How to Spot a Fake Pay Stub: 9 Checks

To spot a fake pay stub, recompute it. Gross pay minus every listed deduction has to equal net pay to the cent, Social Security withholding should land near 6.2 percent of taxable wages and Medicare near 1.45 percent, and the year-to-date totals have to make sense for the pay date and pay frequency. Fabricated stubs are typed by hand from a target salary, so the arithmetic is where they break. Then confirm the money moved: match net pay against the deposits in the borrower's bank statements. A stub that reconciles internally and matches the account is credible. One that does either but not both needs a question before it needs a decision.

Below are the nine checks in the order that catches the most fraud for the least effort, plus what to do when a check fails, because most failures are not fraud.

How do you spot a fake pay stub?

Start with math, not appearance. Underwriters instinctively look at layout first, and layout is exactly what a stub generator gets right. Any of the several dozen US sites selling instant pay stubs will produce a clean document with a real-looking employer block, correct fonts and a plausible template. What those sites cannot do reliably is compute payroll withholding the way a payroll system does, because the person filling in the form picks a salary and lets the site approximate the rest.

Here are the nine checks, cheapest first:

1. Deductions reconcile to net pay. Gross minus federal income tax, state tax, Social Security, Medicare, and any voluntary deductions must equal net pay exactly.
2. FICA percentages are right. Social Security is 6.2 percent of wages up to the annual cap, Medicare is 1.45 percent with no cap. These are fixed rates, not estimates.
3. Year-to-date gross fits the calendar. Divide year-to-date gross by period gross. The result should be close to the number of pay periods elapsed by the pay date.
4. Consecutive stubs step correctly. Two stubs from the same employer should show year-to-date totals differing by exactly one period of gross pay.
5. No perfectly round numbers. Real gross pay for an hourly worker almost never lands on a whole hundred, and net pay after withholding essentially never does.
6. State tax matches the state. A stub showing state income tax withheld for a borrower in Texas, Florida, Nevada, Washington, South Dakota, Wyoming, Alaska or New Hampshire is wrong on its face.
7. The employer is real and matches. The legal name, address and any EIN shown should match what you can find independently and what appears on the application.
8. Document integrity. Check whether it arrived as a native PDF or a screenshot, whether fonts change mid-document, whether columns align, and whether a check or advice number is present.
9. Deposits corroborate. Net pay should appear in the bank statements as a recurring credit on a matching cadence with an employer or payroll-processor descriptor.

Real vs fake pay stubs: the math check, worked

Take a stub claiming $2,600.00 semi-monthly gross for a salaried employee in Ohio. Run the fixed-rate lines first, because those are the ones with no judgment in them.

Three independent failures: a wrong FICA rate, deductions expressed as flat round percentages, and year-to-date gross implying two more pay periods than the calendar allows. Any one on its own might be an oddity. Together they describe a document someone built rather than one a payroll system produced. Note also that net pay was chosen first ($2,100.00, a round number) and the deductions were reverse-engineered to reach it, which is the single most common signature of a generated stub.

How are pay stubs verified?

Verification happens in three tiers, and they prove different things. Document checks prove internal consistency: the stub agrees with itself. Cross-document checks prove corroboration: the stub agrees with records the borrower did not author, primarily bank statements and the prior year's W-2. Source verification proves the employment relationship, and it requires either a written verification of employment from the employer or a payroll data connection the borrower consents to.

Most lenders run tiers one and two on every file and reserve tier three for exceptions, large exposures, or files where the first two produced a flag. That is a sensible allocation, provided tiers one and two are actually run rather than glanced at. The arithmetic check in particular is almost never done by hand under pipeline pressure, which is why pay stub verification software earns its keep: it recomputes every line on every file and matches net pay to deposits automatically, so the cheap checks stop being the ones that get skipped.

How do lenders verify pay stubs against bank statements?

By matching amount, cadence and descriptor. A salaried borrower paid by direct deposit generates a credit equal to net pay, arriving on a fixed schedule, carrying a descriptor that names the employer or its payroll processor. Line the stub up against the account: if the stub says $2,094.30 on the fifteenth and the last of each month, the statements should show $2,094.30 on or within a day of those dates, twice a month, from the same descriptor.

