5 Year Retrospective on a Model: Low Bond Yields and Safe Portfolio Withdrawal Rates
A year ago I took a look at a 2013 paper from Blanchett, Finke, and Pfau (BFP).
I realised that January is the 5 year anniversary of it being published. Which provides a nice point in time to check the model against the actual performance of the market over those 5 years.
The cumulative returns from 2013 until the (almost) end of 2017 are 103%.
That’s an 82nd percentile sequence of returns, according to the model. That is, 82% of other “possible worlds” had worse stock returns than we actually had. Of course, the model might be wrong. Or the model might be right and we just got lucky. After all, being in the 82nd percentile isn’t impossible.
If we received 103% cumulative returns over the past 5 years, what would “average” returns have looked like according to BFP’s model?
At the 50th percentile cumulative returns were 42.8%. We have received nearly double what “average” should have been.
One other possible way of looking at it is…how many of BFP’s failures occurred after experiencing a “first five years” like we’ve actually experienced? Without even checking, we’d expect this to be quite low, since having a good first five years requires some extremely dire “next 25 years” in order to destroy a portfolio.
And sure enough, 98.8% of portfolios that started out with 103% returns in the first five years went on last the full 30 years using 4% withdrawals.
My own belief is that five years is a long time for “high valuations and low returns” people to cry wolf. We’ve seen returns that, according to their own models, are over 80th percentile. Each year that goes by calls the model further into question. Of course, it is always tough to reconcile the dual (and conflicting) mandates of:
- The market can stay irrational longer that you can stay solvent.
- Change your mind when new evidence presents itself.
