2.1.1 “Grant-Date” vs. “Realized” Pay

While the ultimate value of stock awards and stock options is not known until the stock vests and the options are exercised, these equity awards clearly have a value upon grant. Perhaps the most critical choice facing researchers in executive compensation is whether to measure the compensation associated with equity awards as the amount actually realized upon vesting and exercise, or to assign an “ex ante” grant-date value. Most academic research on executive compensation since the mid-1980s has adopted the ex ante approach, valuing stock awards as the fair market value on the date of grant (i.e. the grant-date stock price multiplied by the number of shares granted), and valuing stock options on the grant date using some variant of the Black and Scholes (1973) formula.

When total compensation is measured using grant-date values, it is routinely referred to as expected compensation to distinguish it from realized compensation as measured at the time the stock vests and the options are exercised.³ However, calling the grant-date pay “expected” is somewhat loose:

Similarly, bonus plans have a “grant-date value” typically measured as the target bonus, paid when the company achieves (usually accounting-based) target performance. However, even when target performance equals expected performance, the target bonus is only the “expected bonus” when the rewards and penalties for surpassing or missing targets are symmetric.

To illustrate the distinction between grant-date and realized pay, suppose that a CEO’s compensation in 2010 and 2011 consisted of a salary of $500,000 paid each year, and 50,000 shares of restricted stock awarded at the beginning of 2010 that become non-forfeitable (“vest”) at the end of 2011. Suppose further that the company’s stock price rose from $10 to $30 over the course of these two years. This CEO’s grate-date pay (which includes the grant-date value of the restricted stock) was $1000,000 in 2010 (consisting of the 2010 $500,000 base salary and the unvested stock with a grant-date value of $500,000) and the 2011 salary of $500,000. But, his realized pay (consisting of his base salary plus the amount realized upon vesting) was $500,000 in 2010 and $2000,000 in 2011 ($500,000 in base salary plus $1500,000 from the vesting of his stock at the end of 2011).

Grant-date and realized pay are both legitimate measures of CEO compensation and each is a legitimate answer to a different question. Compensation committees evaluating the competitiveness of their CEO pay package at the beginning of the year (that is, before performance results are tallied) should focus on grant-date pay levels. In contrast, realized pay levels will (by definition) depend on the company’s current and past performance, and are therefore most useful in evaluating whether ultimate rewards have been commensurate with company performance.

The distinction between grant-date and realized pay is also critical for researchers estimating the link between pay and performance. For example, researchers beginning with (I confess, reluctantly) Murphy (1985) have assessed the relation between pay and performance by regressing total grant-date compensation on measures of corporate performance (using CEO fixed-effects or first-differences to control for unobservable factors affecting pay levels). However, consider two otherwise identical executives, the first paid $1 million annually in base salary and the second paid $1 million annually in restricted shares. Researchers regressing grant-date pay levels on performance would conclude that neither executive is paid for performance, when in fact the second CEOs realized pay is strongly related to performance.

The SEC has helped confuse the distinction between grant-date and realized compensation by conflating elements of each in the “Summary Compensation Table” required in corporate proxy statements. In particular, since 2009, the SEC has required companies to report the grant-date fair-market values of stock and option grants in the Summary Compensation Table, while at the same time reporting the realized (rather than target) payouts from non-equity-based bonus plans. In addition, the SEC rules are particularly confusing for companies that pay annual bonuses partly in cash and partly in stock and options, as is common in financial services. As an example, suppose that a CEO receives a bonus of $10 million in January 2012 for performance in 2011, and that $4 million is paid in cash and the remaining $6 million in stock and options. According to SEC rules, the $4 million cash bonus is included as part of 2011 compensation (and reported in the firm’s 2012 proxy statement), while the $6 million bonus paid in the form of stock and options is included as part of 2012 compensation (and not reported until the firm’s 2013 proxy statement).

Adding to the confusion between grant-date and realized pay was the (thankfully temporary) existence of a third measure mandated by the SEC and included in the Summary Compensation Table in proxy statements issued between 2007 and 2009 (covering compensation paid between 2006 and 2008). Under the SEC’s 2007–2009 reporting requirements, “SEC Total Compensation” included the accounting expense the company records for stock and options during the year under Financial Accounting Standard 123R (FAS 123R) discussed below in Section 3.8.4. Using the accounting expense for valuing options instead of the grant-date value of options was a last-minute change to the reporting requirements made by the SEC in December 2006 without public comment. Under the SEC approach that mandates the use of accounting numbers in the table, the grant-date value of the $500,000 grant vesting in two years is reported as $250,000 in the grant year and $250,000 in the following year—numbers that bear no meaningful economic relationship to anything in the system. Fortunately, the confusion was relatively short-lived: in late 2009 the SEC revised its disclosure rules to include grant-date values rather than annual accounting expenses in the summary pay table.

Another element of the confusion in describing the typical CEO pay package reflects the statistical distinction between averages and medians. Suppose, for example, that there are eleven CEOs in an industry, ten receiving compensation of $1 million and the eleventh receiving $12 million. The average compensation in this industry is $2 million (calculated by summing all compensation amounts and dividing by 11), while the median is only $1 million (calculated as the compensation where half the CEOs are paid more and half the CEOs are paid less). Average and median pay are, again, both legitimate measures of CEO pay, but are answers to different questions. Average pay is relevant in assessing aggregate levels of pay (a reader can multiply the average pay by the number of CEOs and get total compensation paid to all CEOs), while median pay is more relevant in describing compensation for a “typical” CEO.

Figure 1 illustrates the 2011 grant-date and realized compensation for CEOs in firms listed in Standard and Poors S&P 500 (essentially the largest 500 US firms ranked by market value). The data are based on proxy statement information reported in Standard & Poors’ ExecuComp database for the 465 S&P 500 firms.⁴ For both measures, total compensation is comprised of six basic components: (1) base salaries; (2) discretionary bonuses; (3) non-equity incentives (based on both annual and multi-year performance measures); (4) stock options; (5) stock awards; and (6) other pay.⁵ Base salaries and the payouts from discretionary (non-formulaic) bonuses are the same for both grant-date and realized total compensation. However, the definitions of the remaining pay components vary with the measure utilized.

For grant-date pay, non-equity incentives are evaluated at the target level of payout (or, calculated as the average of the minimum and maximum payout if the target is not reported).⁶ The grant-date value of stock options is defined as the company’s estimate of the present value of the options on the grant-date: this value is typically based on Black and Scholes (1973) or similar methodologies and approximates the amount an outside investor would pay for the option. Similarly, the grant-date value of stock awards is calculated as of the grant date using the grant-date market price, which in turn approximates the amount an outside investor would pay for the stock. “Other compensation” includes perquisites, signing bonuses, termination payments, and above-market interest paid on deferred compensation. In addition, “other compensation” includes the change in the actuarial value of pension benefits, which typically constitutes a large percentage of compensation for those executives with supplementary defined-benefit pension plans.⁷

For realized pay, non-equity incentives are defined as actual payouts during the fiscal year, including both amounts paid in formula-based annual bonus plans, and current-year payouts from longer-term plans. Stock options are calculated as the gains realized by exercising options during the year, and stock awards are calculated as the value of the stock (or other equity instruments) as of the vesting date. Other compensation includes perquisites, signing bonuses, termination payments, above-market interest paid on deferred compensation, and the actual payments made to the CEO during the year under pension or retirement plans.

The first two columns in Figure 1 depict average grant-date and realized compensation. The pay-composition percentages are constructed by first calculating the composition percentages for each CEO, and then averaging across CEOs. The average grant-date CEO Pay in S&P 500 firms in 2011 was $11.6 million, compared to average realized pay of $12.3 million. Stock awards are the largest single component of both grant-date and realized pay in 2011. The “Other Pay” component of grant-date pay is large compared to the corresponding component for realized pay, reflecting that the definition of grant-date pay includes the (generally positive) change in the actuarial present value of pension benefits during the year. In contrast, the realized pay for pensions include only pension benefits paid during the year for proxy-named executives (which excludes amounts to be paid after retirement).

The remaining two columns in Figure 1 depict median compensation. The composition percentages for median pay are calculated as the median ratio of each component: median ratios do not sum to 100% (because the sum of the medians is not the median of the sum). Median compensation is typically lower than average pay, since a small number of very-highly paid CEOs will increase the average pay but not the median pay. For example, ConocoPhillips’s CEO James Mulva realized $141 million through exercising stock options in 2011. If the options had not been exercised, his pay would have fallen to “only” $5.3 million, and the average realized compensation for the 465 executives in Figure 2.1 would fall $303,000 from $12.436 million to approximately $12.133 million. Equity awards for the median executive are dominated by stock (rather than option) awards, and together option and stock awards comprise about half of total compensation for the typical executive.

