Takeovers: managerial incompetence or managerial shirking?

Griffin, James M. ; Wiggins, Steven N.

Agency theory identifies managerial shirking as the cause for takeovers, while other explanations focus on low ability managers. This paper formalizes Jensen's free cash flow variant of agency theory by constructing a simple two-period game which captures the distinctive empirical implications of the two theories.

Using data for petroleum firms following the oil price shock of 1979-80, we find that firms undergoing financial restructuring exhibited higher values of Tobin's q. Additionally, evidence of management turnover and workforce cuts emphasizes that takeovers appear primarily designed to address agency concerns.

I. INTRODUCTION

Recent explanations for the causes of takeovers focus on the need to enhance managerial efficiency, but there are two competing theories. Grossman and Hart [1980] and Shleifer and Vishny [1986] postulate that takeovers replace managers of low ability. Jensen [1986], in contrast, argues that takeovers emerge from agency problems, specifically the need to restrain managers from dissipating "free cash flow."[1] To Jensen, takeovers increase managerial effort by raising debt obligations, which commit otherwise free cash flows. Thus a takeover could be required either to replace incompetent managers or to constrain existing managers to more effectively pursue shareholder interests--an agency problem.

Unlike previous research which tends to be either theoretical[2] or empirical[3], this paper formalizes and tests a general model of takeovers that incorporates both types of managerial inefficiencies. We attempt to bring out the distinctive implications of these theories and then test between them. Our theoretical model, building on the earlier work of Grossman and Hart [1980; 1982] and Shleifer and Vishny [1986], uses a game theoretic framework to formalize how free cash flow affects the incentives of managers. It shows how stochastic shocks, financial structure, and the probability of bankruptcy interact to determine the optimal level of managerial effort. One implication of the agency explanation for managerial inefficiency is that an exogenous increase in earnings will decrease managerial effort below efficient levels. To attenuate this agency problem, firms will financially recapitalize by increasing the debt/equity ratio, restoring the effort of existing managers to ex ante efficient levels. Alternatively, when managerial incompetence motivates takeovers, the target firms should be underperforming the industry. The takeover should then lead to changes in management rather than recapitalization.

This paper uses a unique data set from the petroleum industry in the early 1980s to test between these distinctive theories. The heavy takeover activity in this industry provides an excellent sample for empirical tests. By restricting the analysis to firms within the same industry, we control for inter-industry effects which could otherwise vitiate the empirical findings. The broad similarity of the firms also permits us to use Tobin's q as a measure of firm performance since the asset bases and competitive conditions of the firms are highly similar. Another key advantage of using this industry is the availability of independent valuations of individual petroleum firm's assets. This independent measure enables us to reliably measure asset values, which can also limit Tobin's-q analysis.

Section II describes the theoretical model. Section III examines three highly varied types of empirical evidence, each aimed at separating the effects of managerial ability versus managerial effort as explanations for takeovers. Section IV recapitulates the major findings.

II. A GAME THEORETIC MODEL OF TAKEOVERS INCORPORATING AGENCY EFFECTS

Grossman and Hart [1980] pioneered the development of game theoretic models of takeovers, leading to further advances by Shleifer and Vishny [1986]. In borrowing extensively from these earlier works, our objective is to construct a simple model incorporating both types of managerial inefficiencies in order to derive their distinctive empirical implications. The model postulates a simple agency problem between a large shareholder and the manager. The manager prefers to shirk, giving rise to an agency problem because managerial effort is not observable and cannot be contracted over.[4] Further, agents cannot contract over surrogates for effort, such as cash flow, because of the inability to distinguish a priori permanent from transitory shocks in earnings.[5]

The game tree is illustrated in Figure 1. In the initial period the founder forms the corporation, contributes equity, issues debt, hires a manager, and then sells the majority of shares to atomistic buyers. The founder then either sells the remaining block of shares to the large shareholder or assumes the large shareholder's role himself. The manager first chooses a level of effort, and then a first-period earnings shock is observed. Large negative shocks result in bankruptcy, while large positive shocks result in takeovers. A takeover will result either in the replacement of incompetent management or the recapitalization of the corporation to raise managerial effort. After the capital structure is determined at the end of period 1, managers select their effort level (Omega 2) in period 2 and the period 2 random shock (Omega 2) is observed. After period 2, the firm is liquidated. The model addresses two critical questions: Why do takeovers emerge in equilibrium, and what changes are implemented as a result?

