Disclosure of the fair value of stock option compensation is available for the first time in the notes to the financial statements of Canadian companies. However, because of differences in the reporting presentation and the assumptions made to obtain the inputs to option-pricing models, investors should not assume that these disclosures will allow them to compare option-granting behaviour across firms.
In the United States, the debate over accounting for stock-based compensation rages on. The Financial Accounting Standards Board (FASB) and U.S. legislators entered into a new round in June 2003, when Representatives David Dreier and Anna G. Eshoo opposed FASB initiatives by proposing a bill that would prevent the expensing of employee stock options for at least three years. In contrast, standard-setters in Canada and at the International Accounting Standards Board released exposure drafts late in 2002 that would require the expensing of stock-based compensation. The release of both drafts was met with little real criticism. And for 2002, the Canadian Accounting Standards Board (AcSB) required Canadian firms to disclose in the notes to their financial statements the fair value of options granted as well as the pro-forma net-income impact of expensing stock-based compensation.
At the end of 2002, we conducted a survey of TSX 60 Canadian firms to determine both the effect of option compensation on reported income and the sensitivity of reported option compensation to the assumptions that companies made in their option-pricing calculations. These disclosures are mandated for Canadian firms with year-ends beginning on or after Jan. 1, 2002, and our study describes the first systematic disclosures on the fair value of option grants by Canadian firms.
For the 47 of the TSX 60 firms disclosing but not expensing employee stock options, we found that pro forma net income including an expense for employee stock options was, on average, eight per cent lower than reported net income. However, when we focused on those firms reporting the cumulative effect of options granted, rather than simply the effect of options granted since Jan. 1, 2002, the average pro-forma net income was 15 per cent less than reported net income. The firms most affected by the recording of option compensation saw their net income decrease by more than 100 per cent. Similarly, a Bear Stearns analysis of U.S. firms suggests that diluted EPS for S&P 500 firms in 2001 would be 20 per cent lower if the employee stock options had been expensed. This level of reduction was driven in part by the fact that 2001 profits were below historic levels, making option compensation a large percentage of a relatively low base; even so, when Bear Stearns normalized the data, it estimated that the expensing of options would still have reduced diluted EPS by 10 per cent.
Additionally, we examined the sensitivity of each company’s disclosed option valuations to variations in their estimates of expected option life and volatility, as well as the appropriate risk-free rate for a sub-sample of the TSX 60 firms. All of these assumptions, even the risk-free rate the firm chooses, are subject to managerial discretion. We found that plausible changes in these estimates could increase the reported fair value of option grants by an amount in excess of 20 per cent. Firms in our sample were particularly sensitive to changes in expected volatility and the risk-free interest rate.
To expense or not to expense
For 2002, companies were only required to disclose option expense information. However, on April 30, 2002, Toronto-Dominion Bank became the first Canadian company to announce that it would recognize employee stock options as an expense, beginning Nov. 1, 2002 (fiscal 2003). CEOA. Charles Baillie noted, “With the introduction of a new Canadian standard on stock-based compensation, we believe the time is right to change the way in which we account for options.” A total of six companies on the TSX 60 expensed options for fiscal 2002: Brascan Corp., CIBC, Domtar, Noranda Inc., Sun Life Financial Services, and TransCanada PipeLines. In the U.S., by March of 2003, more than 80 S&P 500 firms announced that they would expense options.
Forty-seven of the remaining TSX 60 firms reported the effect of option compensation on income through note disclosures. Only seven TSX 60 companies made no disclosures at all-six of them because the timing of their year-end (non-Dec. 31) precluded the requirement to disclose, and one company because the requirements did not apply, since all of its stock-based compensation obligations allowed for settlement in cash.
The effect of option compensation on pro forma income is estimated by first determining the fair value of option grants by applying an option-pricing model and then amortizing that amount over the vesting period of the grant. As a transitional procedure for 2002, firms could choose to report either a single-year or a cumulative effect of option grants. The choice of a single-year effect was allowed to avoid the need to value outstanding grants made in years prior to 2002. A firm reporting a single-year effect would calculate option expense as the fair value of options granted in the current year, divided by the vesting period. A firm reporting a cumulative effect with, for example, a vesting period of four years, would report one-quarter of the fair value of all option grants issued over the current year plus one-quarter of the fair value of options granted over each of the preceding three years. In most cases, this choice was disclosed.
The difference between the single-year versus cumulative expensing can be very significant in some cases. For example, Celestica trades on both the Toronto and the New York stock exchanges, and thus provides a reconciliation between Canadian and U.S. GAAP in its financial statements. Under U.S. GAAP, Celestica reported a cumulative effect of option compensation of over $130 million (CAN.). In contrast, under Canadian GAAP, Celestica was only required to include options granted in 2002. Because many of these were granted late in the year, the reported single-year effect of option compensation in 2002 was only $3.3 million (CAN.). Therefore, single-year disclosure significantly understates the long-run impact of employee stock options on net income.
The six firms that recognized option compensation in income all calculated option compensation resulting from current-year grants only. Not surprisingly, these charges did not result in large changes to net income. Mean and median incomes after deducting option compensation were only two per cent and one per cent lower, respectively, than net income excluding option compensation. Of the 47 firms that disclosed the effects of option compensation on pro forma income in the notes to their financial statements, 26 reported single-year effects, and 21 reported cumulative effects. The mean and median income effects of option compensation for the single-year reporting group were decreases in reported income of three per cent and one per cent respectively. However, the mean and median income effects of option compensation for the group reporting cumulative effects were decreases in reported income of 15 per cent and seven per cent respectively.
Exhibit 1 reports the income effect of option compensation by company category. All amounts are translated into Canadian dollars using an average Canadian/U.S. exchange rate for 2002 of 1.5706. (We calculated this rate by averaging the month-end exchange rates throughout 2002.)
Exhibit 2 highlights the 10 TSX 60 companies whose pro-forma net income was the most affected by the expensing of option compensation. We ranked companies based on the decrease in net income caused by expensing option grants divided by the absolute value of net income in order to incorporate the effect of negative earners in our analysis. Not surprisingly, nine of the 10 most affected companies reported the income effect of all outstanding option grants. Only one company on the top 10 list (Teck Cominco) reported the income effect of grants issued in 2002 only.
Option pay may appear high as a percentage of net income simply because net income is uncharacteristically low for that firm in the current year. For example, Agrium tops the list for option pay expressed as a percentage of income because its 2002 net income is nil. Similarly, Cott Corp., the second company on that list, reported net income in 2002 that was only one-tenth of that reported in the previous year. Financial statement readers may be interested not only in option pay as a percentage of net income but also in the absolute magnitude of option pay. Thus, we also ranked firms by the unscaled amount of option compensation. Three firms made the top 10 list both on the basis of absolute option compensation and option compensation as a percentage of net income: Nortel Networks, Rogers Communications and Barrick Gold. Due to its size, Nortel tops the list of reported total option pay at close to $1,500 million in 2002.
CICA guidance on valuation and recognition
Beyond the single or multi-year effect, the computation of option expense is also affected by a number of assumptions that affect the determination of the fair value of the option and the recognition period. Four key assumptions go into the measurement of the fair value of each option granted using an option pricing model (typically Black-Scholes): i) the expected life of the option, ii) the expected volatility of the underlying stock, iii) the expected dividends on the stock, and iv) the current risk-free interest rate. Exhibit 3 outlines the CICA guidance on the assumptions to be used.
As Exhibit 3 illustrates, the standard focuses on expectations of option life, volatility and dividends, not on historical averages. Thus, firms must make estimates of these amounts, and this has an effect on fair value estimates. Even the risk-free rate is subject to a company’s judgment, as the CICA defines this rate relative to the expected life of the options, which is estimated by the firm. We describe the assumptions made about option life, volatility and the risk-free rate below.
The expected life ranged from 2.5 to 10 years. Each firm’s estimate reflects its experience with option exercise, and the vesting periods of its grants. These numbers are useful for pricing the options so long as the future exercise behaviour corresponds to the firm’s estimate.
The risk-free rate used by companies ranged from 2.8 to 5.9 per cent. Since the CICA Handbook defines the risk-free rate relative to Canadian government bonds with the same maturity, and firms use the rate on the grant date, the fact that there is not a single risk-free rate is not surprising. However, even among grants with the same assumed life, the risk-free rate varies considerably. For example, for grants with an estimated life of four years, the assumed risk-free rate ranges from 2.8 to 5.0 per cent. Interestingly, during 2002 the four-year, zero-coupon rate for Canadian government bonds ranged from 3.9 to 5.5 per cent.
Each firm provides an estimate of the annual volatility of its stock price over the life of the option. This number is firm specific and is not easy to estimate. For example, on June 5, 2003, the Montreal exchange provided estimates of Celestica’s volatility that ranged from 44 to 82 per cent. Furthermore, estimates of Celestica’s volatility obtained through the use of historical data ranged from 65 per cent, using six months of past data, to 75 per cent, using one year of past data. The company actually used an expected volatility of 70 per cent to estimate the value of employee stock options.
Sensitivity of fair values to changes in assumptions
To understand the effect these assumptions could have on reported option compensation, we estimated the sensitivity of the options’ fair value to the assumptions used in the valuation models. For this analysis, we focused on those 20 firms where we could confirm the fair value estimate of the option to within 10 cents of the reported value. (The primary reason we could not compute the fair value of the option for all firms was related to the aggregation of option information in reporting.)
For each firm, we estimated the value of the options grant with the Black-Scholes model using the information about the average strike price, risk-free rate, dividend yield, and volatility provided by the firm. Since options are usually granted at the money, we used the weighted average strike price as our estimate of the stock price on the grant date.
Focusing on these 20 firms, we analyzed the sensitivity of a company’s option expense to its assumptions. This type of sensitivity analysis should be particularly salient to financial statement readers because the CICA recommends that if a firm has a range of possible estimates but cannot identify a “best” estimate, it should use the lower end of the range for expected volatility and life, and the higher end of the range for expected dividends, all of which minimizes the reported value of option grants. We first estimated the weighted average value of the option grants for the 20 firms as being $30.22 million. We then changed the assumed values of the life of the option, the underlying volatility, and the risk-free rate and examined the impact of these changes on the average value of the option grants. (We do not report on the sensitivity of fair values to expected dividend yields because firms tend to change dividend levels only gradually, and thus the assumed dividend yield should be the least subject to a company’s discretion.) For each change, we held the other assumptions constant. These results are reported in Exhibit 4 (see pages 8-10).
We began by varying the assumed life of the grant. As the grant life increases, an option becomes more valuable because this gives the stock price more time in which to increase above the strike price. We increased (decreased) each firm’s assumed maturity of the options by one, two and three years, and left all other assumptions at the level the firm assumed. Exhibit 4 shows an increase in reported fair values of less than five per cent when we increase the assumed option life by three years.
We next examined the effect of the risk-free rate used. As the risk-free rate increases, an option becomes more valuable, as the discount factor effectively reduces the cost of the strike price measured in today’s dollars. We increased (decreased) each firm’s assumed risk-free rate by one, two and three per cent, leaving all other assumptions at their original levels. As Exhibit 4 indicates, the risk-free rate used by TSX 60 firms ranged from a low of 2.8 per cent to a high of 5.9 per cent. The calculated fair value is very sensitive to changes in the risk-free rate. For example, if we increase the risk-free rate by three per cent (the approximate range of rates used by the firms in our sample), the reported fair value increases by an amount in excess of 20 per cent.
Finally, we examined the sensitivity of the value of the option grants to the assumed volatility. As the volatility increases, an option becomes more valuable because the variability in stock price increases the chance that stock value will rise above the strike price. We increased (decreased) each firm’s assumed volatility rate by five, 10 and 15 per cent, leaving all other assumptions at their original levels. Here, as with the expected life, firms must exercise judgment rather than base estimates solely on past performance. The CICA stresses that “an enterprise may need to adjust historical averaged annual volatility to estimate a reasonable expected volatility over the expected life of an option” (CICA Handbook, Section 3870, paragraph A13). The calculated fair value of options is highly sensitive to variations in assumed volatility, with the reported fair value rising in excess of 30 per cent for a 15-per-cent increase in assumed volatility.
As we noted, all of the effects of changing assumptions resulted from varying only one assumption while holding the others constant. To the extent that firms must make judgments on all the assumptions used, the amount of variation in reported option value assigned can only increase. For example, to take the extreme case for all three assumptions used in our example, if we increased the option life by three years, the risk-free rate by three per cent and the volatility by 15 per cent, we would see an increase in the reported fair value of options for the 20 companies analyzed in excess of 90 per cent.
The message for financial statement readers
Those investors who have been calling for more transparency in the financial reporting of option grants have no doubt been eager to peruse recently released corporate financial statements for disclosures on option-based pay. The information on option compensation provided in 2002 financial statements can help investors assess the extent of option grants, though caveats do apply. In order to determine whether the option compensation disclosed in 2002 is likely to be representative of the compensation reported on future income statements, investors must assess whether the calculations are single-year or cumulative. If it is the former, investors must make their own adjustments to increase future forecasted compensation.
Further, investors need to be sensitive to the effect of changing assumptions on reported option value. Because the CICA encourages firms to measure expected option life, volatility and dividend yield, and because measurement of the risk-free rate varies both throughout the year and by expected life of the options, each firm’s estimate of the fair value of option grants is highly subjective. We therefore recommend one additional disclosure that might prove useful to financial statement readers: a sensitivity analysis of the effect of variations in the assumptions used on the reported fair value of option grants.