Printer Friendly

The effects of board composition and direct incentives on firm performance.

* Monitoring by the board of directors is one of several institutions that have developed in modern corporations to resolve the agency problem between top management and shareholders. Yet, the board's effectiveness in fulfilling this monitoring role is not clear. Some (see, for example, ALI [1] and Dunn [10]) believe boards generally fail in their responsibility to monitor management and guide their companies, and have called for regulations requiring that boards be composed of a majority of outsiders. Others, however, argue that market pressures and concern for reputation will lead directors to fulfill their duty (Fama [11], Fama and Jensen [12]). The extent to which boards oversee management and to which this monitoring depends on the composition of the board are important and unresolved empirical questions.

However, to answer these questions, one must consider the board in a broader perspective, as one of the alternative control devices that limit agency problems between top management and shareholders. In particular, managers receive pecuniary incentives to maximize firm value from stock ownership, stock option plans, and adjustments in salary based on performance. The combination of direct pay for performance and monitoring by the board provide whether board effectiveness could be improved by changing its composition is really a special case of a more general question. In particular, can we, as outside observers, understand organizations' control systems well enough to make any general propositions regarding the optimality of particular governance institutions?

Arguments that certain types of observed governance structures are preferable to other observed types imply that some firms have adopted suboptimal governance structures. This argument is implicit in calls for board regulation, as well as proposals to reform executive pay (Crystal [7]), Jensen and Murphy [21] and [22]). Perhaps the most appealing evidence in favor of this suboptimality view comes from M[phi]rck, et al [29], and McConnell and Servaes [28], who find that corporate performance, measured by Tobin's q, rises with ownership at low levels of ownership and then falls with ownership at high levels. This result suggests that there is an optimal governance structure, and that observe firms deviating from it experiencing lower performance.

An alternative view of this problem is that different governance structures are optimal for different firms, for the simple reason that each firm faces its own management problems, and hence finds its own solution. (1) Applied to the question of board regulation, this argument says that binding regulations would actually hurt shareholders by compelling them to adopt board structures that are suboptimal for their firms. Other authors have applied this argument to the question of the effectiveness of different levels of direct incentives and have concluded that the evidence suggests that there is not an observable relation between levels of incentives and corporate performance (Demsetz and Lehn [9], Holthausen and Larcker [19], and Kole [24]).

This paper attempts to measure differences in firm performance caused by board composition and ownership structure. These two variables are intended to measure the direct incentives and monitoring faced by top management. (2) We also control for a number of other variables that are likely to be correlated with corporate performance. We do so to improve the precision of our estimates, as well as to eliminate much of the omitted-variable bias that has undoubtedly affected previous studies of board composition (Baysinger and Butler [2], MacAvoy, et al [26]).

An important improvement of our methods over previous studies in our use of panel data, which allows us to control for possible biases due to the joint endogeneity of our variables. For example, Hermalin and Weisbach [16] find that poor performance leads to changes in board composition, so any cross-sectional regression of performance on board composition will be biased because of changes in board composition resulting merely from past performance. In addition, managerial shareholdings are likely to be related to corporate performance for two reasons: first, managers will exercise their stock options after their stock goes up, but not after it goes down; second, managers with information about good future prospects are likely to buy more stock, while managers with bad information about their own stock are likely to sell. Both of these explanations suggest that we will observe a positive relation between corporate performance and shareholdings that has nothing to do with performance. By using panel data and our instrumental variables approach we control for these potentially spurious relations between ownership, board composition, and performance.

Section I of the paper reviews the theory behind our empirical work, as well as touching on the important methodological issues. The second section of the paper discusses our data. In the third section, we present our specifications and results. The last section is a short conclusion.

I. Theory and Methods

Underlying any study of managerial incentives and firm performance is the theory of agency (Jensen and Meckling [20]). The formal models that have developed this theory (Holmstrom [17] and Shavell [33]) have focused on the situation where a single owner (or principal) designs the optimal incentive contract and offers it to the manager of his or her firm (his or her agent) on a take-it-or-leave-it basis. Although such models have yielded many important insights into the principal-agent relation, it would be naive to view them as accurate descriptions of what takes place in large corporations. Thousands of shareholders simply cannot coordinate to design and negotiate managerial incentives -- free-rider problems, as well as SEC regulations, prevent shareholders from being the active principal that theory supposes them to be. This role must be delegated to the board of directors. Hence, to understand managerial incentives thoroughly, we must understand how the board of directors plays this role.

However, shareholders do not escape agency problems by leaving them to the board of directors, since the directors are themselves agents, whose interests are not necessarily aligned with the shareholders'. This conflict of interests is clearly true of management (inside) directors, whose careers are tied to the CEO's. It is also true, although possibly to a lesser extent, of nonmanagement (outside) directors. The same free-rider problems that commonly prevent the shareholders from being an active principal also prevent the shareholders from actively choosing the board. In practice, the CEO almost always chooses the board (Mace [27]). Thus, potentially, even outside directors are more aligned to management's interests than to the shareholders'.

