Turning blue
Eldred Buck ·
Ok, so if you would like something about turning blue, which isn’t about the US midterm election results, then you’ve come to the right place. Oh incidentally, wasn't that just great to see? Democracy, you gotta love it. Anyway back to more prosaic things.
It’s always nice to reconnect with expert folk and I had a really interesting time discussing future research projects with Prof Anupam Nanda more on that later. We first met, in the time before Covid, when he was at Henley Business School/Reading University and I was on version 2 or maybe 3 of my AI based models. Anyway during this conversation I mentioned that I had actually just recently completed an updated study which was based around the same question as his much-cited 2015 paper, but with a slight nuance. Actually, I had sent a fact sheet on this to a few trade newspapers and news desks at the end of last week, but I think the material was deemed just too unexciting and let’s be frank, the news agenda is pretty bountiful these days. We are all indeed, living in the midst of that well-quoted 'Chinese' curse.
Anyway, I thought I’d summarise the study here for those of you who are interested in a slightly more in depth analysis of real estate, urban economics, cost of living and environmental impacts, so here goes.
The question that we had asked ourselves was this: have the energy price rises since 2021, driven by the conflicts in Ukraine and Iran, changed what homes are worth, across the housing market and which sectors in particular, if any, have been affected most? So, similar to the good Prof's work, but actually a slightly more nuanced sectoral analysis building on that paper's work. I guess at heart it looked like a pretty basic question to the news desks and hence why I'm the latest recipient of the ‘no shit Sherlock’ award. Nevertheless, I've done the study, so let’s go through the good old, who, what, when, where, why and how and then let you be the judge.
To start with, we worked our way through all open market sales with matched EPC data. We took the current running costs for the individual property and the actual savings as itemised line by line in the actual reports themselves. Basically everything that the EPC assessor recommends. If you’ve had to do one recently you’ll know they typically take about an hour or so and cost about £60-150. Obviously it’s more for larger properties. So that was the first step, then we took the fuel prices each certificate was calculated at and moved its costs to the prices in the month the home sold, using the Ofgem price cap and the ONS price indices for gas and electricity.
Efficiency is worth more since the shock
The premium for 10 points of energy score, quarter by quarter, against the price of gas buyers faced at the time.
Before the shock, 10 points of score were worth about 1.0% of the price (1.7% without allowing for age). Since 2023 they have been worth about 1.5% (2.1%). The increase is many times its margin of error, and it did not reverse when gas prices fell back from their 2023 peak: the market has kept the higher value it puts on efficiency. What it means is we can actually measure a change in buyers' motivations that hasn't gone away.
Where it bites hardest
The same measure for each section of the market, chosen before we looked at any results: before the shock (2017–20) and since (2023–26), with the home's age allowed for where the section is not itself defined by age.
| Section | 2017–20 | 2023–26 | Change | Sales |
|---|---|---|---|---|
| All homes | 0.97% | 1.45% | +0.48 ± 0.05 | 4,023,530 |
| Cheaper districts (lowest third of £/m²) | 1.74% | 2.17% | +0.43 ± 0.10 | 1,357,993 |
| Dearer districts (top two thirds) | 0.54% | 1.03% | +0.49 ± 0.06 | 2,559,248 |
| North and Wales | 1.54% | 2.06% | +0.53 ± 0.11 | 1,307,337 |
| Rest of England | 0.75% | 1.18% | +0.43 ± 0.06 | 2,609,958 |
| Solid walls | 1.61% | 2.18% | +0.57 ± 0.10 | 990,401 |
| Cavity and other walls | 0.85% | 1.25% | +0.39 ± 0.06 | 2,928,357 |
| Built before 1930 (age not allowed for) | 1.97% | 2.58% | +0.62 ± 0.11 | 851,217 |
| Built 1930 or later (age not allowed for) | 1.50% | 1.86% | +0.36 ± 0.06 | 3,062,389 |
| Houses | 1.02% | 1.59% | +0.58 ± 0.06 | 3,277,588 |
| Flats | 0.64% | 0.71% | +0.06 ± 0.15 | 559,525 |
| Mains gas heating | 0.90% | 1.42% | +0.53 ± 0.06 | 3,590,288 |
| Electric heating (off gas) | 0.20% | -0.28% | −0.48 ± 0.17 | 275,638 |
| Last certified for a private let | 0.50% | 0.91% | +0.41 ± 0.17 | 404,496 |
Cheaper against dearer districts
North and Wales against the rest
Solid walls against cavity
Did the test hold?
