Currency Risk in a World ETF: The Bet I Never Placed and Still Own

On the same day every month, a fixed amount leaves our household account and buys one world ETF. I set that standing order up years ago and I have thought about it maybe twice since, which was the entire point of setting it up.

Here is what I had never thought about at all: the currency risk in that world ETF. Three currencies pass through one boring transfer, and I held an opinion about exactly none of them. I am paid in koruna. The fund is quoted in dollars. The companies inside it earn their money in something like forty currencies I have never looked at.

I picked that fund on two things, cost and coverage. How I chose it is its own article, and nowhere in that decision does the word "currency" appear. Then this summer, mostly out of curiosity, I opened the index factsheet and read the currency breakdown.

72.85% US dollars. 8.47% euros. 5.73% yen. 3.38% pounds. 3.35% Canadian dollars, 2.33% Swiss francs, and a long tail of change. That is MSCI World as of 31 July 2026.

Data: MSCI World Index currency weights, MSCI factsheet, 31 July 2026.

What bothered me was not the number. It was that I could not have guessed it within thirty points. Largest position I own, bought every month for years, and put on the spot I'd have said "mostly dollars, I suppose" with the confidence of a man who has never checked.

So this article is me checking, properly, and then deciding. Not explaining a mechanism. Deciding.

Your fund's currency label is not your currency risk

Start with the misunderstanding sitting on top of every search result, because it wastes a lot of people's time.

The currency your ETF is quoted in tells you nothing about what you own. Buying the euro-priced line of the same global index instead of the dollar-priced line changes your currency exposure by zero. It is a display setting. The exposure comes from where the underlying businesses earn and hold their money, and swapping the label on the front of the fund does not touch any of it.

A hedged share class is a different thing entirely: the same equity portfolio with a stack of one-month forwards bolted on and rolled monthly, built to behave like the index measured in its own local currencies. MSCI's hedged versions track the local-currency index with correlations between 0.998 and 1.000 and tracking error of 0.31% to 0.89%, according to Vanguard's work on MSCI data going back to 1988. So a hedged fund is not a translation. It is an active decision to strip out the currency and pay for the privilege.

So there are two unchosen exposures riding in the same product. One is currency. The other I've written about before: your diversified index fund is now roughly a quarter ten stocks. Nobody picks either. They arrive in the box.

How much of your return was the exchange rate?

Here is the cleanest demonstration I know, and it is astonishing how rarely anybody puts it in one place. One index, one year, three answers.

In 2025 the MSCI World Index returned 18.44% in local currency terms. That is what the businesses did. A US investor got 21.09%. A euro investor got 6.77%.

Same companies. Same twelve months. A gap of 14.3 percentage points, produced entirely by the exchange rate.

Data: MSCI World Index annual net returns, MSCI factsheets, 31 July 2026.

Now the part nobody remembers, because we only notice currency when it takes something. In 2024 the euro investor got 26.60% on that index while the businesses underneath earned 21.03%. The exchange rate handed the European 5.6 free points and not one person opened a broker app to celebrate. Vanguard names the 2025 extreme even more starkly: a US investor made nearly 22% while a Swiss investor in the same global index made 6.3%.

Data: MSCI World Index net returns, MSCI factsheets, 31 July 2026. Derived figure — each bar is the home-currency net return minus the local-currency net return for the same year, a simple arithmetic attribution rather than MSCI's own decomposition.

Look at the sign flipping. Not the size, the sign. Eight good years and six bad ones for the euro investor, in no pattern anybody could have traded, and the two series are near mirror images of each other because they are two ends of the same exchange rate.

This is also half of an argument Americans have been having with themselves for a decade. International funds finally beat the US index in 2025, MSCI EAFE up 31.22% in dollars against 3.82% the year before. That was not foreign companies suddenly getting better at their jobs. It was the dollar falling 9.4% over the year, 10.8% of it in the first half alone — the worst first half since 1973. I argued the allocation side in the home-bias piece; the currency side is that a chunk of the performance gap people build portfolios around was an exchange rate wearing a costume.

Ten years of hedging, priced in euros

So does paying to remove it work?

MSCI publishes this comparison in euros rather than koruna, which is as close to my own bills as the public data gets. Over the ten years to 31 July 2026, MSCI World hedged 100% to euros returned 11.27% a year. The plain unhedged euro version returned 12.41%. The hedged investor gave up 1.14 percentage points a year.

Fine, that is the price of insurance. Except look at what the insurance delivered. Annualized standard deviation over the same decade: 13.91% hedged, 13.45% unhedged. The hedged version was more volatile. It also had the deeper maximum drawdown going back to 2001, at -55.43% against -53.60%.

