Key takeaways
- MSCI World returned 9.00% in local currency terms in 2016, 7.51% in US dollars and 28.24% in sterling. Same shares, same year, a 20.73 point gap.
- Over the ten years to 31 August 2026 that index ran at 14.86% annualised volatility in US dollars, 13.45% in euros and 11.92% in sterling.
- From 2012 to 2025 the annual currency contribution to returns averaged 0.51 points a year in sterling and -0.68 in US dollars, close to nothing either way.
- In the 2007 to 2009 fall MSCI World lost 54.79% in local terms and 38.28% in sterling. MSCI EAFE lost 55.39% local and 60.41% in US dollars.
- The sterling currency contribution correlated with the local return at -0.65 across 14 annual observations. For the US dollar the figure was plus 0.36.
Two people held the same index in 2016 and booked returns 20.73 points apart
You hold a global equity tracker. Someone in New York holds the same one. At the end of the year you compare notes and the numbers don't match, by a lot.
Neither of you has made a mistake. The return you see is two things multiplied together. There's what the shares did in the currencies they actually trade in, and there's what your own money did against those currencies. MSCI publishes both, side by side, for the same index.
In 2016 the MSCI World Index returned 9.00% measured in local currency terms. It returned 7.51% in US dollars, 10.73% in euros and 28.24% in sterling. Those are net returns on the same 1,280 companies over the same twelve months, and the gap between the best and worst base currency was 20.73 points. The second piece of that arithmetic is the currency contribution to returns, and it's what this page takes apart.
The short answer, stated up front: over 2012 to 2025 currency added almost nothing to the average return in any of the three base currencies, and it changed the risk a great deal. Which direction it changed the risk depended on where the investor lived.
The currency contribution to returns is a ratio, not a subtraction
Take the 2016 sterling case. The local return was 9.00% and the sterling return was 28.24%. Subtracting gives 19.24 points, and that's wrong, because the currency move applied to a portfolio that had already grown.
The right form divides. One plus the base currency return, over one plus the local return, minus one. Run 28.24% and 9.00% through that and the currency contribution to returns comes to 17.65 points, not 19.24. The difference is small here and gets larger as returns get larger.
What is the local return, exactly? It's the index measured as though every exchange rate stood still. MSCI calculates the EAFE local series in 13 different currencies, so "local" isn't one currency at all. It's the weighted basket of the markets the index holds. That matters for what follows: a sterling investor in MSCI World isn't exposed to one exchange rate but to a basket of them, in index weights.
MSCI also publishes a fully hedged version of each index. Its November 2019 methodology describes hedged indexes as representing the return from hedging an equity index in the 1-month forward currency market, with the hedge reset monthly and left alone in between. That hedged series is the practical comparator, because it's what an investor could actually buy. The local series is the cleaner one for decomposition, because it contains no interest differential.
Fourteen years of MSCI World in local terms, dollars, euros and sterling
Everything in this table is MSCI net returns for the same index. The local column comes from the hedged-index factsheets, which print it beside the base currency column, and the US dollar column from the MSCI World Index (USD) factsheet.
| Year | Local currency | US dollars | Euros | Sterling |
|---|---|---|---|---|
| 2025 | 18.44% | 21.09% | 6.77% | 12.75% |
| 2024 | 21.03% | 18.67% | 26.60% | 20.79% |
| 2023 | 23.12% | 23.79% | 19.60% | 16.81% |
| 2022 | -16.04% | -18.14% | -12.78% | -7.83% |
| 2021 | 24.17% | 21.82% | 31.07% | 22.94% |
| 2020 | 13.48% | 15.90% | 6.33% | 12.32% |
| 2019 | 27.34% | 27.67% | 30.02% | 22.74% |
| 2018 | -7.38% | -8.71% | -4.11% | -3.04% |
| 2017 | 18.48% | 22.40% | 7.51% | 11.80% |
| 2016 | 9.00% | 7.51% | 10.73% | 28.24% |
| 2015 | 2.08% | -0.87% | 10.42% | 4.87% |
| 2014 | 9.81% | 4.94% | 19.50% | 11.46% |
| 2013 | 28.87% | 26.68% | 21.20% | 24.32% |
| 2012 | 15.71% | 15.83% | 14.05% | 10.74% |
Compounded across the fourteen years, the local series returned 431.6%, the sterling series 455.9%, the euro series 432.0% and the US dollar series 381.3%. Annualised, that's 12.67% local, 13.04% in sterling, 12.68% in euros and 11.88% in US dollars.
