For the Last Time: Stock Options Are an Expense

Harvard Business Review, March, 2003

 

 
Zvi Bodie; Boston University's School of Management
Robert S. Kaplan;Harvard University Graduate School of Business Administration
Robert C. Merton;Harvard University Graduate School of Business Administration
  

 
The time has come to end the debate on accounting for stock options; the controversy has been going on far too long. In fact, the rule governing the reporting of executive stock options dates back to 1972, when the Accounting Principles Board, the predecessor to the Financial Accounting Standards Board (FASB), issued APB 25. The rule specified that the cost of options at the grant date should be measured by their intrinsic value--the difference between the current fair market value of the stock and the exercise price of the option. Under this method, no cost was assigned to options when their exercise price was set at the current market price.

The rationale for the rule was fairly simple: Because no cash changes hands when the grant is made, issuing a stock option is not an economically significant transaction. That's what many thought at the time. What's more, little theory or practice was available in 1972 to guide companies in determining the value of such untraded financial instruments.

APB 25 was obsolete within a year. The publication in 1973 of the Black-Scholes formula triggered a huge boom in markets for publicly traded options, a movement reinforced by the opening, also in 1973, of the Chicago Board Options Exchange. It was surely no coincidence that the growth of the traded options markets was mirrored by an increasing use of share option grants in executive and employee compensation. The National Center for Employee Ownership estimates that nearly 10 million employees received stock options in 2000; fewer than 1 million did in 1990. It soon became clear in both theory and practice that options of any kind were worth far more than the intrinsic value defined by APB 25.

FASB initiated a review of stock option accounting in 1984 and, after more than a decade of heated controversy, finally issued SFAS 123 in October 1995. It recommended--but did not require--companies to report the cost of options granted and to determine their fair market value using option-pricing models. The new standard was a compromise, reflecting intense lobbying by businesspeople and politicians against mandatory reporting. They argued that executive stock options were one of the defining components in America's extraordinary economic renaissance, so any attempt to change the accounting rules for them was an attack on America's hugely successful model for creating new businesses. Inevitably, most companies chose to ignore the recommendation that they opposed so vehemently and continued to record only the intrinsic value at grant date, typically zero, of their stock option grants.

Subsequently, the extraordinary boom in share prices made critics of option expensing look like spoilsports. But since the crash, the debate has returned with a vengeance. The spate of corporate accounting scandals in particular has revealed just how unreal a picture of their economic performance many companies have been painting in their financial statements. Increasingly, investors and regulators have come to recognize that option-based compensation is a major distorting factor. Had AOL Time Warner in 2001, for example, reported employee stock option expenses as recommended by SFAS 123, it would have shown an operating loss of about $ 1.7 billion rather than the $ 700 million in operating income it actually reported.

We believe that the case for expensing options is overwhelming, and in the following pages we examine and dismiss the principal claims put forward by those who continue to oppose it. We demonstrate that, contrary to these experts' arguments, stock option grants have real cash-flow implications that need to be reported, that the way to quantify those implications is available, that footnote disclosure is not an acceptable substitute for reporting the transaction in the income statement and balance sheet, and that full recognition of option costs need not emasculate the incentives of entrepreneurial ventures. We then discuss just how firms might go about reporting the cost of options on their income statements and balance sheets.
 
Fallacy 1: Stock Options Do Not Represent a Real Cost

It is a basic principle of accounting that financial statements should record economically significant transactions. No one doubts that traded options meet that criterion; billions of dollars' worth are bought and sold every day, either in the over-the-counter market or on exchanges. For many people, though, company stock option grants are a different story. These transactions are not economically significant, the argument goes, because no cash changes hands. As former American Express CEO Harvey Golub put it in an August 8, 2002, Wall Street Journal article, stock option grants "are never a cost to the company and, therefore, should never be recorded as a cost on the income statement."

That position defies economic logic, not to mention common sense, in several respects. For a start, transfers of value do not have to involve transfers of cash. While a transaction involving a cash receipt or payment is sufficient to generate a recordable transaction, it is not necessary. Events such as exchanging stock for assets, signing a lease, providing future pension or vacation benefits for current-period employment, or acquiring materials on credit all trigger accounting transactions because they involve transfers of value, even though no cash changes hands at the time the transaction occurs.

