Operating Efficiency and Corporate Governance

Transcription

1 Operating Efficiency and Corporate Governance Philip H. Dybvig and Mitch Warachka August 2009 Abstract We examine the economic implications of corporate governance. With governance determining the amount a firm invests in capital, book value is endogenous when evaluating corporate governance. In particular, underinvestment by entrenched managers confounds the ability of Tobin s Q to capture the economic implications of weak governance. Therefore, we provide two operating efficiency measures based on a firm s revenue and cost of production. The empirical estimation of these measures indicates that operating efficiency improves with better corporate governance. Keywords: Corporate Governance, Operating Efficiency, Tobin s Q We thank Nina Baranchuk, Long Chen, Alex Edmans, Fangjian Fu, Vidhan Goyal, Mike Lemmon, Michelle Lowry, James Ohlson, Robert Prilmeier, Neal Stoughton, Rong Wang, and Yajun Wang for their helpful comments and suggestions. We also thank Phuong Thanh Doan for her excellent research assistance. Boatmens Bancshares Professor of Banking and Finance, Washington University in Saint Louis, Olin School of Business, Campus Box 1133, One Brookings Drive, Saint Louis, MO phild@dybfin.wustl.edu Singapore Management University, L.K.C. School of Business, 50 Stamford Road, , Singapore. mitchell@smu.edu.sg 1

2 1 Introduction Empirical finance often requires proxies for variables of interest. However, empirical proxies must be chosen carefully since inappropriate proxies can cause a hypothesis to be spuriously rejected or accepted. Indeed, empirical research often involves joint tests of the stated hypotheses and the validity of the chosen proxies. Ideally, empirical proxies would originate from a theoretical model that justifies their use under different assumptions. This paper provides theoretically-motivated empirical proxies for measuring the economic implications of corporate governance. In particular, we improve upon the use of Tobin s Q, the ratio of market value to book value, when evaluating the economic implications of corporate governance. Empirical proxies that measure these implications are of paramount importance when studying the role of governance in corporate activities such as acquisitions or studying the economic impact of government regulations regarding corporate governance. Our framework demonstrates that using Tobin s Q to evaluate governance is flawed due to an endogeneity problem. We then offer a solution to this endogeneity problem by examining managerial decisions regarding a firm s scale and its cost of production. For the single-product firm in our framework, scale is defined as the number of units of production. The operating efficiency measures we propose arise from the maximization of firm value net of the capital invested by shareholders. Our measures of operating efficiency are also consistent with recent empirical and theoretical research. Gompers, Ishii and Metrick (2003) conclude that firms with more share- 2

3 holder rights are better governed since they have a higher Tobin s Q. 1 However, is not clear that Tobin s Q is a suitable proxy for assessing the economic implications of corporate governance. Intuitively, good managers maximize firm value in the numerator of Tobin s Q, while dividing by book value in the denominator controls for the amount of invested capital. Unfortunately, when evaluating governance, book value is endogenous because capital depends on governance. In particular, entrenched managers can enjoy the quite life by operating below their firm s profit-maximizing scale (Bertrand and Mullainathan, 2003), which coincides with an underinvestment in capital that increases Tobin s Q. Higher profit margins, less monitoring, and a lower likelihood of being fired due to negative demand shocks make underinvestment attractive to managers. Our simple single-product theoretical framework assumes managers choose the amount of output their firms produce and control costs. The ideal manager maximizes their firm s market value net of invested capital. Operating at a suboptimal scale and lax cost discipline result in deviations from this objective. For scale to matter, we assume that firms have some monopoly power in their product market. Operating at a suboptimal scale results in a proportional reduction in capital and the cost of production, but causes a less than proportional reduction in revenue since demand curve slope downward. In particular, as marginal revenue decreases with output, operating at a suboptimal scale results in a firm s output price and its marginal profit being higher than optimal. Moreover, with market value net of invested 1 In an earlier study, Morck, Shleifer, and Vishny (1988) interpret the non-monotonic relationship between managerial ownership and Tobin s Q as evidence that managerial ownership has countervailing influences on corporate governance. 3

