Key takeaways
- Across 270 five-year windows starting between January 1997 and August 2021, a £200,000 lifetime tracker paid less interest than a five-year fix taken the same month in 210 of them — 78% of start dates.
- The best start month for the tracker was July 2008, worth £40,629 less interest over five years. The worst was August 2021, worth £44,138 more. That is £84,767 between the extremes, on the same loan.
- The median start date left the tracker £6,255 cheaper over five years, but only 25 of the 270 finished within £2,000 either way.
- Inside the worst window the tracker payment ran from £906 to £1,482 — a 63% rise in five years. The fixed payment didn't move.
- In the first quarter of 2026, 92.3% of new UK mortgage lending was at a fixed rate, and 90.7% of outstanding balances. In the third quarter of 2012 only 31.5% of balances were fixed.
The record depended almost entirely on the month you signed
Fix or track? Here's what the Bank of England's own rate series says, before anyone's opinion gets involved.
Take a £200,000 repayment mortgage over a 25-year term. In every month from January 1997 to August 2021, give a borrower two options. The first is that month's average quoted five-year fixed rate at 75% loan-to-value. The second is that month's average quoted lifetime tracker, which then moves point for point with Bank Rate. Run both for 60 months and add up the interest.
The tracker paid less in 210 of the 270 start months — 78% of them. That looks decisive. It isn't, because the size of the gap moved far more than its direction did.
The best start month for the tracker was July 2008: £40,629 less interest over five years. The worst was August 2021: £44,138 more. Same loan, same two products, £84,767 between the extremes. Only 25 of the 270 start months landed within £2,000 either way. The dispersion isn't noise around an answer. It is the answer.
What the test holds fixed, and what it leaves out
Both borrowers take £200,000 over the same term, on capital repayment. The fixed borrower pays the Bank's average quoted five-year fixed rate at 75% loan-to-value in their start month, unchanged for 60 months.
The variable borrower takes a lifetime tracker — a rate set at a fixed margin above Bank Rate for the life of the loan. It's the cleanest variable product to test, because there's no reversion date in it and no lender discretion over the rate. The margin comes from the quoted tracker rate in the start month, and is then held.
In July 2008 the quoted tracker was 6.26% with Bank Rate at 5%, so the margin was 1.26 points. In August 2021 the tracker was 2.59% with Bank Rate at 0.10%, a margin of 2.49 points. The variable payment is recalculated each month on the remaining balance, which is how a tracker behaves in practice.
Left out: product fees, early repayment charges, and the reversion rate a fixed borrower lands on when the deal ends. The last one isn't small. The Bank's revert-to-rate series stood at 6.60% in July 2026, against 4.61% for a new five-year fix.
July 2008 and August 2021 are the two ends of the record
In July 2008 the five-year fix quoted 6.37% and the tracker quoted 6.26% — 11 basis points apart on day one. Then Bank Rate fell. It was cut to 0.50% in March 2009, and it never exceeded 0.75% again until the end of 2021. The fixed borrower paid £1,334 a month for five years. The tracker borrower's payment fell to £830. Over 60 months the tracker paid £40,629 less interest.
August 2021 ran the same film backwards. The five-year fix quoted 1.39%. The tracker quoted 2.59% — 120 basis points behind before anything had happened. Bank Rate was 0.10% at the start and reached 5.25% by August 2023. The fixed payment stayed at £790. The tracker payment started at £906 and peaked at £1,482, and over five years the tracker paid £44,138 more interest.
Notice what a borrower could actually see at the outset. The starting gap said something about the price of certainty. It said nothing about which product would cost less.
Every start date tested, and the size of each gap
The chart plots the same test at a spread of start dates. A negative bar means the tracker paid less interest over the five years. A positive bar means the fix did.
- January 1997: tracker £11,850 cheaper
- January 2000: tracker £15,494 cheaper
- January 2003: fix £6,640 cheaper
- January 2005: tracker £4,654 cheaper
- June 2006: tracker £16,766 cheaper
- July 2008: tracker £40,629 cheaper
- January 2010: tracker £19,146 cheaper
- January 2013: tracker £1,909 cheaper
- January 2016: tracker £2,813 cheaper
- January 2020: fix £21,539 cheaper
- August 2021: fix £44,138 cheaper
Group them by era and what shows up is a rate cycle, not a product. All 60 start months from 2007 to 2011 favoured the tracker, by £21,996 on average. All 30 available start months from 2017 to 2021 favoured the fix, by £21,300 on average. Neither group picked a better product. They picked a starting date, and the rate-setting committee did the rest.
The tracker payment moved £575 a month; the fixed payment didn't move at all
Interest paid is only half the record. The other half is what the monthly payment did while it was being paid.