The difference between grant-date and realized values, and averages and medians, is especially pronounced for stock options. Figure 2 shows the average and median grant-date values and exercise gains (i.e. realized values) for stock options granted to or exercised by CEOs in S&P 500 firms from 1992 to 2011. As shown in the figure, the average grant-date values (dotted line) and exercise gains (solid line) were remarkably similar leading up to the 2000 burst in the Internet bubble. In contrast, average exercise gains increased while average grant-date values fell leading up to the 2008 financial crisis, reflecting the shift in grants primarily reflecting the shift from options to restricted stock described in more detail below.

Figure 2 shows that median grant values and exercise gains were always below their respective averages. Interestingly, the median exercise gain was zero except for in the 2004–2007 period, indicating that less than half of the S&P 500 CEOs exercised options during most years in the sample (including 2000, when the average gain across all S&P 500 CEOs exceeded $12 million).⁸

Figure 3 shows how average grant-date pay for CEOs has evolved from 1970 to 2011. The data are adjusted for inflation and are based on information extracted from annual Forbes surveys (1970–1991) and Standard & Poors ExecuComp Database (1992–2011).⁹ Non-equity pay includes base salaries, payouts from short-term and long-term bonus plans, deferred compensation, and benefits. Total compensation includes non-equity compensation plus equity-based compensation, including the grant-date values of stock options and restricted stock.¹⁰ Due to changing reporting requirements and data availability some of the estimates of grant-date compensation are approximations, but the trends depicted in Figure 3 are nonetheless historically representative. As shown in the figure, average grant-date compensation increased from about $1.1 million in 1970 to $10.9 million in 2011, down from a peak of $18.2 million in 2000.¹¹ Finally, the figure shows that most of the growth in CEO pay since 1990 is explained by the growth in equity-based pay. Indeed, stock and options constituted only a trivial percentage of pay in the early 1970s, and grew to be the dominant form of pay by the late 1990s.

Figure 4 shows how both the composition and level of grant-date pay evolved from 1992 to 2010. Because of the skewness in the pay distribution (where a small number of CEOs receive unusually high levels of compensation), the median pay in Figure 4 is significantly lower than the average pay in Figure 3 in each year. The pay-composition percentages in the figure are constructed by first calculating the composition percentages for each CEO, and then averaging across CEOs. As evident from the figure, underlying the growth in pay for CEOs since the 1990s is an escalation in stock-option compensation from 1993 to 2001 coupled with a dramatic shift away from options towards restricted stock from 2002 to 2011. In 1992, base salaries accounted for 41% of the $2.9 million median CEO pay package, while stock options (valued at grant-date) accounted for 25%. By 2001, base salaries accounted for only 18% of the median $9.2 million pay, while options accounted for more than half of pay. By 2011, options fell to only 21% of pay, as many firms switched from granting options to granting restricted stock (which swelled to 36% of pay).

In interpreting the time-series in Figure 4, it is important to recognize the selection bias inherent in the S&P 500. In particular, the firms in the index are selected by a committee based primarily on market capitalization and industry representation. For example, during the 1990s the S&P 500 increased its representation of “new economy” firms, as these firms became more highly valued and a more important component of the economy.¹² Indeed, the fraction of the S&P 500 comprised of new economy firms grew from 5.5% in 1992 to over 12% in 2001 (and remained at about 11% for the rest of the sample period). Since new economy firms have traditionally relied on stock options as a major component of pay (see Murphy, 2003), the increase in both the level of pay and the use of options in Figure 4 in part reflects changes in the composition of the S&P 500.

Figure 5 replicates Figure 4 after restricting the sample to only firms included in the S&P 500 in 1992. This sample restriction attenuates the increase in pay levels, which increased by 165% from 1992 to 2000 (instead of 220% as in Figure 4). The figure also suggests that CEO pay continued to increase until 2007 (a starkly different pattern than suggested by Figure 4). However, while Figure 5 mitigates the S&P 500 selection bias in Figure 4, it is subject to a survivor bias: only half of the S&P 500 firms in 1992 were still publicly traded in 2011.

While the analysis in this chapter will generally focus on S&P 500 companies, Figure 6 shows the evolution of the level and compensation for CEO pay below the S&P 500. The data, extracted from ExecuComp, include firms in the S&P MidCap 400, S&P SmallCap 600, and a small number of other firms tracked by S&P. As evident by comparing Figures 4 and 6, the level of CEO pay below the S&P 500 is considerably smaller than pay levels for S&P 500 CEOs. In addition, while median pay for S&P 500 CEOs has more than tripled from 1970–2010, pay for CEOs below the S&P 500 merely doubled. Similar to their S&P 500 counterparts, restricted stock has replaced stock options as the primary form of equity-based compensation.

2.1.2 The “Cost” vs. the “Value” of Incentive Compensation

In constructing measures of total compensation, it is important to distinguish between two often confused but fundamentally different valuation concepts: the cost to the company of granting the compensation and the value to an executive of receiving the compensation. Consider, for example, a company that decides to give a share of restricted stock to its CEO vesting in five years (that is, the CEO is restricted from selling the share of stock for five years, and receives the accumulated dividends [plus interest] upon vesting). Suppose further that the market price of a share of stock is $10. The economic or opportunity cost of the stock grant to the company is the amount the company could have received if it were to sell an unrestricted share to an outside investor rather than giving the restricted share to the CEO. Ignoring the probability of forfeiture and the slight dilution discount associated with issuing a new share, the company could raise $10 by selling the share to an outside investor. Thus, the company’s cost of granting the share is the price of the share on the open market.

Alternatively (but equivalently), by granting the restricted share to the CEO, the company is effectively promising to deliver one share of stock to the CEO in five years. If the company had no shares available to issue, it could satisfy this contract by purchasing a share on the open market in five years at a price that might be higher or lower than $10. If the company wanted to perfectly hedge the “price risk” of its future obligation, it could purchase a share of stock in the open market today (for $10) and deliver it to the CEO in five years. Thus, again, the company’s cost of granting the share is simply the price of the share on the open market.

But, what about the CEO? The CEO would clearly prefer to have $10 today than a promise to receive one share of stock in five years; after all, he could always take the $10 and buy a share of stock today, but will likely have other more-preferred uses for the $10. Moreover, if the CEO is risk averse and undiversified (in the sense that his overall wealth is positively correlated with company stock prices, through existing stock ownership, option holdings, and the risk of being fired for poor performance), the value the CEO places on the share of restricted stock will be strictly less than the fair market value of the share. Note that the CEO’s value will predictably decrease as the CEO becomes more risk averse or less diversified.

Similarly, suppose that the company decides to give the CEO an option to buy a share of stock at a predetermined exercise price. The opportunity cost of granting the option is the amount an outside investor would pay for it. The outside investor is generally free to trade the option, and can also take actions to hedge away the risk of the option (such as short-selling the underlying stock). Black and Scholes (1973) and Merton (1973) demonstrated that, since investors can hedge, options can be valued as if investors were risk neutral and all assets appreciate at the risk-free rate. This risk-neutrality assumption forms the basis of option pricing theory and is central to all option pricing models, including binomial models, arbitrage pricing models, and Monte Carlo methodologies. Ignoring dilution, forfeiture, and early exercise, these now-standard methodologies provide reasonable estimates of what an outside investor would pay, and therefore measure the company’s cost of granting options.

Measures of opportunity cost that ignore dilution, forfeiture, and early exercise will systematically overstate the company’s cost of granting options. Dilution reduces the cost of granting options because companies typically issue new shares when options (technically, warrants) are exercised. While the impact of dilution on any specific option grant is typically immaterial, the impact can be significant when added across all employees receiving options. Forfeiture reduces the cost because executives typically forfeit some or all of their unexercisable options upon resignation or termination.¹³ Most importantly, Ballowing executives to exercise options before they expire reduces the company’s cost of granting options because risk-averse employees—seeking diversification and liquidity—predictably exercise non-tradable options sooner than would an outside investor holding a tradable option.

However, even after appropriate adjustments for dilution, forfeiture, and early exercise, Black–Scholes values do not measure the value of the non-tradable option to a risk-averse executive. In contrast to outside investors, company executives cannot trade or sell their options, and are also forbidden from hedging the risks by short-selling company stock. In addition, while outside investors tend to be well-diversified (holding small amounts of stock in a large number of companies), company executives are inherently undiversified, with their physical as well as human capital invested disproportionately in their company. For these reasons, company executives will generally place a much lower value on company stock options than would outside investors.

