Felix Salmon

How the S&P 500 destroyed $4.5 trillion

By Felix Salmon
February 18, 2011

At the end of 1993, Cisco Systems had a market capitalization of $8 billion. At the end of 2010, it was worth $112 billion. It hasn’t paid dividends, which makes things easy: if you want to calculate the amount of shareholder value that Cisco created between 1993 and 2010, you just subtract the former figure from the latter and get an impressive $104 billion. Right? Wrong. In fact, if you go through the history of Cisco’s stock actions year by year, it turns out that the company has managed to destroy $105 billion over the past 18 years. Microsoft and Intel have both destroyed $72 billion, Time Warner managed to destroy $130 billion, and Pfizer destroyed a whopping $188 billion. Four of the top five value destroyers, however, were financial: AIG, GE, BofA and Citigroup between them destroyed a mind-boggling $739 billion between 1993 and 2010, most of it in 2008.

All these numbers come from my new favorite paper, the product of some serious number-crunching at IESE Business School in Madrid. (Update: I missed some small print in the methodology, so while this paper is interesting it’s not quite as interesting as I thought at first. See below.)

Here’s the abstract:

In the period 1991-2010, the S&P 500 destroyed value for the shareholders ($4.5 trillion). In 1991-1999 it created value ($5.1 trillion), but in 2000-2010 it destroyed $9.6 trillion. The market value of the S&P 500 was $2.8 trillion in 1991 and $11.4 trillion in 2010.

We also calculate the created shareholder value of the 500 companies during the 18-year period 1993- 2010. The top shareholder value creators in that period have been Apple ($212bn), Exxon Mobil (86), IBM (78), Altria Group (70) and Chevron (67). The top shareholder value destroyers in that period have been American Intl Group ($-217), Pfizer (-188), General Electric (-183), Bank of America (-170), Citigroup (-169) and Time Warner (-130). 41% of the companies included in the S&P 500 in 2004 or 2010 created value in 1993-2010 for their shareholders, while 59% destroyed value.

How can the S&P 500 destroy $4.5 trillion of shareholder value over a period when its capitalization rose by $8.6 trillion? The answer is that companies issue stock when it’s expensive, rather than when it’s cheap. During the dot-com bubble, Cisco was an M&A monster, going on a massive acquisition spree and nearly always paying in its highly-rated stock. When that stock crashed, it took down with it all the value invested at the top of the bubble, which was many more shares than were oustanding back in 1993. More generally, companies with high-flying stocks are likely to pay their employees with stock or options. That can account for a lot of value destruction if and when the shares fall to earth.

That said, the S&P 500 would have created shareholder value, on a net basis, between 1991 and 2010 were it not for the annus horribilis of 2008, when $5.8 trillion of value was destroyed. In general, the down years are much bigger than the up years: the best year of all was 2003, when $1.7 trillion of value was created, while substantially more than that was destroyed in each of 2000, 2001, 2002, and of course 2008.

This is one of the main reasons why the returns that real individual investors get from investing in stocks are substantially lower than the theoretical numbers that financial advisers love to show you. And why you prefer to get paid in cash rather than in stock.

Finally, I think this paper demonstrates that Apple is a screaming long-term sell right now. No company, bar Apple, has even created $100 billion of shareholder value over the past 20 years, let alone the $212 billion figure that Apple is currently boasting. It’s an extreme outlier, and the downside is enormous: if you buy Apple shares now, there’s $327 billion of downside.

Google, by contrast, is much less of an outlier, having created a relatively modest $5 billion of shareholder value in its time as a public company. Go check out the number for your own favorite stock in the appendix of the paper, which lists shareholder value creation between 1993 and 2010 for all 633 companies which were part of the S&P 500 either in 2010 or in 2004. Most of the numbers — 59%, to be precise — are negative.

