Bonds for the Long Run

alternative title: Why the Siegel Constant Never Was

We have, throughout this week, alluded to the recent Bill Gross criticism on the Death of the Cult of Equities (See this and this). Elsewhere, several references to Stocks for the Long Run have been made, including the “Siegel constant” – the concept pushed by Wharton professor Jeremy Siegel that equities have consistently produced returns, including reinvested dividends, of 6.6% after inflation.

I want to suggest that these critics have it all wrong. Its not that “The Siegel Constant” is history — as Gross and Eddy Elfenbein have suggested; rather, it is because Stocks for the Long Run (aka SFTLR) was a deeply flawed and erroneous book right from the start.

There are several reasons for this outlying conclusion. The most obvious (but not IMO the most persuasive) has been equity versus fixed income performance. Bonds have — impossibly according to Siegel — outperformed equities. This isn’t merely a short term phenomena: Bonds have outperformed equities over the past 1, 2, 5, 10, 30 and 40 years.

That raises serious questions about the Siegel Constant. My suspicion is it reflects an invalid belief system; it is fair to say that, at the very least, it has not been thoroughly proven statistically.

What’s that again? SFTLR is one of the most widely read books in economic classes and business schools.

Exactly. Like the Efficient Market Hypothesis, or Homo economics, or the rational human, it is one of those squishy, poorly conceived, not well proven concepts that seems to be the underlying basis of some spectacular disasters. And, these issues have been thoroughly disproven by events of the past half century.

But the most damning criticism leveled at SFTLR is that it uses bad data to prove its point. Not only is the fundamental premise wrong, but its flawed in a way that dramatically overstates equity returns.

As Jason Zweig observed, “There is just one problem with tracing stock performance all the way back to 1802: It isn’t really valid.’ Siegel relied on data selectively cherry picked by professors Walter Buckingham Smith and Arthur Harrison Cole. Of over 1000 stocks in existence during 1802-1845, they ignored 97% of them. Hence, Siegel’s data series has an enormous survivorship bias built into it, especially in the 1802-1900 period.

By erroneously front-loading excess returns, the compounding effect over a century is enormous.

Birinyi Associates undertook a comprehensive analysis of Stocks for the Long Run. Their conclusions?

• We find little evidence that the author undertook basic, roll up your sleeves research. The book depends heavily on a 1989 study for the NBER by William Schwert which in turn was a “cobbling” together of a number of other prior studies varying greatly in terms of composition and methodology.

• Earlier researchers suggested that pre-1871 data was not available. Siegel defended his publication by noting that other, more recent, analysts have found the historic data to be available; however, one study’s data was not as complete as suggested, and in one instance had a huge error in data collection.

• Indexes and measures which overlapped those used by Siegel often have differed substantially from those Siegel used.

• The data used in the book showed an average annual price gain of 2.9% for the 200+ years; his total return result, 8.3%, was achieved by an aggressive dividend assumption, in excess of that suggested by other analysts.

Understand what this means: It does not mean that stocks are not a worthwhile investment, nor have no place in an asset allocation portfolio, Rather, what Gross called the cult of equities has radically overstated long term historic returns of stocks.

By artificially goosing equity return data for the 19th century, Siegel may very well has made equities over-owned in the 20th century.

Consider what this error means for traditional investors: The vast majority of classically educated MBAs and Economists have a very significant flaw in their investing assumptions. Imagine how many fund managers are running 100s of billions of dollars using this error as the basis for their money management.

The tenacity of bad ideas is quite astounding. Just because something is wrong, and verifiably so, does not seem to have much immediate impact. It remains a stable of academia, as well as the actual practice of investing investing — despite its questionable truth. We should not be surprised at this. Recall what Max Planck — who won a Nobel Prize for Physics in 1918 for originating quantum theory — famously said: “Truth never triumphs — its opponents just die out. Science advances one funeral at a time.”

Investing and Economics should be so lucky . . .

Previously:
Jeremy Siegel is not having a good year (July 11th, 2009)

Sources:
Does Stock-Market Data Really Go Back 200 Years?
Jason Zweig
WSJ, July 11, 2009
http://online.wsj.com/article/SB124725925791924871.html

What Do Stocks Really Return?
Birinyi Associates, October 06, 2009
http://tickersense.typepad.com/ticker_sense/2009/10/what-do-stocks-really-return.html