Lambert, Larcker, and Verrecchia (1991) and Hall and Murphy (2002) propose measuring the value of a non-tradable option to an undiversified risk-averse executive as the amount of riskless cash compensation the executive would exchange for the option.¹⁴ Suppose that an executive has non-firm-related wealth of w, holds a portfolio S(·) of company shares and options, and is granted n options to buy n shares of stock at exercise price X in T years. Assuming that w is invested at the risk-free rate, r_f, and that the realized stock price at T is P_T, the executive’s wealth at time T is given by

(1)

If, instead of the option, he were awarded V in cash that he invested at the risk-free rate, his wealth at time T would be:¹⁵

(2)

Assuming that the executive’s utility over wealth is U(W), we can define the executive’s value of n options as the “certainty equivalent” V that equates expected utilities (1) and (2):

(3)

Solving (3) numerically requires assumptions about the form of the utility function, U(W), and the distribution of future stock prices, f(P_T). I follow Hall and Murphy (2002) in assuming that the executive has constant relative risk aversion ρ, so that when ρ = l, and when . I adopt the Capital Asset Pricing Model (CAPM) and assume that the distribution of stock prices in T years is lognormal with volatility σ and expected value equal to , where β is the firm’s systematic risk and r_m is the return on the market portfolio.¹⁶

Calculating certainty equivalents from (3) requires data on stock and option grants and holdings (available from corporate proxy statements¹⁷), and also requires unobservable data on executive “safe wealth” (i.e. wealth not correlated with company stock prices) and executive risk aversion. Following Hall and Murphy (2002), I assume that CEOs have relative risk-aversion parameters of 2 or 3, and that each CEO has “safe wealth” equal to the greater of $5 million (in 2011-constant dollars) or four times the current cash compensation.¹⁸ For other inputs, I assume a market risk premium of 6.5%, set the risk-free rate to the yield on 7-year US Treasuries, estimate dividend yields as the average yield over the past 36 months and volatilities based using the last 48 months of stock returns. Dividend yields above 5% are set to 5%, while volatilities below 20% or above 60% are set to 20% and 60%, respectively. As a simplifying assumption, I assume that the term for all options and restricted stock grants equals the term on the largest option grant (or five years if no options are granted), and assume that the executive’s prior holdings of stock and options are fixed throughout the term of the new grants. Finally, I assume (somewhat arbitrarily) that the risk-adjusted value of accounting-based bonuses is worth 90% of target bonuses.

Figure 7 shows the 1992–2011 evolution of risk-adjusted pay for CEOs in S&P 500 firms, assuming constant relative risk aversion of 2 or 3. The bar height depicts median pay without risk adjustments from Figure 4. Several features of the figure are worth noting:

The qualitative results in Figure 7 are robust to alternative definitions of risk aversion, safe wealth, equity premiums, and option terms. Calculating more precise estimates of risk-adjusted compensation for individual CEOs requires unavailable data on outside wealth and unobservable measures of individual risk aversion. In addition, more-precise estimates should allow CEOs to invest outside wealth in the market portfolio (Cai and Vijh, 2005) and allow for early exercise and different vesting and exercise terms of current grants and existing holdings. Nonetheless, the results in Figure 7 highlight that inferences based on reported grant-date compensation do not necessarily extend to risk-adjusted compensation.

2.2.1 The Relation Between CEO and Shareholder Wealth

Most research on CEO incentives has been firmly (if not always explicitly) rooted in agency theory: compensation plans are designed to align the interests of risk-averse self-interested CEOs with those of shareholders. Following this framework, most of the focus has been on the relation between CEO compensation (or CEO wealth) and changes in firm value. Researchers have often used the ratio of equity-based total compensation to total compensation as a measure of incentives. However, the most direct linkage between CEO and shareholder wealth comes from the CEO’s holdings of stock, restricted stock, and stock options. CEO wealth is also indirectly tied to stock-price performance through accounting-based bonuses (reflecting the correlation between accounting returns and stock-price performance), through year-to-year adjustments in salary levels, target bonuses, and option and restricted stock grant sizes, and through the threat of being fired for poor stock-price performance. The CEO pay literature has yet to reach a consensus on the appropriate methodologies and metrics to use in evaluating the “indirect” relation between CEO pay and company stock-price performance. For practical purposes, however, Hall and Liebman (1998) and Murphy (1999) show that virtually all of the sensitivity of pay to corporate performance for the typical CEO is attributable to the direct rather than the indirect part of the CEO’s contract, and the direct part can be measured from information available in corporate proxy statements.

Since agency costs arise when agents receive less than 100% of the value of output, the CEO’s share of ownership is a natural measure of the potential severity of the agency problem. In particular, the CEO’s percentage holdings of his company’s stock measures how much the CEO gains from a $1 increase in the value of the firm, and how much he loses from a $1 decrease. Computing percentage ownership for restricted and unrestricted shares is trivial (simply divide by the total number of shares outstanding). Including stock options in a percentage holdings measure is more complicated, since options that are well out-of-the-money provide few incentives to increase stock prices, while options that are well in-the-money provide essentially the same incentives as holding stock. Therefore, each stock option should count somewhat less than one share of stock when adding the holdings to form an aggregate measure of CEO incentives, and the “weight” should vary with how much the option is in (or out) of the money. In constructing an aggregate measure of CEO incentives, I weight each option by the “Option Delta”, defined as the change in the value of a stock option for an incremental change in the stock price. Option Deltas range from near zero (for deep out-of-the-money options) to near one (for deep in-the-money options on non-dividend paying stock).¹⁹ I call our measure the “effective ownership percentage” to distinguish it from the actual ownership percentage based only on stock (and not option) holdings.

Figure 8 shows the evolution of the median effective percentage ownership for CEOs in S&P 500 firms from 1992 to 2011. The percentage ownership for stock and restricted stock is calculated by dividing the CEOs shareholdings by the total number of shares outstanding. Effective percentage ownership for stock options is measured by weighting each option held by the executive at the end of the fiscal year by “Option Delta” for that option (which varies according to the exercise price and time remaining to exercise), and dividing by the total number of shares outstanding.²⁰ As shown in the figure, stock and restricted stock holdings for the median S&P 500 executive has grown modestly over the 20-year period (reflecting the increased popularity of restricted stock), ranging from 0.12% to 0.15%. Over the same time period, total effective ownership (including delta-weighted options) doubled from 0.35% in 1992 to 0.69% in 2003, before falling to 0.38% in 2011. The drop in ownership in 2008 depicted in Figure 8 primarily reflects that most options held by CEOs at the end of 2008 were substantially out of-the-money and therefore had low incentives and low Option Deltas.

The measure of effective CEO ownership in Figure 8 is essentially the “Pay-Performance Sensitivity” introduced by Jensen and Murphy (1990b). The primary difference is that I am measuring the effective ownership percentage, while Jensen and Murphy measured the change in CEO wealth per $1000 change in shareholder wealth, which equals the effective ownership percentage multiplied by ten. The other difference is that Jensen and Murphy also include indirect incentives from cash compensation and disciplinary terminations. Using data from 1974 to 1986, Jensen and Murphy estimate a median pay-performance sensitivity for stock and options of $2.50 for every $1000 change in shareholder wealth, which corresponds to an ownership percentage of 0.250%.²¹ Therefore, by the end of 2003, pay-performance sensitivities had nearly tripled the data from 1974 to 1986. But, by year-end 2011 the pay-performance sensitivity was slightly above its 1992 level, or about 50% higher than the Jensen–Murphy estimate.

The average market capitalization of firms in the S&P 500 grew (in 2011-constant dollars) from $10.0 billion in 1992 to $35.8 billion in 2000 (before falling to $22.7 billion in 2011), therefore the dollar value of the typical CEOs ownership position is large even if his percentage holding is low. Hall and Liebman (1998) argue that a better way to measure CEO incentives is as the change in CEO wealth for a 1% change in the value of the firm rather than as the ownership percentage. Baker and Hall (2004) provide some theoretical justification for using this measure. In particular, Baker and Hall show that percentage ownership is the right measure of incentives when the marginal product of the CEO effort is constant across firm size, such as a CEO contemplating a new corporate headquarters that will benefit the CEO but perhaps not the shareholders, or an outside takeover bid that will benefit outside shareholders but perhaps not the CEO. But, the Hall-Liebman measure is appropriate when the marginal product of the CEO effort scales with firm size, such as a corporate reorganization (assuming it takes the same amount of CEO effort to reorganize a big firm as a small firm).

Figure 9 shows the evolution of the Hall-Liebman measure—what Frydman and Jenter (2010) call “equity at stake”—from 1992 to 2011. The equity-at-stake measure is calculated as 1% of the effective ownership percentage multiplied by the firm’s market capitalization.²² In 1992, each 1% change shareholder wealth resulted in a $181,000 change in CEO wealth for the median CEO in the S&P 500. The equity-at-stake measure grew to nearly $900,000 in 2000 and again in 2005, before plummeting to $265,000 in 2008 as a result of both the decline in market capitalizations and the decline in Option Deltas.