Although allowing management to choose their own overseers might seem on a par with letting the fox guard the chicken coop, there are, nonetheless, reasons to think that outside directors will exhibit some independence from top management. First, directors have certain legal obligations to the shareholders and they can be held liable for damages if they fail to meet these obligations. Second, directors will have some desire to maintain or establish reputations as good monitors and competent business people (see, e.g. Fama [11] or Fama and Jensen [12]). (3,4) There is evidence, both anecdotal and statistical, that supports this notion that outside directors exhibit some independence from top management. First of all, Mace [27] reports case-study evidence that exceedingly poor performance or obviously bad proposals will be opposed by outside directors. Weisbach [37] finds that outsider-dominated boards are significantly more likely to respond to poor performance by dismissing the CEO. Finally, Brickley, et al [4] find evidence suggesting that outside directors act in the shareholders' interest in their decisions on the adoption of poison pills.

From these papers, it seems clear that outside directors do a better job of acting in shareholders' interests than insiders when it comes to certain aspects of their jobs. However, there are other facets of directors' jobs where insiders are likely to be preferable. For example, Mace [27] and Vancil [35] argue that the board is often used for training and evaluation purposes as part of the CEO selection process. Hermalin and Weisbach [16] find empirical evidence consistent with this argument. Inside directors are important because they convey information to top management and the outside directors (Mace [27]). The net effect of different types of boards on performance is an open question, which we address in this paper.

Our discussion so far has ignored the fact that board composition is endogenously determined. For instance, Hermalin and Weisbach [16] find that poor performance can result in inside directors being replaced by outside directors. Consequently, a chronically poor-performing firm could have a higher than average proportion of outside directors on the board. If we were to include this sort of firm in a regression of firm performance on board composition, we might find that outside directors "caused" poor firm performance. (5)

It is our attention to the determinants of board composition that distinguishes our analysis from that of previous authors. In particular, MacAvoy, et al [26], which finds no evidence that board composition affects performance, and Baysinger and Butler [2], which finds evidence suggesting that board composition affects performance, fail to control for the determinants of board composition, such as ownership. As a result, it is impossible to tell whether their results are spurious or distorted by the influence of uncontrolled-for factors. For example, as we show in Exhibit 1, firm performance is decreasing with top management's stock ownership over certain ranges of ownership. Previous work (Weisbach [37], Hermalin and Weisbach [16]) finds that ownership is inversely correlated with the proportion of outside directors. Thus, the findings of Baysinger and Butler could be due to ownership effects rather than board-composition effects. (6)

The idea that stock ownership by management can reduce the underlying agency problem comes straight out of agency theory: the more stock management owns, the stronger their motivation to work to raise the value of the firm's stock, which is what the other shareholders want. Therefore, there will be less demand for alternative anti-agency measures such as a strong board of outside directors in firms where management owns a large fraction of the stock. On the other hand, we must keep in mind that agency problems need not be monotonically decreasing in stock ownership by top management: large management ownership insulates management from other forces that reduce agency costs such as the threat of takeovers and the discipline of the board (see Demsetz [8] for a discussion). In addition, large management ownership is often a characteristic of family-controlled firms, which are notorious for putting the interests of the family above the interests of the shareholders. (7) Thus, it is possible that agency costs actually increase with ownership over some region.

Another measure of the underlying agency problem is the length of tenure of the board and top management. If management has been a around a long time, this could indicate that agency problems are not too extreme (otherwise managers would have been dismissed by board action, proxy fight, or takeover). Moreover, it could indicate that management has above-average ability, since they have not been fired and they may have been asked to stay on longer than usual because of their good performance. On the other hand, Vancil [35] reports that there is some consensus among CEOs that a CEO should not serve too long, with ten years commonly cited as the "right" tenure. This could be true because CEOs become set in their ways and lose necessary flexibility over time, or long tenure could reflect a proclivity for dominance on the part of the CEO. (8) Consequently, we could see firm performance suffer if management stays on for too long.

From the preceding discussion, it would clearly be a mistake to use a specification that constrained ownership or tenure to affect performance monotonically. Following M[phi], et a [29], we adopt a piecewise linear specification in these variables. Since the impact of outside directors on overall board behavior need not be linear in the proportion of outside directors, we also use a nonlinear specification of this variable.

As discussed above, there is an issue of simultaneity regarding board composition, so we instrument for board composition in our statistical analysis. Ownership is also likely to be a jointly endogenous variable -- to the extent management has better information than the market, we must be concerned with managers buying in anticipation of good performance and selling in anticipation of bad performance. In addition, managers have a tax-based incentive to realize capital gains when the stock has been doing badly, but not when it is doing well.

