Printer Friendly

Corporate control and performance in the 1930s.

I. INTRODUCTION

A fundamental assumption of mainstream microeconomics holds that firms strive to maximize economic profits. Beginning with the work of Berle and Means |1932~, however, researchers have seriously questioned if profit maximization is the sole objective for the large modern corporation given various circumstances. In particular, Berle and Means |1932, 69~ contend that "as the ownership of corporate wealth has become more widely dispersed, ownership of that wealth and control over it have come to lie less and less in the same hands." They argue that the main beneficiaries of this transfer of control ultimately have been the managers because it has put them in the position to pursue their own interests at the expense of profits.

Modern managerialist/neoclassical theory suggests that corporate managers must be exempt from a number of behavioral constraints to successfully deviate from profit maximization and pursue the five P's of increased pay, prerequisites, power, prestige and patronage. First, as Berle and Means note, the executives must be free of a stockholder constraint. Some slack typically exists in this constraint when stock ownership is widely dispersed because an agency problem results from the asymmetry of information regarding true managerial behavior and performance. Simply put, when an individual holds a relatively small percentage of the firm's outstanding stocks, there exists little, if any, willingness (in terms of a financial incentive) to monitor true performance and ability (with regards to corporate power) to either actually or potentially discipline management. Thus, the severity of the stockholder constraint generally depends on the percentage of the firm's outstanding stock held by the dominant stockholder.

Second, according to Williamson |1963~, the product market constraint cannot be fully binding since discretionary expenditures require excess profits. In a competitive market, managers are forced to minimize costs or the firm fails to survive in the marketplace. As a result, the firm must possess some degree of market power for managers to successfully pursue other goals besides profit maximization. Third, some imperfections must exist in the capital markets. Otherwise, any inefficient behavior on the part of management increases the likelihood of a corporate takeover. In some cases, as Marris |1964~ notes, the mere threat of a takeover might sufficiently discipline management. Smiley |1976~ points out, however, that the transaction costs of a takeover gives management some room to exercise discretionary behavior before the risk of a takeover becomes intolerably high.

Fourth, Fama |1980~ argues that imperfections must also exist in the external and internal labor markets. Otherwise, job mobility, within and among firms, creates an incentive for managers to behave efficiently. In addition, better jobs and rewards in the future mean better job performance today. Finally, Jensen and Meckling |1976~ show that managers must be awarded an incomplete incentive contract. If executive compensation completely depends on performance, less inefficient behavior results.

Despite the fact that managers must be, at least, partially free of all five constraints to successfully deviate from profit maximization, a surprisingly large number of researchers, including Larner |1970~, Palmer |1973~, McEachern |1975~ and Neun and Santerre |1986~, has found evidence linking unconstrained managers with lower levels of corporate performance in more modern times.(1) These studies typically focus on the consequence of a loosening of the stockholder constraint on profitability. Researchers either explicitly or implicitly control for the severity of the product market constraint and some slack is assumed in the other three constraints on managerial behavior.(2) With the exception of Neun and Santerre |1986~, however, these studies use a dichotomous measure of stockholder control which is based upon arbitrary and subjective criteria. Moreover, these studies employ an accounting measure of corporate performance. When examining the concentration-profitability nexus, several researchers, such as Smirlock et al. |1984~, have seriously questioned the validity of accounting rates of return as an accurate measure of performance. This same concern should be relevant in this context as well.

This paper substantially improves upon earlier studies testing the Berle and Means thesis in a number of ways. One, this analysis investigates the effect of corporate control on performance during the 1930s. As Stigler and Friedland |1983~ point out, this is the appropriate period to test the Berle and Means thesis since their view of the modern corporation was formulated during this specific time period. In addition, from an empirical perspective, it is easier to identify each firm's relevant product market during the 1930s because firms were considerably less diversified. Two, this paper uses a continuous rather than dichotomous measure of stockholder control, thereby eliminating the ad hoc manner associated with measuring stockholder control in most of the studies. Three, the analysis employs both an accounting rate of return and a Tobin's q ratio, allowing us to examine the sensitivity of the empirical results with respect to both measures of corporate performance.(3)

The empirical results of this paper lend some support for the Berle and Means view of the "modern" corporation in the 1930s. In agreement with their hypothesis, a low degree of stockholder control is found to be associated with a low level of corporate performance, ceteris paribus. However, one important caveat may be in order at this point. Like the other studies on this topic, we treat ownership structure as being exogenously determined and conduct a cross-sectional analysis of the relation between ownership and performance.(4) If ownership structure is truly endogenous, as it may be within a time-series context, misspecification bias may affect the empirical results. This shortcoming should be kept in mind. Nevertheless, Berle and Means were concerned with the effect of stock dispersion on corporate performance and not specifically with the reasons for the vast dispersion of corporate ownership. Consequently, the exogeneity of stockholder control may be a valid assumption when properly testing the Berle and Means thesis.

