The S&P 500: Five Things You Probably Don't Know

The S&P 500 is widely regarded as the best single gauge of the U.S. equities market and as a decent proxy for the total market. But many investors rely on the S&P 500, its statistics, and its progeny—such as the Dividend Aristocrats list—without knowing how they work. This can lead to bad investment decisions.

Here are five things you probably don’t know about the S&P 500.

1. The S&P 500 is Not Static; It is Actively Managed

The S&P 500’s stocks are determined by Standard & Poor’s Index Committee, a unit of Standard & Poor’s (which in turn is a unit of The McGraw-Hill Companies (MHP), Inc.). Precursors of the S&P 500 date back to 1923, when Standard and Poor’s introduced an index covering 233 companies. Today’s 500-stock index was introduced in 1957.

Contrary to a popular misconception, the S&P 500 is not comprised of the 500 largest companies. That said, none of the companies are small. Combined, they cover about 75% of the U.S. equities market in terms of market capitalization. In 2002, the Committee purged foreign-based companies, so the index now represents only U.S.-based companies.

While often called “unmanaged,” there is a fairly continual turnover in the S&P 500. Each year, the Committee changes between about 10 and 40 of the companies in the index. Changes to the index can be made at any time. They are typically announced one to five days before they are implemented.

The Index Committee does not usually reveal the specific reasons when it adds or drops a company. Its guidelines require that companies in the index be industry leaders, widely held, representative of a variety of industries, “liquid” (commonly traded), and fundamentally sound. Companies are removed from the index when they merge themselves out of existence, go bankrupt, become fundamentally unsound, or no longer fairly represent their industry.

So far in 2009, the S&P 500 has dropped nine companies and added an equal number to replace them. Companies dropped from the index so far in 2009 are UST (UST), Sovereign Bancorp (SOV), Weatherford International (WFT), Jones Apparel (JNY), American Capital (ACAS), Tyco (TYC), Noble (NE), Developers Diversified Realty (DDR), and Rohm and Haas (ROH). They have been replaced in the index by Iron Mountain (IRM), Health Care REIT (HCN), Diamond Offshore Drilling (DO), Ventas (VTR), Hormel Foods (HRL), Northeast Utilities (NU), O’Reilly Automotive (ORLY), Time Warner Cable (TWC), and Denbury Resources (DNR). In 2008 35 companies were replaced, meaning that 7% of the index turned over.

Many investment advisers who tout the benefits of “passively holding” an index like the S&P 500 make it sound like the index is an unchanging sample of U.S. companies. On the contrary, it is a frequently changing sample, and in that sense it is not a passive investment at all. Someone who holds the S&P 500 (via a mutual fund or ETF) holds an actively managed set of stocks. The manager is the S&P Index Committee.

2. The P/E of the S&P 500 is Distorted

The P/E of a stock is its current price (P) divided by its earnings-per-share over the past 12 months (E). But what is the P/E of an index?

If the index is equal-weighted (that is, if all stocks in the index contribute equally to the index’s value), then the P/E of the index would be the sum of all the current prices divided by the sum of all the past 12 months’ earnings-per-share.

But what if the index is not equal-weighted? As you probably know, the S&P 500 is “cap weighted,” meaning that the weight of each company is in proportion to its market capitalization. That means that the “price” (or current value) of the S&P 500 is affected far more by price changes in its largest-cap stocks than by its smallest. The largest stock in the index is Exxon Mobil (XOM), which by itself comprises almost 5% of the whole index and wields 934 times the weight of the smallest stock in the index. The top 10 stocks in the index comprise almost 22% of the index’s entire value.

But the index’s P/E is not cap-weighted. To calculate the P/E, S&P sums the earnings-per-share for all 500 stocks over the past 12 months—unweighte... uses that raw number as the E in P/E. The numerator P is the index’s value, which as we have seen is cap-weighted. Thus the calculation of P/E is a hybrid: a cap-weighted P divided by an equal-weighted E for the 500 stocks in the index.

Wharton Professor Jeremy Siegel (author of Stocks for the Long Run and The Future for Investors) pointed out this anomaly in a Wall Street Journal article on February 25, 2009. He called it “…the bizarre way in which S&P…calculates... ratios for their indexes.”

I am not sure in what direction nor by how much this calculation distorts the S&P 500’s “real” P/E, but clearly it distorts it. Say you believe that over the long haul, the average P/E of the whole market has been about 17, or 16, or 15 (the number Siegel uses). It’s an important number, because many people use it to gain a notion of whether the current market is undervalued, overvalued, or priced about right.