II. Data

One measure of profitability is average Tobin's q: the ratio of the firm's market value to the replacement cost of its assets. In the absence of market power, a divergence of q from one represents the value of the assets not include in the denominator of q, such as the value of the internal organization or the value of expected agency costs. A q above one indicates that the market views the firm's internal organization as exceptionally good or the expected agency costs as particularly small. (9)

Our estimates of q are taken from Schaller [32]. To compute the numerator of q, the market value of the firm, we sum the market values of the firm's equity and debt. The market values of the firm's common stock is equal to the number of common shares outstanding times the price per share at the end of the year. We add to this an estimate of the value of the preferred stock, which we derive by dividing the preferred stock dividends by the Standard and Poor's preferred stock yield. The market value of the firm's debt is equal to the sum of the values of the short-term debt and the long-term debt. We assume that the market value of short-term debt is equal to its book value. To estimate the market value of the firm's long-term debt, we use the procedure developed by Brainard, Shoven, and Weiss [3]. We assume that all new issues of long-term debt have a maturity of 20 years, the coupon rate is the prevailing BAA rate, and that the maturity distribution for each firm is proportional to the maturity distribution of aggregate outstanding issues. Given these assumptions, we can calculate the market value of the long-term debt. (10)

The denominator of q, the replacement value of the firm's assets, has three main components: the market value of the capital stock, the market value of inventories, and other assets. To compute the market value of the capital stock, we adjust the book values for inflation and depreciation, assuming that economic depreciation is exponential. (11) For inventory methods other than LIFO, we assume the book value of inventories equals their market value. If LIFO was used, we make an adjustment based on an estimate of the proportion of the firm's inventories which were LIFO. Other assets are intangibles and shares held in other firms. Their market value is assumed equal to its book value.

We control for factors affecting q which are due to time effects and the nature of the firm's industry, and not related to managerial performance. We do this by computing q for all firms on the annual COMPUSTAT industrial file each year in our sample. For each firm in our sample, we then subtract from its q the mean q for all firms in its two-digit SIC code, including those not in our sample. This procedure reduces the variation in q due to industry-specific factors.

Our sample consists of 142 NYSE firms for which we gathered data on board composition and ownership from corporate proxy statements available in Harvard's Baker Library. We classified each director according to his principal occupation. Employees and former employees of the firm we classified as insiders. Nonemployee directors were classified as outsiders. Directors whose status is questionable, such as family members of employees, lawyers, bankers, and investment bankers, were classified as grey directors. (12)

We collected the shareholdings of each director from the sample proxies. The shareholding figures recorded for each director included all shares over which the director had voting power. (13) Although for most directors this was a straightforward calculation, for many directors determining the shares over which they had voting power was a matter of interpreting lengthy footnotes or making judicious decisions when more than one director held voting power over a set of stock. (14)

For those directors who were also their companies' CEOs, we determined their tenures as CEO. For the years within our sample period, we calculated CEO tenure by locating the exact date of any CEO change in the Wall Street Journal Index (the proxies allowed us to detect CEO changes within a given year). We obtained the starting year for CEOs who started before 1971 from the Forbes compensation surveys or from Who's Who in Finance and Industry. (15)

III. Empirical Specification and Results

To examine the effects of the control structure of the corporation on profitability, we regress q, our measure of profitability, on measures of the control structure, as well as other variables which should affect q. We first consider the effects of board composition on q. We estimate piecewise linear equations using board composition. We chose to break our sample at 40% and 60% outsiders, following Weisbach [37]. (16)

We also control for variables ("controls") other than those in the control structure which might affect q. We include expenditures on research and development and advertising, each of which we normalize by size, into the equation. We also include the log of the replacement value of the firm's assets as a direct measure of size.

The OLS equation (pooling years of observations and suppressing the control variables) with the composition variables representing the marginal change in q resulting from more outsiders is as follows (t-statistics are in parentheses):

q = controls - 0.02 COMPL40 - 0.12 COMP4060 (0.06) (0.23)

- 0.18 COMPG60, (0.26)

where the coefficient on COMPL40 is the estimated q with respect to the proportion of outsiders when the proportion is less than 40%, the coefficient on COMP4060 is is the estimated derivative when the proportion is between 40% and 60%, and the coefficient on COMPG60 is the estimated derivative when the proportion is greater than 60%. There is basically no effect of board composition on q. Treating board composition as endogenous using an instrumental-variables approach or estimating on just one year of data makes no meaningful difference to the results.

To estimate the effects of ownership on profitability, we follow M[phi]rck, et al [29] and use a specification that is piecewise linear in ownership levels. Because the marginal effect of additional ownership on profitability may be different in different regions, we divide the ownership levels into four regions: less than one percent, between one percent and five percent, between five percent and 20%, and greater than 20%. (17) Using time-series data on ownership, we are able to control for possible simultaneity between ownership and q. As top management is likely to have inside information about the firm's future prospects, they have an incentive to adjust their portfolios based on their estimate of future performance. Thus, cross-sectional regressions of q on ownership may be misleading as well as statistically incorrect because the results are contaminated by the effect of q on ownership.