Across the 270 windows the tracker payment swung by a median of £263 between its low and its high. In the worst window it swung by £575, from £906 to £1,482 — a 63% rise inside five years. The fixed payment didn't move in any window, by construction.
That's the trade being made. A fix converts an unknown payment into a known one for a set number of years, and the price of the conversion is whatever the fixed rate sits above the tracker on day one. Whether it's worth paying is a question about the household, not the market — specifically, how large a payment rise the household could absorb before something breaks. That's risk capacity rather than risk tolerance, and it's measured against income and cash rather than against a rate forecast. The evidence that people weigh a loss more heavily than an equivalent gain is one reason certainty gets bought at a premium.
The Bank's July 2026 Financial Stability Report scales the same trade from the other side. Nearly 750,000 households paying less than 3% interest are projected to roll off fixed deals during 2026, at an average repayment increase of £170 a month. For the typical owner-occupier leaving a fix over the next two years, the projected increase is £45 a month. A fix doesn't remove a rate change. It moves it to a date you already know.
The strongest objection is that both rates were priced by professionals
There's a serious objection to everything above, and it is that hindsight does all the work. A lender quoting a five-year fixed rate in August 2021 was pricing five years of expected Bank Rate off the swap market, then adding a margin for the risk of being wrong. The tracker margin came off the same desk. Neither number was a forecast a household could have improved on.
That objection is correct, and it's why the 78% figure doesn't travel. The tracker won more often across this sample largely because Bank Rate spent most of it falling or pinned to the floor. It was cut to 0.50% in March 2009 and to 0.10% in March 2020. Its high point in between was the 0.75% set in August 2018. A sample dominated by one direction of travel describes the sample, not the products.
The objection cuts the other way as well. Nobody in August 2021 knew Bank Rate would reach 5.25%. Borrowers who fixed at 1.39% bought certainty and collected a windfall on top of it. Reading that as skill would be reading a coin toss as judgement.
What UK borrowers actually chose, and what the stock is paying now
The FCA's mortgage lending data records the choice at the point of sale, quarter by quarter. In the first quarter of 2026, 92.3% of gross mortgage advances were at fixed rates, and 90.7% of outstanding balances. In the first quarter of 2010, with Bank Rate at 0.50%, just 38.1% of new advances were taken at a fixed rate.
So the British mortgage book has gone from mostly variable to almost entirely fixed: 31.5% of balances in the third quarter of 2012, 90.7% in the first quarter of 2026. The stock of fixed borrowers is also paying less than the variable ones right now — 3.93% on outstanding fixed balances against 5.30% on variable balances in the first quarter of 2026. That gap is a legacy of deals struck when rates were lower. It says nothing about a fix taken today.
The same dataset shows 94.7% of gross advances in the first quarter of 2026 carried a rate less than 2% above Bank Rate, the lowest share since the first quarter of 2023. Most UK mortgage pricing sits in a narrow band above the policy rate, whichever product it is sold as.
The limitations of this test
Start with the sample. 270 start months is one path through one country's rate history, and it contains a 13-year stretch in which Bank Rate never exceeded 0.75%. A backtest is not a forecast, and this one is drawn from an era that may not repeat.
The tracker series is incomplete. The Bank publishes no quoted lifetime tracker rate in some months, and only 30 of the 60 start months from 2017 to 2021 carry one. The most recent era is the thinnest part of the sample, and it happens to be the era that most favoured the fix. The series then stops: the last published lifetime tracker quote is 6.21% in March 2025, against Bank Rate of 4.50% that month. This test can't be run on the products being sold today, only on the ones that were sold up to then.
Both series are averages of rates quoted to new borrowers, not offers made to any individual. A borrower with a larger deposit or a smaller loan would see different numbers. Product fees and early repayment charges are real costs, and neither is captured here.
The test also measures interest over exactly 60 months. Someone who moved house, overpaid or remortgaged early would have a different record, and the early repayment charge is precisely the mechanism that makes a five-year fix expensive to leave.
What would change the conclusion
If Bank Rate settled into a narrow range for a decade, the dispersion above would collapse, and the choice would shrink to the starting gap between the two quoted rates. Most of the £84,767 range in this sample comes from two large, fast moves in opposite directions.
If the fixed-rate premium vanished — if a fix and a tracker were quoted at the same rate on the same day — a fix would be free optionality and the argument would be over. The closest this sample came was July 2008, at 11 basis points, and Bank Rate then fell far enough to hand that window to the tracker anyway.
What the record can't do is pick a rate type. It can say how wide the range has been: £84,767 on a £200,000 loan, driven by a Bank Rate that moved 5.15 points between March 2020 and August 2023. So the number worth knowing isn't the rate on offer. It's the payment a household would face if a move like that happened again, and whether that payment is survivable next to a buffer sized to income volatility. LedgerTouch tracks liabilities beside assets, so a mortgage sits on the same balance sheet as everything else.