Start with the number that makes the standard rule look obvious. For a sterling-based investor holding an equally weighted global bond portfolio from 1975 to 2005, annualised volatility ran at 10.35% unhedged and 5.40% fully hedged. Hedging cut the risk roughly in half. Run the same test on global equities and the gap narrows to 16.78% against 13.86%. Real, but nothing like as dramatic.
Those figures come from Campbell, Serfaty-de Medeiros and Viceira's study of global currency hedging, and they are the arithmetic behind the advice most fund providers give: hedge your foreign bonds, leave your foreign equities alone. Vanguard states the rule in almost those words in its 2026 research on multi-asset currency hedging. It's a good rule. It isn't a law, and the reasons it works matter more than the rule itself.
Currency risk doesn't scale with the asset
A basket of developed-market currencies moves against a home currency at something like 8% a year. That's the volatility Vanguard assumes in its hedging framework, and it's roughly stable across the major pairs. What isn't stable is the volatility of whatever that currency is attached to.
Investment-grade bonds are a low-volatility asset. In the Campbell sample, a fully hedged global bond portfolio ran at 5.40% a year. Bolt an 8% currency exposure onto a 5.4% asset and the currency becomes the dominant source of risk. Unhedged, the same portfolio ran between 8.39% and 12.08% depending on the investor's base currency. The bonds stopped behaving like bonds.
Equities are different. A hedged global equity portfolio ran at 13.86% in the same study, and MSCI's own hedged World indices have sat close to 14% annualised over the past decade. Attach an 8% currency exposure to a 14% asset and the currency is a minority contributor. It can still matter, but it no longer drives the outcome. The BIS made the same point in plainer language in June 2025: equity portfolios are hedged less often than bond portfolios because equity return volatility dominates currency return volatility.
It's worth being clear about what a hedge does mechanically. Selling each foreign currency forward converts a foreign-currency index into something that tracks the local-currency version almost exactly. Vanguard measured this from January 1988 to December 2025: MSCI World hedged to dollars, yen, sterling and Australian dollars ran at 13.9%, 13.8%, 13.9% and 14.0% volatility against 13.8% for the local index, with tracking errors of 0.31% to 0.89%. A hedged fund isn't a bet on anything. It's the underlying market with the exchange rate stripped out.
The formula the rule is approximating
There's a one-line answer sitting underneath all of this. The variance-minimising hedge ratio equals one, plus the currency-asset correlation multiplied by asset volatility divided by currency volatility. Vanguard derives it in the appendix of its 2026 paper. Write it out and the rule of thumb falls straight out.
Take the bond case. Asset volatility of about 5%, currency volatility of about 8%, and a correlation near zero. The ratio of volatilities is 0.6, so even a fairly strong correlation barely moves the hedge ratio away from 100%. Now take the equity case. Asset volatility of 14%, currency volatility of 8%, so the ratio is 1.75. A correlation of minus 0.2 drops the optimal hedge ratio to 65%. A correlation of minus 0.5 takes it to zero.
That's why the answer is asset-dependent without anyone needing a currency forecast. Vanguard's modelled hedge ratios follow the same curve: roughly 70% to 80% for a portfolio with 20% in equities, falling to 10% to 20% for a portfolio with 80% in equities, across five investor domiciles. Hedge the bonds and leave the equities, and you land near those numbers by accident.
What hedging foreign bonds actually buys
The bond side of the rule is about as settled as anything in portfolio construction gets. Vanguard tested global bonds across five markets using monthly data from January 2000 to March 2023 and found the risk-adjusted returns of hedged global bonds were on average 2.9 times those of unhedged global bonds. Not 2.9% better. Nearly three times.
The same study asked the fairer question in reverse: how much extra return would unhedged bonds have needed to justify the extra volatility? For a 60/40 portfolio, between 0.4% and 1.8% a year depending on the currency. The realised currency returns over the period were smaller than that, and more often negative than positive.
Hedging also preserves the job bonds are meant to do. In the worst quartile of months for global equities, hedged global bonds delivered a tight range of outcomes, while unhedged global bonds delivered a much wider one and frequently failed to counterbalance at all. That's the whole argument for holding high-quality bonds, and currency exposure quietly breaks it. It's a useful thing to hold alongside the broader question of which asset classes actually diversified during past crises.
One neat illustration of the mechanics: in the year to 30 June 2017, a euro-area investor in short-dated euro bonds earned minus 0.07%, while a US investor holding the same portfolio hedged back to dollars earned 1.63%. The difference was roughly the gap between Fed and ECB policy rates. Hedging doesn't remove the interest-rate differential — it embeds it in the forward contract.
The equity answer depends on where you live
Here's where the rule earns its keep, and where it gets uncomfortable. MSCI publishes hedged and unhedged versions of the same World index, so the comparison is clean. As at 31 July 2026, over ten years, the MSCI World hedged to sterling had annualised volatility of 13.95%. Unhedged in sterling it was 11.93%. For a UK investor, hedging global equities raised volatility by two percentage points.
Now the same index from a dollar base. Hedged to the dollar, 13.87%. Unhedged in dollars, 14.86%. Hedging lowered volatility by one point. Same index, same decade, opposite sign — and the only thing that changed was the investor's home currency.
The mechanism is straightforward. Sterling tends to fall when global risk appetite falls, so foreign assets gain in sterling terms exactly when domestic ones are suffering. In 2016 the MSCI World returned 6.01% hedged to sterling and 25.62% unhedged in sterling. In 2022 it was minus 18.49% hedged and minus 9.32% unhedged. The dollar behaves the other way: it strengthens in a flight to quality, so unhedged foreign equity amplifies a dollar investor's losses. Deepest drawdown since 2001 for the sterling investor was 48.22% unhedged against 56.16% hedged. For the dollar investor it flips: 57.82% unhedged against 54.44% hedged.
