The Actual Diversification Payoff of International Stocks
Diversification only works when the assets combined do not move in lockstep, so the real question for international stocks is not whether they outperform, but how much less than perfectly they correlate with U.S. stocks. This piece quantifies that tradeoff directly, showing how a blended portfolio can end up less volatile than either piece held alone.
The core idea: correlation, not returns, drives the benefit
Combining two assets into a portfolio reduces total risk below the simple weighted average of their individual volatilities whenever their correlation is less than a perfect 1.0, and the lower that correlation, the larger the reduction, entirely independent of whether either asset's expected return is higher or lower than the other's. This is the mathematical core of diversification, and it is why adding international stocks can improve a portfolio's risk-adjusted profile even when international expected returns are similar to, or even modestly below, domestic expected returns.
The mechanism works because a correlation below 1.0 means the two assets do not experience their good and bad periods at exactly the same time in exactly the same magnitude. When domestic stocks have a weak stretch, international stocks, imperfectly correlated, are statistically likely to have a somewhat less weak, flat, or even positive stretch over the same period, and vice versa, so the combined portfolio's swings are smoothed relative to either piece held alone, even though both pieces remain individually volatile.
The math: portfolio variance with two assets
Portfolio variance for two assets is portfolio variance = w1² × variance1 + w2² × variance2 + 2 × w1 × w2 × correlation × sd1 × sd2, where w1 and w2 are the portfolio weights, variance1 and variance2 are each asset's own variance, and sd1 and sd2 are each asset's standard deviation. Suppose U.S. stocks have an expected volatility of 16% (variance 0.16² = 0.0256) and international stocks have a higher expected volatility of 18% (variance 0.18² = 0.0324), with a correlation of 0.65 between them, broadly representative of recent multi-decade history.
A portfolio split 75% U.S. and 25% international has variance 0.75² × 0.0256 + 0.25² × 0.0324 + 2 × 0.75 × 0.25 × 0.65 × 0.16 × 0.18. Term by term: 0.5625 × 0.0256 = 0.0144; 0.0625 × 0.0324 = 0.002025; the correlation term is 2 × 0.75 × 0.25 × 0.65 × 0.16 × 0.18 = 0.375 × 0.65 × 0.0288 = 0.375 × 0.01872 = 0.00702. Summing: 0.0144 + 0.002025 + 0.00702 = 0.023445, giving a portfolio standard deviation of √0.023445 ≈ 15.31%.
A second example: finding the risk-minimizing weight
Push the same framework further to find how much international allocation minimizes total portfolio risk, holding the same 16%, 18%, and 0.65 correlation assumptions. Testing a heavier international weight, 60% U.S., 40% international: 0.60² × 0.0256 + 0.40² × 0.0324 + 2 × 0.60 × 0.40 × 0.65 × 0.16 × 0.18. Term by term: 0.36 × 0.0256 = 0.009216; 0.16 × 0.0324 = 0.005184; correlation term: 2 × 0.60 × 0.40 × 0.65 × 0.0288 = 0.48 × 0.65 × 0.0288 = 0.48 × 0.01872 = 0.008986. Sum: 0.009216 + 0.005184 + 0.008986 = 0.023386, standard deviation √0.023386 ≈ 15.29%, marginally lower than the 75/25 split's 15.31%.
Testing a still heavier tilt, 50% U.S., 50% international: 0.50² × 0.0256 + 0.50² × 0.0324 + 2 × 0.50 × 0.50 × 0.65 × 0.0288. Term by term: 0.25 × 0.0256 = 0.0064; 0.25 × 0.0324 = 0.0081; correlation term: 0.50 × 0.65 × 0.0288 = 0.00936. Sum: 0.0064 + 0.0081 + 0.00936 = 0.02386, standard deviation √0.02386 ≈ 15.45%, now higher than both the 60/40 and 75/25 results. This shows the risk-minimizing international weight sits somewhere near 40%, meaningfully above where most diversified investors typically hold international stocks, though this exercise deliberately minimizes only volatility and says nothing about which weight maximizes expected return or best matches an investor's actual goals, since a pure risk-minimization target is not automatically the right target for a growth-oriented, long-horizon portfolio.
It is worth extending the variance calculation one more step to see how sensitive the diversification benefit is to the correlation assumption itself, since correlation is the one input in the formula that has moved most over time historically. Rerun the original 75/25 U.S./international split, same 16% and 18% volatilities, but with correlation raised from 0.65 to 0.85: variance becomes 0.0144 + 0.002025 + (2 × 0.75 × 0.25 × 0.85 × 0.0288). The correlation term is now 0.375 × 0.85 × 0.0288 = 0.00918, and total variance is 0.0144 + 0.002025 + 0.00918 = 0.025605, a standard deviation of √0.025605 ≈ 16.00%, essentially identical to the all-U.S. portfolio's own 16% volatility. At a correlation of 0.85, the diversification benefit of this specific 75/25 blend has almost entirely disappeared, illustrating numerically just how much the rising correlation trend discussed below has eroded, though not eliminated, the benefit calculated at the more historically typical 0.65 figure used in the main example.
