Quick answer: Partly, but mostly not Canada’s own borrowing. The Bank of Canada has held its rate at 2.25% since October 2025, yet the 5-year Government of Canada bond that prices fixed mortgages rose from 2.67% on February 27, 2026 to 3.69% on September 23, and was still 3.50% on October 22. Most of that 2026 jump is an oil-and-inflation spike: markets now expect Bank of Canada hikes. Underneath it, the floor has moved up since 2023 because investors want extra pay to hold long-term bonds while governments worldwide, led by Washington’s $1.99 trillion deficit, borrow at record scale. The spike can come off. The floor won’t come off fast. AI borrowing is real, but there’s no measured evidence yet that it’s pushing rates up.
Renewing or buying in the next year? Pick a time below for a 20-minute call. I’ll run your numbers, fixed and variable, at today’s rates.
Calendar not loading? Book here instead.
Why did my fixed rate go up when the Bank of Canada didn’t move?
Because the Bank of Canada doesn’t set your fixed rate. It sets the overnight rate, and that moves prime, variable mortgages and HELOCs. A 5-year fixed is priced off the 5-year Government of Canada bond yield, plus a lender spread. That bond trades every day against every other government bond in the world.
This year shows the split as clearly as I’ve seen it. The Bank of Canada held at 2.25% at every decision in 2026. The 5-year GoC yield climbed about a full point to its September peak. On the lender sheets I quote from, the best uninsured 5-year fixed was 5.29% on October 19; by October 21, with the bond easing, it was 4.74% uninsured and 4.64% insured, roughly 1.1 to 1.3 points over the bond. I covered the mechanics in why fixed rates rise while the Bank of Canada is on hold. This piece asks the next question: who is doing the borrowing that pushes bond yields up, and how much of it is governments versus AI?
| Month | BoC policy rate | 5-yr GoC bond | US 10-yr | Uninsured 5-yr fixed |
|---|---|---|---|---|
| 2019-01 | 1.75% | 1.88% | 2.71% | — |
| 2019-07 | 1.75% | 1.46% | 2.06% | — |
| 2020-01 | 1.75% | 1.50% | 1.76% | — |
| 2020-07 | 0.25% | 0.35% | 0.62% | 2.33% |
| 2021-01 | 0.25% | 0.43% | 1.08% | 1.96% |
| 2021-07 | 0.25% | 0.86% | 1.32% | 2.34% |
| 2022-01 | 0.25% | 1.58% | 1.76% | — |
| 2022-07 | 2.07% | 3.00% | 2.90% | 5.34% |
| 2023-01 | 4.30% | 3.05% | 3.53% | 5.24% |
| 2023-07 | 4.90% | 3.85% | 3.90% | 5.99% |
| 2024-01 | 5.00% | 3.42% | 4.06% | 5.75% |
| 2024-07 | 4.70% | 3.37% | 4.25% | 5.07% |
| 2025-01 | 3.23% | 3.00% | 4.63% | 4.81% |
| 2025-07 | 2.75% | 3.03% | 4.39% | 4.52% |
| 2026-01 | 2.25% | 2.93% | 4.21% | 4.59% |
| 2026-07 | 2.25% | 3.16% | 4.60% | 4.81% |
| 2026-10 | 2.25% | 3.58% | 5.20% | 5.00% |
What’s the popular explanation?
The popular take: governments are borrowing like never before, so bond yields have to rise, and your mortgage pays for it. A newer version adds AI: tech companies are borrowing hundreds of billions for data centres, competing with governments for the same savings, and that drives rates up too.
It’s a good story, and part of it is true. But it lumps together forces that move at very different speeds, and it blames the wrong government.
A bond yield has two parts. One is what investors expect the central bank’s rate to average over the life of the bond. The other is the term premium: extra pay for tying money up for years instead of rolling it over in short-term bills. Government borrowing works mainly through the second part. Oil, inflation and rate hikes work mainly through the first. Separate them and the picture changes.
The spike: how much of 2026 is oil and expected hikes?
