Corporate Governance and Firm Diversification

Financial Management (Financial Management Association),  Spring, 2000  by Ronald C. Anderson,  Thomas W. Bates,  John M. Bizjak,  Michael L. Lemmon

• E-mail
• Print
• Link

IV. The Components of Corporate Governance and Firm Diversification

To provide further evidence on how diversification is related to governance structure, we examine compensation policy, ownership and board structure, and CEO turnover in a multivariate framework and discuss our findings below. Much of the analysis that follows focuses on the cross-sectional differences in governance characteristics between single- and multi-segment firms. Because of the panel nature of our data, however, simple OLS pooled time-series cross-section regressions are likely to overstate the statistical significance of our results due to serial and cross correlation in the error terms. To address these issues, we use Fama-MacBeth (1973) regression procedures. Specifically, we run cross-sectional regressions in each year. The coefficient estimates and p-values are then calculated from the time-series mean and standard error of the yearly coefficients, respectively. We have also investigated the time-series relation between corporate governance and diversification using panel regressions with fixed -effects by firm and year with similar results (results not reported). These within-firms test lack power, however, because there is little time-series variation in our proxy for diversification. We examine some time-series evidence regarding changes in diversification in Section VI.

A. CEO Compensation and Turnover

The sensitivity of CEO pay to firm performance is an important facet of corporate governance (e.g., Jensen and Murphy, 1990) and may be correlated with the degree of diversification for at least two reasons. On the one hand, if diversification is the result of a failure in internal governance mechanisms, then a lower sensitivity of pay-for-performance in diversified firms will be indicative of managerial rent expropriation. On the other hand, as suggested by Paul (1992), relative to alternative governance mechanisms, tying pay to firm performance may be an increasingly costly way to align managerial incentives in diversified firms.

The regressions in Table IV illustrate the relation between diversification and the sensitivity of salary and bonus, stock option value, and total compensation to firm performance. [7] Following the methodology of Jensen and Murphy (1990) and Crawford, Ezzell, and Miles (1995), we regress the first difference of dollar compensation on the change in shareholder wealth, the change in accounting income, a dummy equal to one for multi-segment firms, and interaction terms involving the multi-segment dummy. In using first differences, we deleted observations for first-year CEOs because these managers are likely to receive a different pay package than that of the outgoing CEO. Changes in shareholder wealth and accounting income are measured in thousands of dollars.

In Table IV, we find that the coefficients on the interaction between the multi-segment dummy and market performance is negative and significant at the 0.10 level or better in models 1 and 3. These results imply that salary and bonus, and total compensation, are less sensitive to stock-price performance in diversified companies. The magnitude of the coefficient estimates in Model 3 indicate that total compensation for CEOs in single-segment firms increases by $1.40 per $1,000 change in shareholder wealth. For diversified firms total compensation increases by only $1.11 (per $1,000 change in shareholder wealth). The difference of $0.29 is both statistically and economically significant, representing a drop in pay-performance sensitivity of approximately 20% in diversified firms relative to single-segment firms. [8] Our evidence shows that managers in diversified firms are subjected to less performance-based pay when compared to single-segment firms. As argued previously, however, these findings by themselves do not allow us to effectively discriminate between the agency cost hypothesis and the efficient contracting hypothesis.