Key takeaways
- The inflation uplift on TIPS and index-linked gilts runs three months behind. Over calendar 2022 the lag credited a TIPS holder 7.75%, while US consumer prices rose 6.45% December on December. The lag paid 1.29 percentage points more that year, not less.
- An iShares US TIPS fund with an effective duration of 6.56 years returned -12.68% in 2022. The 0-5 year version, duration 2.33 years, returned -2.86%. Same index, same uplift, 9.82 percentage points apart.
- UK RPI rose 13.44% over 2022 and the lagged uplift on a linker was 14.17%. The iShares index-linked gilt fund, duration 13.08 years, still returned -34.55% — 10.72 points worse than the conventional gilt fund beside it.
- The reason is the real yield. The US 10-year real yield went from -1.04% at the end of 2021 to 1.58% at the end of 2022, a rise of 262 basis points, and duration multiplies that.
- Over the five years to 30 June 2026 the index-linked gilt fund annualised -8.18% a year. The short-dated TIPS fund annualised 3.27%.
The uplift is a multiplier on your principal, and it arrives three months late
You bought an inflation-linked bond because inflation was coming. Inflation came. You lost money. Here's the mechanism that produced that, and it isn't a pricing error.
An index-linked gilt doesn't pay you extra interest when prices rise. It grows the principal, and the coupon is a fixed percentage of a growing number. The UK Debt Management Office puts it plainly: "The semi-annual coupon payments and the principal repayment are adjusted in line with the UK Retail Prices Index ... with a lag." That index is the RPI, the older of the two UK inflation measures.
The adjustment is done with a single number called the Index Ratio. The DMO defines it as "the ratio of the reference RPI applicable to that date divided by the reference RPI applicable to the original issue date of the gilt." TIPS work the same way. Under the US Treasury's offering circular, "Interest payments are based on the security's inflation-adjusted principal at the time we pay interest. We make this adjustment by multiplying the par amount of the security by the applicable Index Ratio."
The lag lives inside that reference index. Both markets use a three-month delay, for the dull reason that a price index for a month isn't published until well into the next one. The DMO's rule: "The reference RPI for the first calendar day of any calendar month is the RPI for the calendar month falling three months earlier." The US rule is word-for-word the same idea, in 31 CFR Part 356: "The Ref CPI for the first day of any calendar month is the CPI for the third preceding calendar month." Days in between are filled in by straight-line interpolation.
Gilts weren't always built this way. The DMO's formulae document records that "Index-linked gilts first issued from April 2005 employ the three-month lag indexation technique first used in the Canadian Real Return Bond ... market, rather than the eight-month lag methodology previously used." The older design left a much bigger gap between the inflation you lived through and the inflation you were paid for.
In 2022 the lag paid more than the year's inflation, not less
The usual complaint about the lag is that it short-changes you. Run the arithmetic on the year everyone remembers and the opposite happened.
Start with the US. The Index Ratio on 1 January 2022 used the October 2021 CPI-U, which the Bureau of Labor Statistics reports as 276.589. The ratio on 1 January 2023 used the October 2022 CPI-U, 298.012. Divide one by the other and a TIPS principal grew 7.75% across calendar 2022.
Now measure the year itself. CPI-U was 278.802 in December 2021 and 296.797 in December 2022, a rise of 6.45%. The lagged uplift beat contemporaneous inflation by 1.29 percentage points.
The UK story is the same shape. October 2021 RPI was 312.0 and October 2022 RPI was 356.2, so the Index Ratio grew 14.17% over 2022. RPI itself went from 317.7 in December 2021 to 360.4 in December 2022, a rise of 13.44%. The lag was worth an extra 0.73 points.
That's not luck so much as arithmetic about acceleration. The lag doesn't shrink your inflation compensation. It shifts the window three months earlier. When inflation is speeding up through the middle of a year, the shifted window catches more of the surge and less of the calm. When inflation is falling, the same mechanism works against you, and you spend a quarter being paid for price rises that have already stopped.
What the lag really costs is the last three months before you're paid
The genuine cost of the lag isn't annual. It's terminal, and it sits at the end of the holding period.
Because the reference index stops three months short, the final months of inflation before a redemption never make it into the payment. The DMO is specific about the timing: "the redemption payment and the final dividend payment on an index-linked gilt with a three-month lag will typically be fixed around 2-6 weeks before the redemption date." So a holder knows the cash amount weeks in advance, and the inflation of those last weeks is somebody else's problem.
Treasury flags the trading consequence in its own risk disclosure, which is a more candid sentence than most fund literature manages: "The calculation of the index ratio incorporates an approximate three-month lag, which may have an impact on the trading price of the securities, particularly during periods of significant, rapid changes in the index."
Read that as a description of a mismatch, not a defect. Over a full life from issue to redemption, a linker pays you the inflation from three months before issue to three months before redemption. That's the same length of window. It just isn't the same window as your own spending.
An index-linked bond is still a bond, and 2022 proved it four times over
Here is what actually happened to money in 2022, from four fund factsheets dated June 2026.
The iShares $ TIPS UCITS ETF holds US inflation-linked government bonds across maturities. Its effective duration is 6.56 years and it returned -12.68% in 2022. The iShares $ TIPS 0-5 UCITS ETF holds the same instruments, capped at five years to maturity, with an effective duration of 2.33 years. It returned -2.86%.
Both funds received the same 7.75% uplift, because it comes from the same reference CPI. The gap between them — 9.82 percentage points — is duration, and nothing else of consequence.
The UK version is starker. The iShares Index-Linked Gilts UCITS ETF reports an effective duration of 13.08 years and returned -34.55% in 2022. The conventional gilt fund next to it, iShares Core UK Gilts, reports 7.21 years and returned -23.83%. The inflation-protected fund lost 10.72 points more than the unprotected one, in the year UK inflation was highest.