Siegel said that with the earnings crash in Q4 2008, when so many companies had huge write-downs, the distortion overstates the P/E ratio. That is, it makes stocks look over-valued. S&P’s final calculation for Q4 2008 was that the 500 stocks had aggregate earnings of negative $23.25. When Siegel wrote that final number was not yet available, but he said,

As the fourth-quarter earnings season draws to a close, there are an estimated 80 companies in the S&P 500 with 2008 losses totaling about $240 billion. Under S&P's methodology, these firms are subtracting more than $27 per share from index earnings although they represent only 6.4% of weight in the index. …If one applies market weights to each firm's earnings using the same procedure that S&P employs to compute returns, the results yield a more accurate view of the current profit picture. Market weights produce a reported earnings estimate of $71.10 for 2008 -- nearly 80% higher than the unweighted procedure. The reason for this stark difference is that the firms with huge losses generally have extremely low market values and hence have a much smaller impact on the total earnings in the index.

Looking forward, S&P’s current estimates are that the S&P 500 will have a P/E of over 4186 at the end of June, and a negative P/E at the end of September, 2009. Both of those estimates result from the overwhelming effect of the negative earnings number for Q4 2008. If Siegel is correct, those P/E numbers are not useful.

3. The Dividend Aristocrats List is Only Updated Once per Year

Many dividend investors treat the Dividend Aristocrats list as an imprimatur of dividend safety and assured dividend growth. They shouldn’t.

S&P updates the list once per year, each December. That means the list ignores events that happen early in the year, such as announced dividend cuts. The annual updating can be very misleading, as S&P describes their Dividend Aristocrats thusly:

S&P 500 Dividend Aristocrats is designed to measure the performance of S&P 500 index constituents that have followed a policy of consistently increasing dividends every year for at least 25 consecutive years.

The poster child for this year’s most misleading company on the Dividend Aristocrats list is GE. The company announced early in the year that they intend to cut their dividend later in 2009. That doesn’t matter to S&P. GE still appears on the list as a Dividend Aristocrat.

Don’t believe this does not fake people out. I have read two articles on Seeking Alpha in the past couple weeks that touted GE as a reliable dividend-raising company. Presumably, the authors each mistakenly relied on the Dividend Aristocrats list to reach that conclusion.

4. The Dividend Aristocrats List Requires Membership in the S&P 500

The list leaves out companies not in the S&P 500. For one thing, that means it ignores foreign-based companies. In fact, the only way a company can get dropped from the Dividend Aristocrats list during the year (that is, prior to the annual December update) is if the company itself is dropped from the S&P 500 during the year.

5. The Dividend Aristocrats List Does Not Include Some Companies Whose Dividend Rises Each Year

The following is not intuitive, but it is true. If a company raises its dividend in the middle of every other year, its total dividend payout will be higher each year. Say that Company A pays $1.00 per quarter in Year 1, for a total payout of $4.00. It continues that dividend in Year 2 until the third quarter, when it raises its dividend to $1.25. The total dividend paid in Year 2 will be $4.50, and S&P will correctly pick this up as a dividend increase. But in Year 3, say that the company does not declare a dividend increase all year, holding the dividend at $1.25 per quarter. The total payout in Year 3 will be $5.00, an increase over Year 2. But to S&P, Company A did not increase its dividend in Year 3, although obviously its payout went up.

If a company follows a similar pattern of raising its dividend every other year, its annual payout will increase every year. But S&P will not count that company as a Dividend Aristocrat even when the total consecutive years of rising dividends reaches 25 straight years. This probably does not match what most people think of when they think of a company raising its dividend each year.

Disclosure: The author is long SPY, an ETF that tracks the S&P 500.

Comments

[Insiderman:]

For cap weighting the index, you have to do it the way you describe:

"But the index’s P/E is not cap-weighted. To calculate the P/E, S&P sums the earnings-per-share for all 500 stocks over the past 12 months—unweighte... uses that raw number as the E in P/E. The numerator P is the index’s value, which as we have seen is cap-weighted. Thus the calculation of P/E is a hybrid: a cap-weighted P divided by an equal-weighted E for the 500 stocks in the index."

When I make an earnings forecast for each company in the S&P 500 and sum those results, I want to be able to multiply by some function of the P/E to know the value of the index.

May 27 09:15 AM

[Ashu Joshi:]

Thx for the detailed information David. Very informative, it is was good read!

May 27 09:59 AM

[Lightway:]

Great article David, thank you for the information !

May 27 12:15 PM

[Fred Voetsch:]

Perhaps one of the worst articles using the most convoluted logic I have read in quite some time.

May 27 06:17 PM

[Vox Rationalis:]

"But the index’s P/E is not cap-weighted. To calculate the P/E, S&P sums the earnings-per-share for all 500 stocks over the past 12 months—unweighted...Thus the calculation of P/E is a hybrid: a cap-weighted P divided by an equal-weighted E for the 500 stocks in the index."