The key to the takeover is the large shareholder who maintains a large enough interest in the corporation to justify monitoring the firm's performance. Monitoring results in two kinds of information. One, the large shareholder is fully informed regarding the firm's realized profits and stochastic earnings shocks. (Atomistic shareholders know only the ex ante distribution and what they can learn from watching behavior.) The large shareholder also observes the performance of the firm relative to other industry firms, and can use this information to determine the quality of the manager. The large shareholder uses this information to decide whether or not to stage a takeover.

Formally, let the production technology produce total revenues (R) as follows:

(1) [R.sub.t] = [P.sub.t] [Alpha Omega.sub.t] g(K) + [Epsilon.sub.t]

where P is product price, [Omega] is the manager's level of effort, and [Alpha] is the manager's native ability, which is unknown to the founder. K is the exogenously determined capital stock, and g(.) that portion of the standard production function that depends on capital (g'(.) > 0, g"(.) < 0). Manager effort then enters multiplicatively, and there are two kinds of managers, high ability (Alpha = [Alpha.sup.H]) and low ability (Alpha = [Alpha.sup.L]). There is also some ex ante probability the manager has high ability.[6] The term E is a random shock to earnings with zero mean, is bounded over the range [THE FORMULA HAS BEEN OMITTED] and is distributed according to the twice continuously differentiable distribution function F(.). Discounting is implicitly imbedded in all prices, so all dollar figures are period 0 dollars.

The founder starts the corporation and then sells an exogenous fraction (1- [Beta) > 1/2 of the shares to atomistic shareholders in "the market" for diversification reasons.[7] Following Shleifer and Vishny [1986], the founder retains B of the shares or they are sold as a block to a large shareholder who monitors the corporation.[8] This monitor cannot dictate policy unless he acquires at least a 50 percent share in the firm. Instead the founder/large shareholder first influences manager performance through the initial debt/equity ratio, and later through a possible recapitalization or management replacement after a takeover. Hence the model provides a simple, stylized view of how the large shareholder can influence management through capital structure or by replacing incompetent managers.

Capital is financed through a combination of debt and equity. Letting the price of capital be one, the sum of debt (D) plus equity (E) must equal capital (K) invested: D + E = K, which acts as the initial capital stock constraint. Debt payments enter the manager's decisions through the cash flow constraint. There must be enough revenues in each period to meet debt obligations, and so the manager's income, Y depends on his ability to make debt payments. To ensure the appropriate effort in each period the founder will schedule debt as installment debt, which requires the manager to generate adequate revenues in each period. Hence the accumulated interest and a fraction of the principal must be repaid at the end of the each period. Let Dt represent the principal and interest that becomes payable at the end of period t. Recalling that all dollar figures are in period 0 dollars, we have

[D.sub.1] + [D.sub.2 = D], and

[THIS FORMULA HAS BEEN OMITTED]

where [NR.sub.t] are net revenues. If the manager cannot make such payments, then the corporation will go bankrupt. Bankruptcy occurs if the net cash position becomes negative. Ignoring for now a possible takeover, and letting Pt be the net cash position, we have

[THIS FORMULA HAS BEEN OMITTED]

Bankruptcy occurs if Pt becomes nonpositive.

Letting [...] be the [...] just (algebraically) small enough to drive the firm's value to zero:

[THIS FORMULA HAS BEEN OMITTED]

Together, these elements yield the profit maximization problem of the founder:

where [P.sub.R] is the probability of a takeover, R is the cost of a takeover, and the other variables are defined above. Equity financing is subtracted from profits to represent the opportunity cost of investing equity in an alternative investment. To ensure an interior solution there are positive bankruptcy costs, and for convenience these costs are assumed to consume the entire residual if bankruptcy occurs.

