Expected versus realized returns.
All right, perhaps rebalancing wasn't quite as simple as I suggested.
Michael Edesess reminds me that Tuesday's column rests on an unreliable pillar. In writing that rebalancing is profitable for assets that have similar expected rates of return, I implicitly assumed that over a sufficiently long time period, actual performance would approximately match forecast performance. Whoa now. Responds Edesess, "Expected returns being the same in no way ensures that realized returns will be the same."
Well, I certainly can't argue with that. The counterargument to rebalancing when expected rates of return are similar is:
1) If I find two assets that have identical (or very similar) past returns, and I run the numbers, then I will find a rebalancing bonus.
2) If I can correctly predict that two assets will have identical (or very similar) future returns, then I will get a rebalancing bonus.
3) But since I can neither invest in the past nor predict the future, I may rebalance away from my best-performing asset, the one that does not mean-revert, and I will therefore end up with a lower-performing portfolio. (During the period, I also will be putting a lot of money into my worst-performing asset.)
Fair enough. However, the issue is more of an additional wrinkle than a refutation.
It is true that with a single-direction, ongoing trend, rebalancing will lower a portfolio's return in the manner described. The investor will continually shift money from the better-performing asset into the laggard. But with a more complex performance pattern, rebalancing can profit even if long-term returns diverge.