Diversification is not a free lunch

Harry Markowitz famously said that diversification is the only free lunch in investing. While the logic of diversification is indeed undeniable, the choices it requires can be quite painful.

13 JUNE 2023

Harry Markowitz famously said that diversification is the only free lunch in investing. While the logic of diversification is indeed undeniable, the choices it requires can be quite painful. Diversification does not fully escape the practitioners’ observation that “no pain, no premium”. Today, both the pain and the premium may be as high as they have ever been.

Consider an investor attempting to build a diversified portfolio in 1979. Businessweek had just published its “death of equity” cover following a long period of poor equity returns in the rising inflation environment of the 60s and 70s. Around about the same time, and for similar reasons, U.S. government bonds were widely regarded as “certificates of confiscation”. US stocks had underperformed emerging markets for a decade. On the winning side, value stocks had vastly outrun growth stocks, commodity trading advisors had performed brilliantly, six of the ten largest US companies by revenues where in energy, gold had crushed the equity markets, with the Dow-to-gold ratio standing at around 2. (It is now 20.) Tasked with building a diversified portfolio, our fictitious 1979 investor would likely have included large shares of emerging market equity, value, commodities, and gold. Clients and bosses would have gladly accepted this allocation. 

Fast-forward to 2023, when a generation of investors has built careers on portfolios consisting largely of US stocks and government bonds (with some deviations to account for home bias), heavy on tech and light on value and emerging markets. For example, popular strategies to track Bridgewater’s well-known All-Weather portfolio consist for 85% of US stocks and US government bonds, an allocation that our fictitious 1979 investor would probably have considered more like “sunshine or light rain” than all-weather. 

Stock Market

How did a 60-40 portfolio (with modest variations) of US assets (or highly correlated markets) become so dominant in the West despite its seemingly limited diversification? An obvious reason is the exceptionally high Sharpe ratios of US stocks and bonds since 1979. A second reason is that these two assets did in fact offer good diversification because of their negative correlation during a period in which central banks aggressively lowered interest rates in response to any downturn in economic or lending activity. The mathematics of diversification greatly rewards negative correlations. For example, the reduction in volatility in an equal-weight portfolio of 500 assets, each having the same variance and cross-correlation 0.5 with all the others, is roughly matched by holding a 50-50 portfolio of just two assets with correlation equal to -0.5. As a case in point, in 2008 the US stock market fell 27%, but 10-year Treasuries were up 20%. Like being paid to own a put option on the stock market. 

The trouble, however, is that correlations can be unstable; some of the most spectacular blow-ups in financial history involved large moves in cross-correlations. The poor return of the 60-40 portfolio in 2022 was due to the SP500 and 10-year Treasuries both losing 18%. Now imagine that our 1979 investor is transported to today, and upon observing this recent data, does not lose a moment to trade his diversified portfolio. His career risk would be very high. In spite of higher inflation, large and established US funds are having a very difficult time defending even modest allocation to international equities,[1] not to mention emerging markets, commodities, value, gold, energy, and mining (including mining required for renewable energy). 

The broader point is that a decade of underperformance by any asset or factor implies a dose of psychological pain and very real career risk for investors diversifying into it, despite the increased expected risk-adjusted return. While momentum is well-documented at horizons up to approximately 12 months, winners and losers in any given decade do not, on average, repeat their performance (if anything the evidence points to modest mean reversion), whether looking at industries, countries, or asset classes. There is therefore no statistical base on which to build a portfolio on the winners of the last decade or two, and yet this is clearly the most comfortable portfolio to hold. In defending international diversification, Asness et al. (2023) feel that “rarely has doing the right thing been so hard”.

Is doing the right thing (from a diversification perspective) finally about to be rewarded? Valuation multiples (such as price-to-sale) in the US (which today accounts for 54% of the MSCI world index) were at near-record lows in 1979-1982, and are now at near-record highs, making a repeat of real returns since 1979 almost impossible. Inflation has been mostly above the ten-year yield since 2009 and, as Carmen Reinhart (2011 and 2022) points out, historical parallels suggest negative real yield may stay with us for quite some time. It is also unlikely, given current inflation rates, that central banks will be able to lower interest rates and expand balance sheets as aggressively as they have done in the past. Building diversified portfolio in this context may require hard thinking and some uncomfortable choices, but rarely has doing the hard thing been so right.   



Cliff Asness, Antii Ilmanen, and Dan Villalon, (2023), “International diversification: Still not crazy after all these years”, Journal of Portfolio Management, 49, 6.

Carmen Reinhart, and Belen Sbrancia, (2011), “The liquidation of government debt”, NBER Working Paper 16893.

Carmen Reinhart, (2022), “Will central banks do what it takes?”, Project Syndicate, December 2022.