To solve the model, one begins with the effort level of the manager ([Omega.t]). The manager's utility function is defined over effort (Omega.sub.t) and income (Y.sub.t) and is assumed to take the following additive and separable form:

[THIS FORMULA HAS BEEN OMITTED]

Effort is generic and includes the various activities that create disutility for the manager but raise firm profits. These activities encompass both direct effort and the disutility of sacrificing possible personal perquisites. The return to effort is that it shifts the density function of earnings to the right, lowering the probability of bankruptcy. Summarizing the manager's expected utility function is

[THIS FORMULA HAS BEEN OMITTED]

where F(...) is the probability the firm goes bankrupt. Given F(...) and income (Y.sub.t), the manager will choose a level of effort that maximizes expected utility as defined in equation (6). To ensure an interior solution for effort, it is assumed that f(.), the density function of E, is symmetric and increasing in [Epsilon] over the relevant range.[9] The singleperiod problem for the manager is to maximize (6):

[THIS FORMULA HAS BEEN OMITTED]

The second-order conditions for equation (7) show how the founder/large shareholder can influence the manager's level of effort through the financial structure of the firm. Specifically, the debt load increases the fixed financial commitments of the firm and decreases the absolute magnitude of the shock that will lead to bankruptcy. Intuitively, with more debt, fewer things can go wrong and allow the firm to remain viable. Thus a less severe negative shock will lead to bankruptcy. By the second-order conditions, increased debt means that f(.) is higher at any given [Omega *], which increases the marginal returns to managerial effort. Hence the large shareholder can increase debt to encourage more effort from the manager, and then use compensation (Y) to entice managers to accept the higher probability of bankruptcy and higher effort level.

Ignoring for now managerial abilities, consider what happens when there is an unexpectedly large [ ] in the first period. In this case the firm will be flush with cash-- a "free cash flow" situation in Jensen's language. The original debt encumbrances (D2) will become insufficient to generate efficient managerial effort, which the founder knows because he can observe earnings.[10] By expending takeover costs, R, and buying a controlling interest, the large shareholder can force the issue of new debt. The net returns to such an activity are the increased second-period net cash flows from increased effort, less the costs of the takeover: (.....) [P.sub.2]a g(K) - R, where [....] is the effort after the takeover, [Omega 2*] the effort without a takeover.[11] Hence, takeovers should occur at those firms experiencing substantial positive earnings shocks. This theory, then, shows how takeovers can occur even when small shareholders attempt to free ride. The large shareholder's stake provides him with a return on monitoring activities, and his earnings on those shares also repay the cost of the takeover.

Stepping beyond the formal model, the threat of a takeover could also lead managers to recapitalize the firm on their own to forestall the takeover. Such strategies are entirely compatible with takeovers motivated by agency concerns. The takeover itself, however, ceases to be necessary. The theoretical problem with expanding the strategy space in this way is to explain why managers do not always recapitalize to forestall a takeover. Presumably the explanation is that managers learn as takeovers take place in similar firms, but a full explanation is left for later work. The empirical analysis below, however, groups "voluntary" recapitalizations together with observed takeovers.

The theory provides distinctive implications when a takeover is motivated by managerial incompetence. Specifically, takeovers should occur in firms whose performance is poor relative to industry standards. In addition, the takeover should not particularly lead to financial recapitalization, but instead the replacement of top management. Hence the two theories provide distinctive predictions regarding the firms that will be targets for takeovers and the changes that will occur.

Two primary hypotheses then summarize the theoretical results:

HYPOTHESIS 1. When a takeover is motivated by agency (free cash flow) reasons, the firm's products should be higher than previously anticipated, with prospects for substantial future positive quasi-rent flows.[12] The threat of takeovers should result infinancial restructuring, which could take the form of increased debt, a spin-off of the source of the free cash flows, or increased regular dividends. Replacement of top management is unimportant.

HYPOTHESIS 2. When a takeover is designed to replace a comparatively ineffective incumbent, the firm taken over should be underperforming its rivals in the industry. Such takeovers will result in the replacement of top management.