II. EMPIRICAL MODEL

To test the Berle and Means thesis, this study uses financial and accounting data for 181 of the largest two hundred nonfinancial corporations in the United States during 1939.(5) Both regulated and unregulated firms are included in the empirical analysis. Most of the data were collected and used by Goldsmith and Parmalee |1940~ in a study commissioned by the Temporary National Economic Committee. Central to the study, Goldsmith and Parmalee identified the distribution of stockholdings among the largest twenty stockholders and the market value of the outstanding stocks of each corporation. The rest of the data are from Moodys Industrial Manual.

Following the study by Neun and Santerre |1986~, we hypothesize a nonlinear relation between the percentage of stock owned by the dominant stockholder and corporate profitability. This nonlinear relation, depicted in Figure 1, implies that some critical minimum percentage of stockholdings, |DS.sub.1~, is necessary for the dominant stockholder to exert minimal influence on managerial behavior and performance. Given the relatively small percentage of outstanding shares owned by the dominant stockholder in range 1, managers do not perceive the dominant stockholder as a viable threat to their job tenure since the dominant stockholder lacks the willingness to monitor performance and the ability (in terms of corporate power) to discipline management. As a result, corporate managers can satisfy their expense preferences through discretionary expenditures and provide a minimum profit rate of |R.sub.min~.

Beyond this minimal percentage requirement in range 2, however, managers are likely to react incrementally to increasing amounts of dominant stockholder control by providing a higher rate of return. Over this range, the typical manager's job tenure becomes increasingly threatened as the dominant stockholder increases the intensity of monitoring activities and possesses greater ability to take over the corporation. Thus, a higher rate of return is offered by management to placate the dominant stockholder at the margin. Finally, at some maximum percentage of stockholdings, |DS.sub.2~, the dominant stockholder effectively acquires full control and managers provide the maximum return to stockholders or |R.sub.max~. Additional shareholdings beyond this point in range 3 have no further impact on profitability, ceteris paribus.

This nonlinear relation between stockholder control and performance is estimated by a piece-wise linear regression technique. Accordingly, the value of DS relevant to each range of Figure 1 is generated in the following manner where |DS.sub.a~, is the actual fraction of the outstanding common stock owned by the dominant stockholder:

(1) |RANGE.sub.1~ = |DS.sub.a~ if |DS.sub.a~ |is less than~ |DS.sub.1~

= |DS.sub.1~ if |DS.sub.a~ |is greater than or equal to~ |DS.sub.1~

(2) |RANGE.sub.2~ = 0 if |DS.sub.a~ |is less than or equal to~ |DS.sub.1~

= (|DS.sub.a~ - |DS.sub.1~)

if |DS.sub.2~ |is greater than~ |DS.sub.a~ |is greater than~ |DS.sub.1~

= (|DS.sub.2~ - |DS.sub.1~)

if |DS.sub.a~ |is greater than or equal to~ |DS.sub.2~

(3) |RANGE.sub.3~ = 0 if |DS.sub.a~ |is less than or equal to~ |DS.sub.2~

= (|DS.sub.a~ - |DS.sub.2~)

if |DS.sub.a~ |is greater than~ |DS.sub.2~.

For example, suppose the values of |DS.sub.1~ and |DS.sub.2~ in Figure 1 are .05 and .45, respectively. If the dominant stockholder actually holds 4 percent of the outstanding stock, the value of |RANGE.sub.1~ equals .04 while |RANGE.sub.2~ and |RANGE.sub.3~ take on values of zero. On the other hand, if |DS.sub.a~ equals .60, |RANGE.sub.1~ takes on the value of .05, |RANGE.sub.2~ equals .40 and |RANGE.sub.3~ has a value of .15. The critical percentages, |DS.sub.1~ and |DS.sub.2~, are unknown and must be estimated. Quandt's |1960; 1958~ "switching of regimes" technique is used to estimate their values. In the context of this analysis, Quandt's technique implies that the optimal percentages for |DS.sub.1~ and |DS.sub.2~ yield the highest adjusted |R.sup.2~ when compared with all other possible percentage figures.(6)

Including various control variables that may influence corporate performance, a piece-wise linear regression equation is estimated in the following form:

(4) R = |B.sub.0~ + |B.sub.1~|RANGE.sub.1~ + |B.sub.2~|RANGE.sub.2~

+ |B.sub.3~|RANGE.sub.3~ + |B.sub.4~HHI + |B.sub.5~ ln ASSET + |B.sub.6~GROWTH + |B.sub.7~AGE + |B.sub.8~POWER + |B.sub.9~UTILITY + |B.sub.10~RAIL,

where:

