The Slow Fuse: Mapping Britain’s Mortgage Refinancing Wall, 2026 to 2029
Download PDF
Executive Summary
Monetary policy reaches British households mainly through mortgages, and in Britain most mortgages are now fixed for two or five years. A change in Bank Rate therefore takes years to work through: it affects the rate on new deals almost at once, but the rate on the stock of existing mortgages only as old deals expire. This paper measures that delay with the latest data and asks what the rate increases of 2026 will cost households over the next three years.
- The stock of mortgages reprices at about 2.7 per cent a month, or 28 per cent a year, toward the rate on new lending. The half-life of a gap between new and outstanding rates is about 25 months. The direct pass-through of Bank Rate to the stock, through variable-rate and tracker loans, has fallen from 0.38 before 2016 to about 0.09 since, as fixed-rate lending has come to dominate.
- In September 2026 the effective rate on new mortgages was 4.65 per cent, and quoted two-year fixed rates reached 5.39 per cent in October after the war-driven rise in gilt yields. The stock paid 4.04 per cent. Closing that gap on 1.56 trillion pounds of mortgages would add about 13 billion pounds a year to household interest payments, and the process is only half complete.
- If new rates stay near 4.9 per cent and Bank Rate at 4 per cent, the effective rate on outstanding mortgages rises to 4.4 per cent by end-2027 and 4.6 per cent by end-2029, adding about 5.3 billion pounds a year to interest payments by end-2027 and 9.2 billion by end-2029. A second hike, with new rates at 5.3 per cent, raises those figures to 8.2 and 13.7 billion. A ceasefire that brings new rates back to 4.2 per cent limits them to about 1 to 2 billion.
- In aggregate the burden is manageable. Mortgage interest absorbed 2.9 per cent of household disposable income in mid-2026, against 5.0 per cent at the 2008 peak, and the central projection takes it to about 3.3 to 3.4 per cent.
- The shock is concentrated. A borrower who fixed a 150,000 pound repayment mortgage for five years in 2021 at about 1.6 per cent pays around 607 pounds a month; refixing at 5 per cent raises the payment to about 841 pounds, an increase of nearly 40 per cent. Borrowers on two-year deals taken in late 2024 face increases of only 2 to 3 per cent. The refinancing wall of 2026 and 2027 is a wall for the pandemic cohort, not for the mortgage market as a whole.
The policy implication follows. The November hike will have its full effect on households only in 2028 and 2029, long after the energy shock that motivated it is likely to have faded. A second hike would add a durable 3 to 4.5 billion pounds a year to household interest bills, concentrated on the cohort already facing the sharpest payment shock.
1. A Policy That Acts With a Delay
Twenty years ago, most British mortgages paid a variable rate that moved with Bank Rate within weeks. Today, according to the effective rate data published by the Bank of England, the large majority of new lending is on fixed rates, mostly for two or five years. This structure has made households less exposed to sudden rate rises, but it has also made the transmission of monetary policy slow and uneven. The rate increases of 2022 and 2023 were still feeding through to household budgets in 2026, three years later, and the hike of November 2026 will not be fully felt until the end of the decade.

Figure 1 shows the gap that results. Quoted fixed rates rose sharply in 2022 and 2023, fell back in 2024 and 2025, and rose again in 2026 as gilt yields climbed with the energy shock: the two-year fixed rate at 75 per cent loan-to-value reached 5.39 per cent in October 2026, the five-year rate 5.19 per cent. The effective rate on new mortgages, which reflects the deals actually taken out, was 4.65 per cent in September. The effective rate on all outstanding mortgages, which is what households actually pay, has risen far more slowly: from 2.0 per cent at the end of 2021 to 4.04 per cent in September 2026.