To be complete, it is also important to briefly address the overall equilibrium of the game. There are five major steps in establishing this equilibrium: (i) determination in period 0 of the founder's initial choice of financial structure and a wage, (ii) the manager's choice of effort in the first period, (iii) the large shareholder's decision to takeover, (iv) the atomistic shareholder's decision to tender, and (v) the manager's choice of second-period effort. To determine existence, it is necessary to solve these problems in reverse order, first determining the optimal second-period effort of the manager, and then work backwards.13

Given the setup outlined above, it can be shown that the decision to stage a takeover generates a sequential equilibrium in pure strategies supported by credible out-of-equilibrium beliefs.[14] In the equilibrium the large shareholder offers the atomistic shareholders a premium that is determined by his takeover costs and his retained share of the corporation, [BETA]. This offer provides nonnegative expected profits to the founder, and atomistic shareholders rationally assess the information content of the offer. The important result is that takeovers can be motivated either by a desire to discipline the manager, or a desire to replace an inept manager.

III. EMPIRICAL TESTS AND EVIDENCE

There are three major advantages to using the petroleum industry over the period 1980 to 1984 for the focus of our empirical tests. First, the oil price increase of 1979-80 constituted what observers at the time viewed as a large, permanent earnings shock. Thus petroleum firms over this period likely met Jensen's free cash flow conditions. Second, the petroleum industry provides a large class of similar firms involved in takeovers. This permits us to control for interindustry effects, which in standard cross-sectional analysis may be correlated with the regressors, leading to biased and spurious results. Instead, interfirm differences among petroleum firms can be more plausibly attributed to differences in managerial effort or ability. Third, independent estimates of asset values are available for petroleum firms, facilitating the use of Tobin's-q analysis.

We examine three classes of empirical evidence. First, we examine how managerial effort (Omega.sub.t) and ability (Alpha) in equation (2) are related to net earnings flows as measured by Tobin's q. Next, we examine the effects of takeovers on the management team and firm's capitalization. Finally, we examine changes in post-takeover behavior for evidence of reduced managerial rent dissipation.

For petroleum firms the period of our data set, 1980 to 1984, was tumultuous, characterized initially by spiralling oil prices and earnings, frustrated efforts to develop new reserves, and then takeovers. As illustrated in Table I, multi-billion dollar acquisitions were commonplace with the disappearance of such giants as Conoco, Gulf, Getty, and Cities Service. Ostensibly to avoid corporate takeovers, Phillips, Unocal, Exxon, Arco, and others financially recapitalized or repurchased significant quantities of their own stock. The data set contains thirty firms in the same industry at approximately the same point in time (1984).[15] The sample contains eight takeover targets, fifteen firms who "voluntarily" restructured either through repurchase of their own shares or those of another firm, and seven other firms that were neither restructured nor taken over.

A Direct Test of the Determinants of Net Earnings Flows

The empirical tests of these competing explanations of managerial inefficiencies rest on two linkages. The first linkage follows from equation (2) and can be simply summarized as follows:

[THIS FORMULA HAS OMITTED]

where net revenues (NR t) depend, inter alia, on managerial ability (...) and managerial effort (Omega.sub.t). Since both alpha and [Omega.sug.t] are unobservable, it is necessary to link them to observable variables. Specifically, from the preceding theoretical model, the reduced-form equation implies that Pt effort will depend on the firm's relative debt (D) and takeover costs (R) as follows:

Omega.sub.t = g(D,R)

where [Delta.g / Delta R < 0.

Turning to the managerial ability hypothesis, the model does not specify measures for ability. Accordingly, we postulate that a key performance index is the management's cost function in replacing reserves. In the early 1980s, managers experienced real difficulty in replacing lowcost reserves as domestic development costs rose from $2.75 per barrel in 1978 to $11.29 per barrel in 1982.[16] Consequently, we use two variables to reflect the cost function of reserve replacement. First, we use the firm's average finding cost (AC) as a primary measure. Separate from the effect of average finding costs is the scale on which new reserves are found. Even though a firm may be a low-cost oil finder, the inability to replace a large fraction of reserves is an indicator of poor management. Hence, managerial ability (alpha]) is measured by average finding costs (AC) and the percentage of reserves replaced (%REPL):

Given our empirical scheme that relates net earnings in equation (8) to measures of managerial effort or ability in equations (9) and (10), it is critical to select a measure of earnings (NR in equation (8)) suitable for interfirm comparisons. The natural measure is Tobin's q, which normalizes firm performance by the value of assets. Further, given the similarity of assets within the petroleum industry, where entry barriers are low and product differentiation is relatively unimportant, interfirm differences in q are likely to be due to heterogeneities in managerial ability or effort.

