Why Jeremy Siegel's S&P Earnings Analysis Is Wrong 10 comments
-
Font Size:
-
Print
- TweetThis
S&P 500 earnings topped out at about $84 a share in June 2007, while corporate profit margins were 44% above their average since 1980. At the time, these numbers were inflated by global bubbles in real-estate, commodities, liquidity, and growth expectations–a lot of global bubbles.
Since 2007, most of the bubbles deflated, stocks plunged, profit margins reverted to their mean (i.e. declined from above to below average), which in turn caused earnings to collapse. S&P 500 earnings estimates for 2008 were revised down to $28–a 67% drop from the highs. Suddenly, investors found that at 20-plus times earnings, the market is not that cheap, even after stocks’ precipitous drop.
As every action movie needs a hero, every cyclical bear market needs its bull. A bull we’ve got– Jeremy Siegel, a Wharton professor, wrote an op-ed article in the Wall Street Journal proclaiming that for the most part, aforementioned factors were not the main causes of the earnings collapse. Instead, flawed computation of the earnings of the S&P 500 is at fault. He explains that because Standard and Poor’s, the creator of the S&P 500 index, doesn’t weigh earnings by market cap but simply aggregates them together, earnings misrepresent the true reality of corporate profitability and thus overstate the market’s true valuation.
Siegel suggests weighing earnings by (today’s) market capitalization instead. When he does just that, all of a sudden reported earnings skyrocket to $71.10 a share. His conclusion is that the market is dirt cheap. It is brilliant, it is uplifting and it is just plain wrong.
If you exclude the losers, the market will always be cheap. In September 2007, General Electric (GE) was the second largest company in the S&P 500 index, Citigroup (C) was number five, and Bank of America (BAC) was number six. Today GE is number nine, its weight in the index declined from 3.15% to 1.42%; Bank of America and Citigroup are numbers 41 and 111, respectively. Financials (GE is a semi-financial stock, as a good chunk of its earnings–or now losses–comes from GE Capital) have dropped almost from a 20% weight in the September 2007 index to an obscurity today as their losses ballooned. Some of these losses are driven by market distortion but a majority of them are real. And, yes, those fat profits of late 2007 will be remembered, longed for, but not repeated for a long time.
By favoring the largest companies in the index (Siegel’s suggestion) you will always over-report the index’s earnings. Higher earnings will drive valuations (market cap) higher and will result in higher weights in the index–earnings numbers will matter more. At the same time, companies that suffer losses will be punished by the market and thus their market cap weight will be lower and (negative) contribution to overall earnings will be less. This is exactly what happened to Citigroup, Bank of America, GE and many other losers.
There is a good reason why S&P doesn’t weight earnings by market cap. Siegel’s suggestion makes as much sense as a university, for instance, calculating an average GPA score for the school by attributing higher weight to cum laude and summa cum laude students.
Market capitalization weighting makes no sense. Ironically, Mr. Siegel starts his article explaining how bizarre market weighted calculations are (S&P computes S&P 500 performance based on market cap, but computes earnings by adding all earnings for companies in the index, not by market cap). He explains that the performance of mega cap companies like Exxon (XOM) completely overshadows the rest of the index where performance of smaller companies becomes irrelevant. I agree. Stock market indexes that are constructed based on market capitalization don’t represent a reality of actual portfolio construction. I have yet to meet a money manager who determines his portfolio weights based on the market capitalization of companies in his portfolio.
What makes his message intellectually discordant is that Mr. Siegel has been promoting an index fund company (and I believe he is on the company’s board too)–WisdomTree. WisdomTree’s sole purpose is the creation of index funds not based on market capitalization but on fundamentals (i.e. price to dividends and price to earnings).
Mr. Siegel in his op-ed said that a disproportionate amount of losses come from companies that represent a very small market capitalization in the index. He is right, but their capitalization is small now, it was not small a year and a half ago. These companies used to be enormous. They are small because of their losses. If you were to recalculate S&P earnings based on the weights of the index in September 2007, you will likely get an earnings number that is far below Siegel’s suggested $71.10.
