An economic, organizational and behavioral model of the determinants of CEO tenure

An economic, organizational and behavioral model of the determinants of CEO tenure

Journal of Economic Behavior and Organization 14 (1990) 363-379. AN ECONOMIC, ORGANIZATIONAL OF THE DETERMINANTS John L. FIZEL, Kenneth North-...

1MB Sizes 0 Downloads 0 Views

Journal

of Economic

Behavior

and Organization

14 (1990) 363-379.

AN ECONOMIC, ORGANIZATIONAL OF THE DETERMINANTS John

L. FIZEL,

Kenneth

North-Holland

AND BEHAVIORAL OF CEO TENURE

K.T. LOUIE

and

Marc

MODEL

S. MENTZER*

7he Pennsylvania State University at Erie, School of Business, Erie, PA 16S63, USA Received

February

1989, final version

received

May 1989

This paper seeks to determine the extent to which chief executive ollicer (CEO) reward systems align the interests of CEOs with interests of stockholders and, therefore, counter the agency problem. The theoretical framework used differs from previous research in two major areas. First, the model focuses on the determinants of CEO tenure rather than CEO compensation, an explicit acknowledgement that most firms do not rely solely on firm performance to determine executive compensation, but rather depend on the ‘going rate’ for CEOs. Second, the model integrates the theoretical constructs from economic, organizational, and behavioral theory. This interdisciplinary model should provide a broader perspective of the environment in which CEOs operate and minimize the pitfalls associated with prior research which has generally adopted the perspective of only a single discipline. The empirical results from this study show that the power structure of an organization and certain key behavioral traits of CEOs are important in the CEO reward system, independent of firm performance. Moreover, these factors even seem to supercede the influence of corporate performance in their impact on CEO tenure.

1. Introduction With the ascendance of the large corporation in the U.S. economy came a proliferation and dispersion of stock ownership which, many theorists contend, have generated a separation of ‘ownership’ and ‘effective control’ [Fama (1980) and Jensen and Meckling (1977)]. The basic premise of these theories is that owners and executives have divergent goals, and if there is no mechanism by which executives are dissuaded from acting in their selfinterests, executives will be free to maximize returns to themselves rather than fulfill the profit-maximizing objectives of owners. If, instead, executives are prevented from acting against the interests of owners, regardless of divergent objectives, separation of ownership and control is inconsequential. This paper seeks to determine the extent to which chief executive officer (henceforth CEO) reward systems align the interests of CEOs with interests of stockholders and, therefore, counter the agency problem associated with separation of ownership and control in large corporations. The model used *We gratefully acknowledge the helpful comments of Beng Soon Chong, Claudia Campbell, and Anthony Krautmann, and the competent research assistance of Nicole Huff and John Walker. We retain the responsibility for any remaining errors. Ol67-2681/90/$03.50

J.E.B.0

C

0

199GElsevier

Science Publishers

B.V. (North-Holland)

364

J.L. Fizel et al., A model of the determinants of CEO tenure

to evaluate CEO rewards differs from previous research in two major areas. First, the model focuses on the determinants of CEO tenure (i.e., length of service as CEO) rather than CEO compensation, an explicit acknowledgement that most firms do not rely solely on firm performance to determine executive compensation but rather depend on the ‘going rate’ for CEOs. If one recognizes compensation to be determined exogenously, the firm’s reward system with respect to the CEO is more accurately characterized as one of contract renewal or termination rather than one of increasing or decreasing compensation beyond the market-dictated compensation range [see Fama (1980), Crain, Deaton and Tollison (1977) and Masson (1971)]. Second, the model integrates the theoretical constructs from economic, organizational, and behavioral theory. The interdisciplinary model should provide a broader perspective of the environment in which CEOs operate and minimize the pitfalls associated with prior research which has generally adopted the perspective of only a single discipline. 2. Literature review An understanding of the factors which comprise CEO reward/incentive systems is vitally important to both theorists and practitioners. And although this interest has spurred the introduction of theories from several disciplines concerning the determinants of CEO rewards, there has been surprisingly little interdisciplinary research. 1 In this section, the central arguments and empirical findings to each of three theoretical perspectives - economic, organizational, and behavioral - are presented.2 But first, the theoretical rationale for using CEO tenure in lieu of CEO dollar compensation is provided. 2.1. Definition

of CEO reward schemes

Typically, analyses of CEO reward systems focus not on tenure but rather on CEO dollar compensation [Ehrenberg and Milkovich (1987)]. Such studies assume that compensation is endogenous, viewing compensation to be solely a function of the executive’s contribution to firm performance. What these studies fail to recognize is that there may be a labor market external to the firm in which demand and supply set an upper and lower bound on executive salaries [Fama (1980), Crain, Deaton and Tollison (1977) and Masson (1971)]. Without independence between the ‘internal’ and ‘Exceptions include Finkelstein and Hambrick (1988), and Ehrenberg and Milkovich (1987). However, these are survey articles and not empirical analyses. ‘The boundaries between the defined perspectives are not necessarily definitive nor is the presentation of the literature within each perspective exhaustive. We do believe, however, that the essence of the existing empirical literature on CEO rewards is captured in our presentatim.