R = measure of profitability as of 1939 (discussed below),

|RANGE.sub.i~ = captures the amount of stock held by the dominant stockholder in each range of Figure 1 (explained above),

HHI = Herfindahl-Hirschman index of the distribution of the remaining stocks not held by the dominant stockholder,

ln ASSET = logarithm of the book value of assets in 1939 (measure of firm-size),

GROWTH = growth of sales from 1934-1939,

AGE = age of the firm,

POWER = dummy variable taking on the value of one if an unregulated firm possessed a high degree of market power,

UTILITY = dummy variable taking on the value of one if the firm operated in the utility industry,

RAIL = dummy variable taking on the value of one if the firm operated in the railroad industry.

Following current practice, profitability is measured by both the return on equity and Tobin's q ratio. The return on equity for each firm is determined by dividing net income by the book value of equity. Tobin's q ratio is calculated for each firm by dividing the market value of the firm by the book value of assets.(7) To determine the market value of each firm, the market value of the common and preferred stocks is added to the book value of the long-term liabilities. Goldsmith and Parmalee |1940~ identify the market value of the firm's outstanding stock while Moodys Industrial Manual lists the value of the long-term liabilities.

Table I displays some descriptive statistics concerning dominant stock ownership, TABULAR DATA OMITTED Tobin's q and the accounting rate of return on equity (ROE) for the firms in the sample. In this table, dominant stock ownership is broken down into twenty cells of 5 percent. The second column reports the number of firms for which the dominant stockholder's ownership percentage falls within that cell. The performance indicators represent an unweighted average for the firms within each cell.

Not surprisingly, the distribution of dominant stock ownership is skewed toward zero. For example, the dominant stockholder owned less than 10 percent of the outstanding stock in 108 of the 181 observations. In comparison, the dominant stockholder had majority control in only fourteen firms of the sample. A skewed ownership distribution is not unusual and most, if not all, studies in this area have reported a similar distribution.(8) For the entire sample of firms, the dominant stockholder owned approximately 15.6 percent of the outstanding stock, on average. The mean q-ratio and return on equity were approximately 1.01 and .065, respectively. From a simple pairwise comparison, no systematic pattern between the degree of dominant stock ownership and firm performance can be discerned. This is because other factors influencing profitability must be held constant before any isolated pairwise correlation between these two variables can be observed.

The dominant stockholder's ability to influence corporate policy may not only depend on her share of stockholdings but also on the manner in which the rest of the stock is dispersed among the other shareholders. The exact net effect of the nondominant ownership pattern is uncertain at this point and depends on two opposing influences. On the one hand, if the remaining stocks are highly concentrated among one or a few other stockholders, the dominant stockholder's transaction costs of acquiring further control is relatively low. Hence, managers provide a higher profit rate given the greater threat of a takeover by the dominant stockholder if the rest of the stocks are more concentrated.

On the other hand, if stock ownership is concentrated among one or a few other stockholders, a free-rider problem may emerge.(9) Each individual large stockholder might anticipate that the other will incur the personal cost of a takeover. In this case, any one stockholder can realize a greater net gain by holding out for the other to initiate the takeover. If a prisoner's dilemma results, corporate managers have greater ability to pursue non-profit maximizing goals when there exists other large stockholders besides the dominant stockholder.

Whether the "transaction cost" or "free-rider" effect dominates is an empirical issue. Therefore, a Herfindahl-Hirschman index, HHI, is included in equation 4 to capture how the distribution of the stock held by investors other than the dominant stockholder affects corporate profitability. To compute this index, the fractional holdings of the next largest stockholders, up to a maximum of nineteen, are squared and added. A greater concentration of the rest of the stocks is suggested as the index increases in value, so a negative (positive) coefficient on HHI supports the free-rider (transaction cost) effect of concentrated secondary ownership.

Up to this point the discussion has focused on the influence of "external" stockholders on corporate performance. However, managers may also own some of the firm's stocks and as "internal" stockholders face a different set of incentives. Jensen and Meckling |1976~ argue that managers as internal stockholders have an increased incentive to maximize profits, while Morck, Shleifer and Vishny |1988~ postulate that managers with significant holdings become more entrenched and provide less profits. To control for the impact of management ownership, the estimated model was modified in two respects. First, a dummy variable was specified that indicates if a manager was the dominant stockholder. A manager was the dominant stockholder in 11 percent of the firms in the sample. Second, a variable was included to reflect the percent of stock held by all managers if they were among the largest twenty stockholders. Managers, as a group, possessed an average of 2.12 percent of the corporate stocks of firms in the sample. Both of these managerial variables proved to be statistically insignificant, however, and the other results remain essentially unchanged so they are not included in the reported regression results.(10)

Firm-size and growth are typically specified as control variables in profit equations. The relative profitability of larger firms depends on the existence of (dis)economies of scale so the expected value of the coefficient estimate on this variable is unknown, a priori. A positive parameter estimate is expected on GROWTH since a greater growth of sales should lead to increased profitability, ceteris paribus. The age of the firm controls for any difference in investment opportunities between young and old firms. The argument is that younger firms have more investment opportunities and therefore are more profitable. Hence, the age of the firm should inversely influence the two measures of performance.(11) The age of the firm is measured by subtracting the initial year of incorporation from 1939.

