own texts.

Each article ends with the pages that sit next to it on the site: the program page, the data hubs by state and by asset class, the worked examples that use the same method, and any archived article that already covers a neighbouring question, so the series adds to what the site has published rather than repeating it.

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)

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