Key takeaways
- Over the 50 years to May 2026, the FTSE Nareit All Equity REITs Index compounded at 11.98% a year, against 9.04% for the S&P 500.
- In 2022 that index returned -24.95% and the FTSE EPRA Nareit Global index -23.6%. The appraised MSCI UK Annual Property Index returned -7.1%.
- Appraisals anchor on earlier appraisals. Monthly IPD returns from December 1986 to May 2011 showed serial correlation of 0.903 at a one-month lag, decaying to 0.176 at a year.
- The Investment Property Forum's 2007 study put UK property's true annual volatility at 13% to 15%, against the 10.3% the unadjusted index showed for 1971 to 2005.
- UK daily-dealing property funds held around 17% of £17bn in cash at 31 March 2020, against 2.5% for the average equity fund.
The same year, two very different numbers
You're weighing a real estate investment trust — a listed company that owns income-producing buildings — against owning property directly, or through a fund that holds the buildings itself. On paper it's the same asset. Over any short horizon it isn't the same experience.
Here's the answer up front. In 2022 the FTSE Nareit All Equity REITs Index returned -24.95%. The FTSE EPRA Nareit Global index, which tracks listed property worldwide, returned -23.6%, and its ex-US version -22.0%. The MSCI UK Annual Property Index, which measures buildings held directly and valued by professional appraisers, returned -7.1%. Those four numbers are the bars in the chart above.
Very little of that gap is about the buildings. Most of it is about how the number is made. One is a price struck by buyers and sellers every second the market is open. The other is an opinion, formed periodically, by someone who starts from the previous opinion.
Over 50 years listed property out-returned the S&P 500, and over 10 years it lost badly
Start with the record, because both camps quote half of it.
Nareit's fact sheet, dated May 2026, puts the FTSE Nareit All Equity REITs Index at 11.98% a year over 50 years. The S&P 500 returned 9.04% over the same span. That's the number the listed-property case rests on, and it's real.
Now the other half. Over 10 years to the same date, the REIT index returned 6.44% a year against 15.65% for the S&P 500. A decade of underperformance that wide isn't noise. It's what happens when an income asset meets rising discount rates while the index it's compared with is carrying a handful of enormous technology companies.
Income is a large part of why the long-run number looks the way it does. In May 2026 the equity REIT index yielded 3.69%, against 1.02% for the S&P 500. REITs distribute most of what they earn, so more of the total return arrives as cash and less as compounding inside the company.
Listed property has been more volatile than global equities, not less
The volatility isn't subtle either. FTSE Russell's July 2026 factsheet puts five-year annualised volatility for the FTSE EPRA Nareit Global index at 17.4%, against 14.9% for the FTSE All-World. Its five-year maximum drawdown was -31.4%.
Nor was 2022 an isolated print. The same equity REIT index returned 41.30% in 2021, then -24.95% the following year. Buildings do not change hands at prices that swing like that.
That is the opposite of what most people expect from bricks and mortar. It's also the number that makes readers reach for direct property instead, on the grounds that houses and offices simply don't move like that. The next two sections are about whether that's true, or whether it only looks true.
An appraisal is anchored to the last appraisal, and the data shows the fingerprint
Direct property has no continuous price. It has a valuation. A valuer looks at recent comparable transactions, at the building's income, and at what the same building was valued at last time. That last input is the problem.
The Investment Property Forum's 2012 working paper measured the effect on the IPD monthly All Property Total Return Index. From December 1986 to May 2011, its serial correlation fell "from 0.903 with a lag of one month, to 0.760 with a three-month lag, 0.511 with a six-month lag to just 0.176 with a one-year lag".
Serial correlation means this period's return predicts the next one. In a market where prices adjust to information, that shouldn't be possible for long. A reading of 0.903 at one month says the index is mostly repeating itself. The paper attributes it to anchoring — Tversky and Kahneman's finding that people start from a prior value and adjust too little from it.
MSCI, which publishes the UK index, says the same thing about 2022 without embarrassment: "The less frequent valuation cycle skipped over the final few months of capital growth in the first half of 2022 and limited the decline in returns." The buildings fell. The index took longer to say so.
You can see the same effect in the tails. In 2022 the annual UK index "avoided the lows of the 2008 global financial crisis", when returns "dropped below -20%" — even though the higher-frequency versions of the same index were much uglier.
Correcting for smoothing moves UK property's risk back where theory expects it
Researchers have a name for the fix: desmoothing. The idea is that today's valuation is a weighted average of new evidence and the old valuation, so you can strip the old part out.
The Investment Property Forum's 2007 study ran five methods across four sample periods. Its conclusion: "Our central, or preferred, estimate of property's historic standard deviation in annual total returns is 13% to 15%, or 1.3 to 1.5 times that observed in the unadjusted index results." Over 1971 to 2005 the raw index showed 10.3%; the simplest desmoothing method put it at 13.8%.
Why that matters practically: on unadjusted figures, UK property looked less risky than gilts, which the authors called "an implausibly low figure". Feed that into an optimiser and it recommends a property weight no professional would run. Desmoothed, property lands where you'd expect it — above gilts, below equities.
The quarterly evidence points the same way. In the 2012 paper, quarterly IPD returns from 1987 to 2008 had a standard deviation of 3.25% as reported, and 4.81% after regime-based unsmoothing.
The strongest objection: nobody agrees how much smoothing there is
Here's the honest counter-case, and it's stronger than the desmoothing literature usually admits.
