Why I Think Jeremy Siegel's Argument Is Flawed

Jeremy Siegel is defending criticism of his argument that Standard & Poor's is incorrectly calculating market profits. Siegel argues that profits within the S&P 500 index should be weighted according to market capitalization, not on an absolute basis as Standard & Poor's currently constructs the estimate of aggregate earnings. Thus, AIG, which has a tiny market cap but a massive loss should have its loss scaled down to the company's relative market cap when tallying up the earnings estimate for the entire index. Using Siegel's methodology, earnings for the S&P 500 in 2008 would have been $71.50, not $14.97.

I think this argument is flawed.

Aggregate output in the economy is not calculated based on relative aggregate firm values. Aggregate output is aggregate output, irrespective of theoretical firm values. Aggregate profitability is a function of aggregate output, not a function of theoretical aggregate firm value. Thus, profitability should be viewed holistically.

Let us look at a hypothetical economy of two stocks, both represented in an index. Company A earns $1 and has a market capitalization of $100. Company B earns $2 and has a market capitalization of $20. Company A has a price/earnings ratio of 100x while Company B has a P/E of 10x. The two companies earn $3 in aggregate and have a market cap of $120, giving the index a P/E of 40x.

However, according to Siegel, Company A's profits should be scaled according to its market capitalization. Thus Company A's earnings should comprise 83% of aggregate profits ($100/$120) while Company B should comprise 17%. Using Siegel's methodology, multiplying Company A's earnings of $1 by 83% and Company B's earnings of $2 by 17% then totaling the two gives us an aggregate earnings figure of $1.17. Yet actual profits were $3, not $1.17. The P/E of the index using Siegel's methodology would be 103x ($120/$1.17), not 40x. How is this an accurate reflection of the true profitability and valuation of the index?

Such a methodology may have other perverse effects. At the peak of the Technology Bubble, tech accounted for roughly 35% of the market capitalization of the S&P 500 but only 15% of aggregate index earnings. If one were looking for yet another reason to rationalize the insane valuations of technology stocks at the time - as so many were - some enterprising strategist could have argued that using Siegel's methodology would have made the technology sector look cheaper. Since the market cap of tech was 35%, adjusting for market cap, tech's profits should have been scaled up to 35%, bringing sector valuations down. Ergo, tech stocks were cheap!

I do have sympathy with Siegel's underlying premise that write-offs in the financial sector have vastly distort the underlying profitability of the S&P 500. I use $70 as the normalized earnings of the market, which is close to Siegel's figure when he constructs the market cap-weighted earnings. However, valuation is ephemeral and using it to scale aggregate market profitability is not an accurate reflection of aggregate profitability in neither the index nor the economy.

Comments

[Alphameister:]

Whether or not Siegel's proposal is flawed to an extent, I regard it as much less flawed and considerably more insightful than the current method of computing "the market" price-earnings ratio. Just as investors prefer to focus on non-GAAP earnings as more meaningful than GAAP earnings, sophisticated investors will want the insights afforded by Siegel's proposal. If 90% of stocks are very undervalued based on valuation multiples, but the distortions introduced by huge losses at financial institutions hide this reality from unsophisticated investors looking only at the "holistic" 500, the resulting lack of insight is likely to lead many investors to an errant conclusion about whether or not to invest.

Apr 12 10:44 AM

[CaptainJJack:]

I agree with Alphameister.

One of the underlying issues is that P/Es of individual stocks are not regarded as meaningful when earnings are zero or negative
(see the "NA" P/Es for Yahoo stock quotes of Ford or Citi, for example).

I think mechanically applying formulaic ratios will often develop inconsistencies. Dividing by zero, trying to take the square root of a negative number, or dealing with multiple roots often cause issues when trying to apply formulas in the real world.

One simple (at least to me) fix is to apply the same weights used in the index to the P/E ratios directly, possibly omitting the N/A ratios.

Apr 12 11:51 AM

[DougM:]

I agree with Toro. Here's a simpler argument: the S&P index is *already* cap-weighted. If you own the index, you own the companies in it, in proportion to the cap weighting. Here's a thought experiment: suppose you could buy up 100% of the shares of 100% of the companies in the index. You would then "own" 100% of the gains and losses from those companies, via straight-up summation. Now back it off to your actual ownership of less than 100%, and it scales back down the same way - you still end up with a fraction of a straight-up summation, not a weighted average. For Siegel's argument to be correct, you'd have to own a non-cap-weighted index - in fact, you'd have to own an index weighted per his proposed methodology.

Apr 12 02:01 PM

[dr wonder:]

Take earnings to an extreme; let one company have sufficient losses to wipe out the earnings for all other companies in the 500. Now what do you decide about the S&P 500? To me, it's more insightful to know what the distribution of P/Es is for the collection of companies. This could be done graphically and the investor could make their own decision as to the merit of the "out-liers" driving their investment decision. In the future, you might have one company with extreme earnings that drives the argument the other way.

I don't believe that the health of the economy should be dictated by one company either way. In the real world, a company with abnormally high losses should fail and go out of business. What has happened is that some managements no longer feel the responsibility to avoid such losses and that's what gives us cases like AIG. There's something fundamentally flawed with a system that allows a company or a few companies to basically bring down the entire credit/financial system -- if companies will not conduct their business in a responsible manner, then regulation becomes mandatory.

Apr 12 05:24 PM

[Alan Young:]

Neither equal weight nor market weight. If you want the real measure of the market's profitability, you wouldn't consider earnings _per share_ at all, but just total the amount of earnings of all the companies.
Total market cap / total earnings = total market P/E.
Whether this number is particularly useful is another discussion. Compiling the total this way in each sector might be more useful.

Apr 15 05:03 AM