Long-term investors and valuation-based asset allocation

Abstract

Valuation-based market timing demonstrates strong potential to improve risk-adjusted returns for conservative long-term investors. Such timing strategies based on the cyclically-adjusted price-earnings ratio provide comparable returns as a 100% stocks buy-and-hold strategy but with substantially less risk. Meanwhile, market timing provides comparable risks and the same average asset allocation as a 50/50 fixed allocation strategy, but with much higher returns. Also, it is important to consider less extreme timing strategies as well, as defining market timing as either all stocks or all cash does not provide a hedge against the possibility that valuations may depart from their historical averages for extended periods. Finally, comparing the strategies over shorter rolling sub-periods reveals that a valuation-based market timing approach fairly consistently provides risk-adjusted returns superior to a fixed asset allocation strategy.

Keywords:

• market valuations
• cyclically-adjusted price-earnings ratio
• PE10
• stock returns
• market timing
• long term
• tactical asset allocation
• buy and hold

JEL Classification::

• C15
• D14
• G11
• G14
• G17
• N21
• N22

Acknowledgements

Though market-timing strategies are specifically contrary to John Bogle's investment philosophy, the author is extremely indebted to suggestions and reading recommendations provided by countless users in the threads ‘Any Studies on Long-term Market Timing?’ and ‘Valuation-based market timing with PE10 can improve returns?’ at the Bogleheads Forum. I am also grateful to Rob Bennett for motivating this investigation, and to financial support from the Japan Society for the Promotion of Science Grants-in-Aid for Young Scientists (B) #23730272.

Notes

¹ I also replicated the analysis using the returns on 10-year Treasury Bonds for the fixed income category, but the differences were not material.

² Fisher and Statman's data for stock returns and valuation measures for the years 1871–1999 are from the appendix of Wilson and Jones (2002). Fisher and Statman received subsequent data for 2000–2002 directly from Jack Wilson, who has passed away. Regarding the 2000–2002 data, and any more recent data as well, his co-author Charles Jones did not keep a copy of their spreadsheet. Meir Statman was very helpful, but unfortunately he also no longer has the later values either, making it impossible to precisely replicate the original results in Fisher and Statman (2006). The Wilson/Jones dataset is no longer updated, justifying the switch in data sources.