As an alternative to both the Jensen-Murphy and Hall-Liebman measures, Edmans, Gabaix, and Landier (2009) provide theoretic justification for measuring incentives using the “wealth-performance elasticity” (i.e., the percentage change in CEO wealth corresponding to a percentage change in firm value) when the CEO effort has a multiplicative (rather than additive) effect on both CEO utility and firm value. In practice, creating this measure generally requires data not available to researchers (in particular, the CEO’s wealth beyond his portfolio of company stock and options).²³

2.2.2 The Relation Between CEO Wealth and Stock-Price Volatilities

Suspicions that executive compensation policies in financial services firms contributed to the 2008-2009 financial crisis eventually broadened to similar suspicions for companies outside the financial sector. In December 2009, as part of the continued fallout from the crisis, the SEC began requiring all publicly traded companies to disclose and discuss compensation policies and practices that might provide incentives for executives to take risks that are reasonably likely to have a materially adverse effect on the company.

When executives receive rewards for upside risk, but are not penalized for downside risk, they will naturally take greater risks than if they faced symmetric consequences in both directions. For top executives, rewarded primarily with equity-based compensation, the primary source of risk-taking incentives emanates from stock options. The pay-performance relation implicit in stock options is inherently convex, since executives receive gains when stock prices exceed the exercise price, but their losses when the price falls below the exercise price are capped at zero.

Since equity is a “call option” on a leveraged firm (Black and Scholes, 1973), equity-based pay in a leveraged firm can provide similar risk-taking incentives to those provided by stock options in an all-equity firm. Consider, for example, an investment opportunity promising equal chances of a $400 million gain and a $600 million loss (i.e. a net-present value of -$100). Shareholders in a $1 billion all-equity firm will have no incentive to pursue this negative NPV investment, because they will bear 100% of both the gains and losses. But, suppose the firm has only $100 million in equity, and $900 million in debt. Equity holders receive 100% of the upside, but their downside liability is limited to the value of their initial equity stake ($100 million). Thus, from the perspective of the equity holders, the project has a net present value of +$150 million.

The conflict of interest between shareholders and debtholders—dubbed the “Agency Cost of Debt” by Jensen and Meckling (1976)—has led several researchers to measure risk-taking incentives by leverage ratios and to prescribe CEO pay structures that include debt as well as equity.²⁴ However, it is worth noting that it is not leverage per se that creates risk-taking incentives, but rather the limited liability feature of equity. For example, the shareholders in the example in the prior paragraph would have incentives to take the negative NPV project even if the firm was a $100 million all-equity firm; in this case losses greater than $100 million would be borne by the government or society, etc., and not by debtholders. It is also worth noting that the severity of the risk-taking incentives depends on the maximum downside risk compared to the dollar amount of equity, and not the value of equity compared to the overall value of equity plus debt. The level of debt is important only to the extent that is available to fund risky negative NPV projects.

Since the value of a stock option (or the value of equity in a leveraged firm) increases monotonically with stock-price volatilities, options (and limited liability) provide incentives for executives to increase such volatilities. In Section 2.2.1, the calculations for pay-performance sensitivities for stock options depended on the Option Delta, defined as the change in the value of a stock option associated with an incremental change in the stock price. Similarly, the calculations for pay-volatility sensitivities for stock options depend on the Option Vega, typically defined as the change in the value of a stock option associated with one percentage-point increase in the stock-price volatility (e.g. from 30% to 31%). Option Vegas are typically highest when stock prices are near the option’s exercise price.

Following Fahlenbrach and Stulz (2011)’s analysis of executive compensation and the financial crisis, I consider two option-based measures for incentives to increase stock-price volatilities:

Figure 10 shows the time trends in the two measures of pay-volatility sensitivities for the median executive in a S&P 500 firm from 1992 to 2011. The left-hand axis reports the Total Option Vega, which reached its peak in 2003 (when the median CEO gained $243,000 by increasing volatility by one percent), and plummeted in 2008 to $127,000 for a one percent increase in volatility. The right-hand axis reports the percentage change in option values associated with a one percent increase in volatility. This “Vega Elasticity” remained relatively constant from 1992 to 2007 at around 1.0 (indicating that a one percentage-point increase in volatility would increase the value of CEO option holdings by about 1%). The Vega Elasticity jumped to over 5.0% in 2008, falling to 2.0% by 2011.

The differences in the two measures in Figure 10 reflect the effect of stock-market movements and, in particular, the market crash at the end of 2008 and the partial rebound by 2011. When stock prices fell (as they did abruptly in 2008, across all sectors of the economy), the options fell out of the money, which implies that the Option Vega for each option becomes smaller (remember that the Option Vega is highest when the stock price is close to the exercise price). But, it turns out that, as stock prices fall, the value of the options held fall even faster than the Option Vega. As a result, the value of options that are out-of-the-money increases more in percentage terms (but less in dollar or euro terms) as volatility increases.

One troublesome fact apparent from Figure 10 is that the two vega measures—both legitimate measures for risk-taking incentives—move in opposite directions in market downturns. There is no accepted methodology for measuring incentives for risk in executive option portfolios, or in executive equity positions in leveraged firms, or in executive contracts more generally.²⁵ Until the recent financial crisis—when compensation policies were blamed for contributing to the meltdown—there had been little focus on the role of compensation policies in providing incentives to take risks.

Finally, while the current controversy over executive incentives has focused on excessive risk taking, it is worth noting that the challenge historically has been in providing incentives for executives to take enough risk, not too much risk. Executives are typically risk-averse and undiversified with respect to their own companies’ stock-price performance. On the other hand, shareholders are relatively diversified, placing smaller bets on a larger number of companies. As a result, executives will inherently be “too conservative” and want to take fewer risks than desired by shareholders. Stock options (or other plans with convex payouts) have long been advocated as ways to mitigate the effects of executive risk aversion by giving managers incentives to adopt rather than avoid risky projects (see, for example, Hirshleifer and Suh, 1992). Similarly, there is a long history of attempts to document an empirical relation between such convexities and actual risk-taking incentives, and the results have been relatively modest.²⁶

2.3.1 Problems with Non-Linear Pay-Performance Relations

Researchers have long acknowledged that non-linear incentive plans cause predictable problems.²⁸ For example, executives capable of producing well above the upper performance threshold in Figure 11 have incentives to stop producing once they “max out” on their bonuses. In addition, they will do their best to transfer performance results that could have been realized this period into a later period.

Similarly, but potentially worse, is the effect of the discontinuity at the lower performance threshold in Figure 11. Executives who believe they cannot achieve at least this level of performance this year will either stop producing or “save” performance for next year by delaying revenues or accelerating expenses. Moreover, if executives see that they are not going to make the bonus pool this year, they are better off taking an even bigger hit this period (since there is no bonus penalty for missing the lower threshold by a lot instead of a little) so they can do even better next period—what accountants have called the “big bath” phenomenon. On the other hand, executives who are struggling to make the lower threshold, but still believe they can make that threshold, have incentives (provided by the threshold bonus) to do whatever is necessary to achieve the lower threshold. Their actions commonly include destroying value by loading the distribution channel so as to recognize revenues earlier, unwisely reducing R&D and required maintenance expenditures, and (in some cases) outright accounting fraud. Each of these actions shifts reported profits from next period to the current period, but does so at an unnecessary cost to the firm.

In both of these cases, the non-linearities provide incentives for CEOs to “manage earnings”. In particular (and assuming that performance is measured by earnings), the bonus plan in Figure 11 provides incentives to “smooth earnings” (by shifting earnings from next period when below the lower threshold and shifting earnings to next year when above the upper threshold), while occasionally taking a “big bath” (when it is not possible, even with manipulation, to get earnings above the lower threshold).

In addition to earnings management, non-linearities also affect risk-taking behavior. In particular, when the pay-performance relation is concave (so that lower performance is penalized more than higher performance is penalized), executives can increase their total bonus payouts by reducing the variability of their performance. Conversely, convex pay-performance relations increase risk-taking incentives. Financial economists have suggested that boards purposely add convexity to CEO pay contracts to offset the reluctance of risk-averse CEOs to invest in risky (but profitable) projects.²⁹ More recently, some academics (as well as Congress and the popular press) have alleged that convexities in banking bonuses (where positive performance is rewarded, but negative performance is not penalized) led to excessive risk-taking that, in turn, facilitated the 2008–2009 financial crisis.

The problems with non-linearities are mitigated by eliminating caps on the upside, and finding ways to implement and enforce “negative” bonuses on the downside.³⁰ While it is difficult to force CEOs to write checks back to the company after a bad year, negative bonuses can be partially implemented by basing pay on multi-period cumulative performance (Holmstrom and Milgrom, 1987) or by deferring current compensation into bonus banks that can be used to fund future negative bonuses (Stewart, 1991). Another indirect way to impose negative bonuses is by reducing base salaries and offering enhanced bonus opportunities (through reduced bonus thresholds).

2.3.2 Problems with Performance Benchmarks

Bonuses are usually not, in practice, based strictly on a performance measure, but rather on performance measured relative to a performance benchmark (Murphy, 2000). Examples include net income measured relative to budgeted net income, EPS vs. last year’s EPS, cash flow vs. a charge for capital, performance measured relative to peer-group performance, or performance measured against financial or non-financial strategic “milestones”. Performance targets (one form of benchmark) typically correspond to the level of performance required to attain the executive’s “target bonus”.

