An earlier version of the paper was presented at the 2002 Academy of International Business Conference In Shanghai, In July 2002. Mortem Benzene’s and Paul Bangle’s comments on an earlier version are much appreciated. 2 1 Introduction The principal idea of this paper is the following: there is some research, predominantly in the field of management research, suggesting that power (which has been defined In alternative ways) generates higher Income to managers.
However, the tests performed in many of the previous studies (see e. G. , Finniest and Humpback (1989), (1996), Lambert, Larker and Weight (1993), Humpback and Finniest (1995) and Barkeep and Openings (1988)) are not very strong. This is partly because they make use of potentially questionable measures of power, partly due to the study designs adopted. Three types of power measures have been used. The first is the manager’s own shareholdings, which is hypothesized to have an (1996), Lambert, Larker and Weight (1993)).
A second measure is the chief executive’s ability to appoint outsiders on the board (Wade, Reilly and Candidate (1990), Lambert, Larker and Weight (1993), Core, Holocausts and Larker (1999), Hillock (1997)) as proxies by the insiders’ share of board members or by CEO duality the CEO
Need essay sample on "Management"? We will write a custom essay sample specifically for you for only $ 13.90/page
The power measure is typically entered as a right hand side variable into a cross- sectional compensation equation with rather few controls, none of which (except for Frey and Sucker’s (1999) analysis) are individual characteristics. And yet, you would expect that human capital measures like experience, tenure and education would constitute important explanatory variables in a more complete analysis of managerial ay. It is, moreover, most likely that these same individual characteristics are instrumental in explaining differences in the amount of power the managerial employee possesses. Consequently, an interesting question is whether the results according to which there is a positive relationship between power and individual earnings still holds if we also cater for human capital. Some of the factors conducive for managerial power are probably unobserved (to the econometricians), but can be conceived of as individual fixed effects.
Hence, a longitudinal analysis accounting for mime-invariant unobservable, is called for. 1 As a matter of fact, most investigations of CEO or other top managers’ pay do not typically control for individual characteristics, save CEO tenure, which is included in some of the studies. The main reason is simply that they are lacking from the data sources used; for a survey of the literature, see Murphy (1999). 3 As pointed out by Frey and Cheer (1999), there is an additional reason for why introducing human capital is potentially significant.
An alternative hypothesis to the managerial power explanation is that there should be no relationship. This is cause one can conceive of both monetary income and power as goods, and accordingly, both are arguments in the manager’s utility function. If this is the case then the compensating wage differentials hypothesis states that the manager is prepared to trade some additional income in exchange for gaining more power.
In other words, more power does not necessarily imply higher income, but rather the opposite. In equilibrium the marginal value of a given amount of additional power equals the marginal value of a further increase in monetary compensation. This does not, however, imply that executives with more power are not observed to receive Geiger compensation as well, but it is then due to differences in managerial talent and the fact that more talent yields both more power and higher compensation.
Hence, a possible reason for why earlier studies have found a positive relationship is that they have not properly controlled for the factors that are conducive for obtaining more power. This paper addresses this issue. Than in previous studies by: testing it against the compensating wage differentials explanation, using both cross-sectional and longitudinal data, and adopting two alternative measures of managerial power; a frequently used indirect nee, and a more direct power indicator. So, this is what I do.
For a cross-section of 2,146 managerial employees in Danish medium-sized and large private sector firms in year 1995, I estimate a log total compensation equation for managers with age, tenure, gender, education of the manager, log turnover, regional and industry affiliation of the firm and two power measures as explanatory variables. 2 The latter are: (I) the oft used log of the number of subordinates to the manager in question, and an index which is constructed to measure the authority each managerial employee in he firm has regarding the choice of actions to achieve certain goals.
Earlier studies estimation managerial pay equations utilizing the same data source are Eriksson (2000) and Eriksson and Lausanne (1999) 4 I next estimate the same pay equation augmented with individual fixed effects on a panel data set covering the years 1992-95. This is a more powerful test in the sense that it estimates the overpay relationship from changes in these variables and that it on top of the managers’ human capital also accounts for unobserved but time- invariant managerial characteristics. The next section gives a brief description of the ATA used.