The usefulness of Tobin's q is frequently limited by the lack of a reliable measure of the value of a firm's assets. However, the John S. Herold Corporation measures the asset value of petroleum firms' oil and gas reserves, refining capacity, marketing facilities, chemical operations and so forth. These measures implicitly provide an independent, carefully constructed measure of the denominator in Tobin's q. The numerator term includes the market value of common and preferred stock,[17] the market value of long-term debt,[18] and working capital. The values for Tobin's q are reported in column 5 of Table I. Note that with the exception of Mesa Petroleum, the estimates of q are typically well below one. The reason for this disequilibrium appears to be that firms had not yet adjusted to the oil price shocks. This lack of adjustment could have occurred because existing management teams could not cope with the new environment and the need to find new reserves. Or it could be that the price shocks created free cash flow necessitating recapitalization. The important empirical point is that while the industry as a whole may have been in a disequilibrium, it is the interfirm differences in Tobin's q that we seek to explain. Table I reveals that these differences are appreciable, suggesting a rich data set.

An empirical model of managerial effort. The preceding theory predicts that managers facing higher relative debt levels will exert greater effort to avoid bankruptcy. For empirical purposes it is necessary to normalize debt by using the debt/equity ratio (D/E) to account for size differences across firms.19 Managers of larger firms also recognize that, even with highly effective capital markets, size could provide some protection from a takeover. Hence, interfirm differences in the costs of takeover are likely to be related to firm size as measured by stockholders' equity (E). To reflect possible non-linearities in takeover costs as size varies, the squared value of equity (E2) is also included. Combining equations (8) and (9) and positing a simple linear model, we obtain:

An empirical model of managerial ability. The next step is to combine equation (8) and (10) and to formalize the relationship between firm performance (as measured by Tobin's q) and managerial ability (alpha) in equation (8). We posit the following linear model:

Equation (12) indicates that, ceteris paribus, a company's managerial ability is measured by its average finding cost for oil (AC) and its ability to replace a significant percentage of the reserves it has produced (%REPL).

A joint model incorporating both effort and ability. Since the two competing explanations of takeovers are not mutually exclusive, it is appropriate to combine equations (11) and (12) to obtain the following general model:

The joint model provides a flamework to test the special cases as restrictions of the general model.

Empirical results. Table II reports the resuits. The results show that the managerial effort model is both statistically and economically significant. Increases in the debt/equity ratio raise the market value of the firm, which means that managerial effort rises with the probability of bankruptcy.20 Equity, which reflects firm size as a determinant of takeover costs, is statistically significant in both equity and equity squared. These results show that takeover costs increase with size, but at a decreasing rate.[21] Hence the central hypotheses of the agency-theoretic model are supported by the data. Further, the adjusted coefficient of determination of .561 indicates that this model accounts for a substantial fraction of the variation in q.

The results provide considerably less support for the managerial ability model as tested in equations (2), (3) and (4). Equation (2) of Table II estimated the basic management ability equation set forth in equation (12). Average finding cost has the wrong sign, whereas the percentage of reserves replaced has the correct positive sign, but both are only marginally significant and the explanatory power is low as reflected in the small R2 (= .08). In an attempt to improve the management ability model, we introduced a dummy management change variable in equations (3) and (4) of Table II under the assumption that average finding costs and the percentage of reserve replacement may not be a sufficient indicator of managerial ability. If the purpose of takeovers is to replace low-ability managers, then we would expect management changes to be associated with low q firms. Whereas the coefficient of the dummy management change variable has the correct negative sign, it is not statistically different from zero, either when used by itself in equation (3) or together with the other managerial ability variables in equation (4). Indeed, equation (4) indicates statistical insignificance of all three explanatory variables and an adjusted coefficient of determination of only .101, compared to .561 for the managerial effort model. Hence the separate tests provide considerable support for the effort model, but only weak support for the ability model.