4 capital being maximized when marginal profits are zero, operating at a suboptimal scale also causes the ratio of market value divided by capital, which is approximately Tobin s Q, to be higher than optimal. 2 Consequently, by mitigating underinvestment, better governance can decrease Tobin s Q. Conversely, better governance decreases costs, which increases Tobin s Q. Thus, the net impact of governance on Tobin s Q is ambiguous. Our framework also demonstrates that return on assets and return on equity are ambiguous with respect to corporate governance. In particular, a high return on equity can either be attributed to underinvestment or stringent cost controls, signifying weak or strong governance, respectively. The ambiguous influence of governance on Tobin s Q leads us to propose measures of operating efficiency that examine revenue, which captures scale, and cost of production separately. After normalizing revenue and the cost of production by capital, these operating efficiency measures are both decreasing functions of corporate governance and are justified by the maximization of market value net of capital invested by shareholders. Our framework assumes that weak governance leads to underinvestment and higher costs. These assumptions are supported by the empirical evidence in Bertrand and Mullainathan (2003) that entrenched managers enjoy the quiet life and underinvest. Aggarwal and Samwick (2006) confirm that weak governance results in underinvestment and conclude that managers incur private costs when investing such as the need for additional monitoring. John and Knyazeva (2006) also report that poor corporate governance leads 2 Maksimovic and Phillips (2001, 2002) conclude that managers allocate resources to ensure that marginal profits are zero. 4

5 to underinvestment. Moreover, Cronqvist, Heyman, Nilsson, Svaleryd, and Vlachos (2009) conclude that entrenched managers obtain private benefits by paying higher wages, while Core, Holthausen, and Larcker (1999) find that weaker corporate governance results in higher CEO compensation. Giroud and Mueller (2007) provide additional evidence that weaker governance in less competitive industries leads to higher costs. Mergers and acquisitions can increase operating efficiency. Our operating efficiency measures are able to capture the benefits of increased scale and cost-savings resulting from acquisitions. Conversely, mergers and acquisitions can reduce operating efficiency by enabling entrenched managers to assemble a collection of inefficient enterprises. While our framework investigates a single-product firm, the operating efficiency of a conglomerate is the combined operating efficiency of its individual divisions. Thus, a combination of inefficient enterprises is identified as being inefficient by our proxies for operating efficiency. After normalizing firm values by sales (assets), Graham, Lemmon, and Wolf (2002) find that target firms have lower normalized valuations than acquiring firms. This finding challenges the assumption that acquisitions destroy value if they lower an acquiring firm s normalized valuation, and is similar to our observation that increasing output until marginal profits are zero reduces Tobin s Q. While our operating efficiency measures capture the economic implications of corporate governance, they are not intended to replace or justify existing proxies for corporate governance. The corporate governance index of Gompers, Ishii, and Metrick (2003), abbreviated as the G index hereafter, assigns firms a score between zero and twenty-four by counting the number of 5

6 their charter provisions that inhibit the replacement of management. Firms with a higher G index are viewed as having less shareholder rights. The B index of Bebchuk, Cohen, and Ferrell (2009) simplifies the G index by focusing on provisions related to staggered boards, limits to bylaw amendments, poison pills, golden parachutes, and supermajority requirements. To clarify, the G and B indices are not theoretically-motivated by our framework. We also use institutional ownership as a proxy for corporate governance. Shleifer and Vishny (1986) argue that institutional investors improve governance by facilitating takeovers, while Cremers and Nair (2005) report that the G index complements internal governance mechanisms such as the presence of institutional investors. Our operating efficiency measures are estimated using COMPUSTAT data. Fama and French (1997) industry classifications control for differences in firm-level revenue and cost of goods sold across industries. Our empirical analysis finds that both measures of operating efficiency are captured by the G index and institutional ownership. Specifically, a high G index and low institutional ownership, which correspond to weaker governance, are associated with underinvestment and higher costs. Furthermore, Tobin s Q is lower for firms with better operating efficiency. Overall, improved governance increases operating efficiency but not Tobin s Q. The remainder of this paper begins in Section 2 with the introduction of our framework. The operating efficiency measures derived from this framework are then estimated in Section 3. Section 4 contains our conclusions and suggestions for future research. 6

7 2 Theoretical Framework Our intention is to provide a general framework for analyzing the economic implications of corporate governance rather than construct a detailed structural model. In our model, corporate managers are entrusted with two crucial tasks. They determine their firm s scale and control its costs. Therefore, our framework has management determining output denoted y and the (per unit) production cost denoted c, where c 0 > 0 denotes the lowest possible cost of producing one unit of output. With c being constant, the firm s total cost of production equals cy. 3 The amount of capital required to produce one unit of output equals k > 0. A linear production function implies the total capital that requires financing by shareholders equals ky. Consequently, capital depends on a firm s corporate governance through its dependent on y. In empirical applications, capital often refers to book value. 4 The discount rate for future cashflows as well as the rental rate and financing cost for capital all equal r. However, neither managers nor investors are assumed to be risk-neutral. Instead, our framework utilizes risk-neutral probabilities. Core, Guay, and Rusticus (2006) document that the expected stock returns are not sensitive to governance. The price of the firm s output is determined by managerial decisions regarding its level of output since firms are assumed to earn some monopoly 3 Economies of scale imply that c is a decreasing function of output and are compatible with our framework. Conversely, diseconomies of scale imply that c is an increasing function of output, and allows underinvestment to lower c. 4 Property, plant, and equipment (PPE) can proxy for capital and avoids complications arising from intangible assets. Our empirical implementation examines PPE in a robustness test. 7