Index-linked gilts sit at the long end because that's where the buyers are — pension schemes matching liabilities decades out. A linker index is therefore a long-duration index by construction, and a reader who owns one has taken a duration position whether or not they meant to.
Divide the uplift back out and the real yield is doing all the damage
The two forces can be separated with one division.
Take the US fund's 2022 total return of -12.68% and divide it by the 7.75% uplift it received. What's left is a fall of about 18.96% in the real price of the bonds — the price you'd see if inflation had been zero all year. That figure ignores coupon income and the index's own turnover, so treat it as the shape of the answer rather than the answer.
Now check it against the discount rate. The US 10-year real yield — the yield quoted on TIPS, after inflation compensation — was -1.04% on 31 December 2021 and 1.58% on 30 December 2022. That's a rise of 262 basis points. Multiply by the fund's 6.56-year duration and the rule of thumb predicts a 17.19% price fall, against the 18.96% implied above.
The estimate lands slightly short, and the direction is expected. Duration is quoted as of June 2026; at end-2021 real yields were lower and durations correspondingly longer, so the true multiplier that year was bigger. Why the rule of thumb drifts on moves this large, and by how much, is worked through in our piece on what a 1% rate move actually costs.
The 5-year real yield moved further still, from -1.61% to 1.66%, a rise of 327 basis points. That the short fund lost so little despite the bigger yield move is the whole point about duration.
For the UK fund, dividing the -34.55% return by the 14.17% uplift implies a real price fall of roughly 42.67%. Inflation linkage handed the holder a seventh of their capital and the repricing of real yields took nearly half.
The strongest case against reading 2022 as a failure of linkage
There's a serious objection to everything above, and it deserves stating properly.
A mark-to-market loss on a bond you hold to maturity isn't a loss at all. An index-linked gilt bought at issue and held to redemption pays its uplifted principal whatever happened to real yields in between. The 2022 drawdown was a repricing of a claim, not a reduction in the claim, which is the distinction between volatility and the risk you actually feel. Anyone matching a known real liability — a pension in payment, school fees indexed to RPI — got exactly the hedge they paid for, and the screen price was noise.
The objection is weaker for a fund, which never matures and keeps rolling into new bonds. But even there, the repricing that destroyed the capital rebuilt the yield. The US 10-year real yield stood at 2.40% on 7 August 2026, against the -1.04% of end-2021. Buyers now are paid a real return that simply wasn't available then.
The recovery has been very uneven, which is the part critics of the buy-and-hold answer get right. Over the five years to 30 June 2026 the 0-5 year TIPS fund annualised 3.27% and the all-maturity TIPS fund 0.78%. The index-linked gilt fund annualised -8.18% over the same five years, against -4.20% for conventional gilts. Four years after 2022, the long UK linker holder is still a long way from whole.
One structural difference is worth knowing, because it's contested design rather than market outcome. TIPS carry a floor: TreasuryDirect states that "when a TIPS matures, you get either the increased (inflation-adjusted) price or the original principal, whichever is greater." Index-linked gilts carry no such protection. The DMO settled that in its December 2004 consultation on the design: "There will be no deflation floor on the principal repayment value", and the redemption payment "could fall below 100 per 100 nominal."
What these numbers can't tell you
The fund durations are dated 30 June 2026 and the returns are from 2022. Applying one to the other is a rough test, and the error runs one way — real durations were longer when real yields were lower. Nothing here reconstructs what those funds held on 31 December 2021.
The uplift arithmetic assumes the reference index moves from 1 January to 1 January. A real holding starts and ends on other dates, with interpolation between them, so an individual bond's ratio differs from the round-number version above.
The US returns are in dollars and the UK returns in sterling, so the two pairs aren't directly comparable across the Atlantic. A sterling investor in a dollar TIPS fund experienced 2022 differently again, because the dollar rose.
RPI and CPI-U aren't the same measure, and neither of them is measuring your basket. UK linkers uprate against RPI, US TIPS against CPI-U, so the two halves of this piece are not like-for-like. Both indices are taken here as published rather than adjudicated.
Above all, this is one rate cycle in two countries, which is a sample of one. A single year in which the US 10-year real yield rose 262 basis points is a case study and not a forecast, and it says nothing about what real yields do next.
What would change the conclusion
If real yields fall from here, the same duration that cost 2022 pays it back. The arithmetic is symmetric: a 13.08-year fund gains on a fall in real yields roughly as hard as it lost on the rise. The starting point now is 2.40% on the US 10-year rather than -1.04%, so there's further to fall than there was.
If you shorten the maturity, the mechanism barely changes but the outcome does. The 0-5 year TIPS fund got the same 7.75% uplift and lost 2.86% instead of 12.68%. A short linker is close to a pure inflation instrument; a long one is a real-yield instrument that happens to be indexed. Where the yield sits on the curve does the rest of the work, which is the argument in real yields and discount rates.
If inflation decelerates sharply, the lag reverses its 2022 favour. The 1.29 percentage points it added in the US that year become a subtraction when the three-month-old print is higher than today's.
And if the index itself is redefined, the contract changes underneath a UK holder. That risk is already scheduled. The UK Statistics Authority has said of bringing CPIH methods into RPI that "the change we propose can legally and practically be made by the Authority in February 2030", and the Chancellor tied his consent to "the maturity of the final specific index-linked gilt in 2030". That's a legislative event, not a market one, and no duration number captures it.
The number to watch isn't the inflation print. It's the effective duration on your own factsheet, sitting next to the words "index-linked". LedgerTouch tracks that alongside the rest of a portfolio. Multiply it by 2.62 — the move the US 10-year real yield made in a single year — and see whether the answer looks like protection.