Stocks Are Always Risky: The Equity Premium Exists for a Reason

Over 80% of my net worth is in stocks. Equities are risky assets, and the equity risk premium exists because there are no guarantees. Both of those are true at once, which is more than the “stocks are safe if you just hold long enough” slogan that dominates most personal-finance writing admits.

The case for heavy equity allocation and the case for clear-eyed humility about risk are the same case. If stocks were risk-free, there would be no premium to earn. The evidence below covers William Bernstein’s four deep risks, 120+ years of global data from Dimson, Marsh, and Staunton (DMS), the total-loss cases of Russia and China, the cautionary case of Germany, and Pástor–Stambaugh’s result that long-horizon equity variance is higher, not lower, once you account for predictive uncertainty. The conclusion is to own equities globally and without the illusion of safety.

Bernstein’s Four Deep Risks

William Bernstein’s short book Deep Risk: How History Informs Portfolio Design argues that short-term volatility is a poor proxy for the real dangers a long-term investor faces. Bernstein calls volatility “shallow risk” and reserves deep risk for permanent loss of purchasing power over decades. He identifies four sources:

• Inflation. Sustained erosion of real purchasing power. The most common historical deep risk. Stocks hold up reasonably well against inflation over long periods because firms can reprice their output, but there are extended windows where real returns are bad or negative.
• Deflation. A protracted decline in the price level that crushes debt-financed businesses and pushes asset prices lower. Rare in modern economies but devastating when it strikes (Japan 1990s, U.S. 1930s).
• Confiscation. The state takes your assets, or taxes and regulates them to the point of loss. Russia 1917 and China 1949 are the canonical cases of outright confiscation of equity claims. Less extreme versions include wealth taxes, capital controls, and forced portfolio conversions.
• Devastation. War or catastrophe that physically destroys the productive capital stock. Germany and Japan in the 1940s are the clearest examples in developed markets.

The framing separates the losses you recover from the losses you do not. A 30% drawdown that recovers in three years is painful but not deep. A decade like the 1970s, a confiscation, or a devastation can end your financial life even if you held stocks patiently. See Bernstein’s Efficient Frontier essays and the Advisor Perspectives review of his work for a broader treatment. Summitward’s dedicated Four Deep Risks guide maps each deep risk to the asset classes that historically hedged it.

Russia, China, and the Germany Cautionary Tale

“Some stock markets have gone to zero” is repeated often enough that it is worth separating the cases where it is literally true from the cases where it is not.

Russia 1917 and China 1949 are the canonical total-loss cases in the modern equity-history literature. The Bolshevik Revolution extinguished Russian equity claims through outright expropriation. The Chinese Communist Party did the same in 1949. DMS state the assumption plainly: when they build Russia and China into their composite indexes they “assume that shareholders and bondholders in Russia and China suffered total losses in 1917 and 1949,” then re-include both countries once their markets reopened in the early 1990s (Yearbook 2022 summary, pp. 8-9). The wipeout is a gap in a discontinuous series rather than its endpoint, and DMS qualify the assumption themselves. It still describes the experience of anyone holding a concentrated Russian or Chinese position in those years, who effectively lost everything. Russia mattered more than its later obscurity suggests: it was roughly 6% of world equity capitalization in 1900.

Germany is a different kind of warning case. Between 1914 and 1948, German equity investors lived through World War I, the 1923 hyperinflation, global depression, World War II, and the 1948 currency reform. The equity experience was not a clean zero. DMS put the WWI loss at roughly two thirds of value and the World War II and immediate postwar loss at 88% in real terms, with bonds down 91% and fixed-income holders wiped out outright by the 1922–23 hyperinflation. German equities then rose4,094% in real terms between 1949 and 1959, and compounded at 3.2% real per year over 1900–2013 (Yearbook 2014, Germany profile). That is catastrophic real wealth destruction rather than a total loss.

The interruption was long. German exchanges closed at the end of July 1914 and official trading resumed only in January 1918. A price-freeze decree in January 1943 fixed share prices in place, the Reichsamt index was discontinued that June, and free price formation returned only after the 1948 currency reform, when trading reopened in Frankfurt on 14 July 1948. Germany is the case against the idea that deep risk only happens in emerging or authoritarian markets. A single country’s stock market, even a major one, can be broken by events outside any investor’s control.

Jorion and Goetzmann’s 1999 Global Stock Markets in the Twentieth Century (Journal of Finance 54(3)) assembled 39 markets going back to the 1920s. Of the 24 markets trading in 1931, only seven came through the following decades uninterrupted, seven closed temporarily, and ten suffered long-term closure. Eleven of the 39 markets suffered a permanent break. Over 1921–1996 U.S. equities returned 4.3% real per year against a median of 0.8% across the other markets, which is why they conclude the U.S. equity premium “appears to be the exception rather than the rule.” Note that their database contains neither Russia nor China; those come from DMS.