Campbell's data shows the same asymmetry in a longer sample. From 1975 to 2005, a Canadian investor's unhedged global equity portfolio ran at 13.74% against 13.86% fully hedged — the hedge slightly increased risk. A Swiss investor's ran at 19.19% unhedged against the same 13.86% hedged. Commodity-linked currencies behave one way, safe havens the other.
Scale matters too. Vanguard's figures for 31 December 2025 put foreign-currency exposure in a market-cap global equity portfolio at 30% for a US investor and 96% for a UK one. A US investor barely has a currency decision to make. That makes the hedging question inseparable from how much of a portfolio's equities sit outside the home market, since the only way to eliminate currency risk entirely is to stop investing abroad.
And for most non-US investors, the currency decision is really a dollar decision. MSCI's World index carried currency weights of 72.85% dollar, 8.47% euro, 5.73% yen and 3.38% sterling at 31 July 2026. Hedging that index from a sterling base is, three-quarters of the time, a decision about sterling against the dollar. The aggregate position is enormous. The BIS put the notional value of outstanding FX derivatives at $130 trillion at end-2024, with almost 90% of contracts referencing the dollar on one side, and non-US investors held more than $17.6 trillion of US equities and $13.6 trillion of US bonds in March 2025.
The case against hedging equities at all
The strongest objection to any of this isn't about volatility. It's that hedging costs money, and the cost can be far larger than the volatility saved.
Hedging with forwards locks in the interest-rate differential between the two currencies. If your home rate is below the foreign rate, you earn that gap. If it's above, you pay it. The BIS defines the hedging cost precisely this way: the percentage gap between forward and spot rates, which equals the interest-rate differential adjusted for the cross-currency basis. For a Japanese investor over the past two decades, that gap has been brutal.
MSCI's Kokusai index makes it concrete. As at 31 July 2026, since January 2001, the version hedged to yen returned 5.33% a year against 7.33% for the same index in local-currency terms. Hedging cost 2.00 percentage points a year. What did it buy? Ten-year annualised volatility of 14.20% hedged against 15.97% unhedged — about 1.8 points of volatility. The Sharpe ratio since 2001 was 0.42 hedged and 0.55 unhedged. A Japanese investor who hedged global equities paid a lot and ended up worse off on a risk-adjusted basis.
That's not a hypothetical. Japanese life insurers cut their hedge ratio from about 60% in 2021 to 40% in 2024, and Taiwanese life insurers were near historic lows at roughly 65% in late 2024. High dollar rates made hedging expensive, and they responded. It's the same interest-rate machinery that drives the yen carry trade and Japan's intervention record.
Vanguard puts realistic all-in hedging costs at 0.25% to 0.50% a year, made up of transaction costs of up to 0.17% plus fund expenses. It also calculates a break-even: on an equity portfolio with 20% volatility, a 1.9-point volatility reduction would justify paying up to 0.76% a year. That maths works when the currency exposure is genuinely risk-adding. It doesn't work when the carry cost is 2.00 points, as it was for a yen investor.
The counter-argument cuts both ways, though, and being honest about that matters. Vanguard's own rolling ten-year figures show hedging reduced developed equity volatility by an average of 1.7 percentage points across the markets it studied from 1981 to 2025. On average, hedging equities helped. The dispersion is the problem: outcomes ranged from a 7.1-point volatility reduction in Japan for the decade to 2016 to a 3.5-point increase in Australia for the decade to 2010. Nobody knew which they'd get. The case for leaving equities unhedged isn't that hedging never works. It's that the benefit is unreliable, the cost is certain, and the currency is a small share of equity risk either way.
What would change the conclusion
Four things, and the first is the one that's already happened once.
Correlations flip. Before the 2008 crisis the euro looked like a safe-haven currency, which implied euro-area investors should hedge their foreign exposure. That relationship reversed afterwards and has since hovered near zero. Vanguard's 54-year sample shows ten-year currency-equity correlations swinging across a wide band in every market it examined. Any hedging policy that depends on a specific correlation holding can spend a decade being wrong.
Rate differentials converge or invert. The Japanese result above is mostly a story about near-zero yen rates against much higher foreign rates. If that gap closes, hedging yen exposure gets cheap and the calculation changes. The reverse applies to investors whose home rates are high — they've been paid to hedge, and they'd stop being paid if the gap narrowed.
The risk measure changes. Everything above uses volatility of total portfolio returns. Someone with liabilities in a foreign currency, or spending plans abroad, has a different problem entirely, and unhedged foreign assets may be the hedge rather than the risk. It's the same distinction that separates standard deviation from the drawdown an investor actually experiences.
Better forecasting arrives. Campbell's study tested whether varying currency positions with interest-rate differentials beats a fixed policy. It barely did: 12.45% volatility conditionally against 12.51% unconditionally, on an equity portfolio. Their conclusion was that there's little evidence risk-minimising investors should adjust currency positions in response to interest differentials. Anyone claiming a tactical hedging edge is arguing against that result.
Two caveats on the evidence itself. MSCI's hedged Kokusai index launched in November 2005, so figures running back to 2001 include back-tested data. And every number here is sample-dependent — the Campbell window ends in 2005, the Vanguard bond study ends in March 2023, and the MSCI figures are a snapshot at 31 July 2026. A different decade produces different volatility comparisons, which is the whole reason the rule is framed around relative volatility rather than realised correlations.
What survives all of it is the asymmetry. The bond case is robust across every base currency, every study and every window: currency volatility swamps bond volatility, so hedging preserves the asset's function. The equity case is genuinely close, and where an investor lives decides it.