8 rents in their respective product market. The following demand function determines the price (per unit) of the firm s output Price(y) = P 0 y. (1) The a parameter in this downward-sloping demand function links prices with output. A large a parameter indicates that prices are insensitive to output. 5 Our framework involves the following observables: cashflow, market value, book value, revenue, and the cost of production. The reduced-form relationship between these observables and corporate governance is discussed in the following subsections. However, our framework does not solve for the optimal level of governance. 2.1 Tobin s Q With revenue equaling output y times the price in equation (1), the firm s revenue minus its cost of good sold equals y ( P 0 y ) cy. This quantity represents the firm s earnings and results in its market value being M(y, c) = i=1 y ( P 0 ) y cy (1 + r) i = y (P 0 c) y2 r. (2) With book value proxying for ky (capital), Tobin s Q equals Q(y, c) = P 0 c y. (3) kr 5 The a parameter does not determine the optimal size of individual firms nor the number of firms within an industry. 8

9 Observe that Tobin s Q is a decreasing linear function of y as the partial derivative Q y equals 1. Thus, Tobin s Q is less sensitive to output when prices are also less sensitive to output. More importantly, improvements in corporate governance have an ambiguous influence on Tobin s Q. Better governance decreases c while increasing y, which causes Tobin s Q to increase and decrease, respectively. Nonetheless, provided management decisions regarding scale are economically more important than cost discipline, improvements in corporate governance decrease Tobin s Q. This decrease is more likely when prices are sensitive to output (a small). When regressing Tobin s Q on proxies for corporate governance, the addition of book value (or other proxies for capital) as an independent variable does not alleviate its endogeneity since the influence of governance on Tobin s Q remains ambiguous. Furthermore, underinvestment pertains to the difference between a firm s optimal scale and the scale chosen by its managers, although only proxies for the latter are observable. Cashflow depends on its operating leverage and financial leverage. With a fraction 0 f < 1 of the firm s capital being rented or financed by borrowing, its cashflow equals ( y P 0 c y ) frky = y (P 0 c frk) y2, (4) where f rky denotes the rental and financing cost. The cashflow in equation (4) is generated from a capital investment by shareholders of (1 f)ky. 9

10 After dividing this cashflow by (1 f)ky, return on equity (ROE) equals ROE = P 0 c y frk, (5) (1 f)k = r + P 0 c y rk (1 f)k = r ( ) Q f. 1 f ROE is estimated from realized cashflow while the numerator in Tobin s Q involves cashflow expectations over multiple future horizons. Nonetheless, as with Tobin s Q, corporate governance has an ambiguous influence on ROE since better governance decreases marginal revenue and the marginal cost of production. This ambiguity also applies to return on assets (ROA), which is a special case of equation (5) where f = 0 since ROA does not depend on the financing of capital. 2.2 Operating Efficiency Ideally, managers maximize market value minus the capital that shareholders are required to finance max c,y M ky = max c,y y (P 0 c rk) y2. (6) r Equation (6) is a concave function with respect to y that is maximized by y = a (P 0 c 0 rk), (7) 10

11 where c = c 0. However, entrenched managers can underinvest by producing less than y and shirk their responsibility to control costs by having c exceed c 0. Using the maximization in equation (6), we propose two operating efficiency ratios to measure the economic implications of corporate governance. These operating efficiency measures are derived from revenue, defined as y ( P 0 y ) according to the demand function in equation (1), and the cost of production, which equals cy. The first operating efficiency ratio, R y, isolates the impact of corporate governance on managerial decisions regarding scale through revenue Revenue Capital = y ( P 0 y ) ky = P 0 y k P 0 + c 0 + rk 2k, (8) with the lower bound being independent of a after invoking y from equation (7). The normalization by capital ensures that R y is a decreasing function of y, although revenue itself is quadratic. In particular, R y is smaller for firms whose output y is closer to y as these firms have less underinvestment. Observe that the influence of corporate governance on R y is not complicated by production costs. The second operating efficiency ratio, R c, isolates the impact of corporate governance on the cost of production Production Cost Capital = cy ky = c k c 0 k. (9) The normalization by capital ensures that R c is not complicated by man- 11