Why U.S.-Only Evidence Is Misleading

The most-cited statistic in long-term equity writing is Jeremy Siegel’s claim that U.S. stocks have averaged ~6.5-7% real returns since the early 19th century, and that rolling long-horizon windows rarely delivered negative real returns. Those numbers are accurate for the United States. They are also drawn from the most successful equity market in the sample.

The DMS dataset, updated annually in the Global Investment Returns Yearbook, covers 35 countries and 120+ years. The picture it paints is more sobering:

The U.S. real return sits 2.2 percentage points above the developed-markets index, and DMS explicitly document the U.S. as an unusually successful market over the sample. The equity premium itself is substantial: across 19 countries over 1900–2010, DMS put the premium relative to bills at 4.6% per year, and at 4.5% per year on the world equity index (Equity Premia Around the World, 2011). That is real compensation for risk. It is also lower than U.S.-only figures suggest, and it came with wide national variation.

See the public UBS summary of the 2026 Yearbook for the current numbers: UBS Global Investment Returns Yearbook 2026 summary (public).

Long Horizons Do Not Repeal Risk

The popular version of “stocks for the long run” is: hold for 20 years, and you effectively cannot lose money in real terms. The global record refutes this as a universal claim. Three facts from it:

• Of the 39 markets in the Jorion–Goetzmann database, eleven suffered a permanent break. Of the 24 trading in 1931, only seven came through the next two decades without interruption.
• German equities lost 88% in real terms across World War II and its aftermath, on top of losing roughly two thirds of their value in World War I. A German investor who started in 1914 waited until the 1950s to see a recovery.
• An investor who entered 2000 after seeing strong prior 20-year equity results still went on to earn a negative real return on world stocks over the next decade. The “time cures all” intuition is not even true of the world portfolio over a 10-year window that many investors lived through.

Long horizons improve the odds. They do not eliminate the risk. Anyone planning on 20-year rolling windows being reliably above zero is using a U.S.-specific empirical pattern as if it were a law of finance.

Pástor–Stambaugh: Long Horizons Can Be More Volatile

L’uboš Pástor and Robert Stambaugh asked what happens to equity variance once an investor admits they cannot observe the true return distribution, only estimate it from noisy history.

Their answer, from the investor’s perspective, is no. Once you incorporate parameter uncertainty and the fact that predictors of return are imperfect, 30-year variance per year comes out 21% to 53% higher than 1-year variance per year across their range of priors, and about 45% higher under their benchmark priors. (Their first 2008 draft put the range at 21% to 75%; they revised it down in 2009, and the published version drops the range from the abstract entirely.) Long horizons layer in moreuncertainty about the true expected return, not less, and that extra uncertainty more than offsets the diversification of independent annual shocks.

This does not settle every allocation debate, and reasonable economists disagree on how much weight to put on it. It is enough to show that “time makes stocks safe” is an empirical regularity that depends on priors, predictors, and which country you happen to be born into. The paper was published as “Are Stocks Really Less Volatile in the Long Run?” in the Journal of Finance (2012); the 2009 revision is the last draft to state the range.

Why I Still Hold 80%+ in Stocks

All of that is an argument for owning equities with eyes open. I keep over 80% of my net worth in stocks because:

• The equity risk premium is real and historically large.DMS measure it at 4.6% per year across 19 countries over 1900–2010, and at 4.5% per year on the world equity index, which is meaningful compensation for bearing real risk.
• My investing horizon is long. Long horizons do not eliminate risk, but they do tilt the distribution of outcomes in favor of equities. The odds are better if I give them time to work.
• The alternatives have their own deep risks. Cash loses to inflation, long bonds lose to inflation and rate shocks, real estate concentrates in one property and one market. No asset class is deep-risk-free.
• I accept compensated risk. The premium is payment for bearing a risk that does not go away. The math favors me on average, and never with certainty in any given year, decade, or lifetime.

Stocks are not a savings account with a better yield. The next Russia, the next Germany, the next Japan are low-probability, high-impact possibilities, and my allocation is a considered bet.

Global Diversification Lowers Regret

If deep risk is real and country-specific outcomes vary massively, the obvious response is to stop making a single-country bet. The United States was about 62% of world equity market value at the start of 2026. A U.S.-only investor holds one side of that 62-38 split rather than owning the world.

The UBS 2026 Yearbook summary concludes that investors in the vast majority of markets were better off investing globally rather than domestically, particularly when currency risk was hedged. That argues for owning the U.S. market alongside the rest of the world, at its global weight.

A common objection is that large U.S. multinationals give you international exposure through their foreign revenue. The evidence says it helps less than you would think. Rowland and Tesar’s spanning tests (NBER 6733) could not reject the hypothesis that adding multinationals to a domestic portfolio leaves the efficient frontier where it was, while adding an international index shifted it. Vanguard’s home-bias work makes the same point qualitatively: a company’s performance has historically been more highly correlated with its domestic market regardless of where its business operations are. (Philips, Kinniry and Donaldson, 2012). This is contested. Rowland and Tesar themselves find weak evidence that U.S. multinationals specifically did offer some benefit, and Cai and Warnock (NBER 12220) find U.S. investors’ indirect foreign exposure through multinationals is comparable in dollar value to their direct foreign holdings. Foreign revenue is a partial substitute at best. Summitward has a dedicated guide on this point: Global Revenue Is Not Diversification.

