One of the abiding asset allocation challenges in recent years has been how much to invest in non-U.S. vs. U.S. equities. In theory, globally-weighted market capitalization is the best choice for a general asset allocation. It is not unreasonable to think that global capital should be smart enough to reflect the relative value propositions across countries and currencies. For investors who want to bet on overall market efficiency, the Vanguard Total World Stock ETF (VT) is a good choice. The problem, however, is that international allocations have massively underperformed for many years. The Vanguard U.S. Total Stock Market Index (VTI) and the SPDR S&P ETF (SPY), which broadly capture U.S. equity exposure, have substantially out-performed VT over the 1-, 3-, 5- and 10-year periods. Given that VT has a 56% allocation to U.S. equities, this highlights the dramatic under-performance of non-U.S. equities.
Annualized Percentage Returns for VT, VTI, and SPY (Source: Morningstar)
Granted, combining domestic and international equities in VT does provide some risk reduction relative to a U.S. only portfolio (VTI). The trailing five year annualized volatility for VTI is 16.02% vs. 15.36% for VT. Even so, the Sharpe ratio is dramatically higher for U.S. equities than for the global portfolio, so international diversification has not been a great bet over the past decade. The same can be said for the past twenty years.
The enormous question going forward is whether we will experience a reversion to the mean, with international out-performing U.S. stocks in the coming years. How much allocation does it make sense to carry in U.S. vs. non-U.S. equity? On the basis of price-to-earnings and price-to-book, international equities are cheaper than U.S. stocks, but how much does this matter? When you include the buyback yield in addition to dividend yield, the total yield for U.S. equities is actually higher than for the MSCI World Index (see slide 50). I am also hesitant to directly compare U.S. yield to international index yield as a robust measure because consistency of dividend payments is not the same across countries.There is also the demographic headwind (see slide 12) for earnings growth in the Eurozone and Japan due to their ageing populations. The U.S. faces this challenge as well, but we are in better shape than the U.K, Japan, or Europe.
Projections of earnings growth and exchange rates present thorny challenges, too. Research Affiliates maintains a useful online tool that estimates future returns using yield and forecasted earnings growth. Their projections suggest that international equities will out-perform U.S. equities, but they have maintained this projection for years and the markets have stubbornly contradicted the projections.
I have, in the past, been a committed believer in mean reversion across asset classes but I am not so sure anymore. In the long-term, as they say, we are all dead and I don’t yet see evidence of factors that could drive a turnaround in the relative performance of U.S. vs. non-U.S. stocks. It is also hard to discount the argument put forth by Jack Bogle that U.S.-based investors can achieve sufficient international exposure because so much of earnings of U.S. companies are from foreign markets. Considering all of the data, my outlook is that U.S. investors, in particular, will be well-served by being underweight non-U.S. equities.