The results are shown in Exhibit 1. In the first column, we estimate the equation using ordinary least squares for all five years pooled. At levels of ownership less than one percent, q increases with ownership. This result suggests that there is a noticeable reduction in agency costs resulting from increasing ownership concentration at low levels of ownership. At levels greater than 20%, q decreases with ownership. This result suggests that increases in ownership above 20% cause management to become more entrenched, and less interested in the welfare of their shareholders. The marginal effect of additional ownership at moderate levels of ownership (between one percent and 20%) on q seems to be negative at lower levels and positive at higher levels, although neither of these effects is significantly different from zero. Expenditures on both advertising and research and development raise q. Finally, smaller firms seem to have higher q's.

In the second column of Exhibit 1, we reestimate the equations treating the shareholdings variables as endogenous, using their lagged values as instruments. The relation between shareholdings and q becomes larger when we use this technique. The signs on the ownership variables are all the same as with ordinary least squares, but the sizes of the coefficients and the t-statistic are all larger. This larger effect is potentially because of the elimination of the simultaneity problem discussed previously. A specification test (Hausman [14]) to determine


whether ordinary least squares (OLS) or instrumental variables (IV) is the correct specification rejects the hypothesis that there is no simultaneity at the five percent level (chi-square statistic = 10.47)

One objection to these results might be the q's for the same firm in different years may not be independent observations. One solution to this problem would be to include firm-specific dummy variables into the equation. This approach, however, would eliminate all between-firm variation from the data. Since we believe that the primary force driving our results is between-firm variation, we do not adopt this approach. We do, however, reestimate the equation using just one year of data to determine the extent of the potential bias due to nonindependent observations. We choose the year 1977 because it is the central year of our sample.

The results for 1977 are shown in the third column of Exhibit 1 using ordinary least-squares and in the fourth column of Exhibit 1 using instrumental variables. The pattern of the signs is the same for 1977 as it is for the entire sample. Once again, treating the ownership variables as endogenous increases the size of the coefficients. However, using just one year of data, we do not reject the hypothesis that there is no endogeneity problem with a specification test.

In Exhibit 2, we reestimate the q equations using both management ownership and board composition as explanatory variables. In addition, we also control for a number of factors likely to be correlated with managerial entrenchment and the control structure of the firm. Since we expect that one way for a CEO to become entrenched is by being a CEO for a long time, we control for CEO

[TABULAR DATA OMITTED] a fraction of total pay, was 6.5% in utilities in contrast to 19.9% for industrial firms. The inference from this comparison, that stronger incentives exist for stockholder wealth maximization among industrials, is supported by other observations as well. Unlike public utilities, which rarely grant stock options to executives, options account for about 11% of the compensation in industrials. Moreover, the stock holdings of public utility managers are less than 3% of those of industrial managers in Murphy's[27] study. Managers of industrial firms own stock worth several times their total annual compensation (see, e.g., Benston [5], and Agrawal and Mandelker[1]).

In sum, one may question whether the incentives for maximizing shareholders' wealth that Benston [5], Coughlan and Schmidt [12], and Murphy [27] document among industrial managers are as strong in the public utility industry. On the other hand, it should be noted that the long tenures of managers in the public utility industry (see Exhibit 3) may give them a long-term view of profitability, which is consistent with the objectives of stockholders.

Managers in Different Positions. Exhibit 3 shows that CHM-T, chairmen who also hold the CEO and/or president position, get the highest total compensation, and therefore are probably the most important managers. Their annual earnings are $340,000. The comparable group of managers is Murphy [27] is the chief executive officers who earned $904,010. (12) Managers in the CEO-T and CHM-ONLY groups in our study earn nearly $40,000 less than managers in the CHM-T group. (13) Managers who only hold the title of president earn about $111,600 less than managers in the CHM-T group. The small standard deviation of TCOMP for the PRES-ONLY group reflects the relatively small percentage of long-term compensation (which generally has greater volatility relative to its mean than is the case with salary and bonus -- see Exhibit 3) in their total compensation as well as the low variation in the salary and bonus part. The PRES-ONLY group earns about $66,600 more than the VP-ONLY group (significant at the five percent level in a two-tailed t-rest). The levels of salary and bonus, SALB, across different positions generally behave in a manner similar to TCOMP.

There is considerable variation in long-term compensation, LTC, which is important in assessing incentives for stockholder wealth maximization. As expected, managers in the CHM-T group obtained the largest LTC of $43,100. They also have the highest proportion of LTC/TCOMP, 8.6%. Indeed, one would expect such an incentive system since it ties the interests of the top managers more closely to those of the stockholders. These managers also hold a relatively large amount of stock (while the mean stock holdings of the presidents is larger, the difference with CHM-T is statistically insignificant). Interestingly, LTC for CEO-T is somewhat lower than for CHM-T (t-statistic = 1.55) underscoring the important role of chairmen in this industry. One plausible explanation for this finding is that external relations with regulators and the public (consumers), which are likely to be the primary domain of the chairman, may have a more important effect on the firm that the operational tasks under the responsibility of the CEO. (14)

Relative to other managers, the incentives of presidents convey conflicting signals. While they hold large amounts of stock that align their interests with those of stockholders, they receive the smallest proportion of long-term compensation, LTC/TCOMP.

