What is the best mix of stocks and bonds to hold during retirement worldwide? In his January 2015 paper entitled “The Retirement Glidepath: An International Perspective”, Javier Estrada compares outcomes for different stocks-bonds allocation strategies during retirement from a global perspective. He considers declining equity, rising equity and static glidepaths with an annual withdrawal rate of 4% (of the portfolio value at retirement) and annual rebalancing during a 30-year retirement period. He tests the following glidepaths:
- Four declining equity strategies that begin with 100%-0%, 90%‐10%, 80%‐20% and 70%‐30% stocks-bonds allocations and shift toward bonds linearly via annual rebalancing.
- Four mirror-image rising equity strategies that begin with 0%-100%, 10%-90%, 20%-80% and 30%-70% stocks-bonds allocations and shift toward stocks linearly via annual rebalancing.
- Eleven static allocations ranging from 100%-0% to 0%-100% stocks-bonds allocations maintained via annual rebalancing, with focus on conventional or near-conventional 60%-40%, 50%-50% and 40%-60% allocations.
He focuses on the failure rate of these strategies during 81 overlapping 30-year retirement periods during 1900-2009. He also considers average and median terminal wealth/bequest, tail risk, annual volatility (standard deviation of annual returns) and upside potential. He defines tail risk (downside risk) as average terminal wealth for the 1%, 5% or 10% lowest values from the 81 periods. Using annual total real returns for stocks and government bonds for 19 countries (in local currency adjusted by local inflation) and for the world market (in dollars adjusted by U.S. inflation) during 1900 through 2009 (110 years), he finds that:
- For the dynamic strategies:
- For the U.S., compared to rising equity strategies, declining equity strategies generally:
- Have lower failure rates.
- Generate higher terminal wealth and exhibit more upside potential.
- While more volatile (based on standard deviation of annual returns), exhibit the same or lower tail risk (same or better downside protection).
- Results for other countries and the world market are similar.
- For the U.S., compared to rising equity strategies, declining equity strategies generally:
- Among the conventional 60%-40%, 50%-50% and 40%-60% stocks-bonds static strategies:
- The 60%-40% stocks-bonds strategy is best of the three.
- For the U.S., compared to declining equity and rising equity strategies, the 60%-40% static strategy:
- Ties for the lowest failure rate.
- Generates the highest terminal wealth.
- Provides the same or better downside protection.
- Results for other countries and the world market are similar, but attractiveness of the 60%-40% static strategy compared to dynamic strategies is less consistent.
- For the 100% stocks static strategy:
- For the U.S., compared to other static and dynamic strategies, a 100% stocks strategy:
- Ties for the lowest failure rate.
- Has much higher upside potential, indicating that its higher annual volatility measures uncertainty about how high and not how low.
- Provides the same or better downside protection.
- Results for the world market and the average country in the sample are similar.
- For the U.S., compared to other static and dynamic strategies, a 100% stocks strategy:
- In general:
- The above findings hold for a withdrawal rate of 3% of the balance at retirement.
- The “4% withdrawal rule” is not nearly as safe globally as it is for the U.S.
In summary, evidence indicates that 100% stocks and 60%‐40% stocks‐bonds allocations are simple and very effective strategies for retirees worldwide.
Cautions regarding findings include:
- The sample period is not long in terms of number of independent 30-year retirement intervals. In other words, 30-year performance statistics for the next 110 years could plausibly be materially different from those for the last 110 years.
- Results apparently include no retirement fund fees or rebalancing costs, which would erode reported performance.
See “Lifecycle Funds Guard Against Upside Volatility?”, summarizing findings from the same author for the pre-retirement accumulation period rather than the retirement distribution period.
