William Sharpe, a Nobel Prize winner in Economics, wrote a recent paper about how the 4% retirement spending rule is inefficient. MarketWatch had a recent feature discussing his paper-and more than anything about the spending rule itself, the piece made me think about how large the gulf in finance is between theory and reality. As Yogi Berra is reported to have quipped, “In theory, there’s no difference between theory and practice. But in practice, there is.” (There’s a link to Mr. Sharpe’s paper in the MarketWatch article.)
The 4% retirement spending rule is clearly a rule of thumb, and I am sure that most practitioners modify it depending on the client’s circumstances. (We prefer a 3% spending rule, and I’ve seen other rules based on the yields available. For example, one paper I read advocated a spending rule of 125% of the yield on the S&P 500, arguing that you can spend more when yields are high than when they are low.) Mr. Sharpe says the 4% spending rule is too simplistic. He’s right-rules of thumb are supposed to be simplistic. But no one using it is really going to mistake it for the be-all-and-end-all.
An extraordinarily complex retirement spending rule that takes many complicated factors into account is just as likely-or maybe even more likely-to fail. The real world is a much messier place than an ivory tower. Things that seem like good ideas in theory, even to Nobel Prize winners (I’m thinking Long-Term Capital Management here), often fail miserably in practice.
The reason that complicated things never work in real life is that there are too many unknowns in the equation. In modern portfolio theory, market returns and correlations between assets are not stable, so the whole thing is essentially unworkable. A perfect retirement spending rule could be made for each client if the practitioner only knew exactly what their investments would earn each year and how long the client would live. That’s not going to happen, so we are left with rules of thumb.
The most important thing about any modeling approach is how robust it is. If you jiggle around the inputs, does it fail miserably or does it continue to work? Is it based on historical inputs which are guaranteed to change, or does it just adapt without making assumptions? We have strong feelings about this. The fewer factors a modeling approach uses, the less likely it is to be knocked down by some unanticipated factor interaction. We use a single-factor model and test rigorously for robustness (you can read our white paper on Bringing Real-World Testing to Relative Strength here). Academic finance would be much more useful to real investors if they kept in mind another saying: it is better to be approximately right than precisely wrong.
[...] “The reason that complicated things never work in real life is that there are too many unknowns in the equation.” (Relative Strength Investing) [...]
My rule of thumb is when you are pessimistic, you can spend the yield on the 10-year Treasury. In normal times, add 1%. When valuations are low, add 2% at most.
Very simple and market-based, reflecting what a balanced portfolio will yield over time.