In section 3 the main results of my econometric analysis are presented. The paper closes by summarizing its key findings. 2 Data description The data set used in this paper has been constructed from data obtained from the These provide detailed information about each manager’s individual characteristics (age, gender, education and tenure), compensations, currently held position bib title and functional area), and some characteristics of the companies in which he is employed (size, location, type and industry).
In addition, the consulting firm collects information about each manager’s authority. This will be discussed in more detail below. From this data set I extract three samples. The first one is a cross-section of managerial employees in 1995, and the other ones are unbalanced panels containing information about 2,164 (4,927) managers in 574 (107) Danish firms, making up 8,394 (8,724) individual-year observations taken during the four-year period 1992 to 1995. The two panels only differ with respect to the availability of information about the two power variables in the data set.
Thus, they overlap to a large extent. I now turn to discuss how the key variable in the following analysis, managerial rower, is measured. The first and more conventional measure of power is the number of subordinates to each managerial employee. This information is available in the current data set for all managers 3 Total compensation, which is the measure of managerial income I use, is the sum of base salary, paid bonuses and commissions, and the employers’ contributions to pension funds (all three components are available as separate variables).
Compensation does not include stock awards or stock options. The latter were virtually non-existent during the time period under study, so their absence should to affect the results obtained. 5 except for the chief executive officers, for whom the number of subordinates is set equal to the total number of employees in the firm. 4 The disadvantage of this measure is that the difference between the values it takes for the CEO and for other managerial employees is for obvious reasons relatively large. This is, of course, especially the case in large firms.
In addition we know from the literature on CEO pay that chief executives in bigger firms receive higher compensation (Rosen (1992)). 5 This means that the estimates from studies employing the number of subordinates s the power variable may actually be capturing also other influences than managerial power. A prime candidate is managerial talent, which according to Rose’s (1982) talent allocation model increases as we move up in the corporate hierarchy and as the size and complexity of the organization increase. In order to check the robustness of the results, I therefore also estimate the compensation equations with the subordinates measure on a smaller sample that excludes the Coo’s. As a further check I also control for the formal position of the managerial employee and his membership in board of directors (and/or of the concern, if the rim is part of one) and the top management team. CEO duality is forbidden by law in Danish stock companies with a capital requirement exceeding 0. 5 million DISK (about 70. 000 EUROS). 7 Hence, this is not a candidate for an additional power variable. 1999) emphasizes the importance of distinguishing between formal and real authority. Formal authority refers to the power to choose goals and the responsibility for the outcomes of actions taken by the persons themselves as well as by agents at lower levels in the hierarchy. Real authority is the power to determine which actions to choose to achieve certain goals. The two types of authority differ also with respect to delegation; formal authority cannot be delegated, whereas real authority can be. In this paper I use an index measure of real authority as my second power variable. 4 As a matter of fact most of the earlier studies focus solely on Coo’s for whom it is natural to think all the firm’s employees as their subordinates. A notable exception is Lambert, Larker and Weight (1990). 5 Moreover, the labor economics literature has documented an employer size effect on wages for employees in general. See Torsos (1999) for evidence and tests of alternative interpretations. 6 For empirical tests of this hypothesis, see Angel and Pumas (1997). 7 Banks, financial and insurance companies are excepted from this rule.
In the data set there are only seven Coo’s, who also serve as chairmen, and most of them are managing smaller firms. 8 Eriksson (2000) adds both formal and real authority to the compensation equation. Both turn out to be statistically significant. Gradual and Jamaica (1999) examine the influence of managerial responsibility and focus more specifically on how the pay- performance elasticity varies with responsibility. 6 The authority index is constructed from an evaluation system for positions used by the consulting firm, and is based on four factors.