12 agement decisions regarding output. R c is smaller for firms with better governance since c decreases with better governance. 6 Observe that R y and R c are independent of investor expectations regarding future earnings and cashflow. Moreover, as better governance results in lower marginal costs and lower marginal revenue, the net impact of governance on operating profit is ambiguous. Consequently, distinct measures of operating efficiency that separate revenue from the cost of production are necessary. Indeed, the net influence of governance on Tobin s Q is an empirical question that depends on the relative importance of decisions regarding scale versus cost discipline. Furthermore, according to equation (2), the earnings-to-price ratio equals r and is consequently independent of governance. With operating profit being approximately equal to revenue minus the cost of production, return on equity (ROE) is proportional to the difference between our operating efficiency measures; ROE R y R c. This property highlights the ambiguity of ROE with respect to corporate governance. For example, a high ROE can be attributed to a high capital-adjusted revenue, which signifies weak governance, or a low capital-adjusted cost of production, which signifies strong governance. The ambiguous influence of corporate governance on operating profit undermines the ability of return on equity as well as return on assets to capture the economic implications of corporate governance. Acquisitions can improve operating efficiency through synergies and cost- 6 Equation (2) implies Revenue = Production Cost + rm with maximizing M B being equivalent to maximizing ( Revenue - Production Cost ) / r. 12

13 savings. Conversely, acquisitions can reduce operating efficiency by creating a collection of diverse enterprises with insufficient scale. While our framework investigates a single-product firm, the combined operating efficiency of two divisions equals their combined efficiency. Specifically, the combined revenue-based operating efficiency measure for two divisions equals Combined Revenue Combined Capital ) ) y 1 (P 0 y y 2 (P 0 y 2 2 = ky 1 + ky ( ) 2 y1 P0 + c 0 + rk y 1 + y 2 2k = P 0 + c 0 + rk. 2k ( ) y2 P0 + c 0 + rk + y 1 + y 2 2k Thus, for a conglomerate, a lower R y measure continues to signify better operating efficiency. Similarly, the following holds for our cost-based measure of operating efficiency Combined Production Cost Combined Capital = c 1y 1 + c 2 y 2 ky 1 + ky 2 c 0 k = c 0 k. ( y1 y 1 + y 2 ) + c ( ) 0 y2 k y 1 + y 2 Therefore, our operating efficiency measures are valid for conglomerates. Nonetheless, our framework has several limitations. First, managerial decisions regarding product attributes are not considered. Second, capital structure is not examined, although managers may play it safe by using less debt and sacrificing the tax benefits of leverage. Third, managers can purchase excess capital to produce a suboptimal amount of output. This 13

14 overinvestment can be identified through a normalization of capital by output to facilitate a comparison of (per unit) capital k across firms. However, output is difficult to compare across firms due to variability in their specific products. 3 Empirical Implementation Gompers, Ishii, and Metrick (2003) attempt to link corporate governance with stock returns. However, Core, Guay, and Rusticus (2006) find no evidence that governance influences expected returns. Consequently, we focus our empirical analysis on the relationships between corporate governance, operating efficiency, and Tobin s Q. Our empirical analysis involves the level of governance. Although our framework predicts increased output and lower costs after improvements in corporate governance, empirical tests of these hypotheses require eventstudies surrounding changes in governance that are not adequately captured by the G index. 3.1 Data Data is obtained from COMPUSTAT to estimate R y and R c. Revenue (REVT) proxies the numerator of R y in equation (8) while cost of goods sold (COGS) defines the numerator of R c in equation (9). We supplement cost of goods sold with general, selling, and administrative expenses (XSGA) as well as operating expenses (XOPR). However, our empirical results are not sensitive to the proxy for cy in the numerator of R c. 14

15 Book value and market value data are obtained from COMPUSTAT and CRSP, respectively. Firms with non-positive book values are removed from our sample before computing Tobin s Q and return on equity. Furthermore, the lowest book values in each year of our sample, which are constrained to be positive, are sufficiently far above zero to prevent extreme measures of operating efficiency. The G index and B index are available in 1990, 1993, 1995, 1998, 2000, 2002, and 2004 using data from the Investor Responsibility Research Center (IRRC). 7 Quarterly institutional ownership data is obtained from 13f filings with the SEC and are averaged within each year. These three firm-level governance proxies are highly persistent. However, we refrain from estimating Tobin s Q and our operating efficiency measures during the intermediate years of the IRRC database since this expansion can inflate the regression t- statistics. Although the t-statistics we report are Newey-West adjusted, the number of lags in this adjustment procedure is limited by the availability of the G and B indices. A total of 10,823 firm-year observations are available for our empirical study. Four digit SIC codes are obtained from CRSP to implement the 49 industry classifications in Fama and French (1997). The operating efficiency measures R y and R c are standardized within these industries. This adjustment controls for differences in intangible assets and economies of scale across industries. Thus, positive and negative operating efficiency ratios 7 Institutional Shareholder Services produces the Corporate Governance Quotient (CGQ) that evaluates firm characteristics in eight categories; audit, board, charter/bylaws, director education, executive and director compensation, ownership, progressive practices, and state of incorporation. The G index is concentrated in the charter/bylaws category. 15