Given their large absolute size and dominance in the market or regulated status, all of the firms in the sample were most likely insulated from a "perfectly competitive" market environment. Slack in the product market constraint can provide managers with the opportunity to pursue their own self interests if the other constraints are not fully binding, as well. Although some slack in the product market constraint probably existed for the firms in the sample, some firms were likely to be more powerful in the marketplace than others. Therefore, a dummy variable, POWER, is specified that takes on the value of one if an unregulated firm (i.e., the firm does not operate in the utility or railroad industry) was likely to possess a high degree of market power. To determine the degree of market power, a study by Wilcox |1970~ is relied upon. The Wilcox study was originally published as the Investigation of Concentration of Economic Power in 1940 and is part of a larger study conducted by the Temporary National Economic Committee. This larger study examined the overall concentration of economic power during the 1930s.

A firm is classified as possessing a high degree of market power if Wilcox identified the company by name in the late 1930s as a dominant firm operating in a highly concentrated market. On the other hand, a firm is classified as having a low degree of market power if it was a dominant firm operating in a less concentrated market (e.g., Sears, Montgomery Ward) or a nondominant firm operating in a highly concentrated market (e.g., all but the "Big Four" in the steel industry).(12) Simply put, firms with a greater degree of market power had the potential to earn higher returns, ceteris paribus. As a result, the coefficient estimate on POWER should be found positive. An appendix listing the market power classification of each firm in the sample will be supplied on request. Fifty-seven of the 115 unregulated firms are classified as possessing a high degree of market power. For convenience, the following discussion refers to the "more powerful" and "less powerful" subsamples of unregulated firms when drawing comparisons.

The last two variables, UTILITY and RAIL, take on the value of one when the firm operated in the utility (n = 36) or railroad (n = 30) industry. A negative coefficient on these two industry dummy variables suggests that the typical firm in these industries earned less than the average rate of return for the less powerful subsample of unregulated firms. Closely related to this issue, Stigler and Friedland |1962~ found that the market values of regulated and unregulated utility firms were not systematically different in the 1930s. Given the particular focus of their study, however, Stigler and Friedland did not examine if utility firms earned more or less than the average rate of return in other industries. Although all the utility firms in the sample happened to be in states with utility regulations, the regression findings indicate how earnings in the regulated utility industry compared to the average rate of return of the other firms in the sample, ceteris paribus.

III. EMPIRICAL FINDINGS

Ordinary least square estimates of the parameters in equation 4 appear in Table II for both the Tobin's q and return on equity measures of corporate performance. The parameter estimate and its corresponding t-statistic (in parentheses) are shown opposite each explanatory variable. Of most interest are the results associated with the stock ownership variables. The adjusted-|R.sup.2~ reaches its highest value of .21 for the Tobin's q regression equation when |DS.sub.1~ and |DS.sub.2~ are set equal to 10 and 35 percent, respectively. Similarly, the adjusted-|R.sup.2~ attains its maximum value of .26 when |DS.sub.1~ and |DS.sub.2~ equal 10 and 25 percent in the return-on-equity regression equation. As the model predicts, the slope parameter estimates are not statistically different from zero on |RANGE.sub.1~ and |RANGE.sub.3~, while the estimated coefficient on |RANGE.sub.2~ is positive and statistically significant in both regression equations. This evidence clearly validates our premise that the relationship between dominant stock ownership and performance is nonlinear and can best be illustrated by Figure 1.(13)

Taken together, these results indicate that the dominant stockholder must have possessed a minimum of 10 percent of the outstanding stock before she had the incentive to exert even minimal pressure on management to provide a higher rate of return during the late 1930s. The incentive and ability to influence managerial behavior and performance increased incrementally beyond the 10 percent threshold level. Complete control over management was achieved when the dominant stockholder owned approximately 30 percent of the outstanding stocks (i.e., the average of 25 and 35 percent). The statistical insignificance of the parameter estimates on |RANGE.sub.3~ implies that further stockholdings had no marginal impact on profitability. In a previous study employing firm-level data from the 1980s, Neun and Santerre |1986~ estimated a cubic relation between dominant stock ownership and various accounting rates of return, and found minimum and maximum values of 17-19 percent and 52-54 percent. Here, however, piece-wise linear regression provides superior results. The different time period, specification and/or sample of firms may account for the different cut-off points in these two studies.