When bonuses are based on performance relative to a benchmark, executives can increase their bonus either by increasing performance or lowering the benchmark. Performance benchmarks therefore create predictable problems whenever the participants in the bonus plan can affect the benchmark. For example, when benchmarks are based on meeting a budget, executives with bonuses tied to budgeted performance targets have strong incentives to low-ball the budget. Boards (and supervisors throughout the management hierarchy) understand these incentives and generally push for higher budgets than those suggested by executives. The result is a familiar and predictable “budget game” that ultimately destroys the information critical to coordinating the disparate activities of a large complex organization (Jensen, 2003).

As another example, when benchmarks are based on prior performance (such as bonuses based on growth or improvement), plan participants understand that increased performance this year will be penalized by higher benchmarks the next year, and will naturally take account of these dynamics when deciding how hard to work and what projects to undertake in the current year. Similarly, when bonuses are based on performance measured relative to that of colleagues, participants can increase their bonuses by sabotaging co-executives (Lazear, 1989; Gibbons and Murphy, 1990). Benchmarks based on industry peers provide incentives for selecting “weak” industries or peers, or staying too long in a defective industry (Dye, 1992).

The problems with benchmarks based on budgets, prior-year performance, co-workers, and other internally manipulable measures can be mitigated by “externalizing” the benchmark; that is, by basing the benchmark on objective measures beyond the direct control of the plan participants. In Murphy (2000), I showed that companies using external benchmarks (which I defined as benchmarks based on fixed numbers or schedules, industry performance, or the cost of capital) were less likely to manage fourth-quarter earnings than were companies with internal benchmarks. However, I was unable to explain satisfactorily cross-sectional differences in the use of internal and external benchmarks, or why nearly 90% of the sample of 177 firms based benchmarks on budgets or prior-year performance.

2.3.3 Problems with Performance Measures

The problem of inappropriate performance measures is illustrated succinctly by the title of Steven Kerr’s famous 1975 article, “On the folly of rewarding A, while hoping for B” (Kerr, 1975). Paying salespeople commissions based on revenues, for example, provides incentives to increase revenues regardless of the costs or relative margins of different products. Likewise, paying rank-and-file workers “piece rates” based on units produced provides incentives to maximize quantity irrespective of quality, and paying a division head based solely on divisional profit leads that division head to ignore the effects of his decisions on the profits of other divisions. Similarly, paying CEOs based on short-run accounting profits provides incentives to increase short-run profits (by, for example, cutting R&D) even if doing so reduces value in the long run.

Conceptually, the “perfect” performance measure for a CEO is the CEO’s personal contribution to the value of the firm.³¹ This contribution includes the effect that the CEO has on the performance of others in the organization, and also the effects that the CEO’s actions this year have on performance in future periods. Unfortunately, the CEO’s contribution to firm value is almost never directly measurable; the available measures will inevitably exclude ways that the CEO creates value, and include the effects of factors not due to the efforts of the CEO, or fail to reveal ways that the CEO destroys value. The challenge in designing incentive plans is to select performance measures that capture important aspects of the CEO’s contributions to firm value, while recognizing that all performance measures are imperfect and create unintended side effects.

While companies use a variety of financial and non-financial performance measures in their annual CEO bonus plans, almost all companies rely on some measure of accounting profit such as net income, pre-tax income, or operating profit. Accounting profit measured over short intervals is not, however, a particularly good measure of the CEOs contribution to firm value, for several reasons. First, CEOs routinely make decisions (such as succession planning or R&D investments) that will increase long-run value but not short-run profit. Second, accounting profits (like equity-based measures) are invariably influenced by factors outside of the control of the CEO, including the effects of business cycles, world oil prices, natural disasters, terrorist attacks, etc. Third, while the measures of accounting profits typically used in bonus plans take into account both revenues and expenses, they ignore the opportunity cost of the capital employed. The use of these accounting measures provides incentives to invest in any project that earns positive accounting profits (not just those that earn more than the cost of capital), and provides no incentives to abandon projects earning positive accounting profits that are less than those required to cover their cost of capital.

Exacerbating the problems with accounting-based performance measures in bonus plans is the fact that they are often expressed as ratios (e.g., earnings per share, return on assets, return on equity, return on capital, etc.). Executives participating in such plans can increase their bonus either by increasing the numerator (accounting profits) or by decreasing the denominator (e.g. shares, assets, equity, invested capital). For example, a CEO paid on the basis of return on capital would prefer a $100 million project earning a 40% return to a $1 billion project earning a 25% return, even though the latter creates more wealth (as long as the cost of capital is less than 22%).

3.5.1 America, Land of the Freeze

In August 1971, in an ultimately (and predictably) unsuccessful attempt to control inflation, President Nixon imposed a 90-day freeze on commodity prices and wages (including executive pay). In December 1971—in what was called Phase Two of the Nixon wage-and-price controls—the Pay Board established by Congress imposed a limit of 5.5% for increases in executive pay (the limit being binding for company-defined groups of executives, but not necessarily for individual executives).⁶⁰ The Nixon wage-and-price controls were not the first time that levels of executive compensation were explicitly limited by legislation, but were the first time such controls were imposed in a peacetime economy. In particular, the World War II-era Stabilization Act of 1942 froze wages and salaries (for executives as well as other labor groups) at their September 15, 1942 level. The Stabilization Act expired in 1946, but was replaced during the Korean War by the Salary Stabilization Boards established in May 1951 as part of the Defense Production Act of 1951. Similar to the Nixon controls, the Korean War Salary Board set a 6% limit on pay increases for each company’s executives taken as a group; the limits were lifted when the Board was quietly disbanded in July 1952.⁶¹

In a debate (and outcome) eerily similar to what would happen two decades later during the Clinton Administration, concerns that the Nixon wage controls would significantly reduce executives incentives led to a series of compromises (or loopholes, depending on one’s perspective).⁶² In particular, while bonuses were generally limited to the amount paid in any one of the last three years plus 5.5%, the limit did not apply to existing sales incentives, commission and production-incentive programs. Moreover, companies could petition to adopt new incentive plans as long as they were directly related to increased productivity. As a result, scores of companies introduced performance-based bonus plans tied to accounting data or revenues, or converted their existing plans into plans exempt from the limits.

Non-qualified stock options were allowed under the Nixon controls only if the plan was shareholder-approved, if the aggregate number of options granted did not increase from the prior three years, and if the exercise price was at least 100% of the grant-date market price. Non-qualified options were treated as wages and salaries under the Nixon controls, and were valued at 25% of the fair-market value of the shares underlying the option.⁶³ This valuation approach represents an interesting (albeit short-lived) historical footnote, since it was imposed a year before Black and Scholes (1973) and decades before companies began routinely placing a value on options when making compensation decisions.

The median continuing CEO in the Forbes 800 received a 4.5% increase in cash compensation in 1971 (below the Nixon limit), 6.0% in 1972, and 8.1% in 1973 (both above the Nixon limit).⁶⁴ Since the government-mandated limits on pay raises applied only to executives taken as a group and not individual executives, companies routinely raised CEO pay by reducing pay (or offering smaller raises) to lower-level executives.⁶⁵ In August 1973, to stop companies from raising CEO pay above the 5.5% limit, the Nixon Administration imposed the 5.5% limit on the more-narrowly defined group of executives identified in company proxy statements. The wage-and-price controls expired in May 1974, in spite of Administration efforts to retain limits on executive compensation.⁶⁶

CEO pay rose significantly after the wage controls were lifted in May 1974. The median continuing CEO in the Forbes 800 received an 11.1% increase in nominal cash compensation in 1974, double the average limit under the Nixon controls. From 1973 through 1979, the median cash compensation for CEOs in the Forbes 800 increased by 12.2% each year (doubling from $162,000 to $324,000), significantly exceeding the average annual inflation rate of 8.5%.

Figure 15 shows the median level and structure of compensation for CEOs in 73 large manufacturing firms from 1964 to 1982, based on data from Murphy (1985) and inflation-adjusted to 2010-constant dollars. To my knowledge, this was the first comprehensive study of executive pay that measured stock options as the grant-date value using the Black and Scholes (1973) approach. In nominal terms (that is, before adjusting for inflation), median CEO pay in the 73 firms in Figure 15 nearly tripled from $148,900 in 1964 to $569,550 in 1982. However, after adjusting for inflation (which averaged over 6.5% annually over this period), median real CEO pay increased only by about 23% over this 18-year period, or about 1.2% per year. Stock options accounted for 2% of total pay for the average CEO in 1964; the use of options had grown to 12% of pay by 1981. Both the level of pay and the use of stock options fell during the 1981–1982 recession.