The difference between the two, the repricing gap in Figure 2, measures the increase still to come for the average mortgage holder. It peaked at about 2.1 points in November 2023, narrowed to around 0.3 points in 2025 as new rates fell and old cheap deals expired, and widened again to 0.6 points in September 2026. On the October quoted rates the gap is closer to 1 point.
Table 1. Data used in this paper
| Series | Bank of England code | Span |
|---|---|---|
| Effective rate, outstanding loans secured on dwellings | CFMHSDE | Monthly, 2004 to Sep 2026 |
| Effective rate, new mortgages (all; fixed) | CFMBJ95; CFMBJ96 | Monthly, 2004 to Sep 2026 |
| Quoted 2-year and 5-year fixed rates, 75% LTV; 2-year 90% LTV | IUMBV34; IUMBV42; IUMB482 | Monthly, 1995 to Oct 2026 |
| Standard variable rate | IUMTLMV | Monthly, 1995 to Oct 2026 |
| Secured lending outstanding | LPMB3TA | Monthly, to Sep 2026 |
| Approvals for house purchase and remortgaging | LPMVTVX; LPMB4B3 | Monthly, to Sep 2026 |
| Household gross disposable income, saving ratio | ONS QWND; NRJS | Quarterly, to 2026 Q2 |
Note: all effective rates are not seasonally adjusted weighted averages for UK monetary financial institutions.
2. How Fast Does the Stock Reprice?
To measure the speed of transmission, the paper estimates a partial adjustment model on monthly data from 2005 to September 2026. Each month, the change in the effective rate on outstanding mortgages is regressed on the gap between the rate on new mortgages and the outstanding rate in the previous month, and on the change in Bank Rate. The coefficient on the gap measures the share of the stock that reprices each month; the coefficient on Bank Rate measures the immediate pass-through through variable and tracker loans.
Table 2. The repricing model
| Parameter | Full sample, 2005 to 2026 | 2005 to 2015 | 2016 to 2026 |
|---|---|---|---|
| Monthly repricing share (gap coefficient) | 0.027 (t = 4.3) | 0.035 | 0.034 |
| Bank Rate pass-through (immediate) | 0.29 (t = 3.6) | 0.38 | 0.09 |
| Implied annual repricing share | 28% | 35% | 34% |
| Half-life of the repricing gap | about 25 months | about 19 months | about 20 months |
Note: dependent variable is the monthly change in the effective rate on outstanding mortgages. HAC t-statistics for the full sample. The best fit is obtained with no lag between new and outstanding rates.
Table 2 reports the results. About 2.7 per cent of the stock reprices each month on the full sample, and a little more in the subsamples, implying that around 30 per cent of the stock moves to current rates each year and that half of any gap between new and outstanding rates is closed in about two years. The striking change is in the immediate pass-through of Bank Rate. Before 2016, a one-point rise in Bank Rate lifted the outstanding rate by 0.38 points at once, through the large stock of variable-rate and tracker loans. Since 2016, the immediate effect has fallen to about 0.09. British monetary policy now reaches households almost entirely through the slow channel of fixed-rate expiries.
This has two consequences. First, the full effect of a rate change on household cash flow takes three to four years to materialise. Second, it falls unevenly: on the households whose deals expire, not on mortgage holders as a whole.
3. Projecting the Refinancing Wall
The estimated model can be used to project the effective rate on outstanding mortgages under alternative paths for new mortgage rates and Bank Rate. Three scenarios are considered. In scenario A, new mortgage rates settle at 4.9 per cent, consistent with the October quoted rates and the November hike, and Bank Rate stays at 4 per cent. In scenario B, a second hike takes Bank Rate to 4.25 per cent and new mortgage rates to 5.3 per cent. In scenario C, a ceasefire brings Bank Rate back to 3.5 per cent and new mortgage rates to 4.2 per cent by mid-2027.

Table 3. Projected effective rate on outstanding mortgages (%)
| Scenario | Dec 2026 | Jun 2027 | Dec 2027 | Dec 2028 | Dec 2029 |
|---|---|---|---|---|---|
| A. New rates 4.9%, Bank Rate 4.0% | 4.18 | 4.29 | 4.38 | 4.52 | 4.63 |
| B. Second hike: new rates 5.3%, Bank Rate 4.25% | 4.27 | 4.43 | 4.56 | 4.77 | 4.92 |
| C. Ceasefire: new rates fall to 4.2% | 4.14 | 4.10 | 4.10 | 4.13 | 4.15 |
Note: starting point 4.04 per cent in September 2026.

