Fixed rates jumped. The variable discount is now 79 points.

The five-year Government of Canada bond yield finished last week at 3.65%. Four days earlier it was 3.44%.

Twenty-one basis points in four sessions. The two-year did the same, 3.13% to 3.35%. When I wrote about this bond in August it sat at 3.36%, three basis points below its twelve-month high, and the question was whether it would break through. It broke through and kept going.

Then Monday the August inflation number landed and gave the bond market no reason to back off.

So here is where an Ottawa buyer actually stands this week.

The pass-through was bigger than the published rates admit

Several major banks raised fixed rates by ten to twenty basis points last week. That understates what borrowers see.

Ron Butler told Canadian Mortgage Trends the published increase is only half of it, because lenders pulled discretionary discounts at the same time. Twenty points on the posted rate plus forty points of vanished discretion is a very different quote than the one you got in July. Clinton Wilkins put the real range at twenty basis points to almost a hundred, and called the pricing all over the place.

On the August Ottawa benchmark of $637,700 with twenty percent down, a $510,160 mortgage, twenty basis points costs $55 a month and about $3,300 over a five year term. Forty points is $110 a month. It also shrinks what you qualify for, which I worked through in August.

The gap between fixed and variable is now 79 basis points

This is the part worth sitting with.

Monday morning, the best five-year fixed on Ratehub's board was 4.09%. The best five-year variable was 3.30%, which is prime minus 1.15.

Seventy-nine basis points. On that same $510,160 mortgage the fixed payment is $2,708 a month and the variable is $2,494. The buyer who takes variable keeps $215 a month, call it $2,580 a year.

The Bank of Canada has not moved. It held at 2.25% on September 2 and prime is still 4.45%. Every bit of this gap opened in the bond market, which the Bank does not set. I explained that mechanism in July.

Three hikes before that trade goes wrong, four before it hurts

Run it forward.

One quarter-point hike puts the variable at 3.55%. Still $148 a month cheaper than the fixed. Two hikes, 3.80%, still $80 cheaper. Three hikes, 4.05%, still eleven dollars cheaper. Four hikes, 4.30%, and now it is $59 a month worse.

So the variable borrower does not lose this trade until the Bank of Canada raises four separate times.

Dave Larock told CMT the bond market is currently pricing about four hikes over the next twelve months. He then added the sentence everybody skipped past: there is not much conviction behind that pricing, and it has been as volatile as everything else lately.

The part that surprised me: even if all four land, one a quarter, the variable borrower pays less every month on the way up. Twelve months comes to roughly $31,140 on variable against $32,502 on fixed. The variable buyer finishes about $1,360 ahead in the year the bond market turns out to be completely right.

That is not a small case. It is also not free.

Now read what is actually inside the inflation number

Headline CPI held at 3.0% in August. That is the print that kept the curve where it is.

Take gasoline out and it is 2.4%. The Bank's own core measures, median and trim, averaged 2.0%, which is the target. Shelter inflation was 1.5%. Grocery inflation was 2.8%, below the headline for the first time since July 2024.

What was up? Gasoline 22.8% year over year. Travel tours 26.1%, and Statistics Canada attributes most of that to a base-year effect from the collapse in Canadian travel to the US in 2025. Fuel oil 43.7%.

This is an energy shock wearing an inflation costume. The Bank said so on September 2, describing CPI as hovering near 3% "mainly because of persistently higher gasoline prices" with little evidence of spillover. The test it set itself was whether energy leaks into everything else. In July, inflation excluding gasoline was 2.2%. In August, 2.4%.

That is some leakage. It is not a lot. The Ontario picture has been running cooler than the national one all year, and nothing in this release changed that.

The risk almost nobody prices properly

Larock's warning is the one I would repeat to any Ottawa buyer this week. Locking a five-year fixed right after a spike caused by a one-off event means you own that rate until 2031. If the Strait of Hormuz reopens and oil normalizes, the bond market unwinds and you are still paying the war premium four years from now.

It cuts both ways and I will not pretend I know which way. What I can tell you is that there is a geopolitical premium baked into today's fixed rate, and five years is a long time to hold somebody else's crisis.

Which is why the shorter terms matter and almost nobody asks about them. Ratehub had the two-year fixed at 3.89% and the three-year at 3.94% on Monday, both cheaper than the five-year. If you want certainty without signing up until 2031, that is where it lives.

Meanwhile, in Ottawa

None of this is a reason to hurry.

Ottawa finished August with 4,496 active listings, the highest August count since 2016. The composite benchmark was $637,700, up 1.0% from a year ago and 0.6% from July. Roughly thirteen hundred listings left the market in August without selling at all.

The rate moved against buyers this month. The market did not. More homes for sale than a year ago and prices going essentially nowhere. A bond chart is not a reason to overpay for a house.

If anything the useful move is boring. Get the pre-approval and take the rate hold. It is free optionality: if rates rise you keep the old one, if they fall you re-price. That was worth close to nothing in a market pricing cuts. It is worth real money in a market pricing hikes.

What I would actually do

Stop waiting for the cut. I said this in August and I believe it more now. The curve is not pricing one, and on September 2 the Governor said he is prepared to go the other way if inflation stays high.

Ask your broker what the three-year costs before you sign a five.

And if you have the income buffer and the temperament for it, a 79 basis point discount is a real argument for variable, for the first time in about two years.

One thing to be clear about: I sell houses, not mortgages. Every figure above is public and checkable, but the rate you personally get depends on your file, your lender and your down payment. Take this to a mortgage broker before you act on it.

If you want to talk about whether the house makes sense at all, which is the part I am useful for, I am at 613-262-6545 or fil@613realtor.ca. Or go look at what is sitting unsold and we can work backwards from there.