16 represent above-average and below-average operating efficiency relative to a firm s industry, respectively. This normalization relaxes the assumption that k and c are identical across different industries. Table 1 reports industry-level averages for Tobin s Q and ROE as well as the proxies for corporate governance and our operating efficiency measures. In Table 1, the operating efficiency measures are not normalized within individual industries. The results in Table 1 confirm that Tobin s Q and ROE as well as operating efficiency vary across industries, while the corporate governance proxies exhibit less industry-level variation. Panel A of Table 2 summarizes the distribution across firms of Tobin s Q and ROE as well as the proxies for corporate governance and operating efficiency. The operating efficiency measures R y and R c are both normalized within each industry. Panel B reports the correlations between these variables and also evaluates their correlations with market value. Besides being highly correlated, both the G and B indices have a slight inverse relationship with Tobin s Q, which suggests that firms with better governance have a higher Tobin s Q. Tobin s Q is also highly correlated with return on equity. Moreover, the positive correlation between the governance indices and operating efficiency indicates that weaker governance lowers operating efficiency. Gompers, Ishii, and Metrick (2003) document that institutional ownership is higher in firms with a higher G index. Institutional ownership is also higher in firms with better operating efficiency. These relationships are explored more thoroughly in the next subsection. The 0.57 correlation between R y and R c is consistent with the cost of production being proportional 16

17 to revenue as both variables depend on output. Moreover, the correlation does not diminish the variation among the cost structures of different firms within the same industry that is central to our analysis. Finally, market capitalization is not highly correlated with the other variables in our study, although larger firms tend to have better operating efficiency. 3.2 Hypotheses and Tests To evaluate the relationship between the G index and industry-adjusted operating efficiency, we estimate the following regression specification R y = α 0 + α 1 G + ɛ, (10) This regression is repeated with R c as the dependent variable instead of R y. In addition, this regression is repeated by replacing the G index as the independent variable with the B index and institutional ownership. A positive α 1 coefficient indicates that a high G index captures the operating inefficiencies associated with poor governance. Unlike the G and B indices, a negative coefficient indicates that firms with less institutional ownership, hence weaker governance, operate less efficiently. Table 3 reports on the coefficient estimates in equation (10). The positive coefficients for the G index and negative coefficients for institutional ownership indicate that these governance proxies both capture operating efficiency. For example, firms with a high G index and low institutional ownership, which are associated with weak corporate governance, appear 17

18 to underinvest and have high production costs. 8 In contrast, the B index yields positive but insignificant coefficients. The results from a multivariate regression involving all three governance proxies confirm that a low G index and high institutional ownership coincide with better operating efficiency. When Tobin s Q replaces R y, a negative coefficient is reported for the B index but not the G index. However, while the B index is better at explaining variation in Tobin s Q, this governance index is not sensitive to operating efficiency. Of the three governance proxies, only institutional ownership is significantly related to Tobin s Q and operating efficiency. 9 We also estimate the following regression specification that examines the relationship between Tobin s Q and operating efficiency Q = β 0 + β y R y + β c R c + ɛ, (11) The above regression is also repeated with ROE replacing Tobin s Q. Negative β coefficients are consistent with Tobin s Q being able to capture operating efficiency. Table 4 reports the coefficient estimates from equation (11). The positive β y coefficient in Panel A indicates that Tobin s Q decreases with better operating efficiency, while Panel B reports a similar relationship between ROE and operating efficiency. The positive β c coefficient likely arises from the high correlation between revenue and costs. More refined data on discretionary costs that are sensitive to corporate governance, as in the empirical 8 Institutional investors may invest in firms with better governance rather than improve governance. 9 The inverse relationship between the G index and Tobin s Q reported in Gompers, Ishii, and Metrick (2003) is weaker after 1998, the last year of their sample period. 18