Even if no cash changes hands, issuing stock options to employees incurs a sacrifice of cash, an opportunity cost, which needs to be accounted for. If a company were to grant stock, rather than options, to employees, everyone would agree that the company's cost for this transaction would be the cash it otherwise would have received if it had sold the shares at the current market price to investors. It is exactly the same with stock options. When a company grants options to employees, it forgoes the opportunity to receive cash from underwriters who could take these same options and sell them in a competitive options market to investors. Warren Buffett made this point graphically in an April 9, 2002, Washington Post column when he stated: "Berkshire Hathaway will be happy to receive options in lieu of cash for many of the goods and services that we sell corporate America." Granting options to employees rather than selling them to suppliers or investors via underwriters involves an actual loss of cash to the firm.

It can, of course, be more reasonably argued that the cash forgone by issuing options to employees, rather than selling them to investors, is offset by the cash the company conserves by paying its employees less cash. As two widely respected economists, Burton G. Malkiel and William J. Baumol, noted in an April 4, 2002, Wall Street Journal article: "A new, entrepreneurial firm may not be able to provide the cash compensation needed to attract outstanding workers. Instead, it can offer stock options." But Malkiel and Baumol, unfortunately, do not follow their observation to its logical conclusion. For if the cost of stock options is not universally incorporated into the measurement of net income, companies that grant options will underreport compensation costs, and it won't be possible to compare their profitability, productivity, and return-on-capital measures with those of economically equivalent companies that have merely structured their compensation system in a different way. The following hypothetical illustration shows how that can happen.

Imagine two companies, KapCorp and MerBod, competing in exactly the same line of business. The two differ only in the structure of their employee compensation packages. KapCorp pays its workers $ 400,000 in total compensation in the form of cash during the year. At the beginning of the year, it also issues, through an underwriting, $ 100,000 worth of options in the capital market, which cannot be exercised for one year, and it requires its employees to use 25% of their compensation to buy the newly issued options. The net cash outflow to KapCorp is $ 300,000 ($ 400,000 in compensation expense less $ 100,000 from the sale of the options).

MerBod's approach is only slightly different. It pays its workers $ 300,000 in cash and issues them directly $ 100,000 worth of options at the start of the year (with the same one-year exercise restriction). Economically, the two positions are identical. Each company has paid a total of $ 400,000 in compensation, each has issued $ 100,000 worth of options, and for each the net cash outflow totals $ 300,000 after the cash received from issuing the options is subtracted from the cash spent on compensation. Employees at both companies are holding the same $ 100,000 of options during the year, producing the same motivation, incentive, and retention effects.

In preparing its year-end statements, KapCorp will book compensation expense of $ 400,000 and will show $ 100,000 in options on its balance sheet in a shareholder equity account. If the cost of stock options issued to employees is not recognized as an expense, however, MerBod will book a compensation expense of only $ 300,000 and not show any options issued on its balance sheet. Assuming otherwise identical revenues and costs, it will look as though MerBod's earnings were $ 100,000 higher than KapCorp's. MerBod will also seem to have a lower equity base than KapCorp, even though the increase in the number of shares outstanding will eventually be the same for both companies if all the options are exercised. As a result of the lower compensation expense and lower equity position, MerBod's performance by most analytic measures will appear to be far superior to KapCorp's. This distortion is, of course, repeated every year that the two firms choose the different forms of compensation. How legitimate is an accounting standard that allows two economically identical transactions to produce radically different numbers?
 
Fallacy 2: The Cost of Employee Stock Options Cannot Be Estimated

Some opponents of option expensing defend their position on practical, not conceptual, grounds. Option-pricing models may work, they say, as a guide for valuing publicly traded options. But they can't capture the value of employee stock options, which are private contracts between the company and the employee for illiquid instruments that cannot be freely sold, swapped, pledged as collateral, or hedged.

It is indeed true that, in general, an instrument's lack of liquidity will reduce its value to the holder. But the holder's liquidity loss makes no difference to what it costs the issuer to create the instrument unless the issuer somehow benefits from the lack of liquidity. And for stock options, the absence of a liquid market has little effect on their value to the holder. The great beauty of option-pricing models is that they are based on the characteristics of the underlying stock. That's precisely why they have contributed to the extraordinary growth of options markets over the last 30 years. The Black-Scholes price of an option equals the value of a portfolio of stock and cash that is managed dynamically to replicate the payoffs to that option. With a completely liquid stock, an otherwise unconstrained investor could entirely hedge an option's risk and extract its value by selling short the replicating portfolio of stock and cash. In that case, the liquidity discount on the option's value would be minimal. And that applies even if there were no market for trading the option directly. Therefore, the liquidity--or lack thereof--of markets in stock options does not, by itself, lead to a discount in the option's value to the holder.