Nonsense. Download the S&P earnings spreadsheet here: www2.standardandpoors..... Note EPS numbers in cells D49 (the current quarter's estimate) and D52-54 (previous quarters results). Add the current quarter and the prior three to get $6.91, which results in the P/E of 130 (cell H49).

Now go to the tab called "Issue Level Data." This is a list of those EPS numbers reported to date for this quarter. The sum of the "As Reported" column is $77 and change. CLEARLY, this is not how the numbers on the "Estimates&PEs" page were calculated.

"Wharton Professor Jeremy Siegel..."

Siegel's P/E discussion was completely wrong, though wrong in a different way from how you're wrong.

May 28 03:53 PM

[David Van K...:]

Vox,

Thanks for pointing out that the sum of the stock-by-stock "as reported" earnings per share for all 500 stocks does not equal the "as reported earnings per share" used in the calculation of P/E ratios, despite the similar labels. S&P obviously adjusts the raw numbers before calculating the P/E ratio.

I have done a fair amount of research, and I cannot find the exact adjustment that takes them from the issue-level earnings to the earnings number they use when calculating the P/E ratio. Perhaps you can enlighten me. My hypothesis is that it involves the "divisor" used to normalize the value of the index over time, but I am unable to confirm that.

It is clear, however, that the adjustment is not by market capitalization, so my basic statement stands. Siegel's op-ed piece in the Wall Street Journal touched off a firestorm of debate that continues to the present day. In a nutshell, some people think Siegel is right, while others think he is way off base. Some think he is biased by other business enterprises in which he is involved. Be that as it may, when the chairman of S&P's Index Committee responded to Siegel, he made it clear that S&P aggregates earnings without weighting them (as I described), although, as you point out, clearly some adjustment is made to the raw earnings numbers before they are transported to the P/E calculation.

I also want to thank you for writing a civil comment, proving that people can disagree without being disagreeable. My observation is that many commenters on Seeking Alpha (and similar forums) often seem unable to disagree without being insulting, belittling, or sarcastic towards the people they disagree with. That's too bad. I know that I, and many other contributors and commenters, are simply trying to make fact-based, informative posts that will help investors. When someone makes a mistake, whether of fact or interpretation, it does not require a sarcastic or belittling comment in response. Such comments reflect far more on the commenter than on the target of their wrath.

One interesting fact that I turned up in my further research is that 8 of the 15 largest "detractors" (negative earners) in the 2008 S&P earnings numbers have been removed from the index. Those removed accounted for about 36% of the negative earnings of the top 15 in the Q4 negative earnings number. This underscores my point about the S&P 500 being an actively managed portfolio of stocks.

May 28 09:21 PM

[Vox Rationalis:]

David Van Knapp wrote:

> I also want to thank you for writing a civil comment,

Paraphrasing Quentin Tarantino, if I'm curt with you it's because time is a factor. My initial comments tend to be short and direct, and my follow-ups tend to mirror the attitudes with which I'm confronted. Though I have been known to be an ass from the get-go.

> S&P obviously adjusts the raw numbers before calculating the P/E ratio.

You can't use EPS at all, because the number of shares is different in every company. If BRK was part of the index, would it make any sense to use its EPS of 1626?

Blitzer's response to Siegel hints at the solution:

"[W]e could hypothetically view the S&P 500 as a single company with 500 divisions...[t]he smallest of these divisions could have an outsized loss that wipes out the combined earnings of the entire company."

Just as the P/E of an individual stock is the market cap divided by earnings (not EPS), the S&P 500 P/E is the sum of the market caps of all component companies divided by the sum of earnings of all component companies. Here's the math:

P/E = Price / EPS

EPS = Earnings / shares outstanding

P/E = Price / (earnings/shares)

P/E = Price * (shares/earnings) or (Price*shares)/earnings

Price*shares = Market Cap

P/E = Market Cap / earnings

If you're skeptical of this, pick a company or two and run it.

While a quick search didn't come up with an S&P methodology document, this is the only way that makes any sense whatsoever, and is indicated by Siegel is his wrong-headed op-ed: "S&P adds together, dollar for dollar, the large losses of a few firms to the profits of healthy firms..." Just as it adds together the market caps, dollar for dollar, of the large firms to the small firms.

BTW, Siegel was wrong because he was trying to weight something that's already weighted.

A little more of the logic is here: seekingalpha.com/user/....

May 29 10:17 AM

[ThomasDevlin:]

This is all very interesting! VOX and David (and I agree let's all try to discuss things without the put downs.

So my question to VOX is where can we get a quick and easy update of the current P/E ratio, and better yet an historical graph using consistent methodology?

Thanks!


On May 29 10:17 AM Vox Rationalis wrote:

> David Van Knapp wrote:

Jun 30 07:23 PM