The high average age tenure of the chaimen who did not hold any other position, CHM-ONLY, suggests that these managers may have held the CEO position earlier (21 out of 81 executive-years had chairmen who were older than 65). This may reflect what Vancil [29] refers to as the "relay process of CEO succession" in which the retiring CEO and chairman gradually transfers his powers to the next CEO and thereafter holds only the chairman position. Managers in the CEO-T and CHM-T groups are younger, although their tenure in the firm is also quite long (but shorter than CHM-ONLY). In general, all of the managers have many years of association with the firm, allowing them, perhaps, to take a long-term view that is consistent with increasing stockholder wealth.

IV. Empirical Results

Estimations of the pay-performance relation (2) are reported in Exhibit 4. If the stockholder wealth maximization hypothesis is valid, we expect the coefficient of the industry-adjusted stock return, b, to be positive. The sales maximization hypothesis predicts that the coefficient of the sales growth variable, c, will be positive.

In all the estimations reported in Exhibit 4, the coefficient of [ - R.sub.It), b, is positive, and statistically significant, except for the PRES-ONLY group. The low adjusted R-squares are typical for such regressions (see, e.g., Jensen and Murphy [19, Table 1]).

[tabular data ommitted)

For the group of all top executives, ALL, the coefficient b has a statistically significant value of 0.11, suggesting that a 10% increase in stock returns raises the total compensation of executives in this industry by 1.1% in the same year. The pay-performance elasticity of 0.11 is about two-thirds of what Murphy [27, Table 6] found for industrial firms. While these results are statistically significant and consistent with the objective of providing managers with incentives to obtain better stock performance, their economic impact may be limited as argued by Jensen and Murphy [19].

We earlier conjectured that individuals in the CHM-T group may be the most important, based on their total compensation. The hold one of the largest amounts of stock, and have the highest LTC/TCOMP ratio. Therefore, they have the strongest incentives to pursue stockholders' wealth maximization. Consequently, we expect that the coefficient b will have the highest value for the extimation using this group. The observed value of about 0.21 is nearly twice that for the group of all managers, ALL, and is statistically significant at the five percent level. It is higher than that for all other groups except the group consisting of managers who hold the chairman-only position. The chairmen, CHM-ONLY, are granted a 3.6% increase in total compensation for every 10% increase in the industry-adjusted returns on common stock for that year. (15) The CEO-T group has a coefficient of 0.14, which is significant at the 10% level.

The PRES-ONLY group presents somewhat anomalous findings. The value of the coefficient b is the lowest compared to the other groups, and is insignificant. One observation, consistent with this apparent lack of incentives to maximize stock returns, is that the PRES-ONLY group has the lowest LTC/TCOMP. However, as observed earlier, their incentives to maximize stockholders' wealth may be derived from their stock holdings which are, by far, the largest of any group. We explore this possibility further in the next section. Interestingly, with a larger LTC/TCOMP value, the VP-ONLY group has a positive and highly significant coefficient.

The coefficient c, which represents the sensitivity of pay changes to sales growth is insignificant for four of the groups in the exhibit, implying that there is no relation between compensation and growth in sales for these groups. It is positive and significant for ALL group (t-statistic of 2.15) and for the chairmen-only group (where it is barely significant at the 10% level). A more pervasive finding of positive and significant values of coefficient c would not have necessarily negated the stockholder wealth maximization hypothesis, since growth in sales can be an instrument in increasing stock value. Indeed, we find some evidence of a significantly positive relationship between stock returns and growth in sales. (16)

Generally, Exhibit 4 constitutes weak evidence in favor of the sales maximization hypothesis. In contrast, Murphy [27] reports a highly significant relation of total compensation with sales growth for every top executive position. Perhaps, it is not all that surprising that the compensation of public utility managers does not depend strongly on growth in sales. The demand for a public utility's product is largely determined by the economic and weather conditions within its service area. The product price is set by the public utility commission. Consequently, public utility managers may have somewhat limited oppurtunities to affect growth in sales for their firms. (17)

Next, we repeat the above analysis using the unconstrained regression (3). This allows us to examine separately the impacts of the rates of return of the firm and the industry on executive pay raises. WE find that estimates of [b.sub.1], the coefficient of the firm's stock return, have similar values (and similar levels of significance) as estimates of b (Equation (2)) in Exhibit 4 for each of the managerial groupings. This finding supports the stockholder wealth maximization hypothesis. The estimates of [b.sub.2], the coefficient of the industry's rate of return, are negative for every managerial grouping. (18) This supports the view that, given the performance of the firm's stock, executive pay raises are negatively related to industry performance. We are unable to reject the hypothesis that [b.sub.1] = -[b.sub.2] for every managerial grouping, except for the PRES-ONLY. (19) This finding supports the industry-adjusted performance measure employed in Equation (2).