Investment banks, commercial banks, and insurance companies have now gone far beyond the basic, 30-year-old Black-Scholes model to develop approaches to pricing all sorts of options: Standard ones. Exotic ones. Options traded through intermediaries, over the counter, and on exchanges. Options linked to currency fluctuations. Options embedded in complex securities such as convertible debt, preferred stock, or callable debt like mortgages with prepay features or interest rate caps and floors. A whole subindustry has developed to help individuals, companies, and money market managers buy and sell these complex securities. Current financial technology certainly permits firms to incorporate all the features of employee stock options into a pricing model. A few investment banks will even quote prices for executives looking to hedge or sell their stock options prior to vesting, if their company's option plan allows it.

Of course, formula-based or underwriters' estimates about the cost of employee stock options are less precise than cash payouts or share grants. But financial statements should strive to be approximately right in reflecting economic reality rather than precisely wrong. Managers routinely rely on estimates for important cost items, such as the depreciation of plant and equipment and provisions against contingent liabilities, such as future environmental cleanups and settlements from product liability suits and other litigation. When calculating the costs of employees' pensions and other retirement benefits, for instance, managers use actuarial estimates of future interest rates, employee retention rates, employee retirement dates, the longevity of employees and their spouses, and the escalation of future medical costs. Pricing models and extensive experience make it possible to estimate the cost of stock options issued in any given period with a precision comparable to, or greater than, many of these other items that already appear on companies' income statements and balance sheets.

Not all the objections to using Black-Scholes and other option valuation models are based on difficulties in estimating the cost of options granted. For example, John DeLong, in a June 2002 Competitive Enterprise Institute paper entitled "The Stock Options Controversy and the New Economy," argued that "even if a value were calculated according to a model, the calculation would require adjustment to reflect the value to the employee." He is only half right. By paying employees with its own stock or options, the company forces them to hold highly nondiversified financial portfolios, a risk further compounded by the investment of the employees' own human capital in the company as well. Since almost all individuals are risk averse, we can expect employees to place substantially less value on their stock option package than other, better-diversified, investors would.

Estimates of the magnitude of this employee risk discount--or "deadweight cost," as it is sometimes called--range from 20% to 50%, depending on the volatility of the underlying stock and the degree of diversification of the employee's portfolio. The existence of this deadweight cost is sometimes used to justify the apparently huge scale of option-based remuneration handed out to top executives. A company seeking, for instance, to reward its CEO with $ 1 million in options that are worth $ 1,000 each in the market may (perhaps perversely) reason that it should issue 2,000 rather than 1,000 options because, from the CEO's perspective, the options are worth only $ 500 each. (We would point out that this reasoning validates our earlier point that options are a substitute for cash.)

But while it might arguably be reasonable to take deadweight cost into account when deciding how much equity-based compensation (such as options) to include in an executive's pay packet, it is certainly not reasonable to let deadweight cost influence the way companies record the costs of the packets. Financial statements reflect the economic perspective of the company, not the entities (including employees) with which it transacts. When a company sells a product to a customer, for example, it does not have to verify what the product is worth to that individual. It counts the expected cash payment in the transaction as its revenue. Similarly, when the company purchases a product or service from a supplier, it does not examine whether the price paid was greater or less than the supplier's cost or what the supplier could have received had it sold the product or service elsewhere. The company records the purchase price as the cash or cash equivalent it sacrificed to acquire the good or service.

Suppose a clothing manufacturer were to build a fitness center for its employees. The company would not do so to compete with fitness clubs. It would build the center to generate higher revenues from increased productivity and creativity of healthier, happier employees and to reduce costs arising from employee turnover and illness. The cost to the company is clearly the cost of building and maintaining the facility, not the value that the individual employees might place on it. The cost of the fitness center is recorded as a periodic expense, loosely matched to the expected revenue increase and reductions in employee-related costs.