J.L. Fizel et al., A model of the determinants of CEO tenure

365

‘external’ market for executive services, a CEO’s compensation package would be largely determined by the market price of executives’ services, and would not necessarily be related to firm performance. Bacon (1982), for example, points out that firms often use the compensation of other executives in the same industry as an index of compensation for CEOs. In addition, Finklestein and Hambrick (1988) argue that a bargaining range for compensation may exist, but that compensation has less variability because firms focus on compensation figures which are generally in line with the marketp1ace.j Although Bacon (1982) and Finklestein and Hambrick (1988) argue that firms attempt to set compensation figures which are generally in line with the marketplace, thereby implying the existence of a narrow bargaining range for CEO compensation, the size of the bargaining range is irrelevant. As long as the firm feels that it cannot pay below (and does not need to pay above) the lower (and upper) limits, then this market-dictated salary range, whatever its size, will mean that the link between executive performance and executive pay within the firm has been broken. The only recourse available to the firm with which to punish poor (or reward good) executive performance is to terminate (or renew) the executive’s contract. Further support for the use of tenure has recently emerged in the theory of ‘tournaments’ [Lazear and Rosen (1981) and Rosen (1986)]. This theory likens executive compensation schemes to a series of contests or competitive lotteries. A major objective of this theory is to explain the substantial concentrations of rewards in the top echelons of firms, whether it is in athletic contests or in the labor market for executives. In this framework, senior managers (e.g., corporate vice presidents) are viewed as competing in a tournament to become CEO. Each senior manager may be willing to sacrifice some earnings (associated with individual performance) which are then placed in the ‘prize’ for which they all compete and which eventually becomes the CEO’s salary. Rosen (1986) has demonstrated that in ‘elimination tournaments’ of this sort, a disproportionately large extra reward to the overall winner is necessary to maintain incentives for tournament survivors at each step in the competition so that they do not ‘rest on their laurels’. Additionally, Lazear and Rosen (1981) have shown that in such tournaments, a worker’s incentives to invest in productivity-enhancing skills increases with the spread between the winning and losing ‘prizes’. Consequently, the theory of tournaments provides a possible reason for the large differences observed between the compensation of CEOs and that of other senior executives. However, to the extent that the theory of tournaments suggests that the compensation of chief executives is not necessarily related to short-run firm 3From the worker’s payment since patterns of well-being.

perspective, of economic

Frank (1985) shows formally the importance of relative behavior are often influenced by interpersonal comparisons

366

J.L. Fizel et al., A model of the determinants

of CEO tenure

performance, the theory reinforces our argument that it is theoretically unsound to use CEO pay to gauge the extent to which executives are responsive to stockholder interests. In order to test whether CEOs are indeed subject to constraints on their behavior, it is necessary to examine other incentive mechanisms whereby CEOs are rewarded or punished. We believe that if CEOs do face constraints on their behavior, the primary mechanism whereby this is achieved is through extension or termination of the service contract as CEO. This is our justification for using CEO tenure rather than CEO compensation as the dependent variable. 2.2. Economic studies Economists have focused principally on firm profit and sales performance as determinants of CEO com~nsation. The major hypothesis is that CEOs, acting as agents for stockholders, will maximize profits unless there is a separation of ownership and ‘effective’ control, in which case CEOs might maximize sales or growth [Baumol (1959)]. The studies which focus on dollar compensation have had divergent results. Some studies [e.g. Murphy (1985), Argarwal (1981)] find that compensation is strongly tied to the profitability of the firm, inferring profit maximization as the behavioral objective of executives. Other studies [see Ciscel (1974), MC&ire, Chiu and Elbing (1962)] find that sales are most instrumental in determining compensation and infer sales or growth maximization as the goal of executives. Some studies [Hirschey and Pappas (1981), Ciscel and Carroll (1980)] find both profits and sales to be significant determinants of compensation and infer sales maximization with a minimum profit constraint. Although these studies typically use sales level to capture sales performance, several profit measures have been used. These include the level and/ or change in return on assets, return on equity, earnings per share, stock valuations, and more. The variety in both data and methodologies makes direct comparisons of earlier studies difficult and tends to impede a resolution to the debate concerning the profit maximization hypothesis.4 Divergent results are also obtained when the focus of the analysis changes from dollars to CEO tenure. For example, Crain et al. (1977, p. 1374) found that tenure was very responsive to the profrt performance of firms and concluded ‘that non-profit maximizing behavior in the private sector is not typically tolerated in the market for executive services’. In contrast, Salancik and Pfeffer (1980) found that CEO tenure was positively related to profits only in firms with low levels of CEO discretion. Similarly, Allen and Panian 4For surveys (1987), Murphy