Finally, we repeat the above analysis with growth in salary plus bonus, GSALB, as the dependent variable in Equations (2) and (3), and do not find any significant coefficients. Again, this finding is not very surprising, since LTC (and its incentive effects) are not included in SALB.

V. Additional Tests

In this section, we further examine a number of issues raised in the earlier sections.

A. The Role of Long-Term Compensation, LTC

In understanding the incentives for pursuing stockholders wealth maximization, long-term compensation, LTC, plays an important role, since it consists of stock and other items sensitive to long-term performance. Moreover, SALB, the other component of TCOMP, is less volatile relative to its mean than TCOMP. (20) This is because the salary component of SALB is downwardly rigid and therefore can not move very closely with stock returns or sales growth. Our findings support this view, since we do not find a relation between changes in salary plus bonus, and stock returns or sales growth. This implies that the significantly positive relationship with total compensation that we find in Exhibit 4 is driven by changes in its long-term component, LTC. Hence, we next examine the relation between growth in LTC, stock returns and sales growth:

[Mathematical Expression Ommitted]

where [] is the growth rate of long-term compensation of manager i in year t over year t - 1, and the other variables are as defined in Section II. The estimates of Equation (4) are presented in Exhibit 5. Coefficient b,

[tabular data ommitted]

which captures the sensitivity of GLTC to stock returns, is positive in all cases, except for the PRES-ONLY group. The low proportion of LTC/TCOMP for the PRES-ONLY group (see Exhibit 3) may explain why we find that these managers do not have incentives tied to stock returns. The coefficient b is significant for ALL, the group consisting of all top managers (t-statistic = 2.30).

Importantly, looking at the topmost managers, CHM-T, we find that the coefficient is highly significant (at the one percent level). The size of the coefficient suggests that the long-term compensation of CHM-T managers is very responsive to stock returns; LTC increases by 44.3% for a 10% increase in stock returns. We also find that stock returns have a large impact on the long-term compensation of managers in the CEO-T and CHM-ONLY groups. However, the long-term compensation of managers in the PRES-ONLY and VP-ONLY groups does not depend on stock returns, according to the exhibit. In fact, the high p-values of the regressions for the PRES-ONLY and VP-ONLY groups suggest that stock returns and sales growth may not be important determinants of changes in long-term compensation for managers in these groups.

The coefficient of sales growth is significant for three groups in the exhibit: these groups are ALL, CHM-ONLY, and VP-ONLY. For the VP-ONLY group, the level of significance is 10%. Overall, there is some evidence of an effect of sales growth on long-term compensation, but it is not as strong as the impact that sales growth has on compensation in industrial firms (Murphy [27]).

B. Lack of Direct Compensation Incentives for


For the PRES-ONLY group, we have found so far that neither the growth in total compensation, GLTC, is related to stock returns. In Section IV, we pointed out that it may not be necessary to provide presidents with incentives through TCOMP or LTC, if they have large stock holdings which suffice to align their interests with those stockholders. This implies that for managers with large stock holdings, the coefficient b in Equations (2) and (4) should be small, as is also argued by Jensen and Murphy [19]. On the other hand, for managers with small stock holdings, the coefficient b should be large. To examine this issue, we reestimate Equations (2) and (4) separately for the PRES-ONLY group with high (above median) and low (below median) stock holdings. We do not find support for this hypothesis. The estimate of b in Equation (2) is larger (and statistically significant at the seven percent level) for the for the former subgroup; it is insignificant for the latter subgroup, consisting of PRES-ONLY with below median stock holdings. The estimate of b in Equation (4) is insignificant for both subgroups. the correlation coefficients between stock holding at the beginning of the year and GLTC, and between stock holding and GTCOMP are both insignificant. These findings do not support the view that the stock holdings of PRES-ONLY of public utilities offset the lack of direct compensation incentives.

Another plausible explanation for the lack of direct compensation incentives for PRES-ONLY to maximize stockholder wealth may be that current compensation may depend on past, rather than the current year's, stock price performance. We investigate this possibility by including the previous year's stock returns and sales growth in Equations (2) and (4). We do not find a relation between changes in compensation to the PRES-ONLY group and either of these two variables.

While our findings for the PRES-ONLY group are not consistent with the proposition that they have direct compensation incentives to maximize stockholders wealth, their interests may be aligned through other factors, such as the incentive effects of the competition to be promoted to the CEO position or the threat of dismissal. Nonetheless, the lack of direct compensation incentives for the PRES-ONLY group represents somewhat of a puzzle, which calls for additional research.

