Long-run databases covering more than twenty markets across more than a century show that most delivered positive real returns, with substantial variation in both level and volatility. The United States sits at the upper end of that distribution, which is not disputed. What is disputed is how far that position could have been anticipated, and what it implies about the future. The first problem is survivorship. Markets for which long series exist today are largely those never closed, nationalised or interrupted by war. Several markets of considerable size in the early twentieth century, including Russia, China and parts of eastern Europe, went to zero for foreign holders at some point. Counting only surviving markets systematically overstates the average return available in advance.
The second problem is repricing. Part of the United States return across recent decades came from rising valuation multiples rather than from earnings growth alone. Multiples can rise for a long time and cannot rise indefinitely. Any extrapolation of past returns implicitly assumes multiples continue expanding at a similar rate, and that assumption is rarely stated explicitly. The third problem is selection after the fact. In nineteen hundred there was no objective basis for identifying the United States as a future leader. The largest market was the United Kingdom, and many observers were optimistic about Argentina. Choosing the winner with today's knowledge and then citing its record as evidence about the future is a circularity that is common and hard to notice.
The same data shows another stable pattern worth recording. Across sufficiently long periods, portfolios holding many markets exhibited lower volatility than most individual markets while delivering returns near the average. This result requires no forecasting ability whatever. It is the mechanical consequence of correlations being less than complete. Correlations among markets rose over recent decades, which weakened the diversification benefit without eliminating it. The main causes include the globalisation of multinational revenue, liberalised capital flows, and shared macroeconomic factors. Even so, differences in sector composition, currency and institutional arrangements continue to provide substantial diversification. The ordering of returns varies enormously across periods, which is another easily overlooked fact.
Measured decade by decade, leadership changes frequently, and a market lagging for ten years leading over the following ten is not unusual. The pattern is not stable enough to support any rotation strategy. Its value lies in demonstrating that any single period's ordering does not persist. For investors, the defensible use of these records is setting expectations rather than selecting markets. A reasonable range for long-run real returns, the possibility of a single market going to zero, and the magnitude of the diversification benefit all rest on reasonably firm historical foundations. Which market leads over the next decade does not. One methodological caution applies to all of it.
These databases required decades of work to assemble and involve judgement about which series to link, how to handle interruptions, and how to treat markets that changed definition. Different research groups make different choices, and the resulting figures differ by enough to matter for anybody using them as precise inputs rather than as broad reference. A further point concerns what the record cannot address. These series measure listed equity, and the share of economic activity that is listed varies enormously across countries and across time. A market where most large enterprises remain private or state-held tells you far less about that economy than one where listing is the norm. The index return reflects only what happened to be quoted.