Averaged over fourteen years the currency contribution to returns was close to zero
Run the ratio on every row and you get the annual currency contribution for each base currency. The averages are unremarkable. Sterling added 0.51 points a year, the euro added 0.19, and the US dollar subtracted 0.68. Over fourteen years, none of that is a return story.
The dispersion around those averages is the story. The annual sterling contribution had a standard deviation of 6.58 points and the euro 6.19. The US dollar contribution had a standard deviation of 2.20 points, because MSCI World is 72.14% United States by weight, so most of a dollar investor's holding carries no currency at all. The United Kingdom is 3.53% of the same index, which is why almost the whole thing is foreign money to a sterling investor.
Single years get large. Sterling's contribution was plus 17.65 points in 2016 and plus 9.78 points in 2022. It was -4.80 points in 2025. The euro contribution ran to -9.26 points in 2017 and -9.85 points in 2025. That last one has a visible cause: the euro rose 13.38% against the US dollar over 2025, from 1.0351 at the end of 2024 to 1.1736 at the end of 2025.
So a decision that averaged out to roughly nothing over the sample still decided several individual years. Averages hide that, and a table of annual returns does not.
Currency cut the volatility of unhedged global equities in sterling and raised it in dollars
MSCI reports annualised standard deviation of monthly net returns for each version of the index. Over the ten years to 31 August 2026, MSCI World measured 13.89% in local currency terms. In sterling it measured 11.92%. In euros 13.45%, and in US dollars 14.86%. The chart above plots those four figures.
Read that carefully, because it runs against intuition. The sterling investor has by far the most foreign currency in the position and the least volatility from it. Adding the currency took 1.97 points off the annual volatility of the equity holding rather than adding to it.
The dollar comparison needs a different index to be fair. MSCI World is mostly American, so a dollar investor's genuinely foreign equity is better represented by MSCI EAFE, which excludes the United States and Canada. Over the same ten years EAFE ran at 11.85% in local currency terms and 14.99% in US dollars. Currency added 3.14 points of volatility there, in the opposite direction to sterling.
That is the asymmetry in one line. The same asset, the same period, and foreign currency acted as a diversifier for one investor and an amplifier for another.
The 2007 to 2009 fall is where the asymmetry is easiest to see
Maximum drawdown makes the point more starkly than volatility does. MSCI puts MSCI World's worst peak-to-trough fall in local currency terms at 54.79%, between 12 October 2007 and 9 March 2009. In sterling the same index fell 38.28%, ending three days earlier. That's 16.51 points of loss that a sterling investor simply did not experience, because the pound fell while world equities did.
The euro investor got almost none of that. MSCI World's worst fall in euros was 53.60%, against 54.79% in local terms, and MSCI dates the start of it to May 2001 rather than to 2007. That's a different and much longer episode, which is its own warning about comparing drawdowns across base currencies. The euro is not sterling.
The dollar investor went the other way. MSCI EAFE fell 55.39% in local terms and 60.41% in US dollars over the same crisis. The dollar rose while the assets fell, so 5.02 points were added to the loss.
There is a mechanism behind this rather than a coincidence. The Bank for International Settlements looked at the broad dollar in its December 2020 Quarterly Review and found that a 1 percentage point appreciation shock to the dollar against a broad basket dampens the emerging market growth outlook by over 0.3 ppt, and growth-at-risk by 0.6 ppt. A rising dollar and deteriorating global conditions tend to arrive together. For anyone whose home currency is on the other side of that trade, the currency leg cushions the equity leg.
The correlation matrix, and the small print on fourteen observations
Here are the correlations between the annual currency contributions and the local currency returns, computed from the fourteen rows in the table above.
| Local return | USD contribution | EUR contribution | GBP contribution | |
|---|---|---|---|---|
| Local return | 1.00 | 0.36 | -0.39 | -0.65 |
| USD contribution | 0.36 | 1.00 | -0.87 | -0.49 |
| EUR contribution | -0.39 | -0.87 | 1.00 | 0.47 |
| GBP contribution | -0.65 | -0.49 | 0.47 | 1.00 |
The first column carries the finding. Sterling's currency contribution moved against the local equity return at -0.65, the euro's at -0.39, and the US dollar's moved with it at plus 0.36. The sign of that first column is what "accidental diversifier" means: nobody chose it, and for two of the three it helped.