Update: Thanks to eagle-eyed commenters for catching something very big here: the paper is measuring risk-adjusted returns, not absolute returns. According to the definitions in the paper, “A company creates value for the shareholders when the shareholder return exceeds the required return to equity”. And the required return to equity is defined as the return of long-term treasury bonds plus a risk premium which seems to fluctuate between about 4% and 5%. Remember that long-term Treasuries did very well indeed over the period in question. So the value-destruction figures here are a bit fictitious: it’s not actual money being destroyed, but rather the hypothetical money that you would have got if you’d invested in instruments yielding about 4.3% over Treasuries. I’d love to see the numbers raw, without the risk adjustment.

17 comments so far | RSS Comments RSS

For a visual look at the differences in stock issuance and buyback policies at CSCO, INTC, and IBM, here’s a posting I did last fall: http://rp-pix.com/bg

Posted by tombrakke | Report as abusive

“During the dot-com bubble, Cisco was an M&A monster, going on a massive acquisition spree and nearly always paying in its highly-rated stock.”

When you are confident in the synergies acquire with cash, if not (or cannot) then acquire using stock.

Posted by david3 | Report as abusive

That’s an interesting paper. Their key metric though–shareholder value creation–is created by comparing to an expected cost of equity for each stock that presumably is derived via CAPM. This in turn makes it a bit dodgy in that you need to figure out the right market risk premium and beta for each stock, which is not straightforward. It does seem they are being fairly conservative on this front, but maybe what this paper suggests is that investors’ risk premia are lower than commonly believed. Although the paper doesn’t quite spell it out, it looks like most of these companies have created value relative to risk-free assets.

Posted by right | Report as abusive

An interesting premise, but don’t they have it backwards? Or at least incomplete?

If Microsoft buys back 100M shares at $50, then the market price plummets to $20, then they’ve destroyed $3B of value for their long-term buy-and-hold shareholders. Surely this is indisputable?

If Microsoft issues 100M shares at $50, and the market price plummets to $20, then the prior shareholders are $3B ahead of where they would have been without the share issuance.

This paper seems to be looking at it exclusively from the perspective of the NEW shareholders. Those who buy newly issued shares (or receive them in a merger, or as option grants) may feel shortchanged when the shares they receive are worth less than the nominal sum. But those who controlled the company at the time the shares were issued are likely happy that they were able to pay with overvalued scrip rather than precious cash.

Agreed entirely that this is part of the reason why blindly diversified individual investors (and index funds) can expect weak returns. Index investing puts more cash towards those companies with an inflated share price, less towards those which are undervalued.

Posted by TFF | Report as abusive

This seemed too horrible to be believed. If the world is this bad, we ought to all just pack our bags and go home.

But look at the very high bar that was set:

Created shareholder value = Equity market value x (Shareholder return – Ke)

For 1991 to 2010, Ke was arbitrarily set at 9.6% to include both the return of treasuries and an extra 4.3% above that for risk premium.

In other words a company that hits a terrific 9.6% real return over the period is awarded a big fat zero. That by the way means a 525% real return over 20 years.

Those poor shareholders are crying all the way to the bank!

Posted by DanHess | Report as abusive

“I’d love to see the numbers raw, without the risk adjustment.”

Ah. That would explain quite a bit…

I think anybody here would jump at the chance to receive expected returns that are 4% to 5% ahead of Treasuries, no?

Posted by TFF | Report as abusive

Salmon strikes out again… like high school all over again, huh felix.

Posted by TinyOne | Report as abusive

I see you are indulging yourself in a lie of the second kind, Felix. You have reported the paper’s headline numbers on “value created” without explaining to your readers that in the lexicon of the paper, this differs materially from “value added.” The paper is more forthright, explaining immediately that whereas “value added” is the absolute total increase in value, “value created” is relative to a “required return”, defined as a risk premium over treasuries.

If you believe that treasuries are risk-free, this relative measure is obviously of interest. However, because of the difficulty in measuring the required risk premium, the absolute values are still important. If you don’t believe that treasuries are risk-free, then the method of the paper must be revised to risk-adjust the treasury return in order to calculate the relative numbers; only the absolute ones are valid.

No matter what, an honest journalist would have mentioned the value created in the ordinary sense. For the record, these values are: 1991-2010 +9.9 trillion, 1991-1999 +9.45 trillion, 2000-2010 +0.45 trillion.

