Key takeaways
- Nationwide's series puts UK house price growth at 5.8% a year nominal from Q4 1995 to Q4 2025. On a constant 75% loan-to-value mortgage at the 2-year fixed rate's 1995 to 2026 average of 4.28%, that became 10.2% a year on the deposit.
- The multiplier only works when price growth beats loan-to-value times the mortgage rate. At the August 2026 2-year fixed rate of 4.92%, a 75% mortgage needs 3.69% a year of appreciation to break even on price alone.
- Nationwide reported annual growth of 1.6% in August 2026. At that pace the price-only return on a 25% deposit runs at about minus 8.4% a year, before the rent an owner stops paying.
- UK prices fell 20.2% from Q3 1989 to Q1 1993 and 18.7% from Q3 2007 to Q1 2009. The first of those wiped out a 10% deposit twice over; a 25% deposit lost 80.6% of its value.
- In 2026 Q2, 47.5% of new UK mortgage lending was above 75% loan-to-value and 8.4% was above 90%, the highest shares since 2007 Q4 and 2008 Q2 respectively.
A homeowner's return is the house's return, multiplied by the loan, minus the interest
How much of the money people make on their homes comes from the house, and how much comes from the mortgage? The honest answer is that the mortgage does most of the work, and that it charges for it.
Here's the arithmetic in one line. If a house rises by p a year, the loan is L of its value, and the loan costs r a year, the return on the owner's own money is (p minus L times r) divided by (1 minus L). Jordà, Knoll, Kuvshinov, Schularick and Taylor use exactly this formula in their 2017 Federal Reserve Bank of San Francisco working paper, The Rate of Return on Everything, 1870 to 2015, to turn an unlevered housing return into a levered one. A 75% mortgage divides the equity by four. It also subtracts three quarters of the interest rate from the growth rate before the division happens. That subtraction is the piece most back-of-the-envelope house maths leaves out, and it's the piece that turns the answer negative when rates are high and growth is low.
This guide runs that formula on fetched UK and US data for three windows: 30 years, 10 years and the most recent 12 months. It's the levered companion to our piece on the house as an investment, which compares housing and equities with no mortgage in the picture. It shows what the mortgage did to the number, in both directions.
UK house prices since 1952: 7.0% a year nominal, and 2.3% a year real over the last 30 years
Start with the unlevered asset. Nationwide's long-run series records an average UK house price of £1,891 in Q4 1952 and £278,784 in Q2 2026. Over those 73.5 years that compounds at 7.0% a year in nominal terms. It's the number people quote when they say property always goes up.
The last 30 years are less generous. Nationwide's Q4 1995 average was £50,930 and its Q4 2025 average was £273,077, which is 5.8% a year. The ONS Retail Prices Index rose from 150.1 in Q4 1995 to 407.2 in Q4 2025, or 3.4% a year. Net of that, the real gain was 2.3% a year. The last 10 years are thinner still: £197,044 in Q4 2015 to £273,077 in Q4 2025 is 3.3% a year nominal, and with the RPI at 260.0 in Q4 2015 that's a real return of about minus 1.2% a year. Whether that counts as a housing inflation hedge is a question we've tested separately.
Those are price-only figures. They exclude the rent an owner-occupier consumes by living in the house, along with maintenance, insurance, transaction tax and interest. In that paper's 16-country data, the rent component of UK housing's nominal return averaged 3.94% a year against a 5.44% capital gain. It's left out here on purpose, because rent isn't what a mortgage multiplies. Price is.
Mortgage leverage from 1995 to 2025: 5.8% on the house became 10.2% on a 25% deposit
Now add the loan. The Bank of England's quoted rate on a 2-year fixed mortgage at 75% loan-to-value starts in January 1995 at 8.13% and stands at 4.92% in August 2026. The simple average across every month from January 1995 to August 2026 is 4.28%. Put that rate and the 5.8% price growth into the formula and the return on the owner's money by loan-to-value comes out as the chart shows.
At 0% loan-to-value, a cash buyer, the return is the house's own 5.8%. At 60% it's 8.0%. At 75% it's 10.2%. At 90% it's 19.1%. The mortgage leverage multiplies the spread between price growth and interest, not the price growth itself, and over this window that spread was 1.5 percentage points in the owner's favour on the borrowed 75%.
Three simplifications sit inside those numbers, and each flatters them. The loan is held at a constant share of the house's value, which no repayment mortgage does; amortisation and price growth both cut leverage over time, so the true multiplier fades. The rate is the average quoted to new borrowers, not the rate any one household paid across 30 years. And the interest is paid out of salary, month by month, rather than out of the house. Treating it as a cost of the position is the right accounting. It's still cash a renter didn't spend.
The break-even: leverage helps only when price growth beats loan-to-value times the rate
The formula has a hinge. When p is bigger than L times r, borrowing raises the return on your deposit. When it's smaller, borrowing lowers it, and the more you borrowed the worse it gets. The hinge moves with the rate, which is why the same house can be a good levered asset in one decade and a bad one in the next.
