(FORTUNE Magazine) – Near the end of May, General Electric stock was selling for just under $50 a share. At the same time, an option to buy a GE share for $55, valid until January 2003, was going for $6.90 on the American Stock Exchange. That's right, people were paying real money for the opportunity to buy a share of GE for more than the current price. And no wonder: GE's shares have a storied history of going up, so betting on a modest rise within the next two years hardly seems far-fetched.

That same glorious record might explain why investors didn't blanch when GE handed outgoing CEO Jack Welch the seemingly lavish preretirement gift last September of three million company stock options. Valid until September 2010, the securities give the legendary chief the right to buy GE shares at the stock's then-price of 57 3/8.

Another reason investors didn't blanch: According to the company's income statement, giving Welch all those options didn't cost a dime.

There's a technical term for this financial inconsistency: nuts. You have two stock options with nearly identical strike prices, but one costs money and the other is "free."

It's free in Accounting Land, that is. In the real world it takes just a little quick arithmetic to get at least an idea of what Welch's options might actually be worth. With the other, similar options selling on Amex for $6.90 a pop, a valuation of somewhere in the neighborhood of $20 million seems reasonable. Or the neighborhood might be even ritzier than that. Welch's options, after all, have a distinct advantage over the Amex version: They can be exercised for a full seven years after the latter expire. Now, suppose that GE's share price does during the next ten years what it did during the past ten and increases eightfold, to $400. At that point Welch could exercise his options to buy GE stock for 57 3/8 a share--and pocket $1 billion (before income taxes and broker fees). That hypothetical billion, mind you, is money that GE would be giving up by selling the shares at a discount to Welch instead of at full price on the New York Stock Exchange. But to repeat: GE's earnings statement will never give a hint of that cost.

Admittedly, this is a lot of math to swallow in the first few paragraphs of a magazine article. What's more--we're sorry to say--this story is not specifically about either GE or the always fascinating Jack Welch; we could have substituted data from any of hundreds of companies. So why start this piece with so many numbers? Because they give form, squishy as it may be, to a question that can be maddeningly abstract: What is the true cost of those yachtloads of stock options that companies give out every year?

An innocent little debating point it is not. Stock options have replaced salary and bonuses as the most important element of executive pay at countless companies--and have become the most important element of everybody's pay at a few places. The tables on the following pages, compiled using the latest crop of annual reports, show just how big a deal options have become even at lumbering old-timers like GM and AT&T. At tech companies, the estimated value of the options granted can dwarf earnings and sometimes even revenues. And while the Nasdaq crash may have robbed options of some of their Silicon Valley luster, most options-granting companies have reacted to the down market simply by granting more of them--at new, lower prices.

Unfortunately, when it comes to the question of what all this largesse is costing you the shareholder, the lack of candor is frightening. While a mine of information about options is buried in proxy statements and the footnotes of annual reports, the impact of options is almost entirely invisible in the quarterly financial numbers most investors follow (earnings per share, cash flow, Ebitda). It's not just that companies are clever about hiding the true cost of employee stock options, or that nutty accounting rules help them do it. All of us--investors, Wall Street analysts, journalists--help them do it. By ignoring the numbers that are out there, albeit squirreled away in footnotes, we've become a nation of options-abuse enablers.

Step one to recovery: Admit that options aren't free. They involve taking something away from the existing shareholders of a company and handing it to the company's employees. This transfer of wealth, or at least potential wealth, may be a good idea. It may pay off for investors in the long run. But the transfer is real nonetheless.

Giving out options costs a company's shareholders in two ways. The first is by diluting their stake in the company. When employees exercise their options, a company has to issue new shares. This means there are more shares outstanding, which in turn means the stake of existing shareholders in the company is reduced. So when an option is issued, it amounts to a claim on the company--think of it as someone putting a lien on your house. And the only way to find out about that lien is to look deep in the footnotes of the annual report.

The other price shareholders pay is the opportunity cost their company incurs by selling shares at a low price to employees instead of selling them at full price to investors. If a company were to take all those discounted shares and sell them instead on the open market, it would of course have a lot more cash to spend. And whatever it spent that cash on--machines, consultants, salaries, bonuses--would show up as an expense on the income statement. In economic terms there is no difference between compensating employees by giving them cash and paying them with securities that they can convert into cash. To put it another way: If selling shares to the public and using the proceeds to pay an employee is a cost, then selling those shares to that same employee at a discount (and letting him book the resale profit) is no less a cost.

Hypothetical sounding or not, the latter cost becomes very real when a company repurchases shares in order to keep options from affecting its reported earnings per share. Remarkably, however, the money spent on share buybacks to counteract options dilution does not have to be reported as an expense.

We're not talking chump change here. Sanford C. Bernstein, which does the most in-depth options research of any firm on Wall Street, last year analyzed the dilutive effect of every outstanding option at 436 large-cap companies at the end of 1999. Bernstein estimates that, to keep existing shares from being watered down, these companies would have to spend a staggering $53 billion a year buying back their own stock. That figure is equivalent to an amazing 13% of operating earnings--51% for tech companies.