Also, Siegel’s claim that a small company taking huge losses distorts results of the index lacks intellectual rigor. An original small (I mean not one of the ones that just dropped in price tremendously because of high losses) company usually has low sales and lower profits, thus it is much harder for it to generate large dollar losses as well. In other words, my mother-in-law’s barber shop or Jones Apparel (the smallest company in S&P index) for that matter will never have enough losses to overshadow mega large caps in S&P.
Problems with Siegel’s suggested method don’t end there. Our understanding of historical valuations is based on S&P 500 P/E data going back to the early 1900s which is computed based on the current “flawed” (according to Siegel) way. Siegel’s P/E based on market cap weighting loses comparative reference to the past and thus loses its meaning as we start comparing apples (his computation) and oranges (aggregate, S&P’s computation). For a reasonable comparison with historical valuation levels Mr. Siegel would need to apply his capitalization weighted method to past periods. At 12 times, based on Sigel’s earnings market, may or may not be cheap as past P/Es are computed differently.
We know that $28 in estimated 2008 earnings doesn’t represent the true earnings power of the S&P 500, nor does $70. Of course, charge-offs tremendously distort reported earnings numbers, and no rational person looks at the 2008 $28 as a base case. But let’s be honest with ourselves, 2007 earnings of $84 are only real to us today if we think that the global bubbles of 2007 will reoccur in the very near future.
Siegel may manipulate earnings until they conform to this thesis as much as he likes, but the true number lies somewhere well south of $71, and thus while some stocks are cheap today, the S&P 500 is not.
By the way, Jeremy Siegel’s book, “Stocks For The Long Run,” is well written and provides a very good overview of the performance of different asset classes over the last two centuries. But the book needs a different title, maybe something like “Stocks for the Really, Really, Really, Long Run.” This way, it would not lure investors into a false sense of security when it comes to stocks. It preaches that stocks are always a buy, no matter what valuations as they do better than other asset classes in the long-run, and that a 7% real rate of return is a birthright for stock investors no matter if the stock market is extremely cheap or ridiculously expensive.
This is very true if your time horizon is 30 years or you plan to live forever. It is also true if you can tolerate seeing your portfolio go nowhere for a decade or longer. Unfortunately, most of us don’t have this time horizon. We need to send kids to school, pay for weddings, boats, and other stuff. I don’t know anyone who has the patience to see their portfolio of stocks do nothing for decades.
Despite stocks being a great investment for the really, really long run, they have periods when their returns are unspectacular. I call them range-bound markets and they take place after secular bull markets. The buy and hold (forget to sell approach) that Siegel’s book lures to believe fails to produce acceptable returns during these type of markets as the last decade proved so far.
Related Articles
|
-
The Seeking Alpha 100
See allThe top 100 stock
market authors
selected for publication
Seeking Alpha 100 » -
Fastest Climbers
Authors with the largestSee all
increase in followers
in the last 7 days
Fastest Climbers » -
Top Commenters
Top commenters,See all
as ranked by their peers
Top Commenters » -
Top Instabloggers
The 100 most activeSee all
Instabloggers.
Top Instabloggers » -
Top StockTalkers
TopSee all
StockTalkers.
Top StockTalkers »
This article has 10 comments:
The SP500 is an important and valuable time series and I think it has served us well. If you market weight earnings, you will always bias earnings towards sectors that are doing well; conversely when sectors are doing poorly, they will be underweighted.
The truth is that the economy, along with the companies that comprise it, moves through cycles and the results of these cycle should not be artificially dampened or exaggerated. But there are other problems with Siegel's suggestion.
Do you weight CITI, for instance, at their market cap at the beginning of the quarter, or at the end of the quarter. How this is answered could have a huge impact on the results reported under the Siegel regime.