A backward glance at Table I reveals the possibility of some influential data points. For example, both measures of performance are strikingly higher in the 35-40 percent cell. This is because one of the three firms in this cell, Coca Cola, had a Tobin's q of 5.21 and a return on equity of .43. Upon closer visual inspection, some other possible outliers in the data were also discovered. As a result, a diagnostic technique developed by Belsley, Kuh and Welsch |1980~ was used to systematically uncover influential observations. Based upon several criteria, eight observations were eliminated from the Tobin's q regression and four observations were dropped from the return-on-equity regression that, statistically, could be considered as influential. All but two of the influential observations were associated with above average values for the performance measures.

The equations were then reestimated to determine the overall sensitivity of the results with respect to these influential observations. The magnitude and statistical significance of the regression coefficients that result from using the limited samples were comparable to those reported in Table II. |DS.sub.2~, however, was estimated to be 20 percent rather than the 25 or 35 percent figure found with the full samples. Given the similarity of the slope parameter estimates, the implausibility of |RANGE.sub.2~ being so narrow and that some important information may be lost about the true distribution of the performance variables (since high values are selectively eliminated), the following discussion is confined to the results associated with the full sample. In any event, both the full and limited samples yield similar qualitative results (quantitative differences are mentioned in the footnotes) and general conclusions concerning the Berle and Means hypothesis.
 TABLE II
 Unrestricted Piece-Wise Linear Regression Results

 Dependent Variable(*)
 Tobin's q ratio ROE

Constant 1.38(**)
0.11(**)
 (4.83) (3.99)

|RANGE.sub.1~ -0.020 -0.0019
 (1.34) (1.32)

|RANGE.sub.2~ 0.024(**)
0.0024(**)
 (3.60) (2.42)

|RANGE.sub.3~ -0.009 -0.0003
 (1.81) (0.71)

HERFINDAHL INDEX -0.0010(**) -0.00007
 (2.30) (1.72)

ln ASSET -0.069 -0.008
 (1.37) (1.59)

GROWTH 0.396(**)
0.051(**)
 (2.93) (3.87)

AGE -0.0004 -0.0001
 (0.22) (0.71)

POWER 0.247(**)
0.017(**)
 (2.34) (1.72)

UTILITY -0.104 -0.012
 (0.78) (0.93)

RAIL -0.341(**)
-0.046(**)
 (2.33) (3.29)

|DS.sub.1~ 10 10

|DS.sub.2~ 35 25

Adjusted-|R.sup.2~ .21 .26

F-statistic 5.79 7.38

Observations 181 180

* The absolute value of the t-statistics appear in parentheses
below the estimated coefficients.

** Significant at the 5 percent level or better.


To provide a stricter test of the theoretical model embedded in Figure 1, equation 4 is reestimated for both measures of performance and the coefficient estimates are restricted to be zero on |RANGE.sub.1~ and |RANGE.sub.3~. This is accomplished by eliminating the |RANGE.sub.1~ and |RANGE.sub.3~ variables from the regression equation. The F-statistics for the difference between the unrestricted and restricted models are F(2, 170) = 2.41 and F(2, 169) = 1.11 for the q-ratio and return-on-equity measures of performance. Given the 5 percent critical value of 3.05, the null hypothesis of the restricted model, which more closely conforms to the empirical specification implied in Figure 1, could not be rejected. The restricted piece-wise linear regression estimates appear in Table III. Maximum values for the adjusted-|R.sup.2~s (.20 and .26) are achieved with |DS.sub.1~ and |DS.sub.2~ set equal to the same values as before.

Overall, the results are similar to the unrestricted piece-wise linear regression model in Table II. The parameter estimate on |RANGE.sub.2~ is positive and significant in both equations, as expected. Using the various parameter estimates in Table III and sample means, the profit differential that resulted from different degrees of corporate ownership by the dominant stockholder is calculated. For example, if the dominant stockholder owned 10 percent or less of the outstanding stock, the regression estimates suggest that the predicted Tobin's q was approximately .93 and the predicted return on equity was about 5.8 percent. On the other hand, if the dominant stockholder possessed full control by owning the necessary percent of the firm's outstanding stock (i.e., 35 or 25 percent depending on the specific measure of performance), the Tobin's q was 1.28 and return on equity was 7.9 percent. Thus, if a dominant stockholder with 10 percent of the firm's stocks acquired the additional shares to effectively attain full control, management reacted by increasing the return on equity by 36 percent and the market value of the company by 38 percent.(14)

While managerial behavior appeared responsive to a large absolute increase in stockholder control during the late 1930s, a small percentage change in stockholder control had little impact on corporate performance. For example, using sample averages and the estimated parameter on |RANGE.sub.2~ in Table III, a profit elasticity estimate with respect to stockholder control is computed. The Tobin's q and return-on-equity elasticity estimates are approximately .16 and .15 when evaluated at the mean point along range 2 of Figure 1.(15) Thus, a 10 percent increase in the percentage of stock held by the dominant stockholder would have caused profitability to increase by only 1.5 percent. Obviously, managerial behavior was not very sensitive at the margin to increasing control in range 2 of Figure 1 (i.e., the line of range 2 has a relatively flat slope).

