At the Bogleheads Forum, users VictoriaF and sashwin both expressed a desire to know about safe withdrawal rates for US-based retirees who invest in a world portfolio, rather than just in the domestic US markets. It is a good question, and I thank them for asking it.
I am writing a brief blog entry about it now, with the intentions of later incorporating this into a more complete paper about the role of international diversification in the retirement portfolios for 17 developed market countries.
The methodology for this exercise is the same as that used in my paper, “An International Perspective on Safe Withdrawal Rates from Retirement Savings: The Demise of the 4 Percent Rule?" from the December 2010 Journal of Financial Planning.
The investor has three asset choices: world stocks, world bonds, and US Treasury bills. This is how world assets are defined in Dimson, Marsh, and Staunton's Triumph of the Optimists. The world stock index is a weighted average of the returns provided in 17 developed market countries. The weights are GDP for the years 1900-1967, and market capitalization for 1968-2008. For the world bond market index, GDP weights are used for the entire historical period. For each of the 17 countries, the local-currency returns were converted into US dollars without hedging for currency risk, and then combined into the world portfolio according to their weights. I assume the investor is in the US, so withdrawals are adjusted for US inflation, and the world returns are measured in US dollars.
First, as a reminder, here is the path of maximum sustainable withdrawal rates for the US retiree using only US assets:
The worst-case withdrawal rate (SAFEMAX) was 4.02% in 1969 with the Dimson, Marsh, and Staunton dataset.
Next, here is the path of maximum sustainable withdrawal rates for the US retiree invested in the world portfolio:
The worst-case withdrawal rate (SAFEMAX) was 4.08%, and that happened for the 1911 retiree.
Keep in mind that the US markets make up around half of the world markets, so the correlation between the US and world should be pretty close. Here you can compare the time paths of the US and World-based MWRs:
Finally, here is another figure which shows how the SAFEMAX varies by asset allocation for the US and world portfolios:
Though I tend to support the idea that retirees should not feel obliged to use high stock allocations during retirement, the findings based in this historical data do make a case for high stock allocations with the world portfolio. This issue should be studied some more before any final conclusions can be made.