3.5.2 The Controversy over Perquisites

While cash compensation escalated (at least in nominal terms) during the 1970s, the use of stock options was relatively stagnant. Part of the declining popularity of options reflected the change in tax policies in 1964 and 1969 that made qualified stock options less attractive, coupled with their outright prohibition in 1976 (see below). More importantly, though, was the prolonged stagnation in the stock market, driven in part by the oil-price shocks of 1973 and 1977. In particular, the nominal value of the bellwether Dow Jones average was basically flat from the beginning of 1965 through the early 1980s (falling from 903 in January 1965 to below 800 by mid-January 1982, and only surpassing 1050 on one day over these seventeen years). While executives continued to receive periodic option grants during this time (once every three years was typical), many of the grants replaced options that expired worthless or options that were cancelled and reissued with a lower exercise price.

The void in compensation created by worthless stock options was quickly filled by a plethora of new plans designed to provide more predictable payouts, including: book-value plans (where executives receive dividends plus the appreciation in book values); long-term performance plans (with payouts based on long-term earnings growth targets); and guaranteed bonuses (with payouts guaranteed independent of performance).⁶⁷ In addition, since the Nixon wage-and-price controls restricted salaries but not company-provided benefits, companies began relying to a greater extent on shareholder-subsidized perquisites or perks such as low-interest loans, yachts, limousines, corporate jets, club memberships, hunting lodges and corporate retreats at exotic locations.

By the mid-1970s, perceived abuses attracted the ire of shareholder activists, the SEC and the IRS.⁶⁸ In December 1975, the IRS circulated a draft of proposed regulations specifying which fringe benefits could be excluded from an executive’s taxable income. A long-held general rule excluded from taxable income benefits arising from the ordinary course of business that do not cost the employer anything extra (such as family members accompanying an executive on the corporate jet). The proposed rule imposed tax liabilities for these and other fringe benefits if the benefits were available only to the most highly compensated executives.

The attack on perquisites escalated in 1977 as President Carter famously rallied against companies taking deductions for the three-martini lunch, yachts and hunting lodges maintained to entertain business associates, first-class air travel, fees paid to social and athletic clubs and money spent on sports and theater tickets.⁶⁹ Congress resisted implementing most of Carter’s reforms as part the Revenue Act of 1978 (in large part because it would potentially affect their own consumption of perquisites) but agreed to eliminate deductions for entertainment facilities.⁷⁰

In August 1977, the SEC issued Interpretive Release #5856 stating that the value of perquisites be included as compensation in proxy statements.⁷¹ In justifying the new disclosures, SEC enforcement chief Stanley Sporkin argued that the “excesses just got to the point where it became a scandal”.⁷² The disclosures in the 1978 proxy statements fueled the fire by focusing even more attention on perquisites.⁷³ The information on perquisites was expanded significantly in 1979 proxy statements, when the SEC implemented its first major revision in proxy disclosures since the 1930s. Also in 1979, the IRS issued significant new auditing guidelines aimed at detecting and taxing executive perquisites. McGahran (1988) argues that the new SEC disclosures made it easier for the IRS to detect (and tax) fringe benefits, and presents some evidence that fringe benefits decreased, while cash compensation increased, as a result of the SEC and IRS actions.

The ongoing attack on perquisites was reflected in the contemporaneous early academic literature on agency theory. For example, the “agency problem” introduced by Jensen and Meckling (1976) focused on managerial consumption of non-pecuniary benefits such as “the physical appointments of the office” and “the attractiveness of the secretarial staff”. Similarly, Alchian and Demsetz (1972) conclude that companies allow personal consumption of corporate (or university) property (such as “privileges, perquisites, or fringe benefits”) because the cost of detecting and punishing such “turpitudinal peccadilloes” is larger than the benefits from prohibiting the activity.

3.5.3 There’s no Accounting for Options

The restricted and qualified stock options created by the 1950 and 1964 Revenue Acts were not formally considered compensation and therefore companies did not record an expense for such options for either tax or accounting purposes. The switch to non-qualified options in the 1970s—which were considered compensation for tax purposes—raised a new question: how should options be accounted for in company income statements? One possibility was to follow the tax code by recognizing an accounting expense at the time an option is exercised. But, in spite of its simplicity, this method is inconsistent with the basic tenet of accounting that expenses should be matched to the time period when the services associated with those expenses were rendered. Rather, the tenet suggested that options should be expensed over their term based on the grant-date value of the option. At the time, however (and for a long time to come) there was no accepted way of placing a value on an employee stock option.

In October 1972, the Accounting Principles Board (APB)—the predecessor to the current Financial Accounting Standards Board (FASB)—issued APB Opinion No. 25, “Accounting for Stock Issued to Employees”. Under APB Opinion No. 25, the compensation expense associated with stock options was defined as the (positive) difference between the stock price and the exercise price as of the first date when both the number of options granted and the exercise price become known or fixed. The expense for this spread between the price and exercise price—called the intrinsic value—was amortized over the period in which the employee is prohibited from exercising the option.⁷⁴ Under this rule, there was no charge for options granted with an exercise price equal to (or exceeding) the grant-date market price, because the spread is zero on the grant-date.

The accounting treatment of options cemented the dominance of the traditional stock option (an option granted with a five- or ten-year term with an exercise price equal to the grant-date market price) and discouraged companies from offering more novel option plans. For example, APB Opinion 25 imposes a higher accounting charge for options with an exercise price indexed to the stock-price performance of the market or industry, because the exercise price is not immediately fixed. Similarly, it imposes a higher accounting charge for options that only become exercisable if certain performance triggers are achieved, because the number of options is not immediately fixed. Finally, it imposes an accounting charge for options that are issued in the money but not for options issued at the money—a feature that became especially significant three decades later in the scandals involving backdating.

3.5.4 The Rise (and Fall) of Stock Appreciation Rights

Under Section 16(b) of the Securities Act of 1934, executives must return any profits realized from buying and selling (or selling and buying) shares of their company’s stock within any period of less than six months. This constraint was not problematic for executives exercising restricted or qualified stock options, since the provisions of the 1951 and 1964 Revenue Acts already required executives to hold shares for six months (for restricted options) or three years (for qualified options) before trading. However, the six-month holding period was particularly troublesome for non-qualified options, since executives were required to pay ordinary income tax based on the date the option was exercised and not when the underlying shares were sold.⁷⁵ Given the depressed stock market in the 1970s, the taxes due upon exercise were often greater than value of the shares when they became tradable.

In December 1976, the SEC formally exempted stock appreciation rights (SARs) from the Section 16(b) short-swing profit prohibition.⁷⁶ Executives holding a SAR are entitled to receive the appreciation on one share of stock. Like stock options, SARs had a pre-determined term but executives were generally free to exercise their SARs at any time prior to the end of this term (after some minimum time had elapsed). Prior to the December 1976 ruling, there was considerable debate about whether SARs would be subject to the short-swing rule and therefore the proceeds from the exercise of the rights would have to be returned to the company. After the SEC ruling, SARs provided a way for executives to reap the benefits of exercising non-qualified options without being subject to the six-month holding requirement.⁷⁷As a result of the ruling, many companies replaced their option grants with SAR grants, or issued tandem SARs and options, which allowed the executive to decide which to exercise. For the next fifteen years, SARs became a ubiquitous component of long-term compensation for most executives.

Jumping ahead a bit, in May 1991 the SEC declared that the six-month holding period begins when options are granted, and not when executives acquire shares upon exercise. Therefore, as long as the executive has held the option for at least six months, he is allowed to immediately sell the shares acquired when options are exercised. This new ruling eliminated the primary advantage of SARs over non-qualified options and, as a result, SARs largely disappeared from existence. In addition, the SEC rule effectively encouraged the practice—commonplace today—of selling shares immediately upon exercise.⁷⁸

The rise and ultimate fall of SARs is a tribute to the cleverness of companies in finding ways around rules that disadvantage executives and companies (in this case, the six-month holding requirement).⁷⁹ Moreover, the experience shows how seemingly innocuous government interventions (in this case, the 1976 and 1991 SEC rulings) can have a major impact on the composition of executive compensation.

3.5.5 Qualified Stock Options Resurrected, but No One Cares

The Revenue Acts of 1964 and 1969 significantly reduced the attractiveness of restricted/qualified stock options, but did not prohibit new grants. As part of the Revenue Act of 1976, Congress allowed executives to retain and exercise grants made prior to May 20, 1976, but banned all future grants of qualified stock options. Since existing grants had a maximum five-year term, the last grant of qualified options was set to expire on May 19, 1981.

As 1981 approached, Congress resurrected a new form of qualified options (now called Incentive Stock Options or ISOs) as a last-minute addition to the Economic Recovery Tax Act of 1981.⁸⁰ ISOs carried many of the restrictions common for qualified stock options (holding periods after exercise, minimum exercise prices, etc.), and in addition were limited to $100,000 per executive per year (calculated as the stock price multiplied by the number of options on the date of grant). While ISOs have continued to be popular in the 2000s for middle-level managers (where the $100,000 limitation is not binding) and for companies without taxable profits (where loss of deductibility for ISOs is not costly), virtually all options granted to CEOs and other top executives since 1972 have been non-qualified stock options.