With respect to the control variables, the parameter estimate on HHI is negative and statistically significant in three of the four regression equations reported in the two tables. These findings lend support for the premise that a free-rider problem emerges when stock ownership is concentrated among a few large stockholders. In this case, as each large stockholder held out for the other(s) to initiate a takeover, the probability of a takeover was lessened. As a result, it was easier for management to pursue its own objectives rather than profit maximization.

Firm size, as measured by the logarithm of assets, appears to have had no marginal impact on profitability. In each regression equation, the estimated coefficient on firm size is consistently negative but not statistically significant from zero using a two-tailed test. The parameter estimates on the GROWTH variable are all positive, as expected, and statistically significant. The age of the firm has the expected inverse impact on profitability, but the relation is statistically insignificant.
 TABLE III
 Restricted Piece-Wise Linear Regression Results

 Dependent Variable(*)
 Tobin's q ratio ROE

Constant 1.22(**)
0.095(**)
 (4.58) (3.73)

|RANGE.sub.2~ 0.014(**)
0.0014(**)
 (2.84) (2.07)

HERFINDAHL INDEX -0.0009(**)
-0.00008(**)
 (2.38) (2.05)

ln ASSET -0.057 -0.007
 (1.13) (1.36)

GROWTH 0.38(**)
0.049(**)
 (2.78) (3.73)

AGE -0.0005
-0.0001
 (0.24) (0.65)

POWER 0.26(**)
0.019(**)
 (2.49) (1.85)

UTILITY -0.15 -0.015
 (1.14) (1.21)

RAIL -0.36(**)
-0.049(**)
 (2.45) (3.55)

|DS.sub.1~ 10 10

|DS.sub.2~ 35 25

Adjusted-|R.sup.2~ .20 .26

F-statistic 6.51 8.93

Observations 181 180

* The absolute value of the t-statistics appear in parentheses
below the estimated coefficients.

** Significant at the 5 percent level or better.


The parameter estimate on the market power dummy variable is positive and significant in all four equations for a one-tailed test. The coefficient estimate on this dummy variable in the Tobin's q equation implies that the market value of a more powerful firm was about 28 percent higher than an otherwise comparable firm with a lower degree of market power.(16) Likewise, the coefficient estimate on this same variable in the return-on-equity equation suggests that the more powerful firms earned a 32 percent higher rate of return on equity during this period.

The empirical results connected with the regulated industries are interesting and merit discussion. In all of the regression equations, the parameter estimates are negative and significant on the railroad industry dummy variable. The estimates indicate that during the latter part of the 1930s, the railroad industry earned a rate of profit which was between 38 (Tobin's q) to 78 (return on equity) percent lower than the average rate of return of the less powerful subset of unregulated firms. Wilcox |1970, 90~ points out that although railroads were one of the strongest monopolies during the nineteenth century (especially in some specific product lines), they were "compelled to face severe competition in the twentieth" century which engaged them "in a struggle for existence." Competition from trucking, cars, buses, water transport, airplanes and pipelines eroded the monopoly position that railroads once held. Interestingly, McFarland |1987, 385~ notes that railroads continued to earn a below average rate of return even after passage of the Staggers Act of 1980 which all but eliminated government regulation of railroad rates. Finally, the negative but insignificant parameter estimates on the utility industry dummy variable, in conjunction with the study by Stigler and Friedland |1962~, indicate that both regulated and unregulated utility firms in the 1930s earned a rate of return comparable to the less powerful subset of unregulated firms. Given the underlying premise that all firms are insulated from a perfectly competitive market environment, this result may also imply that the average utility firm earned more than a perfectly competitive or normal rate of return.