Equation (5) in Table II presents estimates of the joint managerial effort/ability model set forth in equation (14). Comparison of equation (5) with other regressions in Table II shows that when the two sets of variables are included, the variables reflecting managerial effort remain significant and of similar magnitude. In contrast, for the managerial ability variables only the percentage of reserves replaced is statistically significant while average finding costs and management change are insignificant.

Thus the best indicator of managerial ability appears to be the ability of the firm to replace reserves lost via production. This is plausible since the principal activities of oil companies are to produce from existing reserves and replenish reserves via reserve additions; some firms such as Gulf and Texaco found themselves in a liquidation mode. The statistical significance of the reserve replacement variable means that the managerial effort model is rejected when tested as a restriction of the joint model. The explanatory power of the joint model, though, is only modestly greater than the managerial effort model (R2 = .608 rs..561). Thus while there is statistical support for the joint model, the managerial effort model clearly provides the bulk of the explanatory power.

Management Turnover Evidence vs. Financial Recapitalization

The most direct evidence of the motivation for takeovers is the kinds of changes that are instituted during takeovers. The simplest evidence regarding the managerial ability model is how often takeovers and restructuring resulted in new management. As shown in Table I, of the thirty firms in the data set, eight firms disappeared through takeovers over the period 1980-84. One target, Gulf Oil, is prototypical. It had been unable to replace its dwindling reserve base through discoveries and apparently committed numerous other management errors.[22]

The data suggest that while takeovers are a necessary condition for a management change, they are not sufficient. Columns 3 and 4 of Table I examine the turnover of top management with rank of senior vice president or higher by comparing the 1979 management team with the 1985 team. Column 3 reports the percentage of top management positions involving a name change. These figures overstate management change since changes in top management occur in any business. To account for the infusion of "new blood" in the management team, column 4 reports the percent turnover in which management positions were held in 1985 by individuals not listed on the 1979 annual report as part of the management team. Asterisks are placed in column 4 beside percentages exceeding 50 percent new management to denote that we consider such a change large enough to label the firm as receiving a new management team. Those statistics reveal that corporate takeovers are an almost necessary condition for a new management team.[23] Takeovers, however, are not sufficient. DuPont chose to leave Conoco's management team in place and, likewise, U.S. Steel left Marathon's management team intact. Hence takeovers do not generally or even typically lead to new management teams; the data regarding management change does not offer particular support for the managerial ability model over the agency theory.

In contrast, the data on recapitalization uniformly favors the agency theory. Takeovers were uniformly cash acquisitions as opposed to stock exchanges, resulting in combined firms with much greater debt/equity ratios. Furthermore, it explains why twelve of the twenty-two nonacquired firms chose to restructure voluntarily, either by repurchasing their own shares or some target's shares (see Table I). Whether by means of takeover or by voluntary restructuring, the result is markedly higher debt-equity ratios and presumably a more effective constraint on management's propensity towards nonprofit maximization.

Restructuring via debt issues or stock repurchases, moreover, would not fend off takeovers designed to replace low-ability managers since high-ability managers would still raise Tobin's q. Yet, in a large share of the takeovers, recapitalization rather than a management change was the primary effect. Thus, the managerial effort model explains the widespread recapitalization and frequent absence of management change. The ability model does not.

Still, the joint managerial effort/ability model would interpret the data in Table I as suggesting that both non-managerial effort and ability are important. Effort and ability vary across firms in a continuum. For the bulk of the cases, restructuring is sufficient to raise managerial effort so that the firm's new combination of effort (Omega) and ability (alpha) are sufficiently high as to not justify the costs of a takeover. For some grossly incompetent management teams (e.g., Gulf), the costs of the takeover were more than compensated by the gains from displacing the existing management. The reason that more management teams were not displaced is that the costs of a takeover exceeded the expected increase in profits. Thus, this interpretation of the data emphasizes that managerial ability differs significantly, but that due to the high cost of takeovers, only the most blatantly incompetent managers are replaced.

