Key takeaways
- Over the ten years to 31 July 2026, an unhedged MSCI World sleeve beat the fully hedged version by 0.26 percentage points a year in sterling, 1.14 in euros and 7.85 in yen.
- That sleeve level gap reaches a portfolio multiplied by the equity weight. At 60% equities it was worth 0.16 points a year to a sterling investor and 4.71 to a yen one.
- MSCI World held 3.38% of its currency weight in sterling at 31 July 2026, so an unhedged 60% equity sleeve put 57.97% of the whole portfolio into foreign currency.
- Hedging raised ten year volatility for the sterling investor, from 11.92% to 13.93%, and cut it for the yen investor, from 15.97% to 14.20%.
- FTSE's World Government Bond Index returned minus 3.22% a year unhedged to sterling over the five years to 31 December 2025, against minus 1.23% hedged.
The gap on the fund page is not the gap that reaches your portfolio
You're looking at two versions of the same global tracker. One is hedged to your currency and one isn't. What you want to know is what the choice does to your portfolio, not what it does to the fund.
The arithmetic is short. A hedged and an unhedged version of one index differ by some annual return gap, and that gap reaches your portfolio multiplied by the weight of the sleeve you put it in. Over the ten years to 31 July 2026, MSCI's own net return figures put that sleeve level gap at 0.26 percentage points a year for a sterling investor and 7.85 points for a yen investor.
Now scale it. Hold 20% of the portfolio in equities and the sterling version of that decision was worth roughly 0.05 points a year. Hold 100% and the yen version was worth 7.85. Same box on the same form. One of them is a rounding error, and the other is the largest single line in the decade's outcome.
Which one you got depended on two things you fixed before you ever opened the fund page: your base currency, and your equity weight.
Portfolio currency exposure is an equity weight question before it's a currency question
The exposure is countable, so start there. At 31 July 2026 the MSCI World index carried currency weights of 72.85% US dollar, 8.47% euro, 5.73% yen and 3.38% sterling. For a sterling investor, 96.62% of that equity sleeve sits in money that isn't theirs.
Multiply by the sleeve weight and you get the portfolio number. A portfolio with 60% in unhedged global equities carries 57.97% of its total value in foreign currency. At 20% equities it's 19.32%. At 80% it's 77.30%. Nothing about the currencies changed across those three lines. Only the weight did, and that's the point: portfolio currency exposure is set by the allocation first, and by the hedging decision second.
The global equity allocation therefore decides the size of the exposure before anyone ticks a box. How much of the portfolio sits abroad at all is a prior question, and one that home bias settles.
A hedged index removes that exposure mechanically rather than by taking a view on the currency. MSCI describes it plainly: the index "is 100% hedged to the GBP by selling each foreign currency forward at the one-month Forward rate". The contract rolls every month. What's left tracks the same shares priced in their own currencies.
Three base currencies, one index, and a ten year gap running from 0.26 points to 7.85
MSCI publishes a hedged version, an unhedged version and a local currency version of the same index side by side, which makes the comparison unusually clean. Everything below is net returns, annualised over the ten years to 31 July 2026.
| Base currency | Unhedged | Fully hedged | Local currency | Gap |
|---|---|---|---|---|
| Sterling | 12.58% | 12.32% | 13.03% | 0.26 points |
| Euro | 12.41% | 11.27% | 13.03% | 1.14 points |
| Yen | 18.11% | 10.26% | 12.99% | 7.85 points |
The sterling and euro rows are MSCI World. The yen row is MSCI Kokusai, which MSCI also calls the World ex Japan index: it covers 22 of the 23 developed markets, leaving Japan out, because a Japanese investor's foreign equity doesn't include their own market. It hedges differently too, by selling the dollar forward rather than each foreign currency in turn. So the yen row holds within itself rather than against the two above it.
The gap splits into a carry cost you can price and a currency move you can't
The local currency column is the same shares with no exchange rate in them at all. Compare each version against it and the gap comes apart into two pieces.