At the August 2026 2-year fixed rate of 4.92%, the break-even appreciation is 3.69% a year for a 75% mortgage and 4.43% for a 90% mortgage. Over 2016 to 2025 the same rate averaged 2.74%, and 2021 alone averaged 1.45%. A 75% borrower in 2021 needed just 1.09% a year of growth to come out ahead on price. A 75% borrower fixing in August 2026 needs more than three times that. The fixed vs variable mortgage choice decides which r goes into this calculation, and for most of the last 30 years it decided more than the house did.
Run the 10-year window on its own and the point lands. Price growth of 3.3% a year against an average rate of 2.74% gives 5.0% a year on a 25% deposit and 8.5% on a 10% deposit. That's a levered return on housing that beat the unlevered one, in a decade when the house itself lost money in real terms. The borrowing did all of it.
At August 2026 prices and rates, the price-only return on a deposit is negative
Nationwide's August 2026 report put annual house price growth at 1.6% and the average price at £275,465. The ONS and HM Land Registry's UK House Price Index for July 2026 put the average at £273,000, up 1.4% on the year. Take the higher of the two, 1.6%, and the August 2-year fixed rate of 4.92%.
A 75% mortgage: 1.6 minus 0.75 times 4.92 is minus 2.09, divided by 0.25 is minus 8.4% a year. A 90% mortgage: minus 28.3% a year. That's a run rate, not a forecast. It says a buyer who fixed this summer is paying more interest on the borrowed share than the whole house is earning in price. The mortgage leverage that produced 10.2% over 30 years produces a loss at these two inputs.
The interest is the largest single item in the cost of home ownership, and most rent-versus-buy comparisons count it. What they usually skip is that it's also the cost of the leverage. You don't get the multiplier without it.
The rent you stop paying is the part of the return that never gets multiplied
None of the numbers above include the shelter the owner consumes. That matters, because it's the component that keeps the levered return positive when price growth is low. If the long-run UK rent component of 3.94% a year in the Jordà paper's data held today, the August 2026 sum would read 1.6 plus 3.94 minus 3.69, divided by 0.25, or about 7.4% a year on a 25% deposit. The rent doesn't rise with leverage, but it's collected on the whole house while the owner's stake is a quarter of it, so it arrives at four times its face value on the equity.
That 3.94% is an 1870 to 2015 average across a changing housing stock, not a measurement of any 2026 property, and it's net of the running costs the authors deduct rather than the ones a given owner pays. And the rent is consumed, not banked. It only counts as a return if you'd otherwise have paid it. The mortgage overpayment vs investing calculation is the place where that consumed return meets the cash alternative.
The other side of the multiplier: 1989 to 1993 and 2007 to 2009 on a 10% deposit
Leverage has no opinion about direction. Nationwide's series shows the UK average price falling from £62,782 in Q3 1989 to £50,128 in Q1 1993, a nominal drop of 20.2%. The RPI rose from 116.0 to 138.7 across the same 14 quarters, so the real fall was 33.2%. The second episode ran from £184,131 in Q3 2007 to £149,709 in Q1 2009, a nominal drop of 18.7% and, with the RPI at 207.1 and 210.9, a real one of 20.2%.
Put the first of those through a 90% mortgage. The house falls to 79.8% of its purchase price and the loan is still 90% of it, so the owner's equity goes from plus 10 to minus 10.2. The deposit is gone, and a second deposit's worth of debt sits on top. On a 75% mortgage the equity goes from 25 to 4.8, a loss of 80.6%. A cash buyer lost 20.2%. Same house, same three and a half years, three very different outcomes, and the only variable is L.
The Jordà paper's 16-country data put the standard deviation of real housing returns at 9.98% a year against 21.94% for equities, and it's that lower volatility that makes housing look like the safer asset in their tables. Divide 9.98% by the owner's 25% stake and the volatility of the equity in a 75%-mortgaged house is about 39.9% a year, nearly twice that of the equity index. The index number is itself an average across a country. A single house in a single postcode moves more than the average.
Nearly half of new UK lending in 2026 Q2 was above 75% loan-to-value
None of this is theoretical. The Bank of England and FCA's mortgage lenders and administrators statistics for 2026 Q2 report that 47.5% of gross mortgage advances were at loan-to-value ratios above 75%, the highest share since 2007 Q4, and that 8.4% were above 90%, the highest since 2008 Q2. Lending to borrowers with a high loan-to-income ratio was 46.0% of the total, and first-time buyers took 27.3% of it. Gross advances in the quarter were £77.4 billion on an outstanding stock of £1,760.6 billion.
So close to half of the people buying in the spring of 2026 took on at least four times leverage, and roughly one in twelve took ten times or more, at a 2-year fixed rate near 4.9% and against annual price growth of 1.4% to 1.6%. The break-even arithmetic above applies to them directly, and not to the stock of existing owners, most of whom borrowed at lower rates and have since been de-levered by amortisation and by price gains. For that same cohort the comparison against renting turns on the rent-to-price ratio on the street they buy in, which across UK local authorities runs from under 3% to over 7%.
