International diversification and home bias: what correlation says

On this page

Key takeaways

  • Investors in most countries hold much more of their home market than its share of the world's stock market value, a pattern documented by French and Poterba in 1991 and known as home bias.
  • Adding foreign stocks lowers a portfolio's volatility only to the extent that they do not move in lockstep with home stocks; the benefit depends on the correlation.
  • US and international stock returns have been highly correlated in MarketHeist's monthly data since the late 1990s, so the volatility reduction from mixing them has been modest; the correlation has also moved over time, and research reports lower readings in earlier decades.
  • Correlations between stock markets tend to rise in global sell-offs, which is when diversification is wanted most; over long horizons, differences in each country's growth and valuation matter more than month-to-month co-movement.
  • MarketHeist measures it on monthly total returns, both over the full shared history and over rolling three-year windows.

Definition

International diversification is spreading a stock portfolio across countries instead of holding only the home market; home bias is the tendency of investors to hold far more of their own country's stocks than its share of world markets. How much spreading lowers risk depends on how closely the markets move together, which has varied over time.

Also known as: home bias, home country bias, international stocks, US vs international stocks correlation, global diversification, ex-US stocks, owning international stocks.

How it works

A portfolio of only home-country stocks bets on one economy, one currency and one set of companies and sectors. Adding stocks from other countries replaces part of that single bet with others. How much risk that removes comes down to one number: the correlation between the two markets' returns. At +1 they move as one and mixing them changes nothing; the lower it is, the more one market's bad months are softened by the other's.

For the volatility of a mix, correlation enters through the cross term of the two-asset formula below. With two markets of similar volatility held half and half, a correlation near 0.9 leaves the mix almost as volatile as either market alone, while a correlation near 0.4, closer to what studies report for the 1970s and 1980s, lowers it noticeably. That is why the rise in measured correlation since then has shrunk the short-run volatility benefit of going global.

Correlation is not the only benefit. It measures month-to-month co-movement, not whether two markets end a decade in the same place. Two markets can be highly correlated month to month and still deliver very different decade-long results, because their earnings growth, valuations and currencies differ: US stocks lagged international ones over the ten years from 2000 to 2009 and led them by a wide margin afterwards. International diversification spreads that long-run, which-country-wins risk, which a monthly correlation does not capture. Asness, Israelov and Liew (2011) put it as diversification failing in short crashes but working over long horizons.

Home bias has several explanations: familiarity, currency risk, higher costs and taxes abroad, and the fact that home companies already earn part of their profits overseas. A useful comparison is the world market itself: an investor holding only home stocks is making a large active bet against the global mix, whether they mean to or not. Home bias is like only ever shopping at the grocery store on your own street: often fine, but you never find out what the rest of the town sells, or what happens if your street closes.

Formula

Volatility of a two-market mix
σp=wh2σh2+wf2σf2+2whwfρσhσf
σₚ
volatility of the mix
w_h, w_f
weights in home and foreign stocks, summing to 1
σ_h, σ_f
volatilities of the home and foreign markets
ρ
correlation of their monthly returns
Half and half, with equal volatilities
σpσ=1+ρ2
σ
the volatility of each market, assumed equal
σₚ / σ
the mix's volatility as a fraction of either market's
ρ
correlation

At ρ = 0.9 the ratio is about 0.97; at ρ = 0.8 about 0.95; at ρ = 0.4 about 0.84; at ρ = 0 about 0.71. The benefit shrinks quickly as correlation approaches 1.

Rolling correlation over a trailing window
ρt(W)=ρ(ht−W+1…ht,ft−W+1…ft)
hₜ, fₜ
monthly total returns of home and foreign stocks in month t
ρₜ(W)
Pearson correlation over the W months ending in month t
W
window length in months (36 by default)

Worked example

The US total stock market (VTI) and total international stocks outside the US (VXUS), monthly total returns from May 1996 to Sep 2026, correlated over the whole window and over each of the 330 trailing 36-month windows.

Over the whole window the correlation between US and international stocks was 0.85: high, though not 1.