The only reasonable justification we have seen for costing executive options below their market value stems from the observation that many options are forfeited when employees leave, or are exercised too early because of employees' risk aversion. In these cases, existing shareholders' equity is diluted less than it would otherwise be, or not at all, consequently reducing the company's compensation cost. While we agree with the basic logic of this argument, the impact of forfeiture and early exercise on theoretical values may be grossly exaggerated. (See the sidebar "The Real Impact of Forfeiture and Early Exercise.")
 
Fallacy 3: Stock Option Costs Are Already Adequately Disclosed

Another argument in defense of the existing approach is that companies already disclose information about the cost of option grants in the footnotes to the financial statements. Investors and analysts who wish to adjust income statements for the cost of options, therefore, have the necessary data readily available. We find that argument hard to swallow. As we have pointed out, it is a fundamental principle of accounting that the income statement and balance sheet should portray a company's underlying economics. Relegating an item of such major economic significance as employee option grants to the footnotes would systematically distort those reports.

But even if we were to accept the principle that footnote disclosure is sufficient, in reality we would find it a poor substitute for recognizing the expense directly on the primary statements. For a start, investment analysts, lawyers, and regulators now use electronic databases to calculate profitability ratios based on the numbers in companies' audited income statements and balance sheets. An analyst following an individual company, or even a small group of companies, could make adjustments for information disclosed in footnotes. But that would be difficult and costly to do for a large group of companies that had put different sorts of data in various nonstandard formats into footnotes. Clearly, it is much easier to compare companies on a level playing field, where all compensation expenses have been incorporated into the income numbers.

What's more, numbers divulged in footnotes can be less reliable than those disclosed in the primary financial statements. For one thing, executives and auditors typically review supplementary footnotes last and devote less time to them than they do to the numbers in the primary statements. As just one example, the footnote in eBay's FY 2000 annual report reveals a "weighted average grant-date fair value of options granted during 1999 of $ 105.03" for a year in which the weighted average exercise price of shares granted was $ 64.59. Just how the value of options granted can be 63% more than the value of the underlying stock is not obvious. In FY 2000, the same effect was reported: a fair value of options granted of $ 103.79 with an average exercise price of $ 62.69. Apparently, this error was finally detected, since the FY 2001 report retroactively adjusted the 1999 and 2000 average grant-date fair values to $ 40.45 and $ 41.40, respectively. We believe executives and auditors will exert greater diligence and care in obtaining reliable estimates of the cost of stock options if these figures are included in companies' income statements than they currently do for footnote disclosure.

Our colleague William Sahlman in his December 2002 HBR article, "Expensing Options Solves Nothing," has expressed concern that the wealth of useful information contained in the footnotes about the stock options granted would be lost if options were expensed. But surely recognizing the cost of options in the income statement does not preclude continuing to provide a footnote that explains the underlying distribution of grants and the methodology and parameter inputs used to calculate the cost of the stock options.

Some critics of stock option expensing argue, as venture capitalist John Doerr and FedEx CEO Frederick Smith did in an April 5, 2002, New York Times column, that "if expensing were... required, the impact of options would be counted twice in the earnings per share: first as a potential dilution of the earnings, by increasing the shares outstanding, and second as a charge against reported earnings. The result would be inaccurate and misleading earnings per share."

We have several difficulties with this argument. First, option costs only enter into a (GAAP-based) diluted earnings-per-share calculation when the current market price exceeds the option exercise price. Thus, fully diluted EPS numbers still ignore all the costs of options that are nearly in the money or could become in the money if the stock price increased significantly in the near term.

Second, relegating the determination of the economic impact of stock option grants solely to an EPS calculation greatly distorts the measurement of reported income. Such fundamental profitability and productivity measures as return on investment, return on capital employed, and economic value added, which are based on accounting income, would not be adjusted to reflect the economic impact of option costs. These measures are more significant summaries of the change in economic value of a company than the prorated distribution of this income to individual shareholders revealed in the EPS measure. This becomes eminently clear when taken to its logical absurdity: Suppose companies were to compensate all their suppliers--of materials, labor, energy, and purchased services--with stock options rather than with cash and avoid all expense recognition in their income statement. Their income and their profitability measures would all be so grossly inflated as to be useless for analytic purposes; only the EPS number would pick up any economic effect from the option grants.