19 studies of Cronqvist, Heyman, Nilsson, Svaleryd, and Vlachos (2009) and Core, Holthausen, and Larcker (1999), may produce a negative coefficient. Nonetheless, the β c coefficient is insignificant in the third specification reported in Panel A, and negative in the third specification of Panel B. Indeed, consistent with the earlier approximation, ROE R y R c, the coefficients for R y and R c are plus and minus one, respectively. Overall, our evidence suggests that underinvestment is economically more important than cost discipline since poor operating efficiency coincides with a high Tobin s Q. As a case study, we examine the disparity between Tobin s Q and our operating efficiency measures for General Motors (GM). In 2006, Tobin s Q for GM more than doubled as its book value decreased, before becoming negative in the 4th quarter. Conversely, both our measures of operating efficiency increased by over 30%. These increases signify a reduction in operating efficiency during this period. Intuitively, our measures of operating performance do not reward managers for reductions in the book value of their firms. 3.3 Robustness Tests Instead of deviations from their respective industry-level averages, our first robustness test industry-adjusts the operating efficiency measures using deviations from their respective industry-level minimum. These deviations are then standardized using each industry s largest deviation from its minimum to ensure that large outliers (very inefficient firms) are not unduly influencing the regression coefficients. Unreported results confirm that this alterna- 19

20 tive industry-adjustment yields similar empirical results to those reported in Table 3. An additional robustness test replaces book value with property, plant, and equipment (PPE) as a proxy for ky in the denominators of R y and R c. The use of PPE as an alternative proxy for capital does alter our conclusions. 10 Furthermore, removing firms in the banking, insurance, real estate, and financial trading industries also does alter our conclusions. Lindenberg and Ross (1981) document considerable variation in Tobin s Q across industries. They conclude that firms with high Tobin s Q have unique products or factors of production that enable them to earn monopoly rents. Therefore, we construct industry-adjusted Tobin s Q measures by normalizing these ratios within industries. However, accounting for industrylevel differences in Tobin s Q does not alter our earlier conclusions. Finally, several caveats on our empirical analysis are worth highlighting. Our analysis relies on industry classifications (peer firms). However, cross-sectional comparisons of operating efficiency across firms are complicated when firms are sufficiently unique to earn some monopoly rents in their respective product market. In addition, negative demand shocks can temporarily lower operating efficiency. Thus, similar to the evaluation of fund manager skill, the assessment of firm-level board performance requires a comparison of operating efficiency with industry peers over a sufficiently long time series to detect statistically significant differences. Furthermore, a 10 Out of a total of 11,279 firms, we remove 369 firms whose book values are either missing or non-positive while 1,179 firms would have to be removed from our sample since their PPE is missing. However, of the 324 firms that have a non-positive book value, 315 have a positive PPE. Thus, PPE does provide an alternative measure of capital for firms with non-positive book values. 20

21 firm s cost structure and demand function can depend on patents and intangible assets that predate its current management. Our relative assessment of operating efficiency ignores the possibility that an entire industry may operate inefficiently. 3.4 Alternative Measure of Operating Efficiency The maximization in equation (6) provides an alternative measure of operating efficiency; the difference between a firm s market value and the capital invested by shareholders. However, discount rate innovations, which can arise from changes in the market s risk premium, affect a market valuations but not our measures of operating efficiency. Indeed, revenues and production costs are less variable than market valuations. The possibility of being acquired can also increase a firm s market value due to an expectation of improved governance. In contrast, our measures of operating efficiency focus on observed operating efficiency. The correlation between Tobin s Q (M/B) and M B is This correlation indicates that our operating efficiency measures cannot be obtained by maximizing Tobin s Q. Unreported results also find that R y and R c are both negatively correlated with M B. Therefore, better governance corresponds with larger firm values net of capital. This finding indicates that our optimization in equation (6) provides an alternative, albeit more noisy, proxy for operating efficiency. 21

22 4 Conclusion Our simple theoretical framework demonstrates that underinvestment by entrenched managers confounds the relationship between Tobin s Q and corporate governance. Furthermore, an important implication of our framework is that stronger corporate governance can decrease Tobin s Q, return of equity, and return on assets as marginal profits decline towards zero. We provide operating efficiency measures derived from revenue and production costs to assess the economic implications of corporate governance. These operating efficiency measures are derived from maximizing market value net of the capital invested by shareholders. According to this maximization, firms with better corporate governance have lower capital-adjusted revenue and lower capital-adjusted production costs. Empirically, the corporate governance index in Gompers, Ishii, and Metrick (2003) and institutional ownership capture cross-sectional differences in operating efficiency as firms associated with better governance exhibit better operating efficiency. However, Tobin s Q is not higher for firms with better corporate governance. A number of accounting issues remain interesting topics for future research such as the treatment of intangible assets and leases that complicate the estimation of book value. Our framework can also be extended by introducing taxes and bankruptcy. In addition, future research can examine whether mergers improve or reduce operating efficiency. 22

