Ask ten brokers what an underwriter is looking for and you'll get ten different half-answers — “good credit,” “stable job,” “enough down payment.” All true, none complete. Underwriting is a structured read: income first, because everything else is measured against it; ratios second, because that's the pass/fail gate; documents third, because the file has to prove what the application claims; and fit last, because the same file can be an easy approval at one lender and a decline at another.
This handbook walks that structure in order, with the Canadian specifics that actually matter — OSFI's stress test, insured-ratio limits, and how documentation expectations shift by employment type and deal type. It's written for brokers and mortgage agents/associates who want to stop reacting to underwriter questions and start anticipating them before a file is ever submitted.
How lender underwriters actually think about a file
An underwriter's job isn't to find reasons to say no — it's to build a defensible case for yes, one that a file review, an audit, or a mortgage insurer will still agree with a year later. That single fact explains almost everything about how they read a package: they want the numbers to be provable, not just plausible, and they want the story on the cover sheet to match the documents underneath it, line for line.
In practice that means a consistent sequence: confirm income is real and durable, calculate the ratios it supports, verify the down payment and its source, then check the property and the deal shape against the specific lender's and insurer's rules. Every one of the sections below maps to one step in that sequence — understand the order and you understand why a file that looks fine to you can still generate a condition list, or why two nearly identical files land differently at two different lenders.
The reframe: underwriting isn't a judgment on the applicant — it's risk-scoring a set of documents against a policy. Give the underwriter a file that scores itself and there's nothing left to ask.
Step 1. Income documentation by employment type
Every income type has its own proof standard, and mixing them up is the single most common cause of a mid-file condition. The table below is the quick-reference version; the detail underneath covers the two types that generate the most back-and-forth.
| Employment type | Core documents | How income is calculated |
|---|---|---|
| Salaried, permanent, past probation | Employment letter (letterhead, dated, position, salary, start date), recent pay stub, T4/NOA for the required years | Base salary as stated; guaranteed pay only unless the lender allows an averaged variable component |
| Hourly / variable hours | Employment letter confirming hourly rate and typical hours, pay stubs, T4s/NOAs for the lender's required history | Typically averaged over the lender's required period (often two years) rather than a single recent stub |
| Commission | Employment letter confirming commission structure, T4s/NOAs, sometimes a letter from the employer confirming the structure is ongoing | Typically a multi-year average, not the most recent (and possibly highest) year in isolation |
| Self-employed | Two years of T1 Generals with statements of business activities, NOAs confirming no tax arrears, business registration/incorporation documents | Typically a two-year average of declared income — the standard framing lenders use, not last year's deposits |
| Rental income | Lease agreements, and the lender's specific offset or add-back method applied to the figure | Varies by lender — some offset a portion of the mortgage payment, others add a percentage of gross rent to income |
Self-employed applicants deserve a specific callout because it's where the most avoidable declines happen. Lenders work from what was declared to the CRA, not what actually moved through the business — a strong bank balance doesn't offset two years of minimized taxable income on paper. If your applicant has been optimizing for a low tax bill, have that conversation before you apply, not after a decline.
Commission and variable income: the number an underwriter uses is almost always lower than the applicant's best year, because the lender is buying certainty, not upside. Set that expectation early in the client conversation.
Step 2. GDS/TDS ratios and the stress test, worked simply
Two ratios decide how much mortgage an income supports. GDS (Gross Debt Service) is housing costs — principal, interest, property tax, and heat, plus a share of condo fees where applicable — divided by gross income. TDS (Total Debt Service) adds every other debt payment (car loans, credit cards, student loans) on top of housing costs, divided by the same gross income. For insured mortgages, the standard maximums are 39% GDS and 44% TDS; conventional (uninsured) deals often allow more room, but every lender sets its own ceiling within that logic.
The stress test — OSFI's Minimum Qualifying Rate under Guideline B-20 — is the layer on top of both ratios. It requires qualifying at the greater of the contract rate plus 2%, or 5.25%, regardless of the rate the client is actually being offered. So a client approved on paper at their contract rate might still fail to qualify at the stress-tested rate, and that gap is exactly why pre-approval conversations should always be run at the higher number, not the advertised one.
Worked simply: take gross annual income, apply the lender's GDS/TDS ceilings to find the maximum monthly payment the file supports, then confirm that payment still qualifies at the stress-tested rate — not the contract rate. If it doesn't clear the stress test, the deal size (or the debt load) has to change, not the math.
For deeper background on how the qualifying rate interacts with amortization and insured-ratio limits, see OSFI's final revised Guideline B-20 directly.
Step 3. Documents by deal type
Beyond the income proof in Step 1, every deal type carries its own document layer. The table below is a starting checklist, not an exhaustive one — individual lenders add their own wrinkles on top.
| Deal type | What's added on top of standard income/ID docs |
|---|---|
| Purchase | Signed purchase agreement with every schedule and amendment, MLS listing, deposit confirmation, down payment source with 90-day account history |
| Refinance | Current mortgage statement, property tax statement, purpose of funds explanation, updated property valuation (appraisal or AVM depending on the lender) |
| Switch / transfer | Current mortgage statement, payout statement from the existing lender, confirmation of no material change in the borrower's financial position since the original approval |
| Rental / investment property | Lease agreements or market rent estimate, existing property expenses (tax, condo fees, insurance), rental-offset method the lender applies |
| New to Canada | Valid immigration document (work permit, PR card, or equivalent), foreign or Canadian credit history where available, sometimes a larger down payment or a specific newcomer program |
The common thread across every deal type is the same: the file should tell a single, consistent story from the application through every attached document. A refinance whose stated purpose doesn't match the debts being paid off, or a rental deal whose lease amount doesn't match the income calculation, generates a condition even when every individual document is technically valid.
Step 4. Lender fit and the decline reasons that actually recur
The same file can be an easy approval at one lender and a hard decline at another, because each lender weighs the same inputs differently — some are more flexible on self-employed income, some cap rental offsets more conservatively, some have hard overlays on specific property types or credit profiles. Matching the deal to the right lender before submission is arguably as much of the underwriting skill as the math itself.
- →Income doesn't match the documentation standard for that employment type — e.g., using a single strong year for a self-employed applicant instead of the two-year average.
- →Ratios were calculated at the contract rate, not the stress-tested rate, so the client was told a number the file can't actually support.
- →Down payment source is unclear — large or unexplained deposits, or gifted funds not visible in the account history matching the gift letter.
- →Debts on the credit bureau don't match debts declared on the application — an undisclosed car loan or line of credit changes the TDS calculation entirely.
- →The deal simply doesn't fit this lender's overlays — property type, credit profile, or deal size outside what that specific institution underwrites, regardless of how clean the file otherwise is.
The fix for most of these: run the ratio and documentation checks before choosing the lender, not after — matching fit upfront prevents the resubmission cycle entirely.
How AI underwriting fits into this workflow
Everything above is a repeatable calculation once the inputs are in hand — which is exactly the kind of work that's slow by hand and fast for a properly built tool. Treadstone's AI underwriting is being built to let brokers and agents run this handbook's exact logic — income by employment type, GDS/TDS, the stress test, lender-fit flags — against a real file in a fraction of the time and cost of underwriting it manually or hiring in-house. It's currently in early access with an email waitlist, not a general-availability product, so it complements rather than replaces a careful human read for now.
If you'd rather have the fulfillment side of a file — document collection, condition chasing, submission packaging — run by a human team today, that's Treadstone's fulfillment service, live now and separate from the AI underwriting waitlist. For the step-by-step version of catching problems before they reach an underwriter, pair this handbook with the Pre-Underwriting File Review Checklist.