Home sales jump 8% the month a two-year mortgage deal ends

Research by ValuQ: we followed 5,953,153 individual homes in England and Wales from one sale to the next, and counted how many months their owners kept them, to find out when people actually move.

There is a jump in home sales at the two-year mark. Sales of the same home rise 8.25 per cent between the 23rd and the 24th month of ownership, the biggest step at any anniversary in the first ten years. Nothing like it happens at three, four, five, six or seven years. And it is not there at all in homes bought in the late 1990s, before fixed-rate mortgages took over. The shape matches a cost that stops applying on a date: the early repayment charge that runs for the length of a fixed-rate deal.

Every sale of a home in England and Wales is recorded by HM Land Registry, along with the date it completed. Match a home's sale to its previous sale and you know exactly how long the owner kept it. Do that 5.95 million times and you can see when Britain moves.

Sales climb steadily through the first two years of ownership, then jump, then fall away. The jump lands in month 24 and peaks in month 25, and the months just before it are slightly quiet. People who want to move appear to wait, and then go.

Home sales cluster at every whole year, because buying and selling are both seasonal. So the fair test is one anniversary against another. Sales rise 8.3 per cent crossing the two-year mark, 3.9 per cent crossing three years, 2.5 per cent crossing four years and 2.5 per cent crossing five. Strip out the ordinary seasonal pull and the two-year step is the one still standing, at 2.4 per cent above trend across the six months that follow it. Every other anniversary from three to nine years sits between minus 1.3 and plus 0.2 per cent.

Two more things point the same way. The step is absent in homes bought between 1995 and 1999, when fixed-rate lending was far less common, and it appears in every cohort bought from the mid 2000s onwards. And it is largest exactly where people borrow most: flats show 3.5 per cent, leasehold homes 3.2 per cent and homes bought just below their local median price 3.1 per cent, against 1.3 per cent for detached houses and 1.0 per cent for homes worth more than one and a half times their local median.

One honest caveat. The sold-price register holds no information about anyone's mortgage. This is a reading of the timing and of who shows it most, not an observation of anyone's loan. It is also still happening: people who bought between 2019 and 2022 show the same two-year step, at three times its margin of error.

An early repayment charge is a fee your lender applies if you pay off your mortgage before the end of your fixed period. Selling your home pays off the mortgage, so selling inside the fixed period can trigger it, unless your deal can be carried across to your next property. The charge stops applying when the fixed period ends, which for a two-year deal is roughly two years after you completed.

1. 1.

2. 2.

3. 3.

4. 4.

No, and this was worth checking rather than assuming. We compared each sale against what the UK House Price Index did over that exact window, then compared like with like within the same district, property type and purchase year. Price growth runs smoothly through the two-year mark with no break. The share of homes selling for less in cash than they cost does the same. Around 9,800 more sales landed in the six months after the two-year mark than the trend implies, and those sellers did no worse on price. The clock changes when people sell, not what they get.

There is no five-year spike anywhere in the data, which is worth saying plainly because the five-year fix is now the more common product. Two things could account for it. Someone who remortgages at two years starts a fresh charge period that no longer lines up with the day they bought, so the clock stops being visible in the register. And choosing to fix for five years is itself a statement that you do not expect to move soon, so there is no held-back demand waiting for the date. This data cannot separate the two, so we report the finding and stop there.

The full research paper, with the complete method, all tables and the code that produced them, is published under a Creative Commons licence at valuq.co.uk/research.