Most of it. The 5-year GoC bottomed at 2.67% on February 27, 2026, in the flight to safety when the Iran war began. Oil then went above US$100 a barrel. Canadian CPI ran at 3.1% in September, mostly on gasoline. Markets now price about 100 basis points of Bank of Canada hikes over the next 15 months, according to Capital Economics’ October 16 Canada Housing Watch. A bond that expects higher policy rates pays a higher yield, and your fixed rate follows.
The New York Fed publishes a model (ACM) that splits the US 10-year Treasury yield into those two parts. From December 31, 2025 to September 30, 2026, the 10-year rose about one point, from 4.24% to 5.26%. By that model, about 0.87 of it came from higher expected short-term rates and about 0.14 from term premium. The Fed Board’s other model (Kim-Wright) gives term premium closer to half. Either way, government debt did not cause most of this year’s move.
The Canadian 2-year bond tells the same story. It barely reacts to debt supply and mostly tracks the expected Bank of Canada path. It went from about 2.5% in February to 3.22% on October 22. When the short end jumps like that, the market is pricing hikes, not deficits. I wrote about how gas prices put hikes back on the table in September.
The floor: how much does government borrowing actually matter?
This is where the popular take gets something right. The term premium was negative through much of 2020 to 2023: investors were effectively paying for the safety of long bonds, helped by central banks buying them. The ACM estimate went from −0.77% in June 2023 to +0.89% at the end of September 2026, its highest month-end since 2014. That shift alone added more than a point and a half to the US 10-year yield, independent of anything the Fed or the Bank of Canada did.
The Bank of Canada’s own researchers say the Canadian term premium “reached its highest level in over a decade,” and that investors “appear to be seeking compensation for purchasing large amounts of government debt,” with central banks no longer soaking it up. They put the correlation between the Canadian and US term premiums at about 0.92. (Bank of Canada, Sparks at Bank 2026-9, March 2026.)
| Month | Term premium |
|---|---|
| 2019-01 | 0.05% |
| 2019-07 | -0.24% |
| 2020-01 | -0.27% |
| 2020-07 | -0.61% |
| 2021-01 | -0.41% |
| 2021-07 | -0.23% |
| 2022-01 | 0.07% |
| 2022-07 | 0.23% |
| 2023-01 | 0.05% |
| 2023-07 | 0.11% |
| 2024-01 | 0.19% |
| 2024-07 | 0.29% |
| 2025-01 | 0.72% |
| 2025-07 | 0.53% |
| 2026-01 | 0.60% |
| 2026-07 | 0.79% |
| 2026-10 | 1.06% |
So the floor is real. But whose borrowing is it? Look at the scale:
| Borrower | Amount | Context |
|---|---|---|
| US federal government | $1.992 trillion deficit (FY2026) | About 6% of GDP. Net interest alone: $1.052 trillion. US Treasury |
| Big Tech (AI build-out) | ~$279 billion in bonds (2026 forecast) | Up from $109B in 2025. J.P. Morgan Asset Management |
| Government of Canada | $65.3 billion deficit (2026-27) | 1.9% of GDP; federal debt about 41% of GDP and flat. Spring Economic Update 2026 |
| Government of Canada 5-year bonds | $80 billion planned issuance (2026-27) | The bond that prices your fixed rate. Part of $298B in gross bond issuance. Debt Management Strategy |
Ottawa’s deficit is real money, but it barely moves our bond yields. The most-cited academic estimate (Laubach, Federal Reserve) puts the effect at about 3 to 4 basis points of long-term yield for each one-point rise in projected debt-to-GDP. Canada’s federal debt ratio rises about 0.4 points this year. That works out to a basis point or two.
What Canada can’t avoid is importing other countries’ term premium. On my own calculation using Bank of Canada and Federal Reserve data from January 2023 to October 2026, the 5-year GoC moves about 85% as much as the US 10-year on a typical day. Our bonds still yield less than theirs (3.64% on the 5-year against 4.94% on the US 5-year), so Canada’s better books do show up. But when global investors demand more to hold government debt, Canadian borrowers pay a share of it whether Ottawa borrows more or not.