VI. Conclusions

Our findings on public utilities are consistent with the view that compensation packages align the interests of most of the top management with those of their stockholders. These findings contrast with those of previous research on public utilities that find no relation (or a negative relation) between managerial compensation and firm profitability. There are three possible explanations for the differences between our findings and those of prior studies. First, while we investigate compensation changes over an executive's career with the firm, prior studies examine only the relationship between the level of executive compensation and corporate profitability across firms in a given year. Our procedure better controls for other executive-specific and firm-specific determinants of executive compensation. Second, we examine the entire compensation package of a manager, whereas prior studies examine only a part of it, mainly salary and bonus. Third, we use market-based measures of corporate performance, instead of the accounting-based measures used in earlier studies. The former tend to be forward-looking, while the latter provide historical information.

We also examine the incentives for managers occupying several different management positions. Chairmen, who also hold the position of CEO and/or president in the firm, are likely to be the most important managers in public utilities, since they have the largest total annual compensation. We find a positive relation between the change in their total compensation and stock performance, consistent with incentives to maximize stockholders wealth. We find similar results for all the managers as a group. This constitutes evidence which contrasts with the long-held view, partially based on past empirical studies, that managers of regulated firms lack incentives to maximize stockholders' wealth.

(1) See Kaysen [20] for a related discussion.

(2) The regulatory procedure is to set the price of the utility's product to cover costs plus a "fair" rate of return on the invested capital (rate base). The debt component is provided with the embedded cost of debt. Appropriate returns on equity are harder to estimate. Over the years, courts have ruled that public service commissions should provide stockholders with an oppurtunity to earn returns commensurate with those of other firms of similar risk.

(3) The public service commission attempts to provide the rate of return mandated by the courts by setting product prices for a given period. Consequently, until prices are set again (following a regulatory lag), the utility can attempt to maximize its profits given the existing fixed prices.

(4) This also seems appropriate given the relative homogeneity of our sample.

(5) Compensation schemes can be quite complicated because they may be used to address a number of agency problems (Lewellen, Loderer, and Martin [2]).

(6) The firms also have similar capital structures, since the t-statistics for differences in the current ratio and the debt-to equity ratio suggest that there are no differences in short- and long-term debt employed.

(7) We repeated the analysis for the years 1975 and 1984.

(8) In our sample of 471 CEO-years, there are only six CEO-years in which this is not the case.

(9) This category includes titles such as executive vice-president, vice-president, corporate secretary, treasurer, and chief financial officer.

(10) In cases where the chairman also holds the president position, the CEO position seldom exists as a separate position.

(11) Murphy's estimate of total pay includes an estimate of the value of options granted during the year. We exclude stock options, since their use is very rare in the public utility industry.

(12) His group CEOs consists of managers that may simultaneously hold other positions as well.

(13) A two-tailed t-test shows the difference to be significant at the one percent level.

(14) Note, however, that CHM-T and CEO-T are overlapping groups.

(15) But the t-statistics in that case is significant only at the 10% level.

(16) For our example of 690 company-years, the Pearson product-moment correlation between stock returns and sales growth is 0.097 (z-statistic = 2.55) and the Spearman rank correlation is 0.036 (z-statistic = 0.94).

(17) Managers can influence demand through activities such as inducing consumers to switch from oil heating to gas or electric heating, or by attracting new industries to their service area. But, compared to industrial managers, they are unlikely to have a substantial impact on sales.

(18) In two cases, CEO-T (t-statistic = -1.20) and CHM-ONLY (t-statistic = -1.49), the coefficient [b.sub.2] is not significantly different from zero, base on calculated t-statistics. But these t-statistics are likely to downward biased, since [] and [R.sub.It] are highly correlated, with a Pearson product-moment correlation coefficient of 0.64.

(19) The calculated t-statistics of [b.sub.1] - [(-b.sub.2)] are generally too low to be explained by the downward bias as a result of the multicollinearity between the rates of return on the firms and the industry.

(20) The coefficient of variation, (standard deviation/mean), for ALL in Exhibit 6 is about 5.5 times as large for LTC than it is for SALB. It is 2.8 and 0.51 for the two components, respectively.


[1] A. Agrawal and G.N. Mandelker, "Managerial Incentives and Corporate Investment and financing Decisions," Journal of Finance (September 1987), pp. 823-837.

[2] R. Antle and A. Smith, "An Empirical Investigation of the Relative Performance Evaluation of Corporate Executives," Journal of Accounting Research (Spring 1986), pp. 1-39.

[3] W. J. Baumol, Business Behavior, Value and Growth, New York, Harcourt-Brace, 1959.

[4] W. J. Baumol, "On the Theory of the Expansion of the Firm," American Economic Review (December 1962), pp. 1078-1087.

[5] G. J. Benston, "The Self-Serving Management Hypothesis: Some Evidence, "Journal of Accounting and Economics (April 1985), pp. 67-84.

[6] A. A. Berle and G. C. Means, The Modern Corporation and Private Property, New York, Macmillan, 1931.

