What has to be true for fixed to beat variable on a late-2026 $500,000 renewal

The letter on the table has two boxes. Five-year fixed at 4.44%. Five-year variable at 3.49%. The balance is $500,000. Twenty-five years left on the amortization. A late-2026 renewal, modeled, not a family.

The gap is 95 basis points. Variable opens $257 a month cheaper: $2,494 against $2,751. Over a five-year term that opening gap is not the answer. The answer is what has to happen to Prime for the cheaper box to stop being cheaper.

September 2 already happened. The Bank of Canada held the overnight rate at 2.25%. Seventh consecutive hold, the last cut on October 29, 2025. Next dates are October 28, with a Monetary Policy Report, and December 9. Neither one writes the coupon on this letter. Five-year fixed money follows the five-year Government of Canada bond, 3.64% as of September 16. Variable follows Prime, 4.45% as of September 9, which is overnight plus the banks' usual 220 bp. Today's variable is Prime minus 96.

This is the file. Then the paths. Then the things that usually decide it anyway.

A $500,000 file, modeled

Uninsured or insured does not change the payment on a straight renewal of the same balance. The engine is Canadian monthly payments, semi-annual compounding, not in advance. $500,000 remaining, 25 years left, five-year term, adjustable payments on the variable leg.

Fixed at 4.44% is the known point: $2,751 a month, $103,587 of interest over 60 months, $438,545 still owed at term end. No branches.

Variable starts at 3.49% — $2,494, $257 lighter — and everything after that is Prime. Those are the live street-card numbers from ratellow.com/api/rates, last updated September 16. Posted five-year at the chartered banks is 6.09%. Do not shop the posted card.

A stay at your current lender, or a straight switch with the same balance and remaining amortization, does not re-open the prescribed stress test. Increase the loan or the amortization and it does: 6.44% on today's fixed, 5.49% on the variable. The variable qualifies you for slightly more. That is a different file if you are adding money.

What has to be true

Compare total interest over 60 months. Payments and ending balances move, so the monthly number lies.

The paths below are judgment, not a model. The Bank of Canada's own Q2 2026 Market Participants Survey, out July 27, had the first hike not arriving before Q2 2027. The Big Six, as of mid-July, were split: Bank 5, Bank 4, Bank 1 and Bank 2 at 2.25% through year-end, National Bank and Bank 3 at 2.75%. Treat the ordering as the argument. Treat the percentages as soft.

The hold, then a gentle 2027 step (~55%)

The Bank sits through October 28 and December 9. Then a quarter point in the spring of 2027 and another by summer. Policy to 2.75%. Prime to 4.95%. Variable walks 3.49% → 3.74% → 3.99%. The payment drifts to $2,624, still under the fixed coupon at the end of the term. Interest $91,004. Variable is ahead by $12,583.

That is the survey reading, and it is the one the street is priced for.

Cuts come back (~25%)

The July Monetary Policy Report had 2026 real GDP near 0.7% and unemployment at 6.5%. Equifax had first-quarter 2026 mortgage delinquency balances up about 32% year over year nationally, 52% in Ontario. If the oil impulse fades the way the Bank wrote it, the labour market is the binding argument.

Three cuts — October, January, April — take policy to 1.50% and variable to 2.74%. The payment falls to $2,303. Interest $64,822. Variable is ahead by $38,765.

Hikes come back (~20%)

CPI printed 3.2% in May on gasoline tied to the Middle East. The five-year bond spiked near 3.18% in early July and sits at 3.64% now. If that impulse broadens, the Bank moves earlier than any base case: four quarter-point steps through 2027, policy 3.25%, Prime 5.45%, variable a full point above today's 3.49%. The payment climbs to $2,758. Interest $101,111. Variable is still ahead by $2,476.

Reading the three paths

Probability-weighted, variable is ahead by about $17,100. Forget that number.

None of the three paths put fixed ahead. The ugly path still saves about $2,500. The good one saves about $39,000.

The break-even is slow. From a 95 bp opening discount, a single permanent step-up in Prime ties the two products at about +115 basis points if it lands in mid-2027, or about +140 basis points if it waits until early 2028. In English: five quarter-point hikes in 2027 that never reverse is still roughly a tie. Four is not enough. That is a better test than guessing October 28.

Nobody forecast 2022 correctly, the Bank included. These percentages are ours.

The things that usually decide it

Which variable you actually hold

Canadian "variable" is two products. A variable-rate mortgage (VRM) keeps the payment still and moves the interest-versus-principal split. An adjustable-rate mortgage (ARM) moves the payment and keeps the amortization on the calendar. The paths above assume an ARM.