It stayed high throughout. Every trailing 36-month window had a positive correlation, and 100% of them were at 0.5 or above. The highest reading was 0.94, in the window ending Apr 2005, and the lowest 0.66, in the window ending Feb 2026.

Readings of 0.9 and above came in windows that covered the sell-offs of 2000 to 2002, 2008 and 2020, when markets everywhere fell together: the moment a diversifier is wanted most is the moment the correlation rose. The latest reading, 0.72, sits below the full-window figure, as US and international stocks have recently moved apart more than usual.

What this means for a portfolio: at correlations of this size, splitting stocks between the US and the rest of the world lowers month-to-month volatility only a little. The case for international stocks rests less on that and more on not betting the whole stock portion on one country's long-run result.

Data through Sep 2026. Reproduce id d64f854efaf8c50c (this guide, its lab setup and the data month).

Interactive lab

The trailing 36-month correlation between US and international stock returns, with the full-window correlation as a reference line. Switch the international side to Europe, Japan or emerging markets, or the window length, to see how stable each relationship has been.

Rolling 36-month correlation: VTI and VXUSPearson correlation of monthly returns over each trailing 36-month window, 330 windows, dated at the last month of the window; May 1996 to Sep 2026. Whole-window correlation 0.85; lowest 0.66 (window ending Feb 2026), highest 0.94 (window ending Apr 2005), latest 0.72.0.000.100.200.300.400.500.600.700.800.90199920032007201120152019202300.536-month correlation
Correlation of the two series' monthly returns over each trailing 36-month window. Dashed lines: 0 and 0.5. May 1996 to Sep 2026.
MeasureVTI and VXUS
Correlation over the whole window0.85
Trailing windows of 36 months330
Lowest 36-month correlation0.66 (window ending Feb 2026)
Highest 36-month correlation0.94 (window ending Apr 2005)
Latest window0.72
Windows above 0100%
Windows at or above 0.5100%

Pearson correlation of monthly returns. Each window ends at the month shown and covers the 36 months up to and including it. May 1996 to Sep 2026.

How MarketHeist computes it

Every choice an implementation has to make, as the code makes it today. The page is checked against the code each time it is built.

ChoiceHow it is computedIn the code
DataMonthly total returns (dividends reinvested). Funds younger than the period shown are extended backward with older funds, spliced on returns, not prices.
WindowThe worked example and the lab use the common window of the two series, May 1996 to Sep 2026 (365 months), ending at the data month of this page; each rolling point uses the trailing window ending that month.
Portfolio streamsA fund is its proxy-extended monthly return stream. A built-in portfolio is read as the MarketHeist MCP reads it: its composed stream with its own rebalancing rule and overlay. A custom mix is rebalanced yearly.
Rolling correlationThe Pearson correlation of the two series' monthly returns over every trailing window of the chosen length, each point dated at the last month of its window; the full-window figure uses every month of the common window.
CorrelationPearson correlation: the sum of cross-products of deviations from each series' mean, divided by the square root of the product of the two sums of squared deviations.pearsonpackages/portfolio/engine/correlation.ts
Returns and windowMonthly total returns in US dollars (dividends reinvested, currency moves included, unhedged), over the months both funds have data, ending at the latest data month. Rolling values use every trailing window of the chosen length, plotted at the window's last month.
Long historiesFunds younger than the window are extended backward with older funds or indexes that track the same market. VXUS, launched in 2011, is extended with Vanguard's older total international stock index fund, so its early years approximate the ETF rather than reproduce it.ASSET_INFOpackages/portfolio/data/proxies.ts
In the portfolio appA portfolio's correlation heatmap uses the same Pearson correlation over the holdings' full common window, and its risk-over-time view averages the pairwise correlations over trailing 36-month windows.rollingRiskDecompositionpackages/portfolio/engine/risk.ts
Engine functionsThe functions the portfolio app and the MarketHeist MCP tools run, called here on the same monthly returns.pearsonpackages/portfolio/engine/correlation.ts