1) The S&P represents the AGGREGATED market capitalization of the included companies
2) Therefore one needs to divide by the AGGREGATED earnings to arrive at a P/E for the S&P as a whole
This is exactly the same way one would calculate a P/E for a large company using the value and the earnings of its divisions.
Of course, one can debate whether one would rather have the non-weighted average P/E (taking a P/E per company and averaging over the number of companies). It would lead to all kinds of strange behaviour in case of mergers, divestments etc. but why not.
Siegel's suggestion, however, to weigh earnings by market cap and still use the aggregated market cap on top of the division does not make any sense at all (from a mathematical point of view).
You know we're not at the bottom until all these fools who want to use some new method to value the market are gone.
Next, we need to expose the use or operating earnings over the more relevant and historically used, as-reported earnings.
www2.standardandpoors....
Please take a look at this S&P data sheet to see how disparate operating earnings and as-reported earnings are. Also notice how P/E ratios jump lower as earnings estimates come into play.
> Next, we need to expose the use or operating earnings over the more...
Whoops! I meant:
Next, we need to expose the use of operating earnings over the more...
www.indexarb.com/index...
Owning the S&P index is exactly like owning a portfolio of these 500 stocks with weights given by the market cap. Notice AIG is ranked 455. It has a 0.02% weighting. It weighting last year is irrelevant. So if AIG loses another $200B in 2009 year and goes bankrupt, the effect of this stock will be at most 0.02% on your portfolio. If C, BAC, GS, MS together lose another $400B and go bankrupt, those stocks will take down the index another 1.78%. So those 5 companies lose $600B and the other 495 companies could make $600B for a net earnings of zero. So how much is a portfolio worth that makes zero earnings? That is a PE of infinity.
The right answer is that those 5 companies should be ignored. Since they all go bankrupt (worst case scenario of course) you just have a permanent loss of 1.8%. After that you have a portfolio of companies making $600B per year which is about $70 per share for the S&P 500 index. Since the S&P 500 is presently at 757, that would be a forward earnings of 10.8 which seems which reasonable.
This argument is completely bullet-proof and obvious. I can't understand how anyone could even argue that Siegel is wrong.
> Sorry, but you are wrong and Siegel is right.
> This argument is completely bullet-proof and obvious. I can't understand
> how anyone could even argue that Siegel is wrong.
Here's how:
The S&P 500 owns the same percentage of the market cap of each included company.
So for example, if the total market cap of all 500 companies is $7,330,339,733,333 (it’s close enough), and the S&P 500 index is at 800 (close enough), the index represents about 0.000000011% of the market cap of each those companies. So the index has about 38.335 of XOM ($348.5B*0.000000011%), or 4.75% or the index. Repeat this math 500 times and you'll get the right answers (within rounding errors).
Earnings work the same way. Since the index represents 0.000000011% of each company in the index, the earnings of the index is 0.000000011% of the earnings of each company. Let's say XOM earned $42.5B last year (it's close enough). The XOM component of S&P earnings is 0.000000011*$42,500,00... or $4.638. Repeat this 499 times and you'll get the whole number.
If I'm not quite describing it well enough, imagine the index represented 100% of each of the 500 constituents, and was 7330339733333 instead of 800. The percentage weights of each company would be the same. Calculating the earnings of the index, we would simply add up 100% of the earnings of all of the 500 companies in it. If we gave additional weight to any company, we would get the wrong answer. Same is true in the real index.
"Morningstar calculates average price ratios for indexes and investment portfolios using the harmonic weighted average. This method compares the total market value of the portfolio to the portfolio’s share of the underlying stocks’ earnings (or book value, cash flow, sales or dividends). Morningstar prefers the harmonic method to an arithmetic weighted average, because outliers can easily skew the results of the arithmetic method."
On Mar 23 10:23 PM Vox Rationalis wrote:
> DaveJ wrote: