Nội dung

The Impact of Employee Stock Options on the Evolution of Compensation in the 1990s Hamid Mehran and Joseph Tracy First Draft: May 2000 Current Draft: March 2001 Federal Reserve Bank of New York. The views expressed in this article are those of the individual authors and do not necessarily reflect the position of the Federal Reserve Bank of New York or the Federal Reserve System. We benefitted from helpful discussions with Alan Marcus, Elizabeth Keating and comments from seminar participants at the Department of Labor and the Georgetown University Law Center Conference on “Contracts with Highly Skilled Workers.” John Campbell, Martin Lettau and John Graham kindly provided us with a portion of their data. We thank Nathanial Baum-Snow and Dan Burdick for their excellent research assistance. We also thank Hewitt Associates for providing summaries of their data and technical assistance, and Thermo Electron for providing us with the information on details of its employee stock options program and reporting. The decade of the 1990s was remarkable in many ways. This decade produced the longest running U.S. economic expansion. A byproduct of this continued economic growth was a sharp tightening of the U.S. labor market. The growing scarcity of available workers raised the concern that accelerating wage demands would develop which might lead to renewed inflation. Figure 1 shows the growth rate of nominal compensation per hour (CPH) and its relationship to the prime age male unemployment rate during the 1990s.1 Two “wage puzzles” exist during the 1990s. The first relates to the period from 1992 to 1995 when compensation growth declined at the same time that the unemployment rate was rapidly falling. One explanation is that “worker insecurity” early in the expansion accounted for the tepid pay demands during this period.2 From 1995 to 1998, compensation growth accelerated as the unemployment rate broke through the four percent barrier. However, the second wage puzzle emerges in 1999 when compensation growth fell back below the five percent level despite the continued tightening in the labor market during the year. What explains these two wage puzzles? In this paper, we explore whether changes in the structure of pay can help explain the behavior of CPH during the 1990s. Labor markets have changed considerably over the last twenty years. Workers today receive a higher portion of their total compensation in nontraditional forms such as profit-sharing and stock options.3 CPH captures profit-sharing and stock options. However, stock options are reflected in total compensation on the date they are exercised, not on the date they are granted. As we discuss below, there can be several years between the grant and exercise dates for stock options. This raises the possibility that the growing use of stock options may be affecting the timing of when tight labor markets are reflected in CPH growth. We bring the existing data to bear on the question of how the use of stock options is affecting the growth in CPH. Given data limitations, we will focus primarily on the second pay 1 CPH is the most comprehensive of all U.S. pay measures. It captures wage and salary income, tips and overtime, paid leave and severance pay, payments in kind, benefits, bonus and profit-sharing payments, and realizations of stock options. 2 3 See Farber (1997) and Manski and Straub (2000). See Bell & Neumark (1993), Bell & Kruse (1995), Cohn (1999), Duca (1998), Epstein (1999) and Lebow et al (1999). puzzle. While we can provide an educated assessment of the likely impact of stock options on the dynamics of CPH, there is a clear need for more data. We find evidence that stock options may have had an appreciable impact on CPH in the late 1990s. When we recalculate compensation to reflect current stock option grants instead of current realizations, we find that there was likely no downturn in the growth in CPH in 1999. The paper is organized as follows. The next section lays out the essential institutional details regarding employee stock options necessary for our empirical work. Empirical models of the stock option grant and realization decisions are presented and discussed in the third section. In the fourth section, we use these estimates to assess the impact of employee stock options on compensation per hour. The final section discusses some general implications of stock options for the labor market. Accounting for Stock Options Employee stock options (ESOs) are the right to purchase a given number of shares of company stock at the “strike” price between the vesting date and the expiration date of the options.4 The vesting period for an option is the interval of time between when a company grants the option to the employee and when the employee can first exercise the option. If the current market price for a vested option exceeds the strike price, then the option is “in-the-money.” If inthe-money options are exercised (that is, the employee decides to purchase the underlying shares), then the gain to the employee is the difference between the current market price and the strike price times the number of shares exercised.5 If the current market price for a vested option is below the strike price, then the option is “out-of-the-money.” While out-of-the-money options would have no current value if exercised, they still have positive “option value,” which reflects 4 See Murphy (1999) for a discussion of the structure of ESO plans. 5 A common practice is for a cashless transaction to occur using the services of a third-party. The third-party makes a short-term loan to the employee to cover the cost of purchasing the exercised options at the strike price. The shares are then immediately sold back out into the market and the loan is paid off, with a fee going to the third-party. -2- the possibility that the future market price of the stock may rise above the strike price prior to their expiration date. ESOs can be structured as either “incentive stock options” (ISO) or “nonqualified stock options” (NQSO). ISOs must satisfy certain restrictions defined by the Internal Revenue Service which do not apply to NQSOs.6 The primary advantages of ISOs to employees is that the income derived from the exercise of the options is taxed as a capital gain rather than as ordinary income, and the tax is levied when the underlying shares are sold rather than when the shares are exercised. Offsetting this tax gain to employees from ISOs is the loss of a tax deduction to the firm. In contrast to ISOs, the income gain from NQSOs is treated for tax purposes as ordinary income to the employee as of the exercise date, and the company can deduct this cost as a labor expense. Employers are required to file quarterly reports (ES202) that list all taxable sources of income paid to their employees including realized NQSOs. The ES202 reports are used as an input into total compensation. However, these reports do not break out the gain from NQSOs from other sources of compensation. NQSOs became the dominant type of ESO following the reduction in marginal income tax rates in 1986. In January 1993, the Security and Exchange Commission (SEC) started requiring public firms to disclose in their proxy statements both the level of stock option grants and the option exercise activity to their top five executives. The SEC also required that companies report their executive compensation for the two previous years in the annual filing with the agency. Starting in 1991, then, it is possible to collect detailed information on public company stock option programs for top management. Firms could value these option grants using any pricing methodology. The dominant method used in practice is the Black-Scholes pricing formula. ESOs differ in many important ways from traded stock options (TSOs). The most important difference is that they are nontradeable. An employee can exercise a vested in-themoney option, but can not sell the option to an investor. An implication is that both the employee’s valuation of the option and the timing of the exercise decision will be affected by the employee’s risk tolerance. An employee with a significant amount of wealth tied up in company 6 Internal Revenue Code Section 422. -3- stock options has a strong interest in diversifying the risk from movements in the value of the company stock. With TSOs, the employee could simply sell some of the options in the market to another investor, an action which transfers but does not diminish the underlying value of the options. With ESOs, the employee would have to exercise the options in order to diversify their risk.7 This creates an incentive for early exercise of vested ESOs, which reduces their overall value by forgoing their remaining option value. Huddart and Lang (1996) show that workers tend to exercise ESOs soon after their vesting dates, and that this early exercise sacrifices roughly half of the value implied by the Black-Scholes pricing methodology (which is designed to price a TSO). Two other distinctions arise when comparing ESOs and TSOs. As mentioned earlier, ESOs are subject to vesting requirements and tend to have a significant time period until expiration. A variety of vesting schedules are used in practice, with the majority of plans incorporating vesting over two to five years.8 In addition, an employee must exercise any vested in-the-money ESOs prior to leaving a firm. Any non-vested or out-of-the-money ESOs must be forfeited upon termination of the employment relationship. This creates an additional reason for early exercise of ESOs.9 7 For ISOs there is a minimum holding period on the underlying stock which compounds the diversification problem. No similar restriction applies to NQSOs. 8 A Hewitt Associates study of 74 plans in 1998 found that 35% of the plans used Cliff vesting (where all shares vest at the same specified time) with one and three years being the most frequent vesting times, 45% of the plans used uniform vesting (where share vest at a uniform rate over the vesting period) with three and four years being the most frequent vesting times and the remaining 20% either used mixed vesting or provided no information (TCM Data 1998). The most common expiration date is ten years after the grant date. 9 This feature of ESOs makes them a useful tool for reducing employee turnover. Mehran and Yermack (1999) document that the probability of a voluntary departure by a CEO is inversely related to the length of the stock option vesting schedule. They also document that the higher the ratio of deferred compensation to current pay, the less likely a CEO is to leave voluntarily. -4- Measuring the Importance of Employee Stock Options Standard and Poor’s (S&P) ExecuComp is our primary data source. ExecuComp includes annual data from proxy statements for the five highest paid executives in three cohorts of firms: the S&P 500, S&P MidCap 400 and S&P Small Cap 600.10 S&P makes some adjustments to the firms in the data base each year. Our pooled sample which covers the period from 1992-1999 comprises a total of around 2,000 companies. We make extensive use three specific items from the ExecuComp data: the total number of new grants to all employees, the number of grants and their value going to the top five executives and the value of exercised options by the top five executives. We calculate the total value of all new grants in a year by scaling up the value of the grants to the top five executives by the ratio of the total number of options granted to the number of options granted to the top five executives.11 The ExecuComp data is useful for examining general trends in the use of employee stock options during the 1990s. For example, over the decade stock options became the dominant component of an executive’s compensation package. We illustrate this remarkable change using two measures of the relative importance of executive stock options. The first is the ratio of the grant value of new options in a year divided by the executive’s base salary and cash bonus. The second is the ratio of the income gain from stock option realizations in a year divided by the executive’s base salary and cash bonus. Figure 2 shows the averages for these two ratios from 1992 to 1999. Early in the 1990s, both ratios indicate that stock options were typically smaller than an executive’s base salary and cash bonus. By 1996, both ratios equaled or exceeded one. Two years later, continued rapid growth in the expansion of executive stock option programs had pushed both ratios above two, with new grants averaging around 250% of an executives base 10 The median real market value is $8.3 billion for the S&P 500 firms, $1.7 billion for the S&P MidCap firms and $0.4 billion for the S&P Small Cap firms. 11 Starting in 1994, ExecuComp recalculates the grant value of a company’s new options using a consistent set of assumptions on the interest rate, implied stock return volatility, and expected duration of the option. Company Handbooks on ESO plans typically do not make any distinction between executive and non-executive stock option plans. Therefore, we to assume that the Black-Scholes value of an option granted to an executive and to a non-executive is the same. -5- salary and bonus. In 1999, the grants ratio leveled off while there was a sharp reduction in the realization ratio.12 An important related question is whether the use of stock options is also filtering its way down through the ranks of company pay structures. The ExecuComp data allow us to track the distribution of the percent of total new stock option grants that are awarded to the top five management. While this is a very restrictive view of the diffusion of stock options down through the corporate ranks, it has the advantage of providing some sense of recent trends. Figure 3 shows the equally weighted 25th, 50th and 75th percentiles of these top five percentages from 1992 to 1999. Despite the dramatic rise in the use of stock options for executives, there has actually been a slight decline in the fraction of new stock option grants directed toward upper management. This indicates that there has also been a commensurate increase over the 1990s in the use of stock options for employees below the top management level. Given the rapid rise in the use of stock options, it is interesting to speculate on the impact that stock options are having on aggregate compensation growth in the private sector. As noted earlier, aggregate compensation reflects NQSOs when they are realized rather than when they are granted. Unfortunately, there is no currently collected data that permit the direct measurement of the total size of stock option grants or realizations in the labor market. The alternative is to estimate total stock option grants and NQSO realizations by year. The growth rate of CPH net of the income from stock option realizations can then be constructed and contrasted with its actual growth rate. In addition, the cash value of new stock option grants can be added into this net of realizations CPH measure to arrive at a more accurate measure of current labor market pay conditions. We now turn our attention to implementing this approach. The private non-farm sector consists of publically traded and private firms. Over the past five years, public firm have accounted for between forty-seven and fifty percent of employment in the private non-farm business sector. The ExecuComp data consists entirely of publically 12 There is no general agreement as to what has caused the popularity of stock options in the 1990's. Murphy (1999) presents a behavioral discussion. Hall and Liebman (1998) examine the role of taxes where under IRC Section 162(m) compensation above one million dollars in not deductible unless it is performance-based. Of the 1,672 firms on ExecuComp in 1998, 1,566 reported paying less than a one million dollar salary for their CEOs. -6- traded firms, and in 1998 covered roughly 46% of the total employment in public firms.13 Detailed characteristics of publically traded firms are available from the COMPUSTAT data, and equity returns for these firms are available from the CRSP data. For private firms, we have no similar data on their characteristics, nor do we have any details of stock option plans in which to draw any inferences. However, a recent Bureau of Labor Statistics study found that the incidence of stock options in privately held firms in 1999 was significantly below that for publically held firms.14 Based on this evidence we will focus our analysis exclusively on public firms. The basic question, then, is how best to use the ExecuComp data to estimate total stock option grants and realizations for publically held firms. The simplest approach would be to assume for each year that all employees in these firms that are outside of the ExecuComp sample are awarded new stock option grants and realize vested stock options at the average rate observed in that year for employees covered in the ExecuComp data. This approach, however, ignores potentially important variations across firms in their use of stock options that relates to the characteristics of the firm. Taking this variation into account may provide a more accurate estimate of the overall impact of stock options on total compensation. Determinants of Stock Option Grants We start with the problem of estimating stock option grants, since the volume of prior stock option grants is likely to be an important predictor of current realizations. For firms in the ExecuComp sample, we can calculate the Black-Scholes value of the total ESO grants made in the year.15 While we are interested in understanding the determinants of the firm’s decision on the total amount of new grants to make in a year, it is useful to look toward the executive compensation literature for guidance on an appropriate empirical specification. 13 This is based on the comparison of COMPUSTAT employment for ExecuComp firms in 1998 to total employment of COMPUSTAT firms in the same year. 14 See BLS, October 11, 2000. 15 ExecuComp reports the Black-Scholes value of new grants to each of the top five executives, and the ratio of their grants to the total new grants made in the year. From this information, we can back out the total Black-Scholes grant value. -7- The literature on executive compensation starts with the premise that optimal compensation policies should address agency problems between the firm’s managers and its equity and debt holders. The two methods for ameliorating these agency problems are monitoring and incentives.16 A general prediction is that stock options will be more extensively used when agency costs are high and monitoring is difficult. In addition, the accounting treatment of stock options discussed earlier suggests that firms may also use stock options for tax or liquidity reasons. We include several variables to control for expected agency costs. Monitoring may become difficult when a firm has significant growth opportunities. Information asymmetries may arise from these opportunities which make evaluating the managers’ investment choices more difficult [e.g., Mehran (1992), Smith & Watts (1992) and Bizjak et al (1993)]. Stronger incentives, then, are needed to compensate for the monitoring difficulties. These additional incentives can be provided by increasing the share of stock options in total compensation. We measure a firm’s growth opportunities using its market-to-book value (MVBV). The prediction is that stock option grants will be positively related to a firm’s MVBV. Monitoring managers is also difficult in an environment where there is a significant amount of noise regarding the firm’s performance [Lambert & Larcker (1987)]. In such an environment, a higher pay-performance sensitivity is warranted. Yermack (1995) proxies this using the ratio of the relative variability of accounting returns as compared to stock returns. We focus just on the variability of the stock returns over the prior year. The prediction is that higher stock return variability will lead to increased use of stock options. However, higher stock return variability also increases the manager’s risk exposure, which should lead to a higher risk premium to compensate the manager for this added risk.17 This risk premium increases the relative price to the firm of using stock options versus cash compensation, which may induce the 16 Our discussion borrows heavily from Yermack (1995). 17 While volatility always raises the option value of TSOs, Lambert, Larcker, and Verrecchia (1991) and Kulatilaka and Marcus (1994) show that increased volatility can lower the value of ESOs, especially for more risk averse employees. -8- firm to substitute away from stock options in its pay structure. The overall effect, then, of stock return variability on the use of stock options is ambiguous. Capital structure may also exert an important influence on a firm’s compensation system. Stock options, by increasing managers’ pay-performance sensitivity, may encourage managers to pursue riskier investment strategies that tend to favor equity holders over debt holders. If this shift in investment strategies is anticipated by bond holders, then increased reliance on stock options will give rise to a debt premium which differentially impacts highly leveraged firms [John & John (1993)]. To reduce this agency cost of debt, highly leveraged firms may choose to scale back their use of stock options. This should lead to an inverse relationship between a firm’s leverage and its reliance on stock options.18 To help control for any firm life-cycle effects on the use of stock options, we control for a firm’s age which we measure as the number of years the firm’s stock has been traded. If young firms tend to be more cash flow constrained, then we would expect them to rely more heavily on stock options. When a firm issues new stock options it typically incurs no current expense, rather the expense is shifted into the future when the stock options are realized. Workers, however, value these new stock options and are willing to accept lower current cash compensation as a consequence. This should lead to a negative relationship between firm age and the granting of stock options. We also directly proxy for cash flow constraints using an indicator variable for whether the firm has a net operating loss in the current year. Our remaining firm-specific variables include measures of recent performance and firm size. We measure firm performance using the firm’s return on assets (ROA). We use the size of the firm’s assets and employment to control for possible scale effects. Finally, we include twodigit industry effects and year effects to control for any remaining differences across industries and time in the pattern of stock option grants. Our estimation results for stock option grants are presented in Table 1. Summary statistics are provided in Appendix B. For most of our control variables of interest, we divide the range of 18 More specifically, the prediction pertains to the relative portion of an executive’s compensation that is stock based. Our dependent variable is the total amount of stock options granted, rather than the ratio of total stock option grants to total compensation. -9-