The correction depends on a coefficient nobody can observe. The 2007 study found that across methods and periods, "the indicated adjustments to property risk vary from lower risk shown by valuation index figures up to a multiple of nearly three times". Published estimates run "from 1.5 to 3.5 times that shown by index figures". That is not a settled parameter. It's a range wide enough to support almost any conclusion you brought with you.
Worse, the methods can break. The 2012 paper's simple autoregressive desmoother turned a 3.25% quarterly standard deviation into 34.11%. The authors' verdict: "The AR series appears to be implausibly volatile." A correction that produces nonsense in one specification should make you cautious about the specification that produces a tidy answer.
And the objection runs the other way too. A REIT is a leveraged operating company whose shares sit in equity indices, get bought by index funds, and reprice when rates move. Some of its 17.4% volatility is genuinely the buildings. Some of it is the stock market having an opinion about REITs. Neither number — the appraised one or the traded one — is the truth about what a building is worth.
Daily dealing had a price, and UK property funds paid it in cash
The liquidity difference is where this stops being a measurement argument.
A REIT trades like a share. In May 2026, US listed REITs turned over $11.7 billion a day. Direct property doesn't. The FCA's 2020 consultation puts it plainly: "A typical commercial property transaction may take between 2 and 4 months to complete under normal market conditions."
UK open-ended property funds tried to bridge that gap by promising daily dealing anyway. The bridge was cash. At 31 March 2020, 18 daily-dealt UK authorised property funds held around 17% of their £17bn in cash — £2.8bn sitting idle. The average equity fund held 2.5%.
The FCA was blunt about what that costs: "Holding these cash balances is inefficient and reduces expected returns to investors." Roughly a sixth of a property fund's exposure wasn't property. A reader who bought one for its property exposure was getting materially less of it than the label implied.
When the funds suspended, the exit was never there
Cash buffers postpone the problem rather than solving it. UK property funds suspended dealing during the 2008 financial crisis, again after the 2016 EU referendum, once more in 2019, and then in 2020 the FCA published its consultation "while almost all relevant funds are suspended for dealing". Managers of a suspended fund have to review the suspension formally at least every 28 days — which tells you these were not measured in weeks.
The FCA quantified the trap. Over the 20 years to 2020, the average three-month return on a daily-dealing property fund was around 1.2%. Funds fell about a quarter of the time, and "around 6% of the time, the fall was greater than 5%". Then the line that matters: "The worst falls occurred during periods of suspension, so investors were unable to access their money over most of the worst performance periods."
Read that twice. The daily liquidity was reliably available except in the moments anyone wanted it. That's not a smaller drawdown than a REIT's — it's the same drawdown with the exit welded shut.
Illiquidity is measurable as risk, not just as inconvenience. Bond and Hwang's 2004 work, cited in the same consultation, models a three-month sale period as increasing property risk by 10% over a seven-year horizon, and a six-month sale period by 38%.
The regulatory sequel is instructive. The FCA consulted in 2020 on a redemption notice period of "at a minimum 90 days, and may be up to 180 days". In its December 2025 consultation it recorded the outcome: "We consulted on proposals in 2020 to require NURS property funds to have mandatory notice periods, but we never finalised the rules." It plans to consult again in 2026. Six years on, the mismatch the regulator identified is still there.
What this evidence cannot tell you
The three indices in the chart are not a like-for-like pair. Nareit's is US listed property, FTSE EPRA Nareit's is global listed property, and MSCI's is UK commercial buildings held directly. Sector mix differs sharply: the listed indices carry data centres, towers and healthcare, which UK direct property indices barely touch. Part of the 2022 gap is composition, and this sample can't separate how much.
The smoothing research is also old. The 2007 study's preferred sample runs to 2005; the serial-correlation series ends in 2011. Valuation practice, data frequency and index construction have all moved since. MSCI now publishes higher-frequency and transaction-linked measures that the 2007 authors didn't have.
The desmoothing coefficient cannot be observed directly, and the estimates range from below the index figure to roughly triple it. Any single desmoothed volatility number, including 13% to 15%, is a choice as much as a measurement.
Finally, none of this measures the property you'd actually buy. A single building is not an index of 22,985 of them, which is what the MSCI UK annual index held at December 2014, and the arithmetic of leverage on a mortgaged property is different again — a point covered in our piece on your house as a portfolio asset. The distinction between measured volatility and the loss you actually live through is the subject of volatility versus drawdown.
What would change the conclusion
If direct property indices moved to transaction-based pricing, the measured gap would shrink on its own. Repeat-sales and nowcast indices already produce volatility closer to the desmoothed estimates. The smoother number is a property of the method, not of the asset, and methods change.
If notice periods arrive, the comparison changes shape rather than closing. A property fund with a 90-day notice period is a more honest product and a worse substitute for a REIT. The FCA's own view is that funds would hold less cash and more buildings, which raises both expected return and risk.
If REIT sector composition keeps drifting, the question stops being about buildings at all. An index dominated by data centres and communications towers is a technology-adjacent income asset that happens to hold freeholds, and comparing it with UK offices tells you less each year.
The thing worth watching isn't which vehicle wins. It's whether the correlation between the two series behaves the way you assumed when you sized them — a habit examined in correlation instability across regimes. If you hold both a REIT sleeve and a directly held property, LedgerTouch will show you one repriced daily and the other repriced when someone gets round to it. The second one isn't calmer. It's quieter.