Settled by the data
Fixed vs variable, year by year
The fixed-or-variable question is usually argued with today's rates and a guess about tomorrow. Here it's settled with what actually happened. For each year a 5-year term was signed, we take what that year's borrowers really got, on both sides, and follow the Bank of Canada prime rate that actually came next, to the dollar of interest each paid over the full term on a $500,000 mortgage. Across the 15 completed cohorts, variable came out ahead in 11.
| Signed | Fixed rate | Variable start | Fixed paid | Variable paid | Winner |
|---|---|---|---|---|---|
| 2007 | 5.19% | 5.32% (P−0.93) | $121,588 | $59,922 | Variable by $61,666 * |
| 2008 | 5.24% | 4.38% (P−0.37) | $122,792 | $59,338 | Variable by $63,454 |
| 2009 | 3.90% | 2.55% (P+0.30) | $90,692 | $72,979 | Variable by $17,713 |
| 2010 | 3.80% | 1.90% (P−0.85) | $88,311 | $49,076 | Variable by $39,235 |
| 2011 | 3.64% | 2.25% (P−0.75) | $84,506 | $50,340 | Variable by $34,165 |
| 2012 | 2.99% | 2.65% (P−0.35) | $69,111 | $58,424 | Variable by $10,686 |
| 2013 | 2.99% | 2.58% (P−0.42) | $69,111 | $57,848 | Variable by $11,263 |
| 2014 | 2.94% | 2.35% (P−0.65) | $67,931 | $56,164 | Variable by $11,767 |
| 2015 | 2.59% | 2.05% (P−0.65) | $59,692 | $58,830 | Variable by $863 |
| 2016 | 2.39% | 2.15% (P−0.55) | $55,000 | $60,631 | Fixed by $5,630 |
| 2017 | 2.69% | 2.08% (P−0.87) | $62,043 | $53,487 | Variable by $8,556 |
| 2018 | 3.19% | 2.46% (P−1.24) | $73,836 | $56,640 | Variable by $17,196 |
| 2019 | 2.69% | 2.85% (P−1.10) | $62,043 | $74,161 | Fixed by $12,118 |
| 2020 | 1.97% | 1.75% (P−0.70) | $45,188 | $94,206 | Fixed by $49,019 |
| 2021 | 1.79% | 1.25% (P−1.20) | $40,999 | $93,901 | Fixed by $52,901 |
Interest paid over the completed 5-year term on a $500,000 mortgage, 25-year amortization. The variable path is simulated as an adjustable-payment mortgage: each month's rate is the real Bank of Canada prime minus that cohort's discount, with the payment recalculated as prime moves. Rates are achievable medians from shared offers, not a market average. A * marks any cohort whose variable sample is thin (fewer than 25 offers), where the year's result leans on fewer data points. This is history, not a prediction; the next five years are their own question. As of 2026-07-17.