So here's a lesson for investors: When a company hands out tens of millions of options every year and contends through its earnings statement that the things cost nothing, it is telling--there's really no other way to put this--a big stinking lie.

So how should we account for options? The simplest way is the way the Internal Revenue Service does it. As far as Uncle Sam is concerned, Jack Welch's hypothetical $1 billion gain from exercising options would count as $1 billion in compensation expense for GE. And this is fun: All that compensation expense (which, remember, doesn't even show up on GE's earnings statement) would be tax-deductible.

This hefty deduction was in the news a lot last year because so many employees at Cisco and Microsoft and a few other prominent tech companies exercised so many options for such huge gain that those companies paid virtually no income tax. Even if these tech giants didn't pay much or any tax, though, their employees did--and, for the most part, at higher rates than corporations pay. Mark Zandi of Economy.com estimates that $85 billion of last year's federal income-tax revenues came from stock-option profits. That's 36% of the year's budget surplus right there. Thanks, all you options millionaires!

But this information really isn't all that helpful to investors. By the IRS reckoning, Cisco and Microsoft each had a terrible fiscal 2000 (both companies' fiscal years end in the summer) because their employees exercised rafts of options as stock prices rose. Barring a spectacular market recovery, fiscal 2001 will presumably see fewer options exercised (or at least fewer massive options windfalls), thus providing a boost to taxable earnings. None of which, we should stress, says a thing about how either company's underlying business is doing.

A better solution would be to value options when they're granted--that way earnings would give evidence of a company's options generosity but wouldn't be entirely captive to stock price swings. If the options companies gave out were the same as the ones traded on exchanges like Amex and the Chicago Board Options Exchange, that would be easy. But they aren't. The longest-term exchange-traded options expire after three years; most have a duration of months. The options that companies grant to employees, meanwhile, usually last ten years but can't be used until they've vested--usually one to four years after they're granted. They also can't be bought or sold.

One can, however, use a mathematical formula developed for options trading--the best known is something called the Black-Scholes model--to estimate the value of an employee option. The Financial Accounting Standards Board (the independent, nongovernmental body that determines what constitutes Generally Accepted Accounting Practices, or GAAP, in the U.S.) was confident enough in the accuracy of Black-Scholes and similar pricing formulas to demand in 1993 that they be used to estimate options costs so that they'd show up on the earnings statement. At which point a nation of CEOs and CFOs and a few thousand of the Silicon Valley rank and file screamed like hell, protesting that if options had to be expensed, companies would stop giving them out! (Think about that for a minute....)

FASB backed down. It added the new options formula to the GAAP pantheon but allowed companies to keep ignoring options on their income statement as long as they disclosed in an annual-report footnote what their earnings would be if options costs were deducted. That, not surprisingly, is what most companies have chosen to do. (In the tables accompanying this article, Boeing is the one company that puts options costs on its income statement.)

The good news for investors is that the Black-Scholes estimates of options costs are now available and reasonably easy to find in companies' 10-K forms. (For those who consume their 10-Ks online at edgar.sec.gov or a related site, the quickest way to find the options footnote is usually to search on the word "Scholes.") Amazingly, though the standard went into effect six years ago, it is only in the 2000 annual reports that the full cost of these securities is being reflected for many companies. That's because the new rule, which calls for expensing options when they vest, applied only to options granted in 1995 and afterward--and many options take five years to vest. Those costs can be doozies (see tables). In fiscal year 2000, GE's earnings drop from $1.27 to $1.24 per share when the estimated cost of options grants is included; Merrill Lynch's, from $4.11 to $3.73; Microsoft's, from $1.70 to $1.48; Cisco's, from 36 cents to 21 cents; Juniper Networks', from 43 cents to negative 56 cents.

There are still other ways to figure options costs. In their proxy statements, some companies estimate how much money their executives will make if they wait ten years to cash in on their options and the company's stock price goes up 5% or 10% a year (in Welch's case, the figures are $108 million and $274 million). In the tables that go with the preceding article--which include options grants prior to 1995--FORTUNE assesses these securities at a third of their exercise price, an unscientific but common practice. All the options-valuation methods are flawed. But even so, they agree on one thing: Jack Welch got more than a gold watch.

Do companies really think options are free? The answer is no. For proof, consider that many rely on Black-Scholes to estimate options costs internally and help decide how many to give out. And yet most struggle mightily to keep any hint of this predicted cost from the public eye. That isn't as easy as it might sound: One options rule that FASB was able to push through is the requirement that any option with a variable exercise price be valued using Black-Scholes and charged against earnings. With the tech-stock collapse of the past year, companies struggling to keep workers from defecting--or at the very least moping--have been sorely tempted to lower the exercise price of existing employee options. Such repricings have always been controversial, as they so clearly go against any notion of aligning the interests of employees and shareholders. But now they're also accompanied by a big hit to earnings.