Systemic: supported. The premium rose in the market as a whole and in almost every section, far beyond the margin of error, and has held since.
More acute in certain sections: in level, yes; in change, no. The premium is two to three times higher in cheaper districts, the North and Wales, and in solid-walled homes, so energy efficiency weighs most on value where homes are cheapest and hardest to heat. That's right, unsurprisingly the chillier and bluer you get, the more you're going to value those heating improvements next time around. But we had decided in advance to count a section as hit more acutely only if its rise beat the market's by more than both margins of error; none quite did. The shock raised the premium by about half a point almost everywhere. The exceptions are flats and homes with electric heating, where the premium didn't rise, or fell. These are often newer flats whose rating is held down by electric heating, and the data can't separate the two, so I'm not claiming anything for them.
What unmade savings cost a home's value
Each certificate gives the home's running costs now and after the improvements its assessor recommends. Comparing like with like, the gap between the two, put at the energy prices of the sale's month, shows how much less a home sells for when those savings have not been made.
| £ per £100 a year of unmade savings | 2017–20 | 2022 Q2 – 2023 Q2 | 2024–26 |
|---|---|---|---|
| Age allowed for | £1,309 | £776 | £1,609 |
| Without age | £2,166 | £1,226 | £2,134 |
When energy prices spiked, the savings on paper doubled but prices did not follow: buyers priced each £100 a year of unmade savings at little more than half their usual amount. However by 2024 the market had fully caught up, and in the stricter version priced it a little higher than before the crisis (£300 more per £100 a year, ± £63). Today each £100 a year of unmade savings takes about £1,600–2,100 off a home's price; for the median home, with £366 a year of savings to make, that is about £5,900–7,800.
Running costs on certificates are out of date
Every certificate's costs are at the fuel prices of its own month. At today's prices (Ofgem cap, October–December 2026), the median home's certificate understates its running costs by about 90%, and the understatement is much the same for every tenure.
| Tenure | Certificate says | At today's prices | Understated by | Unmade saving today |
|---|---|---|---|---|
| Private rented | £786 | £1,237 | 85% | £274 a year |
| Social | £710 | £1,116 | 84% | £151 |
| Owner occupied | £976 | £1,617 | 89% | £348 |
| All homes | £806 | £1,373 | 90% | £227 |
Read all about it: 90% out!
So the bottom line is that, at today's prices, energy certificates understate what homes cost to run by about 90%, and buyers have caught on. Each £100 a year of energy savings a home has not made now takes about £1,600–2,100 off its price, and our analysis of 5.7 million sales in England and Wales shows this clearly. During the 2022–23 price spike buyers priced those savings at barely half that; since 2024 the market has fully caught up. Each 10 points on a home's energy rating adds about 1.5% to its price, up from 1.0% before 2021.
If people are now thinking about those 10 points on an EPC then what else might they think about when buying a home?
And that brings me neatly back to what the good Prof and I were talking about actually. EPC certificates? No, I'll spare you any more on that. What we were discussing was how we would go about measuring the potential impacts of much bigger environmental effects and over much longer forward timeframes. Think Pacific Palisades, the Gironde, Nepal. Think models that are totally outdated, 100 year risks now happening every 3 years and climate as a three body problem.
Anyway, back on my bike.
Eldred
How the figures were made
Sales: every open-market house and flat sale in England and Wales from January 2017 to mid-2026 (HM Land Registry Price Paid Data, standard price paid category), each matched to the energy certificate in force when it sold: 5.77 million of 6.73 million sales. The newest months are left out while late sales are still being registered.
Like for like: for each quarter, each sale is compared only with sales of the same property type in the same postcode district, allowing for floor area, rooms, new build and freehold. The strict comparison also allows for the home's age; the standard one doesn't. Margins of error are 95%.
Running costs at the time of sale: each certificate's costs moved from the fuel prices it was calculated at to the prices in the month of sale, using Ofgem's price cap and the ONS consumer price indices for gas and electricity. Homes heated by oil or LPG (about one in twenty) are left out.
What it doesn't show: efficient homes may also be better in ways no record captures, such as upkeep. Comparing like for like limits this, and the change over time, which such differences explain less well, is the main finding.
Sources: HM Land Registry Price Paid Data; the energy performance certificate register (Get energy performance of buildings data, GOV.UK); Ofgem price cap; ONS consumer price indices