IV. CONCLUSION

Using period data, the Berle and Means hypothesis is tested. Our findings lend some support for their notion that the degree of stock dispersion influenced corporate performance during the 1930s. In fact, if the average fully manager-controlled firm was miraculously transformed into a fully owner-controlled firm during this time, the empirical results suggest that the market value of the firm would increase, at most, by 38 percent. However, a transformation of this kind would require the dominant stockholder to increase her holdings from a minimum of 10 percent to about 30 percent of the outstanding stock. This would represent a sizeable investment. The estimates suggest that it would have cost approximately $45 million in 1938 dollars (or about $450 million in 1990 dollars) for a stockholder to acquire the minimum additional amount of 20 percent of the outstanding stock to gain full control.(17)

The findings also imply that managerial behavior was not very responsive to increasing control when the stockholder constraint was already partially binding. For example, if the dominant stockholder owned 16 percent of the outstanding stock and increased her/his holdings by 10 percent, the market value of the firm would increase, at most, by 1.6 percent. Lastly, it appears that the dominant stockholder's ability to influence corporate performance was greatest when the remaining shares were widely dispersed.
 APPENDIX
 Piece-Wise Linear Regression Results Using the Limited
Sample

 Dependent Variable(*)
 Tobin's q ratio ROE

Constant 1.26(**) 0.10(**)
 (6.47) (4.47)

|RANGE.sub.1~ -0.014 -0.0019
 (1.39) (1.57)

|RANGE.sub.2~ 0.027(**)
0.0022(**)
 (2.65) (1.86)

|RANGE.sub.3~ -0.002 -0.0001
 (0.60) (0.38)

HERFINDAHL INDEX -0.0006 -0.00005
 (1.95) (1.45)

ln ASSET -0.065 -0.007
 (1.90) (1.69)

GROWTH 0.190(**)
0.041(**)
 (2.06) (3.82)

AGE 0.00001 -0.0002
 (0.11) (1.18)

POWER 0.254(**)
0.026(**)
 (3.50) (3.14)

UTILITY -0.027 -0.005
 (0.30) (0.46)

RAIL -0.222(**)
-0.034(**)
 (2.24) (3.14)

|DS.sub.1~ 10 10

|DS.sub.2~ 20 20

Adjusted-|R.sup.2~ .22 .32

F-statistic 5.96 10.05

Observations 173 176

* The absolute value of the t-statistics appear in parentheses
below the estimated coefficients.

** Significant at the 5 percent level or better.


1. But see Qualls |1976, Mueller |1986~ and Demsetz and Lehn |1985~, who find no statistically significant relation between control and performance when using data from recent time periods. Employing a subset of the firms in this study and a different empirical framework, Stigler and Friedland |1983~ find no relation between control and performance in the 1930s.

2. Firms in the Fortune 500 are typically used as the sample because they most likely possess some degree of monopoly power.

3. Morck, Shleifer and Vishny |1988~ use both of these profitability measures to examine the effect of management ownership on firm performance. Their study does not consider the effect of the stockholder constraint on managerial behavior, however.

4. Given the lack of any suitable instrumental variables, as suggested by Demsetz and Lehn |1985~, two-stage least square estimates are not derived.

5. For a few observations, stock values for 1937 had to be used. Nineteen of the original 200 observations could not be used in the empirical analysis because of missing data or because the firms were privately owned so the market value of their stocks is unknown.

6. For the sake of manageability, the percentage figures are changed by increments of 5 percent until the adjusted |R.sup.2~ is maximized.

7. From a theoretical perspective, the replacement cost of the assets is more appropriate than book value; however, replacement cost data are unavailable. Given the relative stability of prices during the late 1930s, replacement costs and book value may only differ by a small amount. Since book value rather than the replacement cost of assets is used, this ratio could also be referred to as Marris's |1964~ valuation ratio. We thank a referee for pointing this out.

8. For example, see Morck, Shleifer and Vishny |1988~.

9. Shleifer and Vishny |1986~ present a theoretical model where the presence of a large minority stockholder provides a solution to the free-rider problem but they "do not consider the strategic behavior between large shareholders" (p. 463).

10. These results will be provided by the authors on request.

11. A referee recommended we include firm-age in the regression equation.

12. Santerre and Neun |1989~ use this same market power classification scheme.

13. An F-test is performed to compare a regression model with a spline to a regression model without a spline. The F-statistic for the Tobin's q equation was F(2, 170) = 5.17 and the one for the return-on-equity equation was F(2, 169) = 2.54, while the 5 and 1 percent critical values are 3.05 and 4.73. This is regarded as evidence that the null hypothesis of no spline can be rejected. In addition, the adjusted |R.sup.2~ for the Tobin's q equation decreases from .21 (with a spline) to .17 (without a spline). For the return-on-equity equation, the adjusted |R.sup.2~ falls from .26 to .25.

14. With the influential observations eliminated, the comparable coefficient estimate (and corresponding t-statistic) on |RANGE.sub.2~ is .018 (2.47) in the Tobin's q equation and .0011 (1.40) in the return-on-equity equation. Accordingly, the return on equity and market value increased by 20 and 21 percent when the dominant stockholder acquired the additional 10 percent of stocks to gain full control (since |DS.sub.2~ equals 20 percent in this case).