Data: MSCI World 100% Hedged to EUR Index factsheet, 10 years to 31 July 2026, net returns.

Lower return and higher volatility, for a fee. That pair of numbers sits in a public PDF and I have never once seen it quoted.

Before anyone gets comfortable, the mirror. For a US investor over the ten years to 28 June 2024, MSCI EAFE hedged to dollars returned 9.08% a year against 4.33% unhedged, with volatility of 12.50% against 15.19%. That is double the return with a fifth less risk, and it is the strongest single fact on the pro-hedging side. Different index, different end date, and I am not pretending otherwise. What it mostly is, though, is one dollar cycle seen from the other end. The American who studied that evidence and hedged at the close of 2024 hedged himself directly into the worst possible year for it.

Now the bit where I disagree with almost everything written on this subject. Every page that tells you hedging is expensive points at the expense ratio, and the expense ratio is the small half of the bill. The fee gap between hedged and unhedged share classes of comparable world funds now runs from three basis points at the cheap end, 0.06% to 0.09%, up to thirty-five at the dear end, 0.20% to 0.55%. Three basis points is nothing. If the fee were the whole story, hedging would be almost free and I would have a much weaker article.

The durable cost is carry, and it never appears on the fund page. A hedge rolls forward contracts, and forward contracts price in the interest rate difference between the two currencies. Hedge from a lower-yielding currency into a higher-yielding one and you pay that gap every year, quietly, forever. Right now that gap for a euro investor hedging dollars is about 1.4 points, with US effective fed funds at 3.63% in July 2026 against an ECB deposit rate of 2.25%. Through most of 2025 it was closer to 2.3 points. UBS published the cleanest illustration of this I have found: over the twelve months to November 2025, a US corporate bond index returned 6.23% in dollars, minus 3.32% unhedged in euros, and plus 4.00% hedged to euros. The gap between the dollar figure and the hedged euro figure is the carry, and it moves whenever a central bank does.

Hedging isn't a fee. It's a floating fee, set by two committees who have never heard of you.

The best argument against me, and I am not hedging anyway

I'd be doing the cheap version of this if I stopped there. The strongest case against my position comes from a fund house that sells hedged products, which does not make it wrong.

WisdomTree ran rolling windows on MSCI World from March 1986 to December 2018 and asked a simple question: how often did currency exposure actually reduce volatility? The answer was 18.44% of three-year windows, 13.17% of five-year, 5.84% of ten-year, and 0% of twenty-year windows. Zero. Over the horizon I keep telling people to think in, currency exposure never once lowered risk. Median incremental volatility from currency came in around 0.70% to 0.96% depending on the horizon.

That is a direct hit on the lazy version of my own argument, the one that says "it washes out over a lifetime." It does not wash out. It adds a bit of noise and, on that evidence, keeps adding it.

Then the argument that actually made me sit down, because it is my argument. A pension fund owes money to retirees in a specific currency, so the fund's job is to hold assets that wobble the same way the liability wobbles. WisdomTree quote the actuary Steve Scoles on this and then extend it to people: what is true for a pension that owes in a currency is true for an individual who consumes in one. I write constantly that you should start from the life you want and work backwards. By that logic, a European whose mortgage, groceries and childcare are all denominated at home, holding 73% dollars, looks badly matched, and the disciplined move is to hedge.

Here is where I get off. The liability-matching frame is borrowed from an institution that owes a fixed nominal amount on a known date, and I do not have one of those. My liability is thirty-odd years of groceries, and groceries are not a bond. They are a real, drifting claim on the output of the global economy, and so is the thing I own. Under 44% of MSCI ACWI's revenue is earned in the United States, against 64% of its market value being domiciled there. The dollar label sits on top of a set of businesses already earning in currencies all over the world. A pension matching a nominal liability should hedge. Someone matching a lifetime of real consumption already holds something that moves with real global prices, and wrapping a nominal hedge around it does not tighten the match, it just swaps one mismatch for a different one and charges carry for it.

So: the volatility argument is real, I have not found a way around it, and I am not hedging anyway. I am knowingly buying a slightly bumpier ride for a lower running cost and one fewer thing to be clever about. The euro investor's actual ten-year outcome, 12.41% against 11.27%, is the version of that trade that already happened.

The hedged fund nobody wanted until the currency had already moved

The reason I feel comfortable accepting the extra noise is what people demonstrably do with the alternative.

