Key takeaways
- In 2022, the year UK CPI inflation peaked at 11.1%, the UK index-linked gilt index fell 34.48% and the US inflation-linked bond index fell 12.60%.
- The Bloomberg Commodity Index returned 27.11% in 2021 and 16.09% in 2022, then gave back 7.91% in 2023 as inflation came down.
- Gold's benchmark price returned -0.43% in 2022 and then 65.00% in 2025, a year in which annual US inflation ran at 2.68%.
- Over 2021 to 2025 US consumer prices rose 24.41%. The US inflation-linked bond index returned 4.63% across the same five years, a real loss of 15.90%.
- Since 1900 the real US dollar gold price has risen 5.2-fold, an annualised 1.3%, against 1.6% a year real for US bonds and 0.5% for bills.
No single inflation hedge asset covered the whole episode
You want one holding that goes up when your shopping bill does. Three candidates usually get named: inflation-linked bonds, broad commodities, and gold. Between 2021 and 2025 all three met a real test, and the honest answer is that none of them did the whole job.
Commodities worked while inflation was rising, then stopped. Gold did almost nothing while inflation was rising and afterwards produced the largest gain of the three. Inflation-linked bonds, the only one of the three contractually tied to a price index, lost real money across the whole five years. Which of these inflation hedge assets you'd have wanted depended on which half of the episode you were standing in, and that half wasn't knowable in advance.
What each sleeve returned, calendar year by calendar year
The figures below are index returns, taken from the benchmark rows of four iShares fund factsheets dated 31 July 2026. Three are in US dollars: the Bloomberg US Government Inflation-Linked Bond Index, the Bloomberg Commodity Index Total Return, and the LBMA Gold Price. The fourth, the Bloomberg UK Government Inflation-Linked Bond Index, is in sterling.
- 2021. US inflation-linked bonds 6.00%, commodities 27.11%, gold -3.75%, UK index-linked gilts 4.01%.
- 2022. US inflation-linked bonds -12.60%, commodities 16.09%, gold -0.43%, UK index-linked gilts -34.48%.
- 2023. US inflation-linked bonds 3.84%, commodities -7.91%, gold 13.80%, UK index-linked gilts 0.61%.
- 2024. US inflation-linked bonds 1.76%, commodities 5.38%, gold 26.59%, UK index-linked gilts -8.59%.
- 2025. US inflation-linked bonds 6.88%, commodities 15.77%, gold 65.00%, UK index-linked gilts 1.33%.
US consumer prices, measured December to December on the Bureau of Labor Statistics index, rose 7.04% over 2021 and 6.45% over 2022. Then came 3.35%, 2.89% and 2.68%. UK CPI peaked at an annual rate of 11.1% in October 2022. The inflation sat in the first two years; everything after that is the disinflation.
That split is what the chart above plots. It shows each of the three dollar sleeves twice, once for the surge and once for the disinflation, as an annualised real return after US CPI. Every sleeve changes sign or nearly does.
Linkers are inflation-linked, and in 2022 duration beat the linkage
Inflation-linked bonds, in the Journal of Portfolio Management's description, "have coupon and principal payments that are indexed to the price level". In a year when consumer prices rise by more than a tenth, that uplift is large. Inside an index fund it is also swamped by something else.
Inflation-linked bonds are priced off real yields, meaning the yield left after the inflation adjustment. When real yields rise, prices fall, and the fall scales with duration. On 3 January 2022 the US 10-year par real yield was -0.97%. On 30 December 2022 it was 1.58%. That is a rise of 2.55 percentage points inside twelve months.
Now put duration next to it. At 31 July 2026 the fund tracking the UK index-linked gilt index carried an effective duration of 12.75 years. The US inflation-linked fund carried 6.57. The UK index is about 1.9 times as rate-sensitive, and in 2022 it fell almost three times as far: 34.48% against 12.60%. Duration explains the direction and much of the gap. The rest sits in how far UK real yields moved, which is not in the data fetched for this piece, so it isn't asserted here.