Global diversification does not protect you from global catastrophe, worldwide inflation, or a world war; those are correlated shocks. It protects you from the risk that one country, one regime, or one domestic narrative ruins your personal outcome. It lowers the probability of regret, not the probability of loss.

Frequently Asked Questions

Are stocks safe if I hold them long enough?

No. Long horizons improve the odds of positive real returns but do not guarantee them. Eleven of the 39 markets in the Jorion–Goetzmann database suffered a permanent break, and German equities lost 88% in real terms across World War II and its aftermath. World stocks earned a negative real return over the 2000s decade.

What are Bernstein’s four deep risks?

Inflation, deflation, confiscation, and devastation. They are all forms of permanent, long-horizon loss of purchasing power, as distinct from short-term volatility (“shallow risk”).

Did the Russian, Chinese, and German stock markets really go to zero?

Russia 1917 and China 1949 are the canonical total-loss cases in the major world equity histories, though DMS impose that as an assumption and re-include both markets from the early 1990s. Germany is a different kind of warning: hyperinflation, war, a five-year price freeze, and the 1948 currency reform produced catastrophic real wealth destruction, but not a clean zero. German equities fell 88% in real terms across World War II and its aftermath, then rose 4,094% in real terms over the following decade. The correct lesson is that even major developed markets can suffer outcomes that make U.S.-only historical returns an unreliable guide to universal investing behavior.

How much of my portfolio should be international?

A defensible default is to own equities in their market-capitalization weights, which on the start-2026 figure above implies roughly 38% outside the United States. Some investors tilt higher for deeper diversification; some tilt lower for tax or behavioral reasons. The key is that you hold a meaningful non-U.S. allocation, not the precise split.

If stocks are risky, why own so many of them?

Because the equity risk premium is real, the alternatives have their own deep risks, and long horizons tilt the distribution in favor of equities even if they do not guarantee the outcome. Accepting compensated risk is not the same as believing in certainty.

What does the equity risk premium actually mean?

It is the historical excess return of stocks over a risk-free asset (typically short-term bills). Across 19 countries over 1900–2010, DMS measure it at 4.6% per year relative to bills, and at 4.5% per year on the world equity index. The premium exists because equities are risky; it is the payment you have historically received for bearing that risk.

Related Guides

• Did Stocks Double During the Civil War? on exchange closures as a failure mode distinct from a falling market, and why war-return tables only contain the markets that survived.
• But What About Japan? works the single-country case in depth: market beta vs. value vs. small value, vintage risk, and a resilience stress test.
• Stocks Usually Win. “Usually” Is Not a Financial Plan. The bond-specific complement: stocks have lost to long Treasuries for 20-, 41-, and 68-year windows even outside total-loss scenarios. Includes a stress-test calculator for your plan.
• The Case for Global Equity Diversification walks the positive case for owning the world, including valuation, correlation, and the lost-decade evidence.
• Global Revenue Is Not Diversification shows why U.S. multinationals with foreign revenue do not substitute for actual foreign equity ownership.
• VTSAX Is Not All You Need applies the global argument to the popular simple-portfolio debate.
• Sequence of Returns Risk covers another form of long-horizon risk that averages hide: the order of returns matters enormously in retirement.
• The Tech Bro Portfolio is a concrete case of what single-country, single-sector concentration looks like in modern indexing.
• The Four Deep Risks of Investing expands Bernstein’s framework with concrete asset-class hedges for each risk.
• The Best Inflation Hedges Are Boring covers the direct CPI-linked instruments (I Bonds, TIPS) that address inflation risk specifically, where stocks historically fail.
• Do Stock Valuations Still Matter? covers the CAPE-based long-horizon expected-return evidence and what to do when valuations are stretched.
• Do 200 Years of Stock Returns Still Matter? covers McQuarrie’s data-archaeology correction to Siegel and a calculator that compares terminal wealth across five historical regimes.

Key Takeaways

• Stocks are risky even over long horizons. Eleven of 39 markets in the Jorion–Goetzmann database suffered a permanent break, German equities lost 88% in real terms across World War II, and world stocks were negative in real terms over the 2000s.
• Bernstein’s four deep risks are the right frame. Inflation, deflation, confiscation, devastation. Volatility is not the worst thing that can happen to a long-term portfolio.
• The U.S. is an unusually successful market. DMS explicitly document this. U.S.-only back-tests describe one fortunate country.
• Global diversification lowers regret. It cannot eliminate global catastrophe or worldwide inflation, but it hedges the risk that any single country, regime, or domestic narrative ruins your personal outcome.
• High equity allocation and clear-eyed risk accounting are compatible. I own 80%+ stocks on the strength of the odds, with the failure cases in view.