International diversification and home bias compared with its neighbours

Ways international stocks can diversify a home-country portfolio
ChannelWhat it reducesMeasured byHeld up in crashes
Month-to-month co-movementShort-run volatilityCorrelation of monthly returnsWeakly: correlations tend to rise in sell-offs
Country and sector mixDependence on one economy's industriesSector and country weightsPartly
CurrencyDependence on the home currencyShare of foreign-currency assetsVaries: the dollar has tended to rise in global sell-offs
Long-run growth and valuationRisk that the home market has a lost decadeMulti-year return differencesNot a crash hedge; works over decades

Limitations

  • The window starts in the late 1990s, so it cannot show the lower correlations that studies report for the 1970s and 1980s, before markets became more integrated.
  • Correlation captures only how monthly returns line up. It says nothing about which market ends a decade ahead, which has been the larger effect of international diversification in practice.
  • Returns are in US dollars and unhedged, so the international side carries currency moves; a hedged version, or the view from another home currency, gives different correlations.
  • A trailing 36-month window is a small sample; neighbouring windows share most of their months, and short-lived moves in the line can be noise.
  • Past correlations are not a forecast. The relationship has changed with integration of markets, currency regimes and sector make-up, and can change again.

Frequently asked questions

What is home bias in investing?
The tendency to hold far more of one's own country's stocks than that country's share of the world stock market. French and Poterba (1991) documented it for the US, Japan and the UK, where investors held most of their equity at home despite the diversification available abroad.
Are US and international stocks highly correlated?
In monthly returns, yes: MarketHeist's data since the late 1990s shows a high and consistently positive correlation between the US total market and international stocks, with the exact figures in the worked example. That limits how much mixing them lowers short-run volatility, though not the long-run benefit of spreading the bet across countries.
Does international diversification still work?
It depends on the horizon. In a global sell-off most stock markets fall together, so it offers little protection in a crash. Over a decade or more, countries' results have differed widely, and holding several spreads the risk that the home market is the one that disappoints.
Why have international stocks lagged US stocks?
Since the global financial crisis, US companies, especially large technology firms, grew earnings faster, US valuations rose more and the dollar strengthened. In the decade before, from 2000 to 2009, the order was reversed. Which region leads has changed over time, and the lab here shows correlation, not that ranking.
Do correlations rise in a crisis?
Studies find that correlations between stock markets are higher in bear markets than in bull markets (Longin and Solnik, 2001). In the data here the highest trailing readings between US and international stocks came in windows covering the sell-offs of 2000 to 2002, 2008 and 2020.

References

  1. Solnik, B. H. (1974) Why Not Diversify Internationally Rather than Domestically?. Financial Analysts Journal 30(4).
  2. French, K. R. and Poterba, J. M. (1991) Investor Diversification and International Equity Markets. American Economic Review 81(2). The home-bias puzzle.
  3. Lewis, K. K. (1999) Trying to Explain Home Bias in Equities and Consumption. Journal of Economic Literature 37(2).
  4. Longin, F. and Solnik, B. (2001) Extreme Correlation of International Equity Markets. Journal of Finance 56(2). Correlation rises in bear markets.
  5. Asness, C. S., Israelov, R. and Liew, J. M. (2011) International Diversification Works (Eventually). Financial Analysts Journal 67(3).

Try it on your own holdings

Open the three-fund portfolio (US stocks, international stocks, bonds) →

Or ask an assistant connected to the MarketHeist MCP server (https://marketheist.io/mcp), which computes the same figures:

Using MarketHeist.io, how correlated have US stocks (VTI) and international stocks (VXUS) been, over the full history and over rolling 36-month windows, and how much does splitting stocks between them lower volatility?

Ask an AI assistant

The prompt opens in a new tab, already filled in. An assistant without the MarketHeist connector answers from its own knowledge, not from the figures on this page.

Historical simulation, not advice. Computed by the MarketHeist portfolio engine from monthly total returns, data through Sep 2026. Markdown version of this page: international-diversification.md.