3.5.6 Bigger is Better (Paid)

Almost half of the cross-sectional variation in cash compensation in the United States between 1970 and 1982 was explained by company size (usually measured by firm revenues), and the highest-paid executives routinely were at the helm of the largest conglomerates and largest steel, automotive, and oil companies. Year-to-year changes in cash compensation were also largely driven by increases in company size. And non-monetary aspects of compensation—including power, prestige, board memberships, and community standing—were also positively linked to increases in firm size. The strong relation between CEO pay and company size gave CEOs substantial incentives to increase company size, while the decline of equity-based incentive plans gave them little incentive to increase company share prices. It is noteworthy that the implicit incentives to increase company revenue help explain the unproductive diversification, expansion, and investment programs in the 1970s, which in turn further depressed company share prices.

Although CEO pay and bottom-line corporate profitability remained relatively stagnant from 1970 to 1982, productivity did not. Spurred in part by the oil-price shocks of 1973 and 1977, this period brought significant technological advances that improved productivity, declines in regulation, and increases in global trade significant enough to constitute what Jensen (1993) calls the “Modern Industrial Revolution”. By the early 1980s, most sectors in the US economy were saddled with increasing excess capacity, implying that the sectors had more capital and labor than were required to maintain current levels of production. The root causes of the excess capacity differed across industries. In the oil sector, for example, the five-fold increase in the inflation-adjusted price of crude oil led firms to launch massive capacity-increasing exploration and development projects in anticipation of continued price increases; the sector was stuck with the capacity when demand dropped and prices tumbled to pre-shock levels. Technological change dramatically increased capacity for computing firms, while increased competition from non-unionized entrants created excess capacity in a variety of industries ranging from steel to groceries.

By definition, investment in an industry with excess capacity is a negative net-present-value project, since the industry already has more capital and labor than can be productively employed. Indeed, firms with excess capacity can either increase output with the same workforce, or maintain current output with a smaller workforce. However, the 1970s conglomerates and other large companies typically chose to neither increase output (given low market demands) nor decrease their workforce (since pecuniary and non-pecuniary rewards for CEOs were both tied to company size). Moreover, by the end of the 1970s, most of these companies were generating huge amounts of cash, far in excess of that required to fund available positive net present value projects. CEOs, loathe to distribute excess cash back to shareholders, responded by wasting huge amounts of free cash flow through unwise diversification and investment programs.⁸¹

3.6.1 Golden Parachutes and Section 280(G)

The executive compensation practices of the 1970s provided few incentives for executives to pursue value-increasing reductions in excess capacity and disgorgements of excess cash. Equity-based compensation (mostly in the form of stock options) accounted for only a small fraction of CEO pay (Figure 15) and the options that existed often were underwater or expired worthless. Annual bonuses—the dominant form of compensation-based incentives—were focused on beating annual budget targets rather than creating long-run value. Performance-based terminations were almost non-existent and—since the vast majority of CEO openings were filled by incumbents rather than outside hires—the managerial labor market was similarly ineffective in disciplining poor performance.

Boards of directors, typically dominated by corporate insiders (in influence if not in numbers), had little reason to reduce corporate waste as long as the companies were delivering positive nominal profits. However, pressures to improve performance and disgorge cash were ultimately introduced by the capital markets, including “hostile takeover” artists such as Carl Icahn, Irwin Jacobs, Carl Lindner, David Murdock, Victor Posner, Charles Bluhdorn, and T. Boone Pickens. At the time, these takeover artists were known pejoratively as “corporate raiders”, though history has shown they were a positive force in creating substantial amounts of value for shareholders of target firms while reallocating resources to higher-valued uses.⁸² Sometimes this wealth was created by the post-merger activities of the raiders (such as firing incompetent incumbent managers and selling non-productive assets). At other times the wealth was created by responses to the takeover threat (such as spending cash to repurchase shares or to purchase competitors, causing resources to leave the sector and allowing shareholders to find more productive uses for their cash).

The takeover market was complemented by the emergence of leveraged buyouts (LBOs): going-private transactions financed by debt using the target firm’s future cash flows as collateral. Debt created value by providing commitments that the firm would pay its cash flows to debtholders, reducing the amounts available for executives to waste (Jensen, 1986a). Debt also taught executives that capital is costly (since the interest cost of debt capital was more obvious than the implicit, though larger and largely unrecognized, cost of equity capital), leading to reductions in inventories and working capital. The emergence of LBOs and leveraged recapitalizations (in which the firm leverages the capital structure while staying public) created substantial amounts of shareholder value in firms with stable cash flows and no productive alternative uses for their cash, characteristics of many of the mature and declining sectors in the early 1980s.

While employment in companies targeted by hostile takeovers or LBOs was modestly reduced (which was productive given the presumptive excess capacity), the individuals most vulnerable to job losses were incumbent executives opposed to the changes in control. Innovations designed to thwart takeovers included greenmail payments (repurchase of the raiders’ stock at above market prices), standstill agreements (bribes so that the raider does not purchase additional stock), staggered boards (where directors serve overlapping terms, making it difficult for a proxy fight to gain a majority), supermajority rules (requiring more than 50% votes to approve a merger), and poison pills (where shareholders get special rights when there is a takeover bid). But, perhaps the most notorious innovation was the “golden parachute” which provided direct payments to executives following a successful change in control. In most cases, the payment required both the change-of-control and the loss of a job (hence, called “double-triggered” since two things had to happen); in other cases (single-triggered) the change-of-control itself was sufficient to trigger the payment, regardless of job loss.⁸³

Whether change-of-control agreements facilitate or thwart takeovers remains a matter of debate and rests in the details. On one hand, as emphasized by Jensen (1986b), such agreements facilitate takeovers by providing bribes to existing managers to acquiesce to the change in control. On the other hand, such agreements can significantly increase the cost of takeovers for prospective acquirers, especially if the agreements cover dozens or hundreds of executives who have no plausible influence over the takeover decision. In any case, the existence of the apparent bribes paid to top executives (but not to shareholders in general) attracted the ire of a Congress already skeptical of hostile takeovers and their benefits.

Change-in-control arrangements became controversial following a $4.1 million payment to William Agee, the CEO of Bendix. In 1982, Bendix launched a hostile takeover bid for Martin Marietta, which in turn made a hostile takeover bid for Bendix. Bendix ultimately found a “white knight” and was acquired by Allied Corp., but only after paying CEO Agee the golden parachute. The payment sparked outrage in Washington, but Congress could not ban golden parachute payments outright, because such a ban would pre-empt state corporation laws. Congress does, however, control the tax laws, which allow corporations to deduct compensation from income only if the payments represent reasonable compensation for services rendered. By defining particular types or dollar amounts of compensation as unreasonable, Congress can directly determine whether compensation is deductible for corporate tax purposes.

Congress attempted to discourage golden parachutes by adding Sections 280(G) and 4999 to the tax code as part of the Deficit Reduction Act of 1984. Section 280(G) of the Code provides that, if change-in-control payments exceed three times the individuals base amount, then all payments in excess of the base amount are non-deductible to the employer. Also, Section 4999 imposes a 20% excise tax on the recipient of a parachute payment on the amount of payment above the base amount. The base amount is typically calculated as the individuals average total taxable compensation (i.e. W-2 compensation, which include gains from exercising stock options) paid by the company over the prior five years.

Because of the complexity of what appears to be a simple rule, modest increases in parachute payments can trigger substantial tax payments by both the company and executive. For example, suppose an executive with five-year average taxable compensation of $1 million receives a golden parachute payment of $2.9 million, which is less than three times the $1 million base amount.⁸⁴ In this case, the entire $2.9 million parachute payment would be deductible by the company, and would be taxable as ordinary income to the executive. In contrast, suppose that the golden parachute payment was $3.1 million, which is more than three times the $1 million base amount. Under Section 280(G), the company would not be able to deduct $2.1 million (of the $3.1 million parachute payment) as a compensation expense, and (under Section 4999) the executive would owe $420,000 in excise taxes (i.e. 20% of $2.1 million) in addition to ordinary income taxes on the full $3.1 million parachute payment.

The new Section 280(G) impacted executive compensation in several ways. First, the new law led to a proliferation in change-in-control agreements, which had previously been fairly rare. The Deficit Reduction Act was signed into law on July 18, 1984. By 1987, 41% of the largest 1000 corporations had golden parachute agreements for their top executives, and the prevalence of golden parachutes increased to 57% in 1995 and to 70% by 1999.⁸⁵ In addition, the standard golden parachute payment quickly became the government prescribed amount of three times base compensation. By 1991, 47.5% of CEO golden parachute arrangements specified a multiple of three times base pay, and by 1999 71% specified three times base pay. Thus, the rule designed to limit the generosity of parachute payments has led to both a proliferation and a standardization of Golden Parachute payments in most large corporations. Apparently compensation committees and executives took the regulation as effectively endorsing such change-in-control agreements as well as the payments of three times average compensation (which quickly became the standard).