The first factor is the complexity of the problems to be solved independently by the position holder; the second is the degree of freedom in decision making; the third is the intensity and complexity of the communication of the results of the tasks performed; and the fourth is the degree of responsibility in managing subordinates or solving specialist tasks. The index classifies Jobs into 15 levels of authority, of which the seven highest are relevant for managerial employees.
This detailed classification is, however, available for a part of the data set only, whereas an index, which aggregates the classification into six levels, is available for the whole data set. Consequently, this is the index I adopt as the second measure of each manager’s power. 3 Estimation results that the individual characteristics like age, education and gender attach significant coefficient estimates that are generally in line with what has been found for other categories of wage earners. 9 For executive pay differentials also some firm attributes tater.
Thus, manufacturing firms pay their managers less than in services and trade (the omitted category) companies. Managerial employees in firms located in the capital area receive higher compensation than their colleagues in other parts of the country. Finally, firm size, as measured by the turnover, increases the managers’ pay. (This holds true also when the number of each manager’s subordinates is entered as an additional explanatory variable. ) More importantly, managerial compensation is found to be increasing in the number of the executive’s subordinates, which is the iris of the managerial power variables.
Consequently, the estimation results support previous findings of a positive pay and power relationship, but the results of the current analysis differ in so far that I have also controlled for some individual characteristics that may be conducive for gaining managerial power. When I estimated the 9 The only insignificant variable is tenure in current Job. It should be noted that this, of course, is not the conventional tenure variable used in earnings equations, which refers to tenure in the firm. The latter information is not available in the data set. Specification in columns 1 and 2 excluding all individual characteristics, the coefficients to the number of subordinates-variable increased from 0. 105 and 0. 063 to 0. 122 and 0. 074, respectively. At the same time the explanatory power of the equation drops by about a third. Thus, omitting human capital and other individual traits tends to inflate the estimated impact of managerial power. This is noteworthy since most previous studies do not enter other than firm level explanatory variables into the compensation equation.
As was noted in the introduction, a potential limitation of analyses using the abbreviated-variable is that this may be picking up the firm size effect via the chief executives for whom the total workforce in the company constitute their subordinates. In order to check the robustness of the results, I have, therefore, re- estimated the earnings equation on a smaller sample that excludes the Coo’s. As can be seen from the second column of Table 1, the coefficient estimate shrinks somewhat in magnitude but remains significantly different from zero.
Turning to the second measure of power, the real authority index, it can be seen that managers’ insemination is increasing in authority, also after controlling for individual traits as well as the size, industry affiliation and location of firms; see columns (3) and (4) of Table 1. 10 The relationship turns out to be quite robust with respect to different sample restrictions. For example, as can be seen from columns (5) and (6), restricting the sample to those firms for which each managerial employee’s number of subordinates can be observed, generates only marginal changes in the estimates. Aroma position in the firm (Coo’s and higher level managers, respectively) as in the events column of the table, decreases the real authority coefficients somewhat, while leaving their relative magnitudes and statistical significances intact. Finally, in Table 2 1 include the executives’ membership in different boards as additional explanatory variables. Moreover, in column (2), firm fixed effects intended to capture idiosyncratic firm level effects on the level of managerial pay, are added to the model.
In both cases, the number of subordinates as well as the board membership indicators turns out to be 10 Excluding the chief executives from the sample leads again to a drop in the ignited of the power coefficients. However, they continue to differ significantly from zero. 8 statistically significant. The main point, which emerges from Tables 1 and 2, is therefore that more managerial power is associated with higher compensation. Next I repeat the empirical analysis on an unbalanced panel data set, which originates from the same data source and employs the same variable definitions as above, but covers the years 1992 to 1995.
The main purpose of the longitudinal analysis is to put the power hypothesis to a harder test by accounting for unobserved differences in individual traits that are important for gaining (both) managerial power and higher earnings. I have implemented two ways of handling unobserved heterogeneity; the first is assuming time-invariant individual fixed effects, and the second is allowing for random effects. In both cases only time-varying independent variables are included in the analysis. 11 In the main the estimation results, displayed in Table 3, show the same pattern as the jurisdictional analysis.