WisdomTree's own Europe Hedged Equity fund is the case study. Morningstar's Daniel Sotiroff found flows into it climbed through the dollar's 2014 to 2015 run and peaked near the end of that run, then reversed as the currency did. From January 2010 to April 2019 the fund returned 7.8% a year. Its investors, weighted by when they actually put money in, earned minus 2.9% a year. A gap north of ten percentage points annually, on a fund that was doing its job the whole time. His Japan-hedged example shows the same shape at a milder 2.2 points. The biggest US-listed hedged developed-market fund today launched on 31 January 2014, into that same cycle.

Two honesty notes. Morningstar's flow-versus-currency exhibit is described in prose and the underlying monthly points are not published, so nobody can plot it, including me. And the whole dollar-weighted-return literature took a serious punch this May, when Fulkerson, Jordan, Riley and Yan argued in the Financial Analysts Journal that bad timing costs fund investors about 0.10% a year rather than the 1.2% Morningstar's own Mind the Gap reports. I think citing that makes the hedged-fund number stronger, not weaker. A ten-point annual gap is an order of magnitude beyond anything the critics attribute to methodology.

And the pattern is running again with the sign reversed. European hedged share-class assets went from $56.8bn in 2017 to $283.8bn in 2025. Morningstar's European "Other equity" bucket, which is mostly hedged share classes, took in EUR 19.2 billion during 2025, the year the dollar had already fallen 9.4%. I am not claiming those investors are wrong. I am pointing out that the money arrived after the move, exactly as it did in 2015, and that buying insurance the day after the fire is not risk management.

When currency risk in a world ETF is worth paying to remove

I would hedge, without hesitating, in four situations.

Foreign bonds and short-dated foreign cash. There is no argument here and nobody serious makes one. Look again at that UBS example: 6.23% in dollars, minus 3.32% unhedged in euros. If you hold foreign bonds unhedged you are not holding a bond, you are holding a currency trade that pays a coupon.

Money with a date on it. Froot's 1993 work is the practical filter: hedging reduces variance at short horizons and stops doing so at long ones, because real exchange rates drift back toward purchasing power parity. A house deposit for next year is short-horizon money and deserves protection. A pot you will not touch until 2050 is being protected from something that historically reverts.

Drawdown. Vanguard's framework scales hedge ratios to how much risk the portfolio itself carries, landing near 70% to 80% for a portfolio that is only a fifth equities. Someone living off a portfolio has sequence risk and currency risk stacking on the same withdrawal, and no decades left for the mean reversion to arrive. Even for an 80% equity portfolio Vanguard models 10% to 20%, not zero — their conclusion is not "never hedge equities," and I won't pretend it is.

And the genuine mismatch: you plan to spend in a currency you do not earn in, whether that is retiring somewhere else or carrying a mortgage in one currency on an income in another, which is a real hole that the "it evens out" line does not fill.

Perold and Schulman made the point decades ago that not hedging is itself a bet, and they are right. My answer is not that I avoided a decision — it is that this is one of a dozen defaults inside my portfolio, and the only question that matters is which defaults I can hold for twenty years without touching them.

What I do about the currency risk in my world ETF

One global fund, unhedged, bought on the same day every month. The exposure is accepted now rather than ignored, which is a smaller change than it sounds and the entire point of writing this.

I look at it once a year, when I rebalance, and not otherwise. Currency is a volatile series bolted onto an already volatile one, and I have watched myself trade volatile things badly enough to know the answer is fewer looks.

Three questions before you change anything:

1. When will I spend this money? Inside five years, protect it. Twenty years out, you are paying carry to smooth a line you will never look at again.
2. Would I have wanted this hedge before the move that is bothering me? If the honest answer is no, this is a reaction to a price, not a risk decision.
3. What does it cost every single year? Both halves. The fee, which may be three basis points, and the rate differential, which may be 1.4 points and is not printed anywhere on the fund page.

The reason almost nobody can answer question three properly is that the answer is scattered — a brokerage account here, a pension there, savings in a bank, maybe a property, and no single statement adding up what currencies you are actually standing in. That is a large part of why I built the multi-currency net-worth view in the tracker the way I did. A spreadsheet does the same job. The tool just meant I finally did it, and the 72.85% is what came back.

I am not changing anything. My daughter is a few months old, the second house is still doing its very unpassive best to eat my Saturdays, and my currency plan is that I keep getting paid in koruna, keep buying the same fund on the same day, and keep spending my attention on the two things that have actually moved my numbers: what I earn and what I don't spend.

The exchange rate will hand me a good year and then take it back. It did both in the last two. Somewhere around 2040 I would like to be teaching math badly on a Wednesday morning, and I am fairly confident the euro-dollar rate will have no opinion about whether I get there.