Our biggest objection to this spurious claim, however, is that even a calculation of fully diluted EPS does not fully reflect the economic impact of stock option grants. The following hypothetical example illustrates the problems, though for purposes of simplicity we will use grants of shares instead of options. The reasoning is exactly the same for both cases.

Let's say that each of our two hypothetical companies, KapCorp and MerBod, has 8,000 shares outstanding, no debt, and annual revenue this year of $ 100,000. KapCorp decides to pay its employees and suppliers $ 90,000 in cash and has no other expenses. MerBod, however, compensates its employees and suppliers with $ 80,000 in cash and 2,000 shares of stock, at an average market price of $ 5 per share. The cost to each company is the same: $ 90,000. But their net income and EPS numbers are very different. KapCorp's net income before taxes is $ 10,000, or $ 1.25 per share. By contrast, MerBod's reported net income (which ignores the cost of the equity granted to employees and suppliers) is $ 20,000, and its EPS is $ 2.00 (which takes into account the new shares issued).

Of course, the two companies now have different cash balances and numbers of shares outstanding with a claim on them. But KapCorp can eliminate that discrepancy by issuing 2,000 shares of stock in the market during the year at an average selling price of $ 5 per share. Now both companies have closing cash balances of $ 20,000 and 10,000 shares outstanding. Under current accounting rules, however, this transaction only exacerbates the gap between the EPS numbers. KapCorp's reported income remains $ 10,000, since the additional $ 10,000 value gained from the sale of the shares is not reported in net income, but its EPS denominator has increased from 8,000 to 10,000. Consequently, KapCorp now reports an EPS of $ 1.00 to MerBod's $ 2.00, even though their economic positions are identical: 10,000 shares outstanding and increased cash balances of $ 20,000. The people claiming that options expensing creates a double-counting problem are themselves creating a smoke screen to hide the income-distorting effects of stock option grants.

Indeed, if we say that the fully diluted EPS figure is the right way to disclose the impact of share options, then we should immediately change the current accounting rules for situations when companies issue common stock, convertible preferred stock, or convertible bonds to pay for services or assets. At present, when these transactions occur, the cost is measured by the fair market value of the consideration involved. Why should options be treated differently?
 
Fallacy 4: Expensing Stock Options Will Hurt Young Businesses

Opponents of expensing options also claim that doing so will be a hardship for entrepreneurial high-tech firms that do not have the cash to attract and retain the engineers and executives who translate entrepreneurial ideas into profitable, long-term growth.

This argument is flawed on a number of levels. For a start, the people who claim that option expensing will harm entrepreneurial incentives are often the same people who claim that current disclosure is adequate for communicating the economics of stock option grants. The two positions are clearly contradictory. If current disclosure is sufficient, then moving the cost from a footnote to the balance sheet and income statement will have no market effect. But to argue that proper costing of stock options would have a significant adverse impact on companies that make extensive use of them is to admit that the economics of stock options, as currently disclosed in footnotes, are not fully reflected in companies' market prices.

More seriously, however, the claim simply ignores the fact that a lack of cash need not be a barrier to compensating executives. Rather than issuing options directly to employees, companies can always issue them to underwriters and then pay their employees out of the money received for those options. Considering that the market systematically puts a higher value on options than employees do, companies are likely to end up with more cash from the sale of externally issued options (which carry with them no deadweight costs) than they would by granting options to employees in lieu of higher salaries.

Even privately held companies that raise funds through angel and venture capital investors can take this approach. The same procedures used to place a value on a privately held company can be used to estimate the value of its options, enabling external investors to provide cash for options about as readily as they provide cash for stock.

That's not to say, of course, that entrepreneurs should never get option grants. Venture capital investors will always want employees to be compensated with some stock options in lieu of cash to be assured that the employees have some "skin in the game" and so are more likely to be honest when they tout their company's prospects to providers of new capital. But that does not preclude also raising cash by selling options externally to pay a large part of the cash compensation to employees.

We certainly recognize the vitality and wealth that entrepreneurial ventures, particularly those in the high-tech sector, bring to the U.S. economy. A strong case can be made for creating public policies that actively assist these companies in their early stages, or even in their more established stages. The nation should definitely consider a regulation that makes entrepreneurial, job-creating companies healthier and more competitive by changing something as simple as an accounting journal entry.