[7] J. C. Bonbright, Principles of Public Utility Rates, New York, Columbia Press, 1961.

[8] J. A. Brickley, S. Bhagat, and R. Lease, "The Impact of Long-Range Managerial Compensation Plans on Shareholder Wealth," Journal of Accounting and Economics (April 1985), pp. 115-130.

[9] T. M. Carroll and D. H. Ciscel, "The Effects of Regulation on Executive Compensation, "Review of Economics and Statistics (August 1982), pp. 505-509.

[10] A. Christie, "On Cross-Sectional Analysis in Accounting Research," Journal of Accounting and Economics (December 1987), pp 231-258.

[11] D.H. Ciscel and T. M. Carroll, "The Determinants of Executive Salaries: An Econometric Survey, "Review of Economics and Statistics (February 1980), pp. 7-13.

[12] A. T. Coughlan and R. M. Schmidt, "Executive Compensation, Management Turnover and Firm Performance: An Empirical Investiga.

[13] P. M. Dechow and R. G. Sloan, "Executive Incentives and the Horizon Problem," Journal of Accounting and Economics (February 1991), pp. 51-89.

[14] E. F. Fama, "Agency Problems and the Theory of the Firm," Journal of Political Economy (April 1980), pp. 288-397.

[15] R. Gibbons and K. J. Murphy, "Relative Performance Evaluation for Chief Executive Officers," Industrial and Labor Relations Review (February 1990), pp. 30S-51S.

[16] M. Hirschey and J. L. Pappas, "Regulatory and Life Cycle Influences on Managerial Incentives," Southern Economic Journal (July 1981), pp. 327-334.

[17] M. C. Jensen, "Agency Costs of Free Cash Flow, Corporate Finance and Takeovers," American Economic Review (May 1986), pp. 323-329.

[18] M. C. Jensen and W. H. Meckling, "Theory of the Firm: Managerial Behavior, Agency Costs and Ownership Structure," Journal of Financial Economics (October 1976), pp. 305-360.

[19] M. C. Jensen and K. J. Murphy, "Performance Pay and Top Management Incentives," Journal of Political Economy (April 1990), pp. 225-264.

[20] C. Kaysen, "The Corporation: How Much Power? What Scope," in The Corporation in Modern Society, Edward S. Mason (ed.), Cambridge, MA, Harvard University Press, 1960, pp. 85-105.

[21] W. G. Lewellen and B. Huntsman, "Managerial Pay and Corporate Performance," American Economic Review (September 1970), pp. 710-720.

[22] W. G. Lewellen, C. Loderer, and K. Martin, "Executive Compensation and Executive Incentive Problems: An Empirical Analysis," Journal of Accounting and Economics (December 1988), pp. 287-310.

[23] C. J. Loomis, "The Madness of Executive Compensation," Fortune Magazine (July 1982), pp. 42-52.

[24] H. G. Manne, "Mergers and the Market for Corporate Control," Journal of Political Economy (April 1965), pp. 110-120.

[25] R. Marris, "A Model of the Managerial Enterprise," Quarterly Journal of Economics (April 1963), pp. 185-209.

[26] J. W. McGuire, J.S.Y. Chir, and A. O. Elbing "Executive Income, Sales and Profits," American Economic Review (September 1962), pp. 753-761.

[27] K.J. Murphy, "Corporate Performance and Managerial Remuneration: An Empirical Analysis," Journal of Accounting and Economics (April 1985), pp. 11-42.

[28] C.W. Smith, Jr., and R.L. Watts, "The Structure of Executive Compensation Contracts and the Control of Management," Working Paper, Graduate School of Management, University of Rochester, 1984.

[29] R. Vancil, Passing the Baton: Managing the Process of CEO Succession, Boston, Harvard Business School Press, 1987.

Anup Agrawal is an Assistant Professor at North Carolina State University, Raleigh, North Carolina. Anil K. Makhija is an Associate Professor and Gershon Mandelker is a Professor at the University of Pittsburgh, Pennsylvania.
COPYRIGHT 1991 Financial Management Association
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1991 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:Corporate Compensation Policy Special Issue
Author:Hermalin, Benjamin E.; Weisbach, Michael S.
Publication:Financial Management
Date:Dec 22, 1991
Previous Article:ESOPs and profit-sharing plans: do they link employee pay to company performance?
Next Article:Executive compensation and corporate performance in electric and gas utilities.

Related Articles
A policy statement on governance.
The vitals of accountability.
The vitals of accountability.
The best and worst boards of 1996: measuring corporate governance.
Board composition and regulatory change: evidence from the enactment of new companies legislation in New Zealand.
Corporate Governance and Firm Diversification.
Board Independence and Compensation Policies in Large Bank Holding Companies.
Stock options and optimal overhang.
Corporate Governance.

Terms of use | Copyright © 2017 Farlex, Inc. | Feedback | For webmasters