23 References Aggarwal, R. K., and A. A. Samwick, 2006, Empire-builders and shirkers: Investment, firm performance, and managerial incentives. Journal of Corporate Finance 12, Bebchuk, L. A., A. Cohen, and A. Ferrell, 2009, What matters in corporate governance? Review of Financial Studies 22, Bertrand, M., and S. Mullainathan, 2003, Enjoying the quiet life? Corporate governance and managerial preferences. Journal of the Political Economy 111, Core, J. E., W. R. Guay, and T. O. Rusticus, 2006, Does weak governance cause weak stock returns? An examination of firm operating performance and investors expectations. Journal of Finance 61, Core, J. E., R. W. Holthausen, and D. F. Larcker, 1999, Corporate governance, chief executive officer compensation, and firm performance. Journal of Financial Economics 51, Cremers, K. J. M., and V. B. Nair, 2005, Governance mechanisms and equity prices. Journal of Finance 60, Cronqvist, H., F. Heyman, M. Nilsson, H. Svaleryd, and J. Vlachos, Do entrenched managers pay their workers more? Journal of Finance 64, Fama, E. F., and K. R. French, 1997, Industry costs of equity. Journal of Financial Economics 43, Giroud, X., and H. M. Mueller, 2007, Does corporate governance matter in competitive industries? Working Paper. 23

24 Gompers, P. A., J. L. Ishii, and A. Metrick, 2003, Corporate governance and equity prices. Quarterly Journal of Economics 118, Graham, J. R., M. L. Lemmon, and J. G. Wolf, 2002, Does corporate diversification destroy value? Journal of Finance 57, John, K., and D. Knyazeva, 2006, Governance and conservatism in investment decisions. Working Paper. Lindenberg, E. B., and S. A. Ross, 1981, Tobin s q ratio and industrial organization. Journal of Business 54, Maksimovic, V., and G. Phillips, 2001, The market for corporate assets: Who engages in mergers and asset sales and are there efficiency gains. Journal of Finance 56, Maksimovic, V., and G. Phillips, 2002, Do conglomerate firms allocate resources inefficiently across industries? Theory and evidence. Journal of Finance 57, Morck, R., A. Shleifer, and R. W. Vishny, 1988, Management ownership and market valuation. Journal of Financial Economics 20, Shleifer, A., and R. Vishny, 1986, Large shareholders and corporate control. Journal of the Political Economy 94,

25 Table 1: Summary Statistics within Industries Panel A and Panel B of this table reports industry-level averages and standard deviations (Std. Dev.), respectively, for Tobin s Q, return on equity (ROE), the G index in Gompers, Ishii, and Metrick (2003), the B index in Bebchuk, Cohen, and Ferrell (2008), institutional ownership (IO), and our operating efficiency measures denoted R y and R c. These summary statistics are computed within the industry classifications of Fama and French (1997). The number of observations denoted N within each industry is also reported over the seven years for which the G and B indices are available: 1990, 1993, 1995, 1998, 2000, 2002, and Our operating efficiency measures R y and R c are defined in equation (8) and equation (9) as P0 y k and c k, respectively. The demand curve P 0 y determines the output price, hence revenue, corresponding to output y conditional on its intercept P 0 and the sensitivity of prices to output denoted a. The cost of production and capital associated with producing a single unit of output are denoted c and k, respectively. Panel A: Average firm characteristics Industry N Tobin s Q ROE G B IO R y R c Agriculture Food Products Confectionery Liqueur Tobacco Toys Entertainment Publishing Consumer Goods Apparel Healthcare Medical Equipment Pharmaceutical Chemicals Rubber and Plastic Textiles Construction Materials Construction Steel Fabricated Products Machinery Electrical Equipment Automobiles Aerospace Shipbuilding and Railroad Defense Precious Metals Mining Coal Petroleum and Natural Gas Utilities Communications Personal Services Business Services Computer Hardware Software Semiconductors Measurement Equipment Paper and Office Supplies Boxes and Shipping Transportation Wholesale Retail Restaurants and Hotels Banking Insurance Real Estate Financial Trading Other Average Std. Dev. (across industries)

26 Panel B: Standard deviation of firm characteristics (within industries) Industry Tobin s Q ROE G B IO R y R c Agriculture Food Products Confectionery Liqueur Tobacco Toys Entertainment Publishing Consumer Goods Apparel Healthcare Medical Equipment Pharmaceutical Chemicals Rubber and Plastic Textiles Construction Materials Construction Steel Fabricated Products Machinery Electrical Equipment Automobiles Aerospace Shipbuilding and Railroad Defense Precious Metals Mining Coal Petroleum and Natural Gas Utilities Communications Personal Services Business Services Computer Hardware Software Semiconductors Measurement Equipment Paper and Office Supplies Boxes and Shipping Transportation Wholesale Retail Restaurants and Hotels Banking Insurance Real Estate Financial Trading Other Average