Agreement across two or three cycles is strong evidence, because the deposit record comes from the bank rather than the applicant. Disagreement is a question, not a verdict. The most common innocent explanations are a split direct deposit sending part of net pay to a second account, a mid-period change to benefits or a 401(k) contribution, a garnishment that started recently, or a borrower who changed banks partway through the statement period. Ask before you decline. The same logic applies one document over, where detecting a fake bank statement comes down to recomputing running balances rather than trusting the layout.

What are the biggest red flags on a fake pay stub?

Ranked by how reliably they indicate fabrication rather than sloppiness:

• Deductions that do not sum to the difference between gross and net. Payroll software cannot make this error. A human filling in a template does it constantly.
• FICA at anything other than 6.2 and 1.45 percent. These rates are statutory. Watch for the Social Security wage cap on high earners late in the year, which legitimately stops the withholding.
• Round numbers anywhere they should not be. Net pay of exactly $3,000.00, federal tax at exactly 10 percent of gross, year-to-date figures ending in double zeros.
• Year-to-date totals that contradict the date. The strongest single check, and the one forgers most often ignore because it requires thinking about the calendar.
• State tax withheld in a state with no income tax, or missing where the borrower clearly lives and works in a state that has one.
• No check number, advice number or pay period range. Every payroll system emits these. Generators frequently omit them.
• Delivered as a photo or screenshot. Not proof of anything by itself, but it removes the PDF metadata and layer structure that would expose editing, and that is often the point.
• Employer details that do not survive a search. An address that maps to a residence, a phone number that is a mobile, or a company with no footprint.

Can you get in trouble for submitting fake pay stubs?

Yes. Submitting fabricated income documents to obtain credit is loan application fraud. Where the lender is federally insured, it can be charged as bank fraud under 18 U.S.C. 1344, which carries substantial penalties. In practice the consequences that arrive first are commercial rather than criminal: the application is denied, the attempt is recorded, and a loan later discovered to rest on falsified income can be accelerated. Lenders also have reporting obligations in various circumstances. Borrowers occasionally treat a fabricated stub as a shortcut around paperwork they consider a formality. It is not a formality, and the exposure is real.

What if the borrower has no pay stub at all?

Then pay stub verification is the wrong tool and you should not force it. Self-employed borrowers, 1099 contractors, and the growing share of applicants earning across marketplaces and creator platforms have no employer issuing a stub. For those files, income evidence shifts to tax returns, 1099 forms, business bank statements and, increasingly, a consolidated record of platform payouts the borrower can produce across the services that actually pay them. The verification question changes from "is this document real" to "what is the durable income", which is a cash flow analysis problem rather than a document authenticity problem.

The check that carries across both worlds is the deposit record. Whether income arrives as a payroll credit or as a dozen platform payouts, it lands in an account, and the account is the record the applicant did not write. That is why bank statement verification sits underneath almost every income decision, and why teams doing volume tend to standardize on income verification software that reads both document types together rather than treating them as separate reviews.

How many pay stubs should a lender collect?

Two consecutive stubs covering roughly 30 days is the common US standard, paired with 60 to 90 days of bank statements. Two matters more than one, because their year-to-date totals should differ by exactly one pay period of gross pay, and that step check is unavailable from a single document. Files with commission, overtime, bonus or seasonal pay need a longer look, usually a full year plus the prior year's W-2, since the question there is not what the borrower earned last month but what they can be expected to earn consistently. Mortgage underwriters calculating qualifying income generally average variable components over 24 months for exactly this reason, while auto lenders working thin files often decide on a single stub plus a deposit check because the exposure and the speed expectation are different.

Where automation actually helps

None of these checks is difficult. They are just tedious, and tedium is what gets dropped when a queue backs up. An underwriter who recomputes withholding on every stub in a 40-file day is not going to finish, so the check silently becomes optional, and optional checks catch fraud at random. Automating the arithmetic, the year-to-date consistency test and the deposit match means all three run on every file at the same cost, and the underwriter's attention goes to the flags rather than to the recomputation. That is the whole argument: not that software sees something a careful human cannot, but that it stays careful on the four hundredth file.