Table 4. Extra household mortgage interest relative to September 2026 (GBP bn per year)
| Scenario | End-2027 | End-2028 | End-2029 |
|---|---|---|---|
| A. New rates 4.9%, Bank Rate 4.0% | 5.3 | 7.6 | 9.2 |
| B. Second hike | 8.2 | 11.4 | 13.7 |
| C. Ceasefire | 0.9 | 1.4 | 1.7 |
| Cost of a second hike (B minus A) | 2.9 | 3.8 | 4.5 |
Note: annualised interest at the end of each year, on the September 2026 stock of secured lending.
In the central scenario the effective rate on outstanding mortgages rises from 4.04 per cent to 4.38 per cent by the end of 2027 and 4.63 per cent by the end of 2029. On the current stock of mortgages, that adds about 5.3 billion pounds a year to household interest payments by the end of 2027 and 9.2 billion by the end of 2029. A second hike would raise the end-2029 figure to 13.7 billion: the marginal cost of the second hike to households is about 3 billion pounds a year by end-2027 and 4.5 billion by end-2029. A ceasefire would contain the increase to below 2 billion pounds.
The timing matters for policy. The energy shock that justified the November hike is concentrated in 2026 and 2027. The effect of the hike on household budgets peaks in 2028 and 2029. A central bank that tightens against a temporary supply shock in a fixed-rate mortgage market is, in effect, imposing a tax on households that arrives after the shock has passed.
4. The Aggregate Burden Is Manageable

Viewed in aggregate, the refinancing wall is not a crisis. Mortgage interest payments, computed as the effective rate times the stock of secured lending, were about 61 billion pounds a year in the second quarter of 2026, or 2.9 per cent of household gross disposable income (Figure 5). That is far above the 1.8 per cent of late 2021, but well below the 4.7 to 5.0 per cent of 2007 and 2008, when the effective rate was close to 6 per cent. In scenario A the ratio rises to about 3.3 per cent by the end of 2027 and 3.4 per cent by 2029; in scenario B to about 3.6 per cent. Rising incomes absorb part of the increase.
Table 5. Mortgage interest burden (% of household disposable income)
| Period or scenario | Effective rate (%) | Interest (GBP bn per year) | Share of disposable income (%) |
|---|---|---|---|
| 2007 Q3 | 5.88 | 46.4 | 4.7 |
| 2008 Q1 (peak) | 5.0 | ||
| 2021 Q4 | 2.02 | 28.1 | 1.8 |
| 2023 Q4 | 3.29 | 47.9 | 2.7 |
| 2026 Q2 | 3.94 | 60.8 | 2.9 |
| Scenario A, end-2027 | 4.38 | 3.3 | |
| Scenario B, end-2027 | 4.56 | 3.4 | |
| Scenario A, end-2029 | 4.63 | 3.4 | |
| Scenario B, end-2029 | 4.92 | 3.6 |
Note: projections assume household disposable income grows 3.5 per cent and the mortgage stock 3 per cent a year.

The housing market shows the strain at the margin rather than in aggregate. Approvals for house purchase fell from about 66,000 a month in September 2025 to 53,000 in September 2026, while remortgage approvals jumped from 34,000 in June to 43,000 in September as borrowers rushed to lock in rates before further increases. Both patterns are consistent with a market adjusting to higher rates without distress.
5. Who Faces the Shock
Aggregate figures hide the distribution of the shock, which depends on when each borrower last fixed. Figure 7 shows the monthly payment on a 150,000 pound, 25-year repayment mortgage for six vintages of fixed-rate deal, before and after refixing at 5 per cent on the remaining balance and term.

