The following is a response to the article, Return Distributions and the Shiller P/E Ratio, by Keith Goddard, which originally appeared on February 2, 2010 and was contained in your section The Ten Best Articles You Probably Missed on December 28, and a Letter to the Editor by George Vrba, which appeared last week.
Dear Editor,
You seem to be confusing Shiller’s pricing model with a market timing model. Shiller’s model was inspired by Graham’s business value concepts which assiduously reject timing – the endeavor to predict the direction of the market in advance. Shiller’s model provides great perspective:
- If you take it as a market timing model, based on the fact that it has “predicted” the last two bear markets, you should be cautious about its ability to do so in the future, now that the model has wide popularity.
- It forces you to resolve the discrepancy between normalized and ttm P/E. We know how that discrepancy (Shiller at 27x vs. ttm at 16x) was resolved after 2007. But we are now looking at a ten-year period containing two major downturns in earnings in the context of a recovery with sharply rising earnings. You have to normalize the normalized PE!! Get 2008-09 out of the data.
- Based on rolling ten-year return data relative to starting P/E, the market appears overvalued on a nominal basis, but only slightly so on a real-return basis, and undervalued vs. bonds.
Models are great tools, but wisdom is required!
Best Regards,
Bruce Thompson
Thompson Wealth Management
Concord, MA