of the literature, see Finkelstein and Hambrick (1985). and Ciscei and Carroli (1980).

(1988), Ehrenberg

and Milkovich

J.L. Fizel et al., A model of the determinantsof CEO tenure

367

(1982, p. 546) report that profits had only a marginal impact on tenure and stated, in conclusion, that ‘pro~tability is not the only goal of the large corporation’. 2.3. ~rganiza~ionai

studies

Organizational studies emphasize factors which are significant to the design and implementation of a CEO compensation package. The principal elements of these analyses are the size and power structure of the firm. Theoretically, the impact of firm size on CEO reward systems is ambiguous since it depends on the relative strengths of two opposing tendencies. On one hand, the gap between management and ownership appears to be a positive function of firm size. That is, stockholders in larger firms are much further removed from actual managerial decisions than stockholders in smaller firms. This may increase the difficulty for any group of dissatisfied stockholders to exert their influence to reduce the dollar compensation of the CEO or to remove the CEO. Furthermore, Galbraith (1971) has argued that a corporation’s ability to control its environment may lead to a greater degree of managerial autonomy and corporate stability. This ability, Galbraith continues, is often positively related to firm size. Together, these arguments imply that larger firm size is associated with more lucrative contracts and/or longer tenure for the CEO. On the other hand, larger corporations are more visible to the public at large including public officials (potential regulators), the news media, and the investment community. Consequently, it is reasonable to expect the degree of scrutiny given a firm, as well as the extent of the firm’s response to such scrutiny, to be a positive function of firm size [Pfeffer and Salancik (1978)]. This would suggest a higher rate of CEO turnover, and hence a lower level of observed tenure, among CEOs of larger firms. There may also be structural differences between large and small corporations that reinforce this effect. For example, Gomez-Mejia, Tosi and Hinkin (1987, p. 52) have argued that firm size may be positively associated with job complexity. They note that as organizational structures become larger and more complex, the CEO’s ‘work specifications’ may become more demanding, In the context of human capital theory [Becker (1964)], this implies that CEO compensation should increase commensurately with the additional work requirements demanded by the job. In a model of tenure, to the extent that firm size is related to job complexity and additional human capital requirements, one would expect a higher rate of turnover and lower average tenure among CEO’s of larger firms, with compensation held constant. Recently, analysts have suggested that the power structure of a firm may moderate both firm performance- and firm size-CEO reward relationships. One genre of these analyses [e.g. Gomez-Mejia, Tosi and Hinkin (1987) and

368

J.L.

Fizel et

al.,

A model of the determinants

of CEO tenure

Dyl (1988)] hypothesizes that CEO rewards will be directly related to firm performance unless the CEO can gain control of the firm. The greater the influence of a CEO, the more likely the CEO will be able to enhance compensation and/or avert replacement, regardless of firm performance. Empirical results tend to support the premise that the power structure of a firm is an important determinant of CEO rewards. For example, both Weisbach (1987) and Salancik and Pfeffer (1980) lind that the more powerful the CEO, the less likely it is that the CEO will be dismissed. Moreover, each of these studies finds that CEO employment is only affected by firm performance in firms with low levels of managerial power.