But we have to question the effectiveness of the current rule, which essentially makes the benefits from a deliberate accounting distortion proportional to companies' use of one particular form of employee compensation. After all, some entrepreneurial, job-creating companies might benefit from picking other forms of incentive compensation that arguably do a better job of aligning executive and shareholder interests than conventional stock options do. Indexed or performance options, for example, ensure that management is not rewarded just for being in the right place at the right time or penalized just for being in the wrong place at the wrong time. A strong case can also be made for the superiority of properly designed restricted stock grants and deferred cash payments. Yet current accounting standards require that these, and virtually all other compensation alternatives, be expensed. Are companies that choose those alternatives any less deserving of an accounting subsidy than Microsoft, which, having granted 300 million options in 2001 alone, is by far the largest issuer of stock options?

A less distorting approach for delivering an accounting subsidy to entrepreneurial ventures would simply be to allow them to defer some percentage of their total employee compensation for some number of years, which could be indefinitely--just as companies granting stock options do now. That way, companies could get the supposed accounting benefits from not having to report a portion of their compensation costs no matter what form that compensation might take.
 
What Will Expensing Involve?

Although the economic arguments in favor of reporting stock option grants on the principal financial statements seem to us to be overwhelming, we do recognize that expensing poses challenges. For a start, the benefits accruing to the company from issuing stock options occur in future periods, in the form of increased cash flows generated by its option motivated and retained employees. The fundamental matching principle of accounting requires that the costs of generating those higher revenues be recognized at the same time the revenues are recorded. This is why companies match the cost of multiperiod assets such as plant and equipment with the revenues these assets produce over their economic lives.

In some cases, the match can be based on estimates of the future cash flows. In expensing capitalized software-development costs, for instance, managers match the costs against a predicted pattern of benefits accrued from selling the software. In the case of options, however, managers would have to estimate an equivalent pattern of benefits arising from their own decisions and activities. That would likely introduce significant measurement error and provide opportunities for managers to bias their estimates. We therefore believe that using a standard straight-line amortization formula will reduce measurement error and management bias despite some loss of accuracy. The obvious period for the amortization is the useful economic life of the granted option, probably best measured by the vesting period. Thus, for an option vesting in four years, 1/48 of the cost of the option would be expensed through the income statement in each month until the option vests. This would treat employee option compensation costs the same way the costs of plant and equipment or inventory are treated when they are acquired through equity instruments, such as in an acquisition.

In addition to being reported on the income statement, the option grant should also appear on the balance sheet. In our opinion, the cost of options issued represents an increase in shareholders' equity at the time of grant and should be reported as paid-in capital. Some experts argue that stock options are more like contingent liability than equity transactions since their ultimate cost to the company cannot be determined until employees either exercise or forfeit their options. This argument, of course, ignores the considerable economic value the company has sacrificed at time of grant. What's more, a contingent liability is usually recognized as an expense when it is possible to estimate its value and the liability is likely to be incurred. At time of grant, both these conditions are met. The value transfer is not just probable; it is certain. The company has granted employees an equity security that could have been issued to investors and suppliers who would have given cash, goods, and services in return. The amount sacrificed can also be estimated, using option-pricing models or independent estimates from investment banks.

There has to be, of course, an offsetting entry on the asset side of the balance sheet. FASB, in its exposure draft on stock option accounting in 1994, proposed that at time of grant an asset called "prepaid compensation expense" be recognized, a recommendation we endorse. FASB, however, subsequently retracted its proposal in the face of criticism that since employees can quit at any time, treating their deferred compensation as an asset would violate the principle that a company must always have legal control over the assets it reports. We feel that FASB capitulated too easily to this argument. The firm does have an asset because of the option grant--presumably a loyal, motivated employee. Even though the firm does not control the asset in a legal sense, it does capture the benefits. FASB's concession on this issue subverted substance to form.

Finally, there is the issue of whether to allow companies to revise the income number they've reported after the grants have been issued. Some commentators argue that any recorded stock option compensation expense should be reversed if employees forfeit the options by leaving the company before vesting or if their options expire unexercised. But if companies were to mark compensation expense downward when employees forfeit their options, should they not also mark it up when the share price rises, thereby increasing the market value of the options? Clearly, this can get complicated, and it comes as no surprise that neither FASB nor IASB recommends any kind of postgrant accounting revisions, since that would open up the question of whether to use mark-to-market accounting for all types of assets and liabilities, not just share options. At this time, we don't have strong feelings about whether the benefits from mark-to-market accounting for stock options exceed the costs. But we would point out that people who object to estimating the cost of options granted at time of issue should be even less enthusiastic about reestimating their options' cost each quarter.