The -0.87 between the dollar and euro contributions is close to mechanical. Much of the euro contribution to a euro investor's global index return is the euro against the dollar, and much of the dollar contribution is the same rate seen from the other side.
Now the caveat, which is large. Fourteen annual observations is a small sample. A correlation of -0.65 on fourteen points has a wide confidence interval, and a single year like 2016 does heavy work in it. Nothing here should be read as a stable parameter, and monthly data over a longer window would give a better estimate than this table can.
The strongest objection: in the 1975 to 2005 data, sterling sat on the other side
The most serious challenge to everything above is that the sign is not fixed. Campbell, Serfaty-de Medeiros and Viceira examined exactly this question in Global Currency Hedging, NBER Working Paper 13088, and their sample runs from 1975 to 2005. Their abstract reports that "the US dollar (particularly in relation to the Canadian dollar) and the euro and Swiss franc (particularly in the second half of the period) have moved against world equity markets".
The dollar, in their data, was the defensive currency. Sterling was not. They place their seven currencies on a spectrum. "The Japanese yen, the British pound, and the US dollar fall in the middle, with the yen and the pound more similar to the Australian and Canadian dollars, and the US dollar more similar to the euro and the Swiss franc." The Australian and Canadian dollars are the procyclical end of their range.
So over 1975 to 2005 sterling leaned the wrong way for a hedge, and over 2012 to 2025 it was the most defensive of the three. Two careful sources, opposite conclusions, and neither is wrong about its own sample. The honest reading is that these correlations are regime-dependent rather than structural. The same paper puts the euro and the pound at 68% correlated with each other, which is a reminder that the currencies are not independent bets either.
There's a second objection worth stating plainly. Currency has no expected return to speak of, so an unhedged position is not compensated for the risk it carries. The 0.51, 0.19 and -0.68 point averages above are consistent with that. If the risk-reducing correlation is unreliable and the return is zero, the case for unhedged exposure rests entirely on a relationship that has already flipped once in the published record.
What these figures cannot tell you
They're index returns, so they carry no fund charges, no platform fee and no spread. Anyone converting money to buy an overseas holding also pays a conversion charge, and the currency conversion costs on UK platforms are a separate and quite different question from the one measured here.
The hedged versus unhedged decision also carries a carry cost, set by short-term interest rate differentials, which the local currency column deliberately excludes. That cost is the other half of the arithmetic, and the size of it against a portfolio depends on how much of the portfolio is in equities at all. Portfolio currency exposure scales with the equity weight, and currency hedging is treated very differently for bonds than for equities for exactly that reason.
The sample is fourteen calendar years for the annual table and ten for the risk figures. Both are backtests over a period with one long equity bull market in it, and neither is a forecast. Calendar years are also an arbitrary cut: a currency move that straddles a December does not appear in either year as a large number.
Finally, the size of any of this depends on how much foreign equity is held in the first place, which is a prior question about home bias rather than about currency. A portfolio with a fifth of its value in global equities has a fifth of the exposure discussed here, and the split is countable from a holdings list before any of this arithmetic is run.
What would change the conclusion
If the dollar stopped behaving defensively. The whole asymmetry rests on the dollar rising when global equities fall. 2025 is a live test of that: world equities rose 18.44% in local terms while the euro gained 13.38% against the dollar. The euro contribution that year was -9.85 points, and since MSCI World is mostly US-listed, that single rate accounts for most of it. A run of years like that would flip the euro row and eventually the sterling one.
If sterling stopped falling in global risk-off episodes. The 2016 contribution of 17.65 points was not a global event. The pound fell 7.84% in a single day when the referendum result landed in June 2016, from 1.4800 to 1.3639, and 16.34% across that year, for domestic reasons. A currency that falls on domestic news can just as easily fall while world equities are rising, which is the same exposure with none of the benefit.
If your foreign holdings are bonds rather than equities. The arithmetic above is a currency contribution sitting on top of a double-digit equity return: 12.67% a year in local terms over the sample. A 6.58 point annual swing looks different against a sleeve returning low single digits, which is part of why the hedging convention for bonds is not the one used for equities.
The figure worth tracking isn't the exchange rate. It's how much of your equity is priced in money that isn't yours, and whether that share moves with or against the rest of the portfolio when things go wrong. Those two numbers are knowable in advance. The correlation that connects them is not.