Evidence Regarding Activities within the Firm

While the most direct evidence regarding managerial inability is changes in top management, decreased managerial rent dissipation provides the most direct evidence regarding the free cash flow explanation. The early agency papers by Manne [1965] and Williamson [1963] argue that managers dissipate rents through excessive investment in staff, wage, and from expanding firm size. Accordingly, we examined employment trends over the period 1980-86 for our sample of firms. If we segregate the sample into those firms involved in takeovers, those firms restructured through stock repurchase (stock repurchase > 5 percent), and all others, we find the following mean reductions in employment: -34.5 percent for firms involved in takeovers; -35.4 percent for firms restructured via own stock repurchase; and -16.3 percent for unaffected firms. These data clearly indicate that all petroleum firms underwent a major retrenchment in employment over the period, but greater reductions were experienced where either takeovers or restructurings occurred. For example, in 1980 the combined employment of Gulf and Chevron was 99,100 employees and by 1986, Chevron (the combined firm) employed only 51,000 workers--a 48.4 percent attrition. Even for firms restructuring via stock repurchase, the reduction was equally severe; Arco and Exxon reduced their respective workforces by 50.1 percent and 42.4 percent.

These results are confirmed in a simple regression relationship between the percentage change in employment over the 1980-86 period and a dummy restructuring variable which equalled one if the firm restructured either through a cash acquisition or repurchase of at least 5 percent of its stock. The results are as follows:

[THIS FORMULA HAS BEEN OMITTED]

Note that even in the absence of restructuring or takeovers, employment fell on average by 16.3 percent, but restructured firms cut employment by an additional 18.7 percent. Thus, these findings support the agency theory hypothesis that managers exert greater "effort" --i.e., cut perquisites-by reducing employment.

In addition, managers may consume perquisites by implementing submarginal exploration projects. Two recent papers by Griffin [1988] and Baltagi and Griffin [1989] examine the linkage between takeovers and restructuring on the investment behavior of petroleum firms. Griffin [1988] uses a panel data set of twenty-five petroleum firms (1979-85) to nest Jensen's [1986] free cash flow within a neoclassical model of U.S. domestic exploration investment. After controlling for neoclassical investment variables, cash flow was still found to exert a significant impact on investment.

Looking at a broader grouping of investment activities, (worldwide exploration expenditures, offshore lease bidding activity, offshore lease acquisitions, and R&D expenditures), Baltagi and Griffin [1989] test the two competing models here--managerial effort versus managerial ability. They find considerable support for the managerial effort model predicated on agency theory. Hence the data offer broad support for the agency theory over the managerial incompetence theory.

IV. SUMMARY. CONCLUSIONS AND POSSIBLE EXTENSIONS

This paper presents a simple model of corporate takeovers motivated by agency_ considerations as well as the conventional managerial incompetence explanation. This model forrealizes the analysis of Jensen [1986] and others who argue the importance of agency explanations of takeovers. This analysis then serves to motivate the empirical tests separating managerial effort from managerial ability explanations for takeovers.

The empirical analysis focuses on individual firms in the petroleum industry. The evidence strongly supports the agency theory explanation for takeovers. The view that takeovers are designed to replace low-ability managers with highability managers may explain some individual cases, but it does not explain the broad evidence for the petroleum industry. This evidence includes an analysis of Tobin's q, changes in management teams, and changes in employment and investment after takeovers. All three sources of empirical evidence generally indicate that takeovers are primarily designed to address agency concerns.

There are several potentially important implications of our findings. First, if agency considerations are of broad empirical importance as a motivation for takeovers, voluntary restructuring by incumbent management may be efficient. Rather than be viewed as a mechanism to entrench incompetent management, voluntary restructuring may serve to avoid the costs of takeovers while constraining managers to exert the profit maximizing level of effort. A second implication of our findings is that agency explanations may help to explain why takeovers and mergers are most pronounced during periods of robust expansion. Positive random shocks are likely to create free cash flow situations during which agency problems are likely to be more acute. Whether these conjectures prove to have widespread empirical support must await further research.