2.4. Behavioral

studies

According to the behavioral perspective, characteristics of the CEO are instrumental in devising an appropriate compensation package for the CEO and in evaluating the performance of the CEO. The CEO’s personal characteristics are important because they may be used as ‘indicators’ of education, experience, and potential success. In the human resources literature, this is often called the ‘screening hypothesis’ because with imperfect information, personal characteristics are often used as proxies for other (unmeasurable) qualities that are associated with successful performance. In a study of the determinants of executive compensation, for example, Hogan and McPheters (1980, p. 1062) argued that in setting compensation, ‘a board of directors will look beyond current performance variables to certain individual characteristics which they feel constitute a forecast of a successful manager’. CEO age is often used to indicate a general level of experience or training, while CEO years of service within a firm is often used as a measure of firm-specific training or acquisition of knowledge and skills that are unique to the CEO’s firm. Both are expected to be directly related to CEO compensation and tenure. However, several studies [e.g. Coughlan and Schmidt (1985) and Weisbach (1987)] have found turnover to be higher for CEOs as age increases. Because voluntary departures of CEOs are often due to retirement, and involuntary departures may be more likely to occur when they can be disguised as voluntary retirements, the screening hypothesis for the age variable may be masked by cases of ‘real’ and ‘forced’ retirement. Another CEO characteristic is the level of compensation.5 Increased dollar compensation may be an incentive which fosters commitment to the organization either by expressing confidence in the abilities of the CEO or by 5This discussion focuses only on models However, we include this theoretical construct section of the paper, that a tenure model compensation as the dependent variable.

where CEO tenure is the dependent variable. because we believe, as we argue in an earlier subis more appropriate than models using dollar

J.L. Fizel et al., A model of the determinants

of CEO tenure

369

reducing the likelihood that a CEO can find a superior (better-paid) alternative position [Coughlan and Schmidt (1985)]. An alternative hypothesis suggests that the complete reward system for executive performance may include compensation and tenure. For example, a high-paying firm may be a low-retention firm. Given the possibilities of such tradeoffs in the reward system, one would then expect a negative relationship between compensation and tenure.

3. Empirical model: Specification

and variable definitions

Each of the theoretical perspectives has made useful contributions to the analysis of CEO reward systems. However, the inconclusive pattern of results may be partially attributable to incomplete specification of the models. If models of CEO reward schemes fail to integrate the arguments of any of the above described perspectives, omitted-variable bias is possible. A model which integrates the economic, organizational, and behavioral determinants of CEO reward systems is presented in this section. This model should minimize the omitted-variable problem and provide a broader perspective of the CEO reward system than prior research. A key concern of this model is the relative significance of the various theoretical constructs in explaining variations in CEO tenure. For example, if the economic variables are more important than the organizational variables such as firm size, indicating that executives are removed when their firms fail to generate adequate profits, it would be an indication that CEO reward systems align the interests of CEOs with interests of stockholders. But if tenure is more closely related to firm size, one may infer that profit maximization is not the primary CEO objective. Similarly, if the organizational power structure variables are significant, CEOs will be more likely to avert replacement regardless of firm performance. Therefore, imbedded in the appraisal of the relative significance of economic and organizational variables is the question of how immune executives of large corporations are to profitability constraints. An assessment of the relative significance of the different theoretical perspectives may also indicate, unlike the scenario just described, complementarity between variables. For example, an economic variable, profits, and a behavioral variable, age, may be significant simultaneously. Such results would indicate that CEO tenure is enhanced by both good firm performance and the experience level of the CEO. 3.1. The dependent

variable: Tenure

Tenure is the number the firm as CEO.

of years as of 1985 that

the individual

had served

370

J.L. Fizel et al., A model of the determinants qf CEO tenure

3.2. Economic variubks

Profit and sales performance are the economic variables in this analysis. While it is important to use profit rates and not profit size, the theoretical and empirical literature on executive compensation is less clear on whether absolute or relative performance measures should be used. Moreover, tenure is cumulative. Therefore a single year of firm performance is not likely to capture all the elements of firm performance imbedded in the historic and multiple CEO retention/termination decisions. To address these issues, we use both absolute and relative measures of the previous five years of firm profit and sales performance. The absolute measures of performance are the five-year average return on equity (ROE), five-year average annual growth rate in earnings per share (H’S), and the five-year annual growth rate in sales (SALES). The first of two sets of relative measures of performance include the difference between firm performance and industry median performance using the five-year average return on equity (ROEDIFF), ftve-year average annual growth rate in earnings per share (~PSD~~F), and the five-year annual growth rate in sales (SALESDIFF). The second set of relative measures of performance are ROEDUM, EPSDUM and SALESDUM, which are dichotomous variables taking a value of one if the firm performance over the previous five years, in the data described above, was greater than the performance of the industry median over those same years. If the CEO reward system aligns the interests of the CEO with those of stockholders, a positive relationship between tenure and profits is expected. If instead, CEOs maximize sales, a positive relationship between tenure and sales would be expected. If both profit and sales variables have positive coefftcients, CEOs can be said to maximize sales with a minimum profit constraint [Baumol (1959)]. But, if each of the performance variables is insigni~cant, then one might ask, as Crystal (1988a, 1988b, p. 62) did, ‘where’s the risk in CEOs’ rewards? Crystal’s response was that ‘often there isn’t any’ and that ‘with a few notable exceptions, the CEOs suffer little or not at all no matter what happens to the company’. 3.3. Organizntional variables The organizational literature has focused primarily on the role of organizational size and power structure in the CEO reward system. Firm size (SIZE) was operationalized by firm sales, assets, and employees measured at their 1985 levels.6 ‘The three empirical constructs results. Consequently, we include empirical results OF the model.