A VRM's frozen payment defers the adjustment. Push the rate high enough and the payment stops covering interest — the trigger rate — and unpaid interest piles onto principal. On this $500,000 opening at 3.49%, the trigger sits near 6.1%, which is Prime around 7.0%, some 255 bp above today. Remote on all three paths. The mechanic is still why industry summaries of the Bank's renewal work put about 10% of variable renewers facing payment jumps above 40% while about 25% see payments fall at least 7%. Same product name. Opposite households. Start with VRM versus ARM, then the trigger-rate math.

Breaking it

A closed fixed mortgage exits at the greater of three months' interest or the interest rate differential. A closed variable exits at three months' interest, flat. IRD is a fixed-rate test.

FCAC's own worked example, not ours: $200,000 at 6% with 36 months left, against a 4% posted rate, is about $3,000 on three months' interest and about $12,000 on IRD. The borrower pays the higher figure.

On this modeled $500,000 file, three months' interest is about $5,550 on the 4.44% fixed and about $4,363 on the 3.49% variable. Only the fixed carries IRD on top. Those are file math, not a government schedule.

If you might move, refinance, or pull equity inside five years, that option is often worth more than 95 bp. It is also why switching from variable to fixed mid-term exists: many lenders will convert you. Confirm yours before you count on it. Variable is a decision you can partly take back. Fixed is not.

The test, if it applies

A straight switch at renewal is largely exempt. Extra principal or a longer amortization brings the test back at the greater of contract plus two points or 5.25% — 6.44% on today's fixed, 5.49% on the variable. How the stress test differs for fixed versus variable.

Forty-two percent is not an argument

CMHC's Spring 2026 Residential Mortgage Industry Report put variable at 42% of newly extended mortgages in February 2026. Most popular single category. First time variable outpaced fixed since 2022. Traditional five-year fixed was 11%.

Read the series, not the print. 57% variable at the January 2022 peak. 5–6% at the mid-2023 trough. 42% by February 2026. Households buy variable when it is cheap and leave when it is not. Forty-two percent is a spread observation. The 2022 cohort that crowded in at the peak bought the top. The long-run record is kinder to variable than that cohort. The crowd is still not a reason.

Run the two boxes

The widget is this file. Variable in the current-rate box at 3.49%. Fixed in the new-rate box at 4.44%. $500,000. Twenty-five years. Change your own balance before you trust anyone's ranking, including ours.

If you want a third offer on the same screen, open the renewal comparison view. Then the full fixed-versus-variable framework and the 2026 worked example.

What you are actually buying

Variable is the better expected-value choice at a 95 bp opening gap, and being wrong now takes a genuine reversal — five hikes that stick. Fixed at 4.44% is not a timid purchase. It is about $12,583 of extra interest on the consensus path for a payment you can put in a spreadsheet. A household that already ate a 15–20% jump, or that cannot take $2,758 if Prime walks the ugly path, is buying sleep. That is rational.

The three-year fixed at 4.24% is the middle door: $2,696 a month, about $2,900 less interest over three years than sitting in the 4.44% five-year, and you are back in the market in 2029. You are also back in the market in 2029.

Pick on your own two numbers. The Bank already held. October 28 will not save a letter you sign on the posted card.

FAQ

Is variable cheaper on a $500,000 late-2026 renewal? It opens cheaper. $2,494 a month at 3.49% against $2,751 at 4.44%, a $257 gap, 95 bp. Over five years it stays cheaper unless Prime rises about 115 bp by mid-2027 and stays there. That is five quarter-point hikes that do not reverse.

How many Bank of Canada hikes would make the 4.44% five-year win? About five quarter-point steps that land in 2027 and never come off. Four is not a tie on this file. A hike path that waits until early 2028 needs closer to 140 bp.

Does choosing variable change the stress test at renewal? Not on a stay or a straight switch of the same balance and remaining amortization. Extra money or a longer amortization brings the prescribed test back: 6.44% on today's fixed, 5.49% on the variable.

What does it cost to leave early? Closed variable: three months' interest, about $4,363 on this modeled $500,000 file. Closed fixed: the greater of three months' interest, about $5,550, or IRD. FCAC's published example on a smaller loan shows IRD running four times three months' interest. Those dollar figures are file math, not a government fee schedule.

Should I take the 4.24% three-year instead? It pays $2,696, $55 under the five-year fixed, and saves about $2,900 of interest over 36 months on this file. You reprice in 2029. A shorter fixed, with a second decision attached.

Last verified September 17, 2026. Street rates from ratellow.com/api/rates, lastUpdated September 16. Overnight held at 2.25% on September 2, seventh consecutive hold. Payments use Canadian monthly compounding, semi-annual not in advance. Scenario interest totals are modeled on an adjustable-payment variable.