Table 6. Payment shock by vintage
| Deal taken | Rate at origination (%) | Expiry | Payment before (GBP per month) | Payment after (GBP per month) | Change |
|---|---|---|---|---|---|
| 5-year fix, June 2021 | 1.60 | June 2026 | 607 | 841 | +39% |
| 5-year fix, December 2021 | 1.59 | December 2026 | 606 | 841 | +39% |
| 5-year fix, June 2022 | 2.90 | June 2027 | 704 | 841 | +19% |
| 2-year fix, September 2024 | 4.58 | September 2026 | 841 | 861 | +2% |
| 2-year fix, December 2024 | 4.60 | December 2026 | 842 | 861 | +2% |
| 5-year fix, September 2023 | 5.24 | September 2028 | 898 | 841 | -6% |
Note: 150,000 pound loan, 25-year term at origination; after expiry the remaining balance is refixed at 5.0 per cent over the remaining term.
The contrast is sharp. Borrowers who fixed for five years during the pandemic, when five-year rates were around 1.6 per cent, face payment increases of close to 40 per cent as their deals expire through 2026 and early 2027. Borrowers whose two-year deals were taken in late 2024, at rates already above 4.5 per cent, face increases of only 2 to 3 per cent. Those who fixed for five years at the 2023 peak will see payments fall when they refix. The refinancing wall of 2026 is a wall for the pandemic cohort: largely households who bought or remortgaged in 2020 and 2021, often first-time buyers stretching to buy at the top of the market.
This concentration has two implications. First, the risk to financial stability is lower than the aggregate interest bill suggests, because the cohort facing the largest increases has also accumulated five years of amortisation and, in most cases, wage growth. Second, the consumption effect may be larger than aggregate figures imply, because the cohort affected is younger, more likely to have children and more likely to cut spending when payments rise.
6. Conclusions
Britain’s mortgage market has turned monetary policy into a slow fuse. The stock reprices at about 30 per cent a year, and the immediate pass-through of Bank Rate has almost disappeared. The rate rises of 2026 will therefore reach household budgets mostly in 2027, 2028 and 2029. In aggregate the burden is manageable: the share of income spent on mortgage interest will rise to about 3.3 to 3.6 per cent, well below the levels of 2007 and 2008. But it is concentrated on the households who fixed at historically low rates in 2021, whose payments are rising by close to 40 per cent.
For the Bank of England, the lesson reinforces the conclusion of the first paper in this series. A second hike would add 3 to 4.5 billion pounds a year to household interest bills, arriving in 2028 and 2029, long after the energy shock it is meant to contain. For the government, the case for targeted support, if any is needed, lies with the 2021 cohort rather than with mortgage holders in general.
Methodological Appendix
Repricing model. The equation is dR(t) = a [N(t-L) – R(t-1)] + b dB(t) + e(t), where R is the effective rate on outstanding mortgages, N the effective rate on new mortgages, B Bank Rate and L a lag chosen by maximising fit over 0 to 12 months (L = 0). Estimated by least squares without a constant, January 2005 to September 2026, with Newey-West standard errors. The half-life is log(0.5) / log(1 – a).
Projections. The model is iterated forward from September 2026 under the scenario paths for N and B. Extra interest is the change in R times the September 2026 stock of secured lending (1,563 billion pounds), held constant; the debt service ratios allow the stock to grow 3 per cent and disposable income 3.5 per cent a year.
Payment examples use the standard annuity formula with monthly compounding. Limitations: the model treats the stock as homogeneous; it does not distinguish lenders, loan-to-value bands or interest-only loans, and it cannot capture borrowers who extend their terms to reduce payments, a common response that lowers the immediate shock but raises lifetime costs.
References
Bank of England (2026), Bankstats: effective interest rates and quoted household interest rates, accessed November 2026.
Bank of England (2026), Financial Stability Report, 2026.
Bank of England (2026), Monetary Policy Report, November 2026.
Office for National Statistics (2026), Quarterly sector accounts, 2026 Q2.
Driss EFFINA, PhD
Independent Researcher, Global Strategy Files
The opening pages of our four country reports: United States, United Kingdom, Japan and China. A free PDF, sent to your inbox.
You will also receive occasional analysis from Global Strategy Files. Unsubscribe in one click, at any time.