of firm size provided identical and, therefore, redundant only the sales proxy for size in the table presenting the

J.L. Fizel et al., A model of the determinants qf CEO tenure

371

CEO power will depend in part on the vigilance of the board of directors in monitoring CEO performance and enforcing the preferences and objectives of stockholders. The vigilance of the board is related to its composition. Inside directors, who are employees of the firm and subordinates of the CEO, may be reluctant to remove a CEO when firm performance suffers. In contrast, outside directors, those with no other managerial function within the firm, are more likely to exercise their authority independently and objectively. The variable OUT, calculated as the percentage of the board of directors composed of outsiders, is included to test the hypothesis that increasing the number of outsiders on a board reduces the CEO’s power to avert replacement. While percentage of outside membership is expected to have an impact on CEO tenure, there may be a minimum level of outside board membership before the board will be able to exert any signi~cant influence on CEO tenure. For example, Weisbach (1987, p. 10) found this threshold level of outsiders to be about 60 percent of the board before CEO succession is significantly affected. The squared term of outside board membership (O~~SQ) is included to account for this possible non-linearity. Also included in the model is a variable, CHAIR, used to capture whether the CEO is also the chair of the board of directors. Because a CEO who is also chair of the board wields considerably more power than one who is not, CHAIR is another measure of the degree of influence the CEO has over the firm’s governance structure. In effect, the CEO/Chair oversees all decisions, including those pertaining to renewal or temination of the CEO’s own contract. Thus, CEOs who also chair their boards may be expected to experience greater tenure on average.

The behavioral literature concentrates on the experience and commitment of the CEO. As described earlier, CEO age is often used as an indicator of the level of general experience attained by the CEO, and CEO service within a firm is used as an indicator of the level of firm-specific experience or knowledge attained by the CEO. Both general experience, measured as age of CEO at promotion (AGE), and firm-specific experience, measured as CEO service to the firm prior to promotion (SERV), are expected to increase tenure. CEO age may also capture a retirement effect. If this effect dominates the impact of general experience, age and tenure would be negatively related. Total dollar compensation to the CEO is used as a proxy for the CEO’s commitment to the firm. Higher compensation may be viewed as a reward for a job well done or increased opportunity cost of seeking an alternative

312

J.L. Fizel et al., A model of the determinants of CEO tenure

position. In either case, a direct relationship and tenure is expected.

between

compensation

(COMP)

3.5. Model specljkation and sample Based on the theoretical arguments of section 2, a model nants of corporate CEO tenure may be specified as follows: Tt = B, + B, ECONOMIC,

of the determi-

+ B, ORGANIZATIONAL,

+ B, BEHAVIORAL,+u,, where = length of tenure as CEO of the firm; = performance of the CEO’s firm ECONOMIC CONSTRUCTS (ROE, ROEDIFF, ROEDUM, EPS, EPSDIFF, EPSDUM, SALES, SALESDIFF, SALESDUM); ORGANIZATIONAL CONSTRUCTS= firm size and power structure (SIZE, OUT, OUTSQ, CHAIR); = characteristics and commitment of BEHAVIORAL CONSTRUCTS the CEO (AGE, SERV, COMP); and = a random error term. u T

Past research on topics relating to CEO rewards has been hampered by the restriction-of-range problem. For example, because the Fortune 500 ranks firms by sales, the companies on that list are relatively homogenous in terms of sales, and sales is less likely to be a significant predictor.’ It is therefore desirable that a data set be used whose composition is comparatively heterogenous, making it easier to identify which aspects of economic performance and organizational size, if any, are significant factors. The Forbes (1985) annual report on executive compensation is ideal for this purpose since its 785 companies represent the union of four sets: The 500 largest companies by assets, by sales, by profits, and by market value. The result is a relatively heterogenous list of companies, including many whose small sales (but large assets) would disqualify them from inclusion in the ‘Hogan and McPheters (1980, p. 1064), for example, report that profitability is significant and sales insignificant in explaining executive compensation when lirms are chosen on the basis of their rank in sales. However, if firms are chosen on the basis of their rank in profits, then sales not profits - become significant in explaining executive compensation.