Is AI spending pushing mortgage rates up?
Not measurably, yet. The spending is enormous. Microsoft, Alphabet, Amazon and Meta are on track for roughly $700 billion of capital spending in 2026, up from about $410 billion in 2025, much of it on data centres, chips and power. Alphabet alone guided to $195 to $205 billion in July. A growing share is borrowed: J.P. Morgan Asset Management counted $194 billion of hyperscaler bond issuance in the first half of 2026 alone.
The case that this raises rates is reasonable. A capital-hungry build-out competes for the same savings governments borrow from, and if AI lifts productivity, the return on capital (and the neutral rate) rises with it. In September, The Economist listed the AI build-out, competing with governments for the same capital, as one reason long-term yields stay high.
But the one study that tests it directly found the opposite so far. Jens Christensen of the San Francisco Fed and Glenn Rudebusch of Brookings looked at what moved market-based estimates of the neutral rate. Fiscal news gave “only a modest upward lift,” and news of major generative-AI model releases “is associated with an overall decline in measures of r*.” (Brookings, August 18, 2026.) And in scale, $279 billion of AI bonds is about a seventh of one year of US federal borrowing.
My read: AI could push long-term rates up over the next several years, especially if the build-out keeps growing at this pace and runs into inflation from chips, power and construction. Saying it’s doing that now goes further than the data does.
Why won’t fixed mortgage rates fall fast?
Because the two pieces come off at different speeds. The spike depends on oil and on whether the Bank of Canada actually delivers the hikes markets have priced. Capital Economics thinks it won’t: its call is that the 100 basis points priced in is “too aggressive,” with two hikes to 2.75% by mid-2027 and mortgage rates dropping back next year. If they’re right, some of this year’s increase reverses.
The floor is a different story. The US is projected to keep running deficits near $2 trillion a year, the Congressional Budget Office has US debt rising from about 100% of GDP to 120% by 2036, and central banks are no longer buyers of last resort. None of that turns around in a year. Here’s what the floor looks like in plain numbers:
| Period | BoC policy rate | 5-year GoC yield | Bond minus policy rate |
|---|---|---|---|
| 2019 average | 1.75% | 1.54% | −0.21 pts |
| 2021 average | 0.25% | 0.94% | +0.69 pts |
| 2024 average | 4.55% | 3.29% | −1.26 pts |
| October 22, 2026 | 2.25% | 3.50% | +1.25 pts |
In 2019 the Bank of Canada sat at 1.75% and the 5-year bond yielded less than that. Today the Bank is only half a point higher, and the bond is about two points higher. Part of that gap is expected hikes, which can fade. The rest is the term premium, and it isn’t going back below zero while the world’s largest borrower is adding $2 trillion a year.
What does this mean for your mortgage?
Don’t plan around 2021 rates. Sub-2% 5-year fixed rates needed a negative term premium and an emergency-level policy rate at the same time. Neither is coming back on a renewal timeline.
Expect some relief, not a reset. If oil cools and the Bank of Canada hikes less than markets expect, the spike part of fixed rates can come down through 2027. That’s a forecast, not a promise, and Governor Macklem’s recent hawkish comments are a reminder that it can go the other way first. You can see current pricing on my mortgage rates page.
Use a rate hold. Most lenders will hold a rate for 90 to 120 days at no cost. If rates fall before you close, you take the lower one. If you’re renewing in the next year, that free option is worth having now.
Fixed versus variable depends on your file. Variable is priced off the Bank of Canada, which hasn’t moved, so it’s currently well below fixed. Whether that’s the right trade depends on whether you can carry a hike or two. That’s a conversation about your numbers, not the bond market’s.
Renewing or buying in the next year? I’ll show you your options side by side, fixed and variable, with today’s numbers.
What’s genuinely uncertain?
Term premium can’t be observed. It’s a model estimate, and the two Fed models disagree by up to half a point on how much of 2026 it explains. Capital Economics reads the first half of the year as more term-premium-driven than the New York Fed’s monthly figures show. I’ve used the more conservative split.