Transaction tax adds a fixed drag on the equity rather than the house. On the July 2026 average price of £273,000, stamp duty land tax at the residential bands in force for the 2026/27 tax year (nothing on the first £125,000, 2% on the next £125,000, 5% on the next £675,000) is £3,650. On a 10% deposit of £27,300 that's 13.4% of the buyer's equity paid on the way in. A first-time buyer at the July 2026 first-time-buyer average of £229,302 pays nothing, because the 2026/27 first-time buyer relief covers purchases up to £300,000.
The US version: Case-Shiller at 4.7% a year against a 6.95% mortgage
The same arithmetic runs on American data, with a higher rate and a higher hurdle. The S&P Cotality Case-Shiller US National Home Price Index, base January 2000 equals 100, stood at 336.663 in June 2026. That's 4.7% a year over 26.4 years, and 1.5% over the 12 months to June 2026 against a reading of 331.605 a year earlier. Freddie Mac's 30-year fixed mortgage average was 6.95% in the week to 17 September 2026.
At an 80% loan-to-value, the US break-even is 0.8 times 6.95, or 5.56% a year of appreciation. The 26-year average of 4.7% doesn't clear it. The 12-month figure of 1.5% falls far short, and produces a price-only return on a 20% deposit of about minus 20.3% a year. The US drawdown is the larger one, too: the index fell from 184.607 in July 2006 to 133.987 in February 2012, a 27.4% decline, which at 80% loan-to-value takes a 20-point equity stake to minus 7.4. The Jordà paper puts the US long-run split at a 3.54% capital gain and a 5.33% rent component, so the American owner's return leans even harder on the unlevered part.
The strongest objection: for the economy as a whole, leverage added only 1.1 points
The authors who supplied the formula are the best witnesses against making too much of it. Jordà and his co-authors lever up a 7.0% real housing return using an economy-wide loan-to-value of 20% and a 2.5% real safe rate, and get 8.1%. Using a 3% real mortgage rate instead gives 8.0%. Their conclusion is that the adjustment "is not consequential for the main conclusions we present in this paper", and that housing and equity returns stay roughly comparable once both are levered. If your leverage looks like the country's, this whole piece is about a single percentage point.
The reply is that nobody's leverage looks like the country's for long. A 20% average spans owners who paid off their loans decades ago and buyers who completed last month at 90%. The same paper says that "houses can be levered much more than equities" and that "the majority of households in advanced economies today hold a leveraged portfolio in their local real estate market". The 8.1% is the right number for the housing stock. The 10.2%, the 19.1% and the minus 28.3% are the right numbers for the person signing the mortgage deed, and they converge on 8.1% only as the loan is paid down.
What this arithmetic can't tell you
The data cannot tell you what one house did, and the formula assumes a constant loan-to-value and an interest-only loan. A repayment mortgage de-levers itself every month, so the true multiplier on a 25-year loan starts at four and ends at one, and the average over the term is a good deal lower than four. Price growth de-levers it faster. That means the 10.2% overstates what a 1995 borrower who never remortgaged actually earned on their money, and the minus 28.3% overstates what a 2026 borrower loses if growth stays at 1.6%.
The rates are quoted averages for new 2-year fixes at 75% loan-to-value. A 90% borrower pays more than that, and a borrower on a legacy fix pays less, so the break-even lines shift by product. Nationwide's index is mix-adjusted and its methodology has changed several times since 1952, which the notes to its own data file record. The deflator is the RPI, because it's the ONS series with a quarterly history back to 1987; a CPI deflator gives a higher real figure in every window without changing a sign. And every figure here is an index, the average of thousands of transactions rather than the one house you'd own, whose return has a much wider distribution around the same mean.
Finally, the 30-year window is one draw. It contains the 2010 to 2021 stretch of the lowest mortgage rates in the series, when the spread between growth and interest was widest. A window starting in 1989 or 2007 gives a different levered number, and one starting in 2026 begins with a hurdle the last one never had.
What would change the conclusion
If the 2-year fixed rate fell back toward its 2016 to 2025 average of 2.74%, the break-even for a 75% borrower would drop from 3.69% to about 2.06%, and the August 2026 growth rate of 1.6% would be within a percentage point of clearing it. The whole sign of the price-only result depends on r, and r has moved by more than 6 points in the series since 1995.
If house price growth returned to its 30-year average of 5.8%, the current rate would be no obstacle. At 4.92% and 75% loan-to-value the levered return would be about 8.3% a year on price alone, before rent. The last 12 months are one observation, and a low one.
If the rent component were measured for today's stock rather than inherited from the long run, the levered total return could move either way. A 2026 yield below 3.94% would push the August arithmetic back toward zero. A higher one would widen the cushion. It's the least well-measured input here and the one that carries the most weight at current rates.
The number worth watching isn't the house price index. It's the gap between that index's annual growth and your loan-to-value times your rate. LedgerTouch holds the mortgage on the same balance sheet as the house, so that gap is a figure you can read rather than one you have to reconstruct.