27 Table 2: Summary Statistics and Correlations Panel A of this table reports on the distribution for Tobin s Q, return on equity (ROE), the G index in Gompers, Ishii, and Metrick (2003), the B index in Bebchuk, Cohen, and Ferrell (2008), institutional ownership (IO), and our operating efficiency measures. The operating efficiency measures R y and R c are defined in equation (8) and equation (9), respectively, as P0 y k and c k. The demand curve P 0 y determines the output price, hence revenue, corresponding to output y conditional on its output P 0 and the sensitivity of prices to output denoted a. The cost of production and capital associated with producing a single unit of output are denoted c and k, respectively. Both operating efficiency measures are normalized within the industry classifications of Fama and French (1997). The correlations between the variables in Panel A along with market capitalization (M) are reported in Panel B. Panel A: Summary statistics Percentiles Variable 10 th 25 th 50 th 75 th 90 th Tobin s Q ROE G B IO R y R c Panel B: Correlations Variable M Tobin s Q ROE G B IO R y R c M Tobin s Q ROE G B IO R y R c 1.00

28 Table 3: Operating Efficiency, Tobin s Q, and Corporate Governance This table reports the coefficient estimates from the regression specification in equation (10), R y = α 0 + α 1 G + ɛ, that examines the relationship between operating efficiency and corporate governance. Below each regression coefficient, t-statistics are reported in parentheses. Besides the G index of Gompers, Ishii, and Metrick (2003), the B index of Bebchuk, Cohen, and Ferrell (2008) and institutional ownership (IO) proxy for corporate governance. The regression is also repeated by replacing R y with R c and Tobin s Q. The operating efficiency measures R y and R c are defined in equation (8) and equation (9), respectively, as P0 y k and c k. The demand curve P 0 y determines the output price, hence revenue, corresponding to output y conditional on its intercept P 0 and the sensitivity of prices to output denoted a. The cost of production and capital associated with producing a single unit of output are denoted c and k, respectively. Both measures of operating efficiency are normalized within the industry classifications of Fama and French (1997). α 0 G R 2 # of obs. R y % 10,823 (-1.97) (2.15) R c % 10,823 (-2.33) (2.55) Tobin s Q % 10,823 (8.57) (-0.75) α 0 B R 2 # of obs. R y % 10,823 (-0.18) (0.23) R c % 10,823 (-0.49) (0.62) Tobin s Q % 10,823 (16.58) (-3.71) α 0 IO R 2 # of obs. R y % 10,823 (4.70) (-6.00) R c % 10,823 (5.22) (-6.68) Tobin s Q % 10,823 (13.30) (2.99) α 0 G B IO R 2 # of obs. R y % 10,823 (0.12) (2.80) (-1.87) (-6.02) R c % 10,823 (0.14) (2.94) (-1.65) (-6.72) Tobin s Q % 10,823 (4.42) (1.28) (-2.64) (3.09)

29 Table 4: Operating Efficiency and Tobin s Q Panel A of this table reports the β coefficients from the regression specifications in equation (11), Q = β 0 + β y R y + β c R c + ɛ, which examines the relationship between Tobin s Q and operating efficiency. Panel B replaces Tobin s Q with return on equity (ROE). The operating efficiency measures R y and R c are defined in equation (8) and equation (9), respectively, as P0 y k and c k. The demand curve P 0 y determines the output price, hence revenue, corresponding to output y conditional on its intercept P 0 and the sensitivity of prices to output denoted a. The cost of production and capital associated with producing a single unit of output are denoted c and k, respectively. Both operating efficiency measures are normalized within the industry classifications of Fama and French (1997). Below each regression coefficient, t-statistics are reported in parentheses with inf denoting a number greater than a hundred million. Panel A: Operating efficiency and Tobin s Q β 0 β y β c R 2 # of obs. Tobin s Q % 10,823 (28.16) (4.18) Tobin s Q % 10,823 (28.15) (4.04) Tobin s Q % 10,823 (28.13) (0.42) (0.33) Panel B: Operating efficiency and ROE β 0 β y β c R 2 # of obs. ROE % 10,823 (13.58) (7.64) ROE % 10,823 (12.78) (6.61) ROE % 10,823 (-2.48) (inf) (-inf)