A few tech companies and dot-coms have capitulated and now have options costs deducted from their earnings. More interesting, though, are the lengths other companies have gone to in order to avoid this terrible fate. Last spring Microsoft granted 70 million new options to employees after its stock price collapsed in the wake of its antitrust defeat in federal court. It would seem to have been a better deal for shareholders (that is, their stake would have been diluted less) if Microsoft had simply lowered the exercise price on 70 million already granted options. But that would have forced the company to start running options expenses through earnings.

Another, more creative method for avoiding an earnings charge for options was pioneered last fall by Sprint, which exploited a loophole in the FASB standard by canceling millions of employee options with the promise to grant new ones in their place six months and one day later. That doesn't technically count as a repricing, so Sprint doesn't have to report an options expense. This has turned out to be a popular technique--other exploiters of the six-months-and-a-day loophole, according to investor watchdog Institutional Shareholder Services, include i2, Inktomi, Real Networks, and Commerce One.

But there is an even more perverse manifestation of the corporate fear of expensing options. Nearly every unbiased observer of the executive-pay scene believes the best way to structure options, especially for CEOs and their ilk, is to link the exercise price to a market or peer-group index. That way options recipients aren't rewarded for a rising overall market that has nothing to do with their own performance--and won't be punished for a market slump. Hardly any companies are willing to touch indexed options, however. Why? Because as variable-priced options, they have to be charged against earnings.

This gets at the heart of what's so wrong about the current state of options accounting. It's not that options are bad in and of themselves. Indeed, at companies like Intel, Microsoft, and Cisco, generous options grants have gone hand in hand with spectacular, sustained growth. For startups with little cash but lots of potential, options may well be the fairest way to pay key employees. And the rise of options in the 1980s and 1990s has clearly played at least some role in the transformation of so many big American corporations from stodgy bureaucracies into more agile, investor-friendly beasts.

But the failure to acknowledge their cost leads to all sorts of strange and counterproductive corporate behavior. Top managers misallocate resources, overpay themselves, and ignore other, possibly better ways of motivating employees--all because giving out options is "free."

One study, by Harvard law professor Christine Jolls, found a very strong statistical link between options use and share buybacks. There is nothing wrong, per se, with companies buying back their own stock--it can be a good way to return excess cash to shareholders, and can signal that management thinks its stock is cheap. But a bigger motivation for buybacks lately has been to combat the dilution caused by the zealous granting of stock options. And this motivation often kicks in at exactly the wrong time--not when the shares are at a bargain-basement price but when they're high and rising. (Because that's when top executives and other options holders are apt to cash in.) That is exactly what happened to the likes of Sun Microsystems, which spent $631 million (a third of net income) buying back 12 million shares when its stock was near its peak last year. At the end of May, Sun could have bought those same shares for $433 million less.

So what is to be done? FASB, burned once, won't touch the subject again for years, maybe decades. Institutional investors are to a limited extent taking matters into their own hands. Increasingly they're voting against options plans that allow repricing or result in a higher-than-average "overhang" (options outstanding plus shares available for future options grants, divided by total shares outstanding). But so far few plans have actually been rejected by shareholders. In the end, the best check on bad corporate behavior comes down to investors' making day-to-day decisions on what to buy and sell. That's how financial markets are supposed to work, right?

To help make those investing decisions, one of the best measures to look at is this: How much of the company is management giving away? The overhang measure mentioned previously figures in both options and shares available for future options grants. For the sake of simplicity, we've taken the number of options outstanding and divided it by shares outstanding for the tables accompanying this article. As becomes clear in the tables, companies in the same industry can be wildly divergent in their options generosity. And guess what: This generosity already appears to affect share-price performance. According to research by compensation consulting firm Watson Wyatt, companies with the highest overhang in their sectors have the worst stock performance. But companies with lower-than-average overhang don't do so well either. It's those in the middle--actually the high end of the middle--that do best.

You can find a lot of other things about options in company reports that may have no statistical significance but sure are interesting. For one thing, does a company make it easy to find information about its options practices? Well, Microsoft is probably the champion here, including in its 10-K a Black-Scholes-adjusted version of its entire income statement. Tech companies in general are surprisingly good at telling the world, albeit it in a footnote, about their options costs. Among the Fortune 25 companies we looked at, Texaco and AIG stand out for crystal-clear options footnotes, while Exxon Mobil's, SBC's, and Duke Energy's are on the muddy side.

Then there's the fascinating question of how many options end up in the hands of the five or six executives whose compensation is detailed in the company proxy statement. It's always a ton; what's interesting is when you divide that number by the total number of options outstanding. Then you find out that at some companies, the big guys have all the fun. Jarringly, some of those big guys work in Silicon Valley, a land supposedly imbued with new-economy egalitarianism.

At Oracle and Siebel Systems, more than 30% of the options are in the hands of top executives, and most of them go to CEO/founders Larry Ellison and Tom Siebel. Both men are brilliant, highly motivated executives who already own, outright, multibillion-dollar stakes in their companies. There is no conceivable reason they should be getting options, and one has to imagine that even their obviously cowed boards of directors wouldn't be giving them any if the things weren't "free." If that's not reason enough right there to start paying attention to the cost of options, what is?

FEEDBACK: jfox@fortunemail.com