We recognize that options are a powerful incentive, and we believe that all companies should consider them in deciding how to attract and retain talent and align the interests of managers and owners. But we also believe that failing to record a transaction that creates such powerful effects is economically indefensible and encourages companies to favor options over alternative compensation methods. It is not the proper role of accounting standards to distort executive and employee compensation by subsidizing one form of compensation relative to all others. Companies should choose compensation methods according to their economic benefits--not the way they are reported.
 
The Real Impact of Forfeiture and Early Exercise

Unlike cash salary, stock options cannot be transferred from the individual granted them to anyone else. Nontransferability has two effects that combine to make employee options less valuable than conventional options traded in the market.

First, employees forfeit their options if they leave the company before the options have vested. Second, employees tend to reduce their risk by exercising vested stock options much earlier than a well-diversified investor would, thereby reducing the potential for a much higher payoff had they held the options to maturity. Employees with vested options that are in the money will also exercise them when they quit, since most companies require employees to use or lose their options upon departure. In both cases, the economic impact on the company of issuing the options is reduced, since the value and relative size of existing shareholders' stakes are diluted less than they could have been, or not at all.

Recognizing the increasing probability that companies will be required to expense stock options, some opponents are fighting a rearguard action by trying to persuade standard setters to significantly reduce the reported cost of those options, discounting their value from that measured by financial models to reflect the strong likelihood of forfeiture and early exercise. Current proposals put forth by these people to FASB and IASB would allow companies to estimate the percentage of options forfeited during the vesting period and reduce the cost of option grants by this amount. Also, rather than use the expiration date for the option life in an option-pricing model, the proposals seek to allow companies to use an expected life for the option to reflect the likelihood of early exercise. Using an expected life (which companies may estimate at close to the vesting period, say, four years) instead of the contractual period of, say, ten years, would significantly reduce the estimated cost of the option.

Some adjustment should be made for forfeiture and early exercise. But the proposed method significantly overstates the cost reduction since it neglects the circumstances under which options are most likely to be forfeited or exercised early. When these circumstances are taken into account, the reduction in employee option costs is likely to be much smaller.

First, consider forfeiture. Using a flat percentage for forfeitures based on historical or prospective employee turnover is valid only if forfeiture is a random event, like a lottery, independent of the stock price. In reality, however, the likelihood of forfeiture is negatively related to the value of the options forfeited and, hence, to the stock price itself. People are more likely to leave a company and forfeit options when the stock price has declined and the options are worth little. But if the firm has done well and the stock price has increased significantly since grant date, the options will have become much more valuable, and employees will be much less likely to leave. If employee turnover and forfeiture are more likely when the options are least valuable, then little of the options' total cost at grant date is reduced because of the probability of forfeiture.

The argument for early exercise is similar. It also depends on the future stock price. Employees will tend to exercise early if most of their wealth is bound up in the company, they need to diversify, and they have no other way to reduce their risk exposure to the company's stock price. Senior executives, however, with the largest option holdings, are unlikely to exercise early and destroy option value when the stock price has risen substantially. Often they own unrestricted stock, which they can sell as a more efficient means to reduce their risk exposure. Or they have enough at stake to contract with an investment bank to hedge their option positions without exercising prematurely. As with the forfeiture feature, the calculation of an expected option life without regard to the magnitude of the holdings of employees who exercise early, or to their ability to hedge their risk through other means, would significantly underestimate the cost of options granted.

Option-pricing models can be modified to incorporate the influence of stock prices and the magnitude of employees' option and stock holdings on the probabilities of forfeiture and early exercise. (See, for example, Mark Rubinstein's Fall 1995 article in the Journal of Derivatives, "On the Accounting Valuation of Employee Stock Options.") The actual magnitude of these adjustments needs to be based on specific company data, such as stock price appreciation and distribution of option grants among employees. The adjustments, properly assessed, could turn out to be significantly smaller than the proposed calculations (apparently endorsed by FASB and IASB) would produce. Indeed, for some companies, a calculation that ignores forfeiture and early exercise altogether could come closer to the true cost of options than one that entirely ignores the factors that influence employees' forfeiture and early exercise decisions.