Every quarter, Canada’s major banks publish interest rate forecasts. Financial media covers every prediction. And Canadians trying to decide between fixed and variable mortgages hang on every word. But how are these forecasts made, how reliable are they, and how should you actually use them?
Before looking at forecasting methodology, here are today’s actual posted mortgage rates for reference:
| Term | Posted Rate |
|---|---|
| 1-Year | 5.49% |
| 3-Year | 6.05% |
| 5-Year | 6.09% |
Posted/benchmark conventional mortgage rates used for stress-test purposes – not the discounted rate lenders typically offer. Always confirm current offers directly with a lender or mortgage broker. Source: Bank of Canada Valet API (series V80691333, V80691334, V80691335), as of August 26, 2026, fetched 2026-08-28.
How rate forecasts are made
The forecasting process
Professional rate forecasters (bank economists, central bank staff, independent researchers) use a combination of:
| Input | What It Tells Them |
|---|---|
| Bank of Canada forward guidance | What the BoC says it expects to do |
| Inflation data and projections | Whether the BoC will need to raise, cut, or hold |
| GDP growth forecasts | Whether the economy is strong enough to handle current rates |
| Employment data | Labour market strength → inflation → rate pressure |
| Bond market pricing | What financial markets have already priced in |
| Overnight Index Swap (OIS) rates | Market-implied probability of future BoC moves |
| US Fed policy | Constrains BoC room to cut or raise |
| Global conditions | Trade, geopolitical risk, commodity prices |
| Housing market indicators | Prices, sales, inventory, construction |
| Fiscal policy | Government spending and deficit plans |
The two main approaches
| Approach | Method | Used By |
|---|---|---|
| Top-down macro models | Econometric models linking GDP, inflation, and rates | Bank of Canada, Big 5 economics teams |
| Market-implied pricing | Derive expected rates from OIS, bond futures, and forward rates | Bond traders, mortgage industry |
Market pricing is often more accurate in the short term (1–6 months) because it aggregates millions of investor opinions. Macro models are used for longer horizons but become unreliable beyond 12 months.
How to read a bank rate forecast table
Here’s a typical format you’ll see from a Big 5 bank economics department:
| Quarter | BoC Overnight Rate | Prime Rate | 5-Year Bond Yield | 5-Year Fixed Rate |
|---|---|---|---|---|
| Q1 2026 (actual) | 2.75% | 4.95% | 2.80% | 4.30% |
| Q2 2026 (forecast) | 2.50% | 4.70% | 2.70% | 4.20% |
| Q3 2026 (forecast) | 2.50% | 4.70% | 2.65% | 4.15% |
| Q4 2026 (forecast) | 2.50% | 4.70% | 2.60% | 4.10% |
| Q1 2027 (forecast) | 2.50% | 4.70% | 2.55% | 4.05% |
This table is a hypothetical illustration of forecast-table format, not live data — see the posted-rate table above for actual current figures.
How to interpret this
| Column | What It Tells You | How to Use It |
|---|---|---|
| BoC overnight rate | Expected policy rate path | Tells you direction for variable rates |
| Prime rate | Expected base for variable lending | Your variable rate = prime +/- your discount |
| 5-year bond yield | Expected direction for fixed rates | Fixed rates typically = bond yield + 1.5%–2.0% |
| 5-year fixed rate | Expected mortgage rate for new fixed mortgages | Compare to today’s rate to decide lock-in timing |
Key things to notice
- How much do the forecasts change quarter to quarter? Small changes suggest confidence. Large revisions suggest uncertainty.
- Is the direction consistent across banks? If all 5 banks forecast cuts, the direction is likely correct (even if the magnitude isn’t).
- How far out do they forecast? Beyond 4 quarters, treat numbers as directional guesses at best.