Posted by Greycap | Report as abusive

Wait, you’re saying Apple has $327B of downside now? So we’re talking about the risk of them going bankrupt? I guess every stock has that risk, but why single them out, especially when their sales and profits are growing rapidly?

And there’s a reason no other company has created as much value as Apple, because no other company has their profits and growth. Yes, they are an outlier, but somebody has to be. No reason to sell them yet (I’m not).

And yeah, TFF, I would jump at 4-5% over treasuries. Wouldn’t trade my Apple shares for them just yet, but would definitely take that over Cisco and Intel.

Posted by KenG_CA | Report as abusive

Well, if you didn’t actually *read* the paper, then I must withdraw my imputation of dishonesty. I apologize.

What I would like to know is what “long term treasury” investment means. Does this mean buying a stripped cashflow that matures at the measurement horizon (e.g. buy in 1991, mature in 1999?) Or does it mean buying a coupon bond that matures at the measurement horizon and doing – what, exactly – with the coupons? Or does it mean buy a long (e.g. 30Y) bond and rolling over each year, realizing P&L? Or does it mean buy a 1Y stripped cashflow each year? Why should any of these investment strategies be considered “risk-free”? Just how were those annual risk premia calculated anyway?

Posted by Greycap | Report as abusive

Bravo Felix!

Be careful to write those “hidden truths” too often… :)

Posted by robb1 | Report as abusive


You destroy more value with each word you write. Either totally wrong or totally boring.


Posted by snarkyfest | Report as abusive

$10,000.00 Price No of Share Split Value
Jan 10-92 0.502 19,920 10,000
Mar 23-92 39,841 2
Mar 22-93 79,681 2
Mar 21-94 159,363 2
Feb 20-96 318,725 2
Dec 17-97 637,450 2
Sep 16- 98 1,274,900 2
Jun -22-99 2,549,801 2
Mar-23-00 5,099,602 2
Dec-24-2010 23.96 5,099,602 122,186,454.18

Just a back of the envelope calculations for Cisco. If someone invested $10000 in 1992 and held on till end 2009. That would be the return he might have (excluding transaction cost and taxes) even if it reduces about 5% in annual return, that still much higher than the 9.6% the paper expects to generate. So could some one quite explain to me on what basis does CISCO destroy value over that period of time.

Posted by brmr | Report as abusive

brmr, I didn’t read the paper in detail, but from what Felix was saying I *think* he was talking about Cisco stock in aggregate, not the experience of any single investor. Cisco was regularly issuing stock throughout the 90s, for acquisitions and for employee options, and those likely did not fare as well as a buy-and-hold investor from 1992.

Posted by TFF | Report as abusive

brmr and TFF: I just recreated the calculations for Cisco and everything is due to the benchmark he is using. This not so much about a buy-and-hold investor from 1992 as a series of 18 annual buy-and-hold investors who bought the entire firm at the beginning of January and sold at the end of December. I get “created shareholder value” for Cisco of –98.9 billion dollars, and I think the difference with the reported figure of $–105b is because I couldn’t be bothered to account for share buybacks.

https://spreadsheets.google.com/ccc?key= 0Avrd27tf6U4KdDJuSVA3c1VkMUR6VUR1VGFPNGZ zLXc&hl=en

Posted by guanix | Report as abusive

Better link: http://1e6.us/pj

Posted by guanix | Report as abusive

It’s not important how much Cisco lost or earned. The question is the investor would have done better in another investment or not. Public companies should not be allowed to have a P/E larger than 25, no matter what there hypothetical growth is. Nobody can guarantee future earning. Investing media shouldn’t be allowed to hype an investment. In 1999 there were analysts predicting the Dow will top 16000. One analyst wrote a book, the Dow will top 40000.

Posted by joefar | Report as abusive

Post Your Comment

We welcome comments that advance the story through relevant opinion, anecdotes, links and data. If you see a comment that you believe is irrelevant or inappropriate, you can flag it to our editors by using the report abuse links. Views expressed in the comments do not represent those of Reuters. For more information on our comment policy, see http://blogs.reuters.com/fulldisclosure/2010/09/27/toward-a-more-thoughtful-conversation-on-stories/