J.L. Fizel et al., A model 01 the determinants of CEO tenure

313

Fortune 500. Also, whereas only manufacturing firms comprise the Fortune 500, the more heterogenous Forbes list includes many service organizations. Because the tenure model focuses on the retention/replacement question with respect to CEOs, firms were chosen for the sample in such a way that there would be a balanced representation of firms which had replaced their CEOs in the period of analysis, as well as firms which had retained their CEOs during the period. To do this, the Forbes compensation data for 1985 and 1984 were compared, which revealed that 634 companies had the same CEO over the period, 72 companies had a new CEO, and 79 companies were removed from the list. The 72 companies in which the CEO was replaced were included in the sample. The sample was stratified by randomly selecting 72 other companies (from among the 634 which had retained their CEOs) representing industries which matched the industries of the firms already selected. Therefore, the initial sample included 144 firms. Ultimately, however, complete information on each variable was available for only 136 of these 144 firms.8

4. Empirical model: Estimation

and analysis

Table 1 presents the results of the ordinary least squares estimation of alternative specifications of the determinants of CEO tenure. Although the principle of pay for performance receives support from both theoreticians and practitioners concerned with CEO reward systems, the results of this study imply that, in reality, the principle is not implemented. The parameter estimates for the economic variables indicate that neither profit nor sales performance is a significant driving force behind CEO retention and termination decisions, and ultimately, tenure. These estimates are robust across all firm performance definitions and model specifications. The results in table 1 also show that firm size coefficients are consistently negative and statistically significant. This suggests that the greater visibility of the large firm and/or the greater job demands for CEOs of large firms increase the turnover rate among CEOs. Moreover, the results refute the countervailing hypothesis that larger firms permit a greater degree of managerial autonomy and, hence, longer CEO tenure. Perhaps the most significant feature of the size variables is that their influence on tenure is dramatically different than their influence on compensation. Most compensation studies have found assets, employee numbers, and especially sales and percentage change in sales to be insignificant or positive, *The entire data set was culled from the following sources: (1) Business Week, ‘Scoreboard Special’, Annual Special Issues (1984-1985); (2) Forbes, ‘The Forbes 500s’. Annual Issues (1984 1985); (3) Forbes, ‘Who Gets the Most Pay’, June 3, 1985, 135 (12), pp. 116153; (4) Forbes, ‘Annual Report on American Industry’, January, 1985; and (5) Moody’s Manuals series (1981-1985).

314

J.L. Fizel et al., A model of the determinants

I

I

I

of CEO tenure

I

vi-

22 I

/

I

I

66

I

0.001* (1.85)

0.002* (1.79)

TCOMP

0.246 0.193 4.588

“Absolute value of t-statistics ***Significant at 0.01 level. **Significant at 0.05. *Significant at 0.10.

R squared Adj R squared F-statistic are reported

0.250 0.196 4.655

-8.813 (1.42)

0.108*** (2.41)

0.105** (2.34)

SERV

- 8.522 (1.36)

0.181** (2.52)

0.178** (2.47)

AGE

CONSTANT

2.018** (1.73)

1.989** (1.70)

- 24.766** (2.03)

CHAIR

-24.578** (2.01)

0.106*** (2.36) 0.002** (1.96)

0.105** (2.34) 0.002** (1.95)

0.246 0.192 4.577

-8.155 (1.34) 0.247 0.193 4.597

-8.417 (1.37)

0.171** (2.42)

0.175** (2.47)

0.172** (2.45)

0.245 0.192 4.554

0.254 0.201 4.775

- 6.948 (1.14)

0.001** (2.23)

0.001** (2.08) - 7.798 (1.28)

0.103** (2.32)

0.104** (2.32)

0.182*** (2.58)

1.772% (1.52)

1.97s** (1.69)

1.929** (1.66)

19.334 (1.27)

-0.0002*** (3.38)

2.003** (1.71)

22.407 (1.48)

-0.0003*** (3.39)

- 22.459* (1.83)

- 24.206** (1.97)

22.048 ( 1.46)

-0.0003*** (3.27)

- 24.887** (2.04)

-24.154** (1.94)

22.012 (1.45)

-0.0002*** (3.27)

in parentheses.