The accuracy problem
Rate forecasts are consistently wrong
| Forecast Period | What Was Predicted | What Actually Happened |
|---|---|---|
| Early 2020 | Rates to hold steady at 1.75% | COVID: rates crashed to 0.25% in weeks |
| Late 2020 | Rates to stay low for years | Correct — but no one predicted the 2022 surge |
| Early 2022 | BoC to raise rates “gradually” to 2%–2.5% | BoC hiked to 5.00% — more than double the prediction |
| Early 2023 | Rate cuts by late 2023 | BoC hiked AGAIN (to 5.00%) — cuts didn’t start until June 2024 |
| Early 2024 | 3–4 cuts in 2024 | Actually delivered 5+ cuts — faster than forecast |
Why forecasts fail
| Reason | Explanation |
|---|---|
| Unpredictable shocks | Pandemics, wars, trade disruptions, and financial crises are not in models |
| Herding bias | Forecasters cluster around consensus to avoid being the outlier — even when outlier risks are high |
| Anchoring | Forecasts tend to be small adjustments from current rates, missing big moves |
| Political and fiscal surprises | Government policy changes (tariffs, housing rules, spending) are hard to predict |
| Feedback loops | Rate changes themselves affect the economy, which changes the future rate path |
| Global contagion | US Fed surprises, European crises, or Chinese slowdowns ripple into Canadian rates unpredictably |
Market-implied rate expectations
An alternative to bank forecasts is market pricing — derived from financial instruments that trade based on expected future rates.
Overnight Index Swaps (OIS)
OIS contracts price in the market’s expectation for the Bank of Canada overnight rate at specific future dates. These are updated continuously and represent the collective view of thousands of professional traders.
| Advantage | Limitation |
|---|---|
| Real-time, constantly updated | Reflects the most likely path, not the full range of outcomes |
| Aggregates many opinions | Can shift dramatically on new data |
| Good predictor for 1–3 months | Less reliable beyond 6 months |
| Captures risk pricing | Not directly available to retail consumers |
How to check market expectations
- WealthNorth rate updates — we translate market pricing into plain language
- Bloomberg or Refinitiv — OIS rates and forward curves (professional terminals)
- CME Group — some Canadian rate futures (limited)
- Major bank research notes — often include OIS-implied rate paths
How to use rate forecasts in your mortgage decisions
Rule 1: Use direction, not specific numbers
If all major banks forecast the BoC overnight rate declining by year-end, the direction is probably correct — variable rates will likely fall. But don’t count on a specific number (e.g., exactly 2.25% by Q4).
Rule 2: Consider the range of outcomes, not just the base case
| Scenario | Probability (typical) | What It Means |
|---|---|---|
| Base case (consensus forecast) | ~50%–60% | The most likely path |
| Upside scenario (rates drop more) | ~15%–20% | Economy weakens, BoC cuts faster |
| Downside scenario (rates rise) | ~15%–20% | Inflation rebounds, BoC holds or hikes |
| Tail risk (extreme move) | ~5%–10% | Recession → emergency cuts, or crisis → rate spike |
Rule 3: Make the decision that works in most scenarios
| Strategy | Works If Rates… | Fails If Rates… |
|---|---|---|
| Lock in fixed now | Rise or stay stable | Drop significantly (opportunity cost) |
| Choose variable, expect cuts | Fall as forecast | Rise unexpectedly (payment shock) |
| Short-term fixed (3-year) | Are lower at renewal | Are higher at renewal |
| Split mortgage (half fixed, half variable) | Move in either direction | Neither — moderate outcome in all cases |
Rule 4: Never make a rate bet you can’t afford to lose
If your budget can only handle payments at the forecasted future rate and not at a higher rate, you’re speculating — not planning. Always stress-test your budget for rates 1%–2% higher than the forecast.
Rate forecast decision framework
| Your Situation | Recommended Approach |
|---|---|
| Tight budget, can’t absorb payment increases | Fixed rate — certainty trumps potential savings |
| Comfortable budget, can handle +1%–2% | Variable — benefit from expected cuts, absorb risk |
| Uncertain timeline (may sell in 2–3 years) | Short-term fixed or variable — avoid long commitments |
| Renewing from ultra-low rate (2020–2021) | Fixed — budget for higher payment, lock in stability |
| Renewing from high rate (2022–2023) | Variable — may benefit from normalization, already accustomed to high payments |