The numbers are about to move. The Bank of Canada decides on October 28. Alphabet, Microsoft, Meta and Amazon report quarterly results between October 21 and 29 and may raise their capex guidance again. The federal budget lands November 5 with new deficit and borrowing figures. I’ll update this piece after the budget.
AI could cut either way. A capital-hungry build-out argues for higher rates. A productivity boom that lowers inflation argues for lower ones. Fed officials have said the payoff could take five to ten years. Nobody knows which force wins on a 5-year mortgage horizon.
Oil is the swing factor. A quick end to the oil shock would take the spike down faster than any forecaster expects. A longer one would keep the Bank of Canada leaning toward hikes.
Frequently asked
Does government borrowing raise mortgage rates in Canada?
Partly, through the term premium: the extra yield investors demand to hold long-term bonds when governments borrow heavily. The Bank of Canada says the Canadian term premium is at its highest in over a decade and moves closely with the US one. But Canada's own federal deficit, about $65 billion or 1.9% of GDP in 2026-27, moves Canadian bond yields by only a basis point or two. Most of the pressure is imported from US borrowing, which ran a $1.99 trillion deficit in fiscal 2026.
Why did fixed mortgage rates go up in 2026 when the Bank of Canada didn’t raise rates?
Fixed rates are priced off the 5-year Government of Canada bond, not the Bank of Canada's policy rate. That bond rose from 2.67% on February 27, 2026 to a peak of 3.69% on September 23, mostly because higher oil prices and inflation led markets to expect Bank of Canada rate hikes. The Bank itself held at 2.25% all year.
Is AI spending pushing interest rates up?
Not measurably yet. Microsoft, Alphabet, Amazon and Meta are on track for roughly $700 billion of capital spending in 2026, and AI-related bond issuance is forecast near $279 billion. But that is about a seventh of US federal borrowing, and an August 2026 study by the San Francisco Fed's Jens Christensen and Brookings' Glenn Rudebusch found news of major AI model releases associated with lower, not higher, estimates of the neutral interest rate. AI could push rates up later; the evidence doesn't show it now.
What is the term premium?
It's the extra yield investors want for locking money into a long-term bond instead of rolling over short-term bills. It can't be observed directly and is estimated with models. The New York Fed's ACM estimate for the US 10-year went from −0.77% in June 2023 to +0.89% in September 2026, its highest month-end since 2014.
Will fixed mortgage rates go down in 2027?
Some relief is possible. Markets price about 100 basis points of Bank of Canada hikes over the next 15 months; Capital Economics expects fewer and sees mortgage rates dropping back in 2027. That would reverse part of the 2026 spike. The higher term premium is likely to stay, so a return to 2019 or 2021 fixed rates isn't a realistic planning assumption.
Why do US bond yields affect Canadian mortgage rates?
Government bonds of similar quality trade against each other globally. On a typical day from 2023 to 2026, the 5-year Government of Canada yield moved about 85% as much as the US 10-year. When investors demand more to hold US government debt, part of that reaches Canadian bonds and then Canadian fixed mortgage rates.
Should I choose fixed or variable right now?
It depends on your file. Variable rates follow the Bank of Canada, which has held at 2.25%, so variable is currently about a point below fixed. Fixed protects you if the Bank hikes as markets expect. A rate hold of 90 to 120 days costs nothing and lets you take a lower rate if fixed rates fall before you close.
This is commentary, not personalized mortgage or investment advice. Figures are as of the dates cited: Bank of Canada policy rate 2.25% (unchanged since October 29, 2025); Government of Canada yields as of October 22, 2026 (Bank of Canada Valet); US 10-year Treasury as of October 21, 2026 (Federal Reserve H.15 via FRED); best available 5-year mortgage rates as of October 21, 2026, from lender sources tracked by Mortgages for Less. Term-premium figures are model estimates from the Federal Reserve Bank of New York (ACM) and the Federal Reserve Board (Kim-Wright). Your rate and eligibility depend on your file. Josh Tagg is a licensed mortgage broker with Mortgages for Less (Alberta) and Indi Mortgage (Ontario).