APPENDIX

Data Description

AC = Five-year weighted average of world
- wide total finding costs excluding re
- visions. Units: $ per barrel of oil
 equivalent. Source: Arthur Andersen
 and Co. Oil and Gas Reserve Disclo
- sures: 1980-1984 Survey of 400 Public
 Companies (Houston: Arthur Ander
- sen, 1985) Table 3.
D = Sum of long-term book debt + capital
 lease obligations + current liabilities.
 (Units: 10005's). Source: Annual Re
- ports.
DMC = Dummy management charge for
 firms replacing over 50 percent of top
 management (senior vice-president
 or higher) by individuals not listed as
 part of 1979 management team. See
 Table I. Source: Annual Reports 1979,
 1985.
DRES = Dummy restructuring variable equal
 to 1 for all firms which either repur
- chased at least 5 percent of their stock
 over the period year-end 1983 to 1985
 or acquired a firm. See Table I.
 Source: Annual Reports.
E = Total year-end shareholders equity.
 Units 10005's. Source: Annual Re
- ports
%EMP= Percentage change in total employ
- ment over the period 1979 to 1985.
 Source: Annual Reports.
%REPL=Percentage of worldwide additions to
 reserves (via additions and revisions)
 relative to production over 1980
- 1984. Source: Arthur Anderson, Oil
 and Gas Reserve Disclosures. Table 15.
q = Tobin's q measuring the market value
 of the firm relative to the asset value
 as computed by the John S. Herold
 Corporation, Industry Appraisals, se
- lected issues. The numerator, market
 value of the firm is the market value
 of common shares + estimated mar
- ket value of preferred shares + esti
- mated market value of long-term
 debt + (current assets - current liabil
- ities). Source: Annual Reports and J.
 S. Herold. Also see fn. 18, 19.

REFERENCES

Adelman, Morris. "Finding and Developing Costs in the U.S.A., 1945-1985.| Working Paper, Massachusetts Institute of Technology, 1986.

Baltagi, Badi and James M. Griffin. "Alternative Models of Managerial Behavior:. Empirical Tests for the Petroleum Industry." Review of Economics and Statistics, November 1989, 579-85.

Baumol, William J. Business Behavior, Value, and Growth. New York: Macmillan, 1959. Cummins, Clint et al. "The R&D Master File." Stanford, National Bureau of Economic Research mimeo, 1985.

Griffin, James M. "A Test of the Free Cash Flow Hypothesis: Results from the Petroleum Industry." The Review of Economics and Statistics, February 1988, 76-82.

Grossman, Sanford and Oliver Hart. "Takeover Bids, the Free-Rider Problem, and the Theory of the Corporation." Bell Journal of Economics, Spring 1980, 42-64.

"Corporate Financial Structure and Managerial Incentives," in The Economics of Information and Uncertainty, edited by John J. McCall. Chicago: University of Chicago Press, 1982, 107-40.

Hirshleifer, David and Sheridan Titman. "Share Tendenng Strategies and the Success of Hostile Takeover Bids." Journal of Political Economy, April 1990, 295-324.

Jarfell, G.A., J.A. Brickley, and J.M. Netter. "The Market for Corporate Control: The Empirical Evidence Since 1980.'' Journal of Economic Perspectives, Winter 1988, 49-68.

Jensen, Michael C. ''Agency Costs of Free Cash Flow, Corporate Finance and Takeovers.'' American Economic Review, May 1986, 323-29. Manne, Henry. "Mergers and the Market for Corporate Control.' Journal of Political Economy, April 1965, 110-20.

Marris, Robin. ''A Model of Managerial Enterprise.'' Quarterly Journal of Economics, May 1963, 187-209.

Morck, Randall, Andrei Shleifer, and Robert Vishny. "Characteristics of Targets of Hostile and Friendly Takeovers," in Takeovers: Causes and Consequences, edited by Alan J. Auerbach. Chicago: University of Chicago Press, 1989.

Palepu, Krishna G. "Predicting Takeover Targets: A Methodological and Empirical Analysis.' Journal of Accounting and Economics, 8(1), 1986, 3-35.

Scharfstein, David. ''The Disciplinary Role of Takeovers." Review of Economic Studies, April 1988, 185-99.

Scherer, E M. 'Corporate Takeovers: The Efficiency Arguments." Journal of Economic Perspectives, Winter 1988, 69-82.

Shleifer, Andrei and Robert Vishny. "Large Shareholders and Corporate Control." Journal of Political Economy, June 1986, 461-88.

Williamson, Oliver. "Managerial Discretion and Business Behavior." American Economic Review, December 1963, 1032-57.