22.527 (1.49)

22.192 (1.47)

OUT

OUTSQ

- 0.0002*** (3.17)

-0.0002*** (3.07)

SIZE

Y m

376

J.L. Fizel et al., A model of the determinants of CEO tenure

but not negative as found here. Indeed, the compensation studies often conclude that the compensation-enhancing effects of size/sales imply support for the premise that growth/sales maximization takes priority over profit maximization as the economic objective for CEOs. This is clearly not the case when using CEO tenure. Increasing the size of the firm increases the riskiness of CEO retention and reduces tenure. Organizational power structure variables are also integral to the tenure model. The ‘outsider’ variable (along with its squared term) produced coefficients which support the contention that CEO tenure is a non-linear function of the percentage of outside board members. In contrast to Weisbach (1987) who found that outsiders on a board did not influence CEO turnover until they comprised at least 60% of the board, these results indicate that boards composed of at least 1% outsiders will tend to reduce CEO tenure. Because boards rarely consist of more than 20 to 25 members, the pragmatic implication is a monotonic decline in tenure as the percentage of outsiders increases. In addition, the coefficients on the CHAIR variable indicate that a CEO who is also chair of the board of directors is likely to remain in office fonger than CEOs who are not chair of their firms’ boards. These results again indicate the highly important role of organizational power structures on CEO retention and removal. In general, as CEO control over corporate governance increases (resulting in greater executive discretion), the average level of tenure also increases. Finally, the estimated coefficients for the behavioral variables were consistent with a priori expectations. Age at promotion and length of service to the firm prior to promotion were both positively related to CEO tenure and statistically significant in all specifications of the model. The results lend support to the ‘screening hypothesis’ that age reflects general experience and length of service reflects arm-speci~c experience, both of which should enhance the tenure of a given CEO. Although Hogan and McPheters (1980) had found that general experience also increased CEO dollar compensation, they did not find that firm experience was a significant factor in influencing compensation. The coefficient on the compensation variable (COMP) was also positive and significant in each of the model specifications. These coefficients are consistent with the hypothesis that higher compensation implies higher opportunity costs for giving up a CEO position and, hence, provides incentives for greater CEO commitment to the organization. The coefficients also suggest that firms do not trade off compensation and tenure in the CEO reward system. The results of this study support the need for an integrated framework to study the determinants of CEO tenure. Variables derived from the theoretical constructs of each of the three perspectives were statistically significant.

J.L. Fizel et al., A model of the determinants of CEO tenure

317

Moreover, most signs of significant variables were consistent with a priori hypotheses suggesting that the different pers~ctives are complementary. However, the results do indicate that the organizational and behavioral factors had a larger role in explaining the variations in tenure than did the economic variables. As a result, one should be particularly cautious in accepting the results of studies which have focused exclusively on firm performance variables. 5. Summary and conclusions This paper investigated the performance of chief executive officers (CEOs) and the factors that motivate their performance by examining the determinants of tenure for the CEOs of 136 large U.S. corporations. One key methodological contribution of this paper is the use of tenure instead of CEO compensation as a proxy to gauge CEO performance. Although most empirical work uses compensation, these studies neglect the role of external labor markets in the dete~ination of CEO rewards. Once one accounts for the possibility that these external markets may dictate a limited range within which compensation may vary, the primary means by which a CEO is able to be rewarded (punished) for superior (inadequate) performance is through contract renewal (termination). Consequently, tenure becomes more appropriate than compensation as a means of measuring CEO performance. A second methodological contribution of this paper is the development of a model of CEO tenure which integrates hypotheses from the economic, organizational and behavioral literature. The results of this study indicate that serving as both CEO and chair of the board, age, length of service with the firm, and total compensation all have a positive effect on a CEO’s length of tenure. On the other hand, larger firms and those firms with boards composed of a greater proportion of outside directors tend to have CEOs that experience shorter lengths of tenure. The results have three important implications in light of earlier research. First, the use of a tenure model provides many results which are contrary to those found in compensation models. Compensation models often find economic variables such as profit performance, sales performance, or both to be significant explanatory variables. These variables were insignificant in the tenure model. Two likely reasons for the contrasting performance of the tenure and compensation models are that: (1) compensation is a theoretically faulty measure of executive performance, as described previously, and (2) compensation studies often rely exclusively on profit and sales variables and are beset by omitted-variable biases. In any case, future research should aim at reconciling these conflicting results. Second, the organizational power structure is a major explanatory factor in determining executive tenure.