Across the full five years the UK index returned -36.49% while UK consumer prices rose 28.30% on the ONS index. In real terms that's a loss of 50.50%, over exactly the stretch that contained the 11.1% peak. If you had bought index-linked gilts in a fund because inflation was coming, the inflation came and the fund halved your purchasing power.
The pattern is old. Neville, Draaisma, Funnell, Harvey and Van Hemert, writing in the Journal of Portfolio Management in August 2021, found annualised real returns during US inflationary regimes of "-3% for the 2-year, -5% for the 10-year, and -8% for the 30-year bond". Longer is worse when inflation rises. That holds whether or not the bond is indexed, because the indexation changes the cash flows and duration decides the price.
Commodities did the job during the surge, then stopped
Across 2021 and 2022 the Bloomberg Commodity Index returned 47.56% while US consumer prices rose 13.94%. That's a real gain of 29.50% over two years, or 13.8% a year. Nothing else here came close during the inflation itself.
Across 2023, 2024 and 2025 the same index returned 12.35% while prices rose 9.18%. Real gain: 2.90% over three years, or 1.0% a year. It didn't lose money. It stopped doing anything special.
That's the shape the long-run work predicts. The same Journal of Portfolio Management study examined eight US inflationary regimes between 1926 and 2020. Commodities in aggregate had "a perfect track record of generating positive real returns during our eight US regimes, averaging an annualized +14% real return". In noninflationary times the same basket managed "around +1% real".
Read those two numbers together and the picture changes. Treating commodities as an inflation hedge means accepting long stretches of roughly nothing in exchange for the years when the price level moves. The 2021 to 2025 record is that bargain in miniature: two very good years, three ordinary ones. What decides whether the trade is worth taking isn't the 14%, it's how much of your holding period you expect to spend in the 1%. If you want the mechanics of why a commodity index return is not the same as the price of oil, the roll and collateral legs are covered in our note on commodity index returns.
Gold made its money after the inflation, not during it
Over 2021 and 2022 gold returned -4.16% in dollars while US prices rose 13.94%. In real terms that's a loss of 15.89%, worse than commodities by a wide margin and only a little better than the US inflation-linked bond index, which lost 18.69% in real terms over the same two years.
Then, over 2023 to 2025, gold returned 137.70% while prices rose 9.18%. Real gain: 117.70%, or 29.6% a year. Gold produced the best five-year number of the three by a distance, and produced nearly all of it once annual US inflation had dropped to 3.35% and below.
The long series says this is normal rather than surprising. The UBS Global Investment Returns Yearbook 2026 states plainly that "the relationship between gold and inflation is weak". Of the 28 years since 1900 in which inflation exceeded 3%, gold's return was negative in 13 of them, which is close to a coin toss. The Journal of Portfolio Management authors cite Erb and Harvey's finding that gold's apparent ability to hedge unexpected inflation "is driven by a single observation: 1979".
The long-run record puts gold between US bonds and bills
The Yearbook draws on the Dimson, Marsh and Staunton database, which starts in 1900. Since then the real US dollar gold price has risen 5.2-fold, "an annualized return of 1.3%".
Set that against the same database's US series over the same span. Real returns ran 6.6% a year for equities, 1.6% for bonds and 0.5% for bills, with inflation averaging 2.9% a year. Gold's long-run real return lands between government bonds and Treasury bills, and closer to the bills. It beat cash and it trailed the asset it is meant to protect you against holding.
The start date decides the answer, though. In the 54 years since the Bretton Woods system ended, annualised real gold returns were 4.7% in the US, 5.8% in the UK and 4.3% in Switzerland. That is a different asset from the one a 1900 start describes, and the difference is a change in the monetary regime rather than an inflation hedge working. Anyone quoting a gold return without its start date is quoting a choice, not a fact. Our piece on how much gold in a portfolio works through what that does to a weighting argument.
What a 5% inflation hedge sleeve did to the arithmetic
Sleeve returns are not portfolio returns, and the gap between the two is where most of this argument actually lives. On a 5% weight, held without rebalancing, here is what each sleeve contributed to a portfolio's total return in 2022, the worst year of the episode.
- UK index-linked gilts: -1.72 percentage points.
- US inflation-linked bonds: -0.63 points.