15. The comparable elasticities for the limited samples are .10 and .09, respectively.

16. The predicted values of firm performance are determined by substituting the appropriate dummy variable sequence along with the mean values of the remaining independent variables into the fitted restricted piece-wise regression equations.

17. This figure represents 20 percent of the market value of the typical fully manager-controlled firm in the sample ($222.5 million) and is a lower bound since the market value will most likely increase as the dominant stockholder signals her intention to acquire more stock in the company. If |DS.sub.2~ equals 20 percent, as the limited sample suggests, the cost of attaining full control equals $22 million.

REFERENCES

Belsley, David A., Edwin Kuh and Roy E. Welsch. Regression Diagnostics. New York: John Wiley, 1980.

Berle, Jr., Adolph A., and Gardiner C. Means. The Modern Corporation and Private Property. New York: The Macmillan Co., 1932.

Demsetz, Harold, and Kenneth Lehn. "The Structure of Corporate Ownership: Causes and Consequences." Journal of Political Economy, December 1985, 1155-77.

Fama, Eugene F. "Agency Problems and the Theory of the Firm." Journal of Political Economy, April 1980, 272-84.

Goldsmith, Raymond W., and Rexford C. Parmalee. The Distribution of Ownership of the 200 Largest Nonfinancial Corporations. The Temporary National Economic Committee, Monograph No. 29, U.S. Government Printing Office, 1940.

Jensen, Michael C., and William H. Meckling. "Theory of the Firm, Managerial Behavior, Agency Costs and Ownership Structure." Journal of Financial Economics, October 1976, 305-60.

Larner, Robert J. Management Control and the Large Corporation. Cambridge: Dunellen, 1970.

McEachern, William A. Management Control and Performance. Lexington, MA: Heath and Company, 1975.

McFarland, Henry. "Did Railroad Deregulation Lead to Monopoly Pricing? An Application of Q." Journal of Business, July 1987, 385-400.

Marris, Robin. The Theory of 'Managerial' Capitalism. London: Macmillan, 1964.

Morck, Randall, Andrei Shleifer and Robert Vishny. "Management Ownership and Market Valuation." Journal of Financial Economics, January/February 1988, 293-315.

Mueller, Dennis C. Profits in the Long Run. Cambridge: Cambridge University Press, 1986.

Neun, Stephen P., and Rexford E. Santerre. "Dominant Stockownership and Profitability." Managerial and Decision Economics, September 1986, 207-10.

Palmer, John. "The Profit-Performance Effects of the Separation of Ownership from Control in Large U.S. Industrial Corporations." Bell Journal of Economics and Management Science, Spring 1973, 293-303.

Qualls, P. David. "Market Power and Managerial Behavior," in Essays in Industrial Organization in Honor of Joe S. Bain, edited by Robert Masson and P. D. Qualls. Cambridge, MA.: Ballinger, 1976, 89-104.

Quandt, Richard E. "The Estimation of the Parameters of a Linear Regression System Obeying Two Separate Regimes." Journal of the American Statistical Association, December 1958, 873-80.

-----. "Tests of the Hypothesis That a Linear Regression System Obeys Two Separate Regimes." Journal of the American Statistical Association, December 1960, 324-30.

Santerre, Rexford E., and Stephen P. Neun. "Managerial Control and Compensation in the 1930s: A Reexamination." Quarterly Journal of Business and Economics, Autumn 1989, 100-18.

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

Smiley, Robert. "Tender Offers, Transactions Costs and the Theory of the Firm." Review of Economics and Statistics, February 1976, 22-32.

Smirlock, Michael, Thomas Gilligan and William Marshall. "Tobin's q and the Structure-Performance Relationship." American Economic Review, December 1984, 1051-60.

Stigler, George S., and Claire Friedland. "What Can Regulators Regulate? The Case of Electricity." Journal of Law and Economics, October 1962, 1-16.

-----. "The Literature of Economics: The Case of Berle and Means." Journal of Law and Economics, June 1983, 237-68.

Wilcox, Clair. Competition and Monopoly in American Industry. Westport, CT.: Greenwood Press, 1970.

Williamson, Oliver. "Managerial Discretion and Business Behavior." American Economic Review, December 1963, 1032-57.
COPYRIGHT 1993 Western Economic Association International
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1993 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Santerre, Rexford E.; Neun, Stephen P.
Publication:Economic Inquiry
Date:Jul 1, 1993
Words:6818
Previous Article:Frank Knight on risk, uncertainty, and the firm: a new interpretation.
Next Article:Pot-bellies, cattle breeds and revealing signals.
Topics:

Terms of use | Copyright © 2016 Farlex, Inc. | Feedback | For webmasters