Second, Section 280(G) (and the corresponding Section 4999) gave rise to the “tax gross up”, in which the company offset the tax burden of the 20% excise tax by paying an additional amount for the tax (and the tax on the additional amount).⁸⁶ The percentage of agreements that included gross-up provisions increased from 38% in 1991 to over 82% by 1999.⁸⁷ This gross-up concept was subsequently applied to a variety of executive benefits with imputed income taxable to the executive, such as company cars, club memberships, and personal use of corporate aircraft.

Third, Section 280(G) also provided incentives for companies to shorten vesting periods in stock option plans, and incentives for executives to exercise stock options even earlier than they would normally be exercised. Consider two otherwise identical executives with golden parachutes paying three times base compensation and holding identical options. Suppose that one of the executives exercises a year prior to the change in control, while the other holds until the change in control. Since base compensation under Section 280(G) includes gains from exercising options, the first executive can receive a higher parachute payment before triggering the excise tax, thus increasing the benefits from early exercise. Moreover, unexercisable stock options routinely become vested (or exercisable) upon a change in control, and the value of these options is defined by the IRS as part of the parachute payment subject to the excise taxes. Therefore, companies and executives can reduce change-in-control related tax liabilities by shortening the time until options become exercisable, and by exercising early and therefore reducing the incentive effects of those plans.

Similarly, unvested restricted stock routinely becomes vested upon a change in control, and a portion of the value of these shares upon vesting is defined by the IRS as part of the parachute payment subject to the excise taxes. Thus, companies can also reduce change-in-control related tax liabilities by shortening the vesting period for restricted stock.

Finally, but perhaps most importantly, the 1984 tax laws regarding Golden Parachutes appear to have triggered the proliferation of Employment Agreements for CEOs and other top-level executives in most large firms since the mid-1980s. In particular, Section 280(G) applies only to severance payments contractually tied to changes of control, while individual CEO employment agreements typically provide for severance payments for all forms of terminations without cause, including (but not limited to) terminations following control changes. Therefore, companies can circumvent the Section 280(G) three-times-base-compensation limitations (at a potentially huge cost to shareholders) by making payments available to all terminated executives, and not only those terminated following a change in control. Indeed, Graef Crystal (when he was still a leading compensation consultant) predicted the unintended consequences of the enactment of these tax provisions in his 1984 opinion piece in the Wall Street Journal:

 But will Congress’s new reforms really curb those who want to offer excessive compensation? Not necessarily. Congress has, as usual, made an opening move in a corporate chess game and neglected to consider its opponents countermoves. Instead of having a contract that covers only a change of control, some companies may now implement all-embracing employment contracts that guarantee a person employment (or what he would have earned had he continued to be employed), for say, five years, and under all circumstances. You won’t see one word in that contract about payments in the event of a change of control, and the net effect will be to give the executive more than he would have had had Congress not given free rein to its passions.⁸⁸

In summary, although Section 280(G) was meant to reduce the generosity of parachute payments, the government action appears to have increased the prevalence of: (i) change-in control plans; (ii) tax gross-ups; (iii) early exercise of stock options; (iv) short vesting periods for restricted stock and stock options; and (v) employment agreements. Each of these outcomes both reduces the incentive effects of incentive compensation for CEOs and other executives and increases the costs of these plans to their firms.

3.6.2 The Shareholder Awakening

The emerging market for corporate control had pronounced effects on the US stock market. After nearly two decades of stagnation, the Dow Jones Industrial Average rallied from below 800 to over 2700 between mid-1982 and mid-1987 (i.e. appreciating nearly 30% per year for five years). While the largest beneficiaries were shareholders in firms that became takeover targets, the rally was broad based and lifted share prices across a wide range of firms and industries. However, executives vigorously (and often successfully) fought takeovers in the 1980s by adopting anti-takeover provisions and by lobbying for political protection (Holmstrom and Kaplan, 2001). Therefore, in spite of the gains to shareholders (or perhaps because of the redistribution in wealth resulting from these gains), hundreds of bills were introduced in Congress to curb takeovers and highly leveraged transactions (Fischel, 1995).

Court decisions and legislation in the late 1980s (coupled with the October 1987 stock market crash) brought the hostile takeover market in the US to a virtual halt. The high-yield debt market was crippled by the indictment and subsequent guilty pleas of Michael Milken and Drexel Burnham Lambert and by restrictions on high-yield debt holdings imposed on savings institutions, commercial banks, and insurance firms, and by major punitive changes in the US bankruptcy law that made it uneconomic to reorganize troubled firms outside of bankruptcy.

But the lessons of the wealth creations learned from the takeover wave resonated with shareholders. In 1985, Robert Monks founded Institutional Shareholder Services to provide proxy-voting advice to institutional shareholders. In 1986, corporate raider T. Boone Pickens founded the United Shareholders Association focused on improving governance and compensation. Academics increasingly argued that traditional management incentives that focused on company size, stability, and accounting profitability destroyed rather than created value, and recommended that executive pay be tied more closely to company value through increases in stock options and other forms of equity-based incentives. These pressures began having an impact: non-equity-based CEO pay continued to grow in real terms after the mid-1980s, but became a smaller part of the total compensation package. For the first time since the 1950s, stock options re-emerged as the dominant form of incentives compensation.

Figure 16 shows the median level and average structure of CEO compensation from 1980-1992, based on Hall and Liebman (1998). Total grant-date compensation is defined as the sum of salaries, bonuses, and the grant-date value of stock options using the Black and Scholes (1973) formula. The annual sample size varies between 365 and 432 firms, and is representative of the population of the large US firms. The percentage composition is defined by dividing the average salary and bonus (or options) by the average total compensation for each year.⁸⁹ As shown in the figure, inflation-adjusted median pay levels doubled from 1980 to 1992 from $946,000 to $1,800,000. The increase in pay primarily reflects the increase in stock option grants, which accounted for nearly half of total aggregate CEO pay by 1992.

Although the takeover and LBO market had been largely shut down by political forces, investors and executives began recognizing that value is created through reducing excess capacity or by reversing ill-advised diversification programs. As emphasized by Holmstrom and Kaplan (2001), stock options allowed executives to share in the value created by internal restructurings: “Shareholder value became an ally and not an enemy”.

3.6.3 Controversial Pay Leads to Sweeping New Disclosure Rules

Between October 13–19, 1987, the Dow Jones Average dropped nearly 800 points (from 2508 to 1738), losing 30% of its value in a week. Executive stock options, which had only recently become an important part of pay, were suddenly underwater. Companies responded by repricing existing options or by significantly increasing the size of their post-crash option grants (Saly, 1994).

The October 1987 crash turned out to be short-lived: by August 1989 the Dow Jones reached an all-time high of 2735, hitting 3000 by July 1990. Stock options issued both before and after the crash were well in the money and becoming exercisable. Large manufacturing firms—still sorting out the excess capacity issues of the 1970s—were downsizing and laying off workers, to the delight of shareholders but attracting the ire of Congress, labor unions, and the media. The combination of valuable options, robust stock markets, and the 1990-1991 recession provided the perfect recipe for a populist attack on executive pay.

The CEO pay debate achieved international prominence during the 1990-1991 recession. The controversy heightened with the November 1991 introduction of Graef Crystal’s (1991) exposé on CEO pay, In Search of Excess, and exploded following President George H. W. Bush’s pilgrimage to Japan in January 1992 (accompanied by an entourage of highly paid US executives). What was meant to be a plea for Japanese trade concessions dissolved into accusations that US competitiveness was hindered by its excessive executive compensation practices as attention focused on the huge pay disparities between top executives in the two countries.⁹⁰

In response to growing outrage, legislation was introduced in the House of Representatives disallowing deductions for compensation exceeding 25 times the lowest-paid worker, and the Corporate Pay Responsibility Act was introduced in the Senate to give shareholders more rights to propose compensation-related policies. The SEC preempted the pending Senate bill in February 1992 by requiring companies to include non-binding shareholder resolutions about CEO pay in company proxy statements,⁹¹ and announced sweeping new rules in October 1992 affecting the disclosure of top executive compensation in the annual proxy statement. Among other changes, the SEC’s new 1992 disclosure rules required companies to produce (a) a Summary Compensation Table summarizing the major components of compensation received by the CEO and other highly paid executives over the past three years; (b) tables describing option grants, option holdings, and option exercises in greater detail; (c) a chart showing the company’s stock-price performance relative to the performance of the market and their peer group over the prior five fiscal years; and (d) a report by the compensation committee describing the company’s compensation philosophy. Overall, the new rules dramatically increased the information available about stock option grants and holdings, and the performance graph cemented the idea that the objective of the firm was to create shareholder value.