Hedged minus local is the carry, which is what the forward contract cost. It ran at minus 0.71 points a year in sterling, minus 1.76 in euros and minus 2.73 in yen. That ordering isn't an accident. The BIS puts the mechanism plainly: for investors outside the United States, hedging costs "rise as the difference between short-term dollar interest rates and the equivalent rate in local currency widens". Through 2022 it measured that gap widening from 0.7% to 3.5% for the euro against the dollar, and from 0.3% to 5.5% for the yen. By the end of 2022 the yen hedger was paying more than the euro hedger, which is the ordering the carry column shows. That's the currency hedging cost, and it's the half of the decision you can price in advance.
Unhedged minus local is the currency move: minus 0.45 points in sterling, minus 0.62 in euros and plus 5.12 in yen. The first two are small. The third is the yen's decade, and it isn't priced into anything you can read before you buy.
At a 20% equity weight the decision was a rounding error. At 100% it wasn't.
For a portfolio rebalanced back to its weights, a return gap on one sleeve reaches the total at roughly that sleeve's weight. It's a first order approximation, and at these magnitudes it's close enough to work with. Portfolio currency exposure and the size of the hedging decision move together, because both are the sleeve weight multiplied by something.
The chart above plots the 60% and 100% columns of that arithmetic. Here is the fuller version.
| Equity weight | Sterling base | Euro base | Yen base |
|---|---|---|---|
| 20% | 0.05 points | 0.23 points | 1.57 points |
| 60% | 0.16 points | 0.68 points | 4.71 points |
| 100% | 0.26 points | 1.14 points | 7.85 points |
Read down a column and the size of the currency decision changes by a factor of five. Read across a row and it changes by more than an order of magnitude. The sterling investor at 20% equities and the yen investor holding nothing but equities were filling in the same box on the same kind of form. One was deciding about 0.05 points a year. The other was deciding about 7.85.
The bond sleeve runs the other way, which is why the standard rule is asymmetric
All of that is the equity sleeve. The bond sleeve behaved differently, and the direction was consistent. FTSE Russell's figures for the World Government Bond Index, published in February 2026 and measured to 31 December 2025, put the five year annualised return at minus 3.22% unhedged to sterling against minus 1.23% hedged. In euros it was minus 2.74% against minus 2.55%. In yen it was plus 4.87% unhedged against minus 4.49% hedged.
The yen row flips again, for the same carry reason as before. What's consistent is the spread. In 2025 alone a dollar investor's unhedged holding of that index returned 7.55%, while a euro investor's lost 5.18%. The hedged versions returned 3.79% and 1.63%. FTSE Russell's reading is that "unhedged returns showed wide dispersion", while "currency-hedged results were more tightly clustered across base currencies".
The reason is the size of the asset sitting under the currency. Dimson, Marsh and Staunton measure it across 20 foreign countries in the 2026 Global Investment Returns Yearbook, covering 1900 to 2025 and the shorter run from 1972. Currency risk, they write, "on average added around 6 percentage points to total risk whether we focus on equities or bonds, although currency risk adds proportionally more to the risk of bond portfolios". Six points bolted onto an equity sleeve is a minority of its risk. Six points bolted onto a government bond sleeve is a far larger share of it, which is what proportionally more means here. That asymmetry is the basis of the standard currency hedging rule.
At the portfolio level it has a consequence that's easy to miss. A 20/80 portfolio holds most of its money in the sleeve where currency risk is large relative to the asset. An 80/20 portfolio holds most of its money in the sleeve where it isn't.
The counter case: measure risk instead of return and one row changes sign
Everything so far is return, and return is the friendlier measure here. Run the same three bases on volatility and the answer stops agreeing with itself.
Over the same ten years to 31 July 2026, hedging raised annualised volatility for the sterling investor, from 11.92% unhedged to 13.93% hedged. That's 2.01 points of extra volatility bought with the 0.26 points of return given up. For the euro investor hedging added 0.46 points, 13.45% against 13.91%. For the yen investor it cut 1.77 points, 15.97% down to 14.20%.