What market history shows about the benefit
Long-run historical studies of realized U.S. and international developed-market correlation show it has generally trended upward over the past several decades, from figures often below 0.5 in earlier multi-decade windows to figures more commonly in the 0.7 to 0.85 range in recent years, a pattern consistent with increasing global economic and financial integration. This rising correlation has genuinely reduced, though not eliminated, the volatility-reduction benefit calculated above, since the same variance formula produces a smaller improvement as correlation climbs toward 1.0.
At the same time, realized correlation is not a single fixed number; it varies meaningfully with the measurement window and, crucially, tends to spike higher during periods of acute market stress, when nearly all global equity markets sell off together, and to settle back toward more typical levels during calmer stretches. This means an investor evaluating the diversification benefit purely from a single crisis-period correlation figure will systematically underestimate the benefit available over a full market cycle, while an investor relying only on a calm-period figure will overestimate the cushion available during exactly the periods when it matters most.
Applying this in a real portfolio
For a practical portfolio, the takeaway is not to chase the precise risk-minimizing weight calculated above, since that number depends on volatility and correlation estimates that shift over time and carry real estimation uncertainty, but to recognize that a meaningful international allocation, commonly recommended in the range of 20% to 40% of total equity exposure, improves risk-adjusted outcomes through this correlation mechanism regardless of whether international stocks happen to outperform or underperform U.S. stocks in any given year. The benefit shows up in smoother portfolio-level returns over time, not necessarily in a higher headline return every single year.
This matters especially for investors approaching or in retirement, where portfolio-level volatility directly affects sequence-of-returns risk, the danger of large withdrawals coinciding with a market downturn. A blended U.S. and international allocation that reduces overall portfolio volatility, even modestly, reduces that sequence-of-returns risk correspondingly, an effect that compounds in importance the closer an investor gets to relying on the portfolio for regular income.
This sensitivity result carries a practical implication worth stating directly: an investor who built their international allocation decision around correlation data from an earlier, lower-correlation era should periodically redo the calculation with current figures, since the conclusion can shift from a clear risk-reduction benefit to something closer to a wash, as it did moving from 0.65 to 0.85 in the example above. This does not mean international diversification has stopped working in any absolute sense, since even at the higher correlation the portfolio's risk was not made worse by adding international stocks, only that the reduction is smaller than an outdated, lower-correlation calculation would predict, and expectations about how much smoothing international exposure will provide during a typical period should be set accordingly.
Actionable breakdown
- Add international exposure to reduce risk, not to chase higher returns.
- The benefit comes from correlation below 1.0, not relative performance.
- Even a higher-volatility asset can lower total portfolio risk when blended.
- Expect the diversification benefit to shrink somewhat during global crises.
- Correlations rise together in severe downturns across most equity markets.
- Judge the benefit over a full market cycle, not one crisis period.
- Recheck correlation and volatility assumptions periodically.
- These figures drift over decades as markets integrate further.
- A calculation done ten years ago may understate today's correlation.
- Combine developed and emerging markets for broader diversification.
- Emerging markets have historically shown somewhat lower correlation.
- A blend captures more of the available diversification benefit.
It is also worth distinguishing the volatility-reduction benefit calculated throughout this piece from a related but different concept, the reduction in maximum drawdown, the peak-to-trough decline a portfolio experiences during its worst stretch. Lower standard deviation does not automatically translate into a proportionally smaller maximum drawdown, since drawdowns depend heavily on how correlation behaves specifically during the crisis period itself, not on the average correlation used in the variance formula. A blended U.S. and international portfolio with a meaningfully lower long-run standard deviation than an all-U.S. portfolio can still experience a maximum drawdown close to, or in some historical episodes even slightly deeper than, the all-U.S. portfolio's own worst drawdown, precisely because correlations tend to rise together during the acute crisis periods that produce the deepest drawdowns, the same pattern discussed above in the context of global financial stress.
Common pitfalls
The first pitfall is abandoning international allocations after a stretch of U.S. outperformance, which defeats the purpose of a diversification strategy that is designed to pay off unevenly across time, not in every single calendar year. A second pitfall is expecting the correlation benefit to hold at its long-run average precisely during a crisis, when correlations across global equity markets have historically risen together, reducing, though not eliminating, the diversification cushion exactly when investors want it most.
A third pitfall is treating a strong recent run in either U.S. or international stocks as predictive of the next decade's leadership, when historical rotation between the two has generally been difficult to forecast in advance. A fourth pitfall is chasing the precise mathematically risk-minimizing weight as if it were a fixed, permanent target, when the underlying volatility and correlation inputs shift over time and a reasonable range, rather than one exact number, is the more realistic goal.
The bottom line
International diversification can lower total portfolio risk for a similar expected return, purely through correlation below 1.0, even when the individual foreign markets are themselves more volatile on a standalone basis.
One last practical note: none of the variance arithmetic above requires an investor to forecast which of the two assets will outperform in any given year, which is precisely why the diversification case is more robust than a return-forecasting case would be. An investor uncertain whether U.S. or international stocks will lead the next decade can still act confidently on the diversification math today, since the volatility-reduction benefit does not depend on correctly guessing that direction, only on the two assets continuing to move somewhat independently of one another, a far weaker and more durable assumption than any specific forecast about which one will win.
Related reading: international investing, asset allocation, what you are missing owning only U.S. stocks, extra risks in international funds, testing whether diversification is still worth it.