378

J.L. Fizel et al., A model oJ the determinants oJ CEO tenure

Therefore, studies of executive reward systems should account for the organizational environment in which rewards are determined and, more specifically, acknowledge the input the executive has in the decision-making process of the organization. This study found that the more influence the CEO has in organizational governance, the greater the average CEO retention rate. Finally, behavioral variables are significant factors in CEO retention/replacement decisions. Apparently, in a world of imperfect information, directors evaluate CEOs by readily accessible indicators of future CEO performance. These factors include items such as age and service to the firm. The overall implication of this study is that the power structure of an organization and certain key behavioral traits of CEOs are important in the CEO reward system, independent of firm performance. Moreover, these factors seem to supercede the influence of corporate performance in their impact on CEO tenure.

References Allen, M.P. and S.K. Panian, 1982, Power, performance and succession in the large corporation, Administrative Science Quarterly 27, 538-547. Argarwal, N.C., 1981, Determinants of executive compensation, Industrial Relations 20, 3646. Bacon, J., 1982, Corporate directorship practices: The compensation committee, Report no. 829 (The Conference Board, New York). Baumol, W.J., 1959, Business behavior, value and growth (Harcourt-Brace, New York). Becker, G., 1964, Human capital (National Bureau of Economic Research, New York). Ciscel, D.H., 1974, Determinants of executive compensation, Southern Economic Journal 40, 613-617. Ciscel, D.H. and T.M. Carroll, 1980, The determinants of executive salaries: An econometric survey, The Review of Economics and Statistics 62, 7-13. Coughlan, A.T. and R.M. Schmidt, 1985, Executive compensation, management turnover, and lirm performance: An empirical investigation, Journal of Accounting and Economics 7, 43-66. Crain, W.M., T. Deaton and R. Tollison, 1977, On the survival of corporate executives, Southern Economic Journal 43, 1372-1375. Crystal, G.S., 1988a, The wacky, wacky world of CEO pay, Fortune 6,68%78, June. Crystal, G.S., 1988b, Where’s the risk in CEO’s rewards? Fortune, 19, 62-66, Dec. Dyl, E.A., 1988, Corporate control and management compensation, Managerial and Decision Economics 9, 21-26. Ehrenberg, R.G. and G.T. Milkovich, 1987, Compensation and performance, National Bureau of Economic Research, Inc., Working paper no. 2145. Fama, E. 1980, Agency problems and the theory of the firm, Journal of Political Economy 88, 288-306. Finkelstein, S. and D.C. Hambrick, 1988, Chief executive compensation: A synthesis and reconciliation, Strategic Management Journal 9, 543-558. Frank, R.H., 1985, The demand for unobservable and other nonpositional goods, The American Economic Review 75, 101-l 16. Galbraith, J.K., 1971, The new industrial state, 2nd ed. (Mentor, New York). 1987, Managerial control, performance and Gomez-Mejia, L.R., H. Tosi and T. Hinkin, executive compensation, Academy of Management Journal 30, 51-70. Hirschey, M. and J.L. Pappas, 1981, Regulatory and life cycle influences on managerial incentives, Southern Economic Journal 48, 327-334.

J.L. Fizel et al., A model of the determinants of CEO tenure

319

Hogan, T.D. and L.R. McPheters, 1980, Executive compensation: Performance versus personal characteristics, Southern Economic Journal 46, 106&1068. Jensen, M.C. and W.H. Meckling, 1977, Theory of the firm: Managerial behavior, agency costs and ownership structure, Journal of Financial Economics 3, 305-360. Lazear, E.P. and S. Rosen, 1981, Rank-order tournaments as optimum labor contracts, Journal of Political Economy 89, 841-864. Masson, R.T., 1971, Executive motivations, earnings and consequent equity performance, Journal of Political Economy 79, 1278-1292. McGuire, J.W., J.S. Chiu and A.O. Elbing, 1962, Executive incomes, sales and profits, American Economic Review 52, 7333761. Murphy, K.J., 1985, Corporate performance and managerial remuneration, Journal of Accounting and Economics 7, 1142. Pfeffer, J. and G.R. Salancik, 1978, The external control of organizations (Harper and Row, New York). Rosen, S., 1986, Prizes and incentives in elimination tournaments, The American Economic Review 76, 701-715. Salancik, G.R. and J. Pfeffer, 1980, Effects of ownership and performance on executive tenure in U.S. corporations, Academy of Management Journal 23, 653-664. Weisbach, M.S., 1987, Outside directors and CEO turnover, Working paper (University of Rochester, New York).