- Gold: -0.02 points.
- Commodities: 0.80 points.
And across the full five years, 2021 to 2025:
- UK index-linked gilts: -1.82 points.
- US inflation-linked bonds: 0.23 points.
- Commodities: 3.29 points.
- Gold: 6.39 points.
The spread between the best and the worst of the three dollar sleeves over five years came to 6.2 percentage points of total portfolio return. That is real money and it is far smaller than the sleeve headlines suggest, because a 5% weight scales every one of those numbers down. Picking the winner of this table was worth a few points of portfolio return across an entire inflation cycle, not the 65.00% or the -34.48% the individual lines show.
The strongest objection: an index is not the instrument
Here is the serious case against everything above, and it is specific to linkers.
An individual index-linked gilt held to redemption delivers its real yield whatever the price does in between. The 34.48% fall in 2022 is a mark-to-market loss on a fund tracking a 12.75-year-duration index, and a fund is worth the index price on the day. A holder of one bond, bought at a known real yield and held to its redemption date, got that real yield. A fund holder got the index.
So the fair reading is narrower than "linkers failed". Linker funds failed as a short-horizon hedge in 2022, because over a one-year window their duration risk is bigger than their inflation linkage. Whether they fail over a holding period matched to their duration is a different question, and this five-year table cannot answer it.
The paper that supplies the century of regime evidence disagrees with the 2022 reading too. It found US inflation-linked bonds "robust when inflation rises", with a "2% annualized real return" across the five most recent inflationary regimes. Its explanation for why the next episode might look different was published in August 2021, before the episode happened: "the starting TIPS yield in our inflationary regimes was +2.4%, whereas it is currently -0.9%. The low yield means that TIPS will be a very expensive inflation hedge going forward."
That is the reconciliation between the century and the five years. Inflation-linked bonds worked historically when they were bought at a positive real yield. In 2021 that wasn't on offer. The 2022 loss isn't evidence that the instrument is broken; it's evidence that the entry price was.
What this evidence cannot tell you
One episode is one episode. Five calendar years and a single inflation shock is a sample of one, and a ranking built on it describes rather than forecasts. Every backtest here, the century of regime data included, is history.
The currencies don't match. Three of the four series are in US dollars and the UK index-linked series is in sterling, so a sterling holder's gold and commodity returns differ from these by the exchange rate move. That move isn't in the data fetched for this piece, so it isn't stated.
Index returns aren't investor returns. These are index and net asset value figures with income reinvested, before platform fees, dealing costs, tax and any timing difference. A real holder got less, and how much less depends on the account.
The regime evidence carries its own limitations. The Journal of Portfolio Management study's inflation-linked series is synthetic before 1997, because the US Treasury didn't issue the instrument until then. And the Yearbook's caution about gold cuts both ways. If the link between gold and inflation is genuinely weak, then gold's 2023 to 2025 run may have had little to do with inflation either, which would make it the winner of this table for a reason the table doesn't capture.
What would change the conclusion
The entry real yield. On 26 August 2026 the US 10-year par real yield was 2.34%. On 3 January 2022 it was -0.97%. The Journal of Portfolio Management paper put the average starting yield across the historical regimes where inflation-linked bonds worked at 2.4%. On that measure the sleeve now enters a future shock on terms much closer to the ones that worked than to the ones that didn't, and 2022 stops being the relevant precedent.
The kind of shock. The commodity result comes mostly from supply-driven episodes. The paper's authors say as much, pointing at the 1970s oil embargo and the structural changes since, and warn that "caution needs to be exercised in interpreting the data". Inflation led by services rather than goods doesn't obviously reward a basket of futures on physical things.
What gold was actually pricing. If the 65.00% in 2025 was about something other than the price level, gold tops this table by coincidence, and the next inflation shock has no reason to repeat it.
A portfolio tool shows what a sleeve is worth today and how far its weight has drifted; LedgerTouch does that continuously. What no weight tells you is the real yield you bought at, and on this evidence that is the number that decided which of these three inflation hedge assets paid.
So the thing to watch isn't the CPI print. It's the real yield on the day the position goes on.