Widen the window and the ranking reverses outright. Measured from 31 January 2001, the deepest fall the unhedged Kokusai index ever took was 65.70% peak to trough, in the crash that bottomed in March 2009, against 55.26% for the hedged version. That's 10.44 points deeper, and it sits outside the ten years of returns above, which is the point rather than a technicality. The decade that hands the yen investor 7.85 points a year simply doesn't contain the episode where the choice hurt most. The distance between a volatility figure and a loss someone lived through is its own subject, covered in volatility vs drawdown.
The one cost that doesn't depend on which way the currency went
There's a charge attached to hedging that shows up whatever the exchange rate does, and it's printed on the factsheet. Take the iShares Core MSCI World UCITS ETF. Its unhedged US dollar accumulating share class carries a total expense ratio of 0.20%. Its sterling hedged distributing share class carries 0.30%. Both factsheets report fund net assets of 147,709.09 million dollars as at 31 July 2026, because it's one fund holding one set of shares.
The hedged share class therefore costs 0.10 points a year more. The two classes differ in whether they pay income out as well as in the hedge, so that gap isn't a pure hedging fee. It is the price difference a buyer faces. Scale it the way everything else here scales: a 60% equity sleeve makes it 0.06 points a year on the portfolio, and a 20% sleeve makes it 0.02. Set against a sterling gap of 0.26 points, a fixed charge that size is a real share of the whole decision. Set against a yen gap of 7.85 points, it's noise.
What these figures cannot tell you
Ten years is one sample, and this one has a single dominant fact in it. The dollar was 72.85% of MSCI World's currency weight at 31 July 2026, so most of what the unhedged column measures is the dollar against one other currency at a time, over a single decade. A different decade produces a different table, and the yen row in particular sits in the tail of that distribution rather than the middle.
The indices aren't identical either. The yen row uses MSCI Kokusai and the other two use MSCI World. The hedged sterling index launched on 1 May 2006 and the hedged Kokusai index on 9 November 2005, and MSCI states that data before a launch date is back tested. The ten year window is live data. The figures running back to 2001 are not entirely.
An index is also not a holding. Every figure above is an index return before the charge a fund adds, and what an investor keeps depends on fund domicile and on the account the fund sits in.
The bond figures are measured to 31 December 2025 and the equity figures to 31 July 2026, which is why no combined portfolio return is computed from the two of them here. One limit matters more than the rest. Return gaps scale with weight; volatility does not, because portfolio volatility depends on how the sleeves move against each other, and a factsheet can't tell you that. Every risk figure above is a sleeve figure, not a portfolio figure.
What would change the conclusion
If rate differentials converge. The carry line moves towards zero and the decision collapses into the currency move alone. The minus 2.73 points a year a yen hedger paid is the short-term rate gap the BIS describes, charged month after month through the forward.
If the dollar's decade runs backwards. The plus 5.12 points a yen investor collected from currency was the yen falling. A stretch in which it rose turns that number negative, and takes the unhedged advantage with it.
If your spending isn't in your base currency. Every figure here measures returns in the currency you report in. Someone whose costs are denominated abroad has a different problem, and unhedged foreign assets may be the hedge rather than the risk.
If you change what you're measuring. The sterling row says hedging added 2.01 points of volatility and cost 0.26 points of return. Both are decade specific. The structure isn't: the weight of the sleeve sets the size of the decision, whatever a given decade does with the sign.
The number worth tracking isn't on the fund page at all. It's the portfolio currency exposure you actually carry, the share of your own money sitting in currencies you don't spend, and it moves every time the allocation moves rather than only when you switch share class. LedgerTouch reports that share continuously; a spreadsheet reports it whenever you rebuild it. Either one answers the question the fund page can't, which is how big your currency decision currently is.