ECCE Research Note 06-01: CORPORATE GOVERNANCE AND THE COST OF EQUITY CAPITAL: EVIDENCE FROM GMI S GOVERNANCE RATING

Transcription

1 ECCE Research Note 06-01: CORPORATE GOVERNANCE AND THE COST OF EQUITY CAPITAL: EVIDENCE FROM GMI S GOVERNANCE RATING by Jeroen Derwall and Patrick Verwijmeren

2 Corporate Governance and the Cost of Equity Capital: Evidence from GMI s Governance Rating Jeroen Derwall 1 European Centre for Corporate Engagement, The Netherlands Maastricht University, The Netherlands RSM Erasmus University, The Netherlands Patrick Verwijmeren RSM Erasmus University, The Netherlands ECCE Research Note Version 2.0: January Corresponding author: Jeroen Derwall, Assistant Professor of Finance, Maastricht University, Faculty of Economics and Business Administration, Department of Finance, Tongersestraat 53, 6200 MD Maastricht. The Netherlands. We acknowledge the support of Gavin Anderson and Governance Metrics International. This research was completed at Maastricht University on behalf of the European Centre for Corporate Engagement (ECCE). The usual disclaimer applies. 2 Suggested reference: Derwall, J. and P. Verwijmeren (2006), Corporate Governance and the Cost of Equity Capital: Evidence from GMI s Governance Rating, ECCE Research Note 06-01, European Centre for Corporate Engagement.

3 INTRODUCTION This research note describes how the corporate governance attributes of publicly listed companies are received by financial markets. At the heart of our study is the relation between firms corporate governance ratings and their cost of equity capital. We present empirical evidence to quantify the effects of investors concerns for corporate governance on expected returns in today s financial landscape. Using a panel of U.S. firms with diverse corporate governance ratings over the period , we investigate whether differences in corporate governance translate into differences in expected equity returns, hence, into variation in the cost of equity. Our study uses governance ratings provided by Governance Metrics International (GMI) to assess the association between the quality of corporate governance and the cost of equity capital of public firms. While prior empirical research revolves around the association between corporate responsibility measures and a firm s market value, only few studies isolate the cost of equity component that is critical to portraying equity markets attention to risks associated with poor corporate governance. We use the ex ante cost of capital implied in contemporaneous stock price and analyst forecast data for this purpose. To our knowledge, Ashbaugh-Skaife et al. (2006) is the only academic study directly related to this research note. Their evidence suggests that several corporate governance constructs are associated with a lower cost of equity, consistent with the idea that better governance reduces the agency and information risks against which investors price protect. We follow and extend their approach using GMI s overall corporate governance rating and using an alternative cost of equity measure. More specifically, we first document that firms with better overall corporate governance enjoy a lower cost of equity capital. Second, we find that better governance is associated with lower systematic risk, as measured by a firm s beta. Third, we relate corporate governance to idiosyncratic (i.e., firm-specific) risk. Unlike earlier studies, we use the Fama-French three-factor market equilibrium model to measure idiosyncratic risk because several studies suggest that idiosyncratic risk measured by more parsimonious models, such as the traditionally used CAPM, comprises both extra-market risk and firm-specific risk. We find that better governance is associated with lower idiosyncratic risk even if we control for the two extra-market risk factors developed by Fama and French (1993). 2

4 DATA Measuring Corporate Governance We use the database from Governance metrics international (GMI) to evaluate corporate governance practices of U.S. firms. GMI s ratings system has been developed following extensive research in multiple markets and consultations with institutional investors, corporate officers and directors and governance specialists. GMI provides assessment of 3,800 companies, including complete coverage of the MSCI World and MSCI EAFE Index. We use the 2005 GMI Home Market Overall Rating in this study on U.S. firms. The overall corporate governance scores developed by GMI bundles the quality of corporate governance along various well-known dimensions, such as board quality and independence, executive compensation and remuneration schemes adopted by the firm, the degree to which firms have anti-takeover provisions and other mechanisms that affect shareholder rights, and the quality and transparency of information the firm provides. A higher score reflects better corporate governance; see Bauer et al. (2005) for more details on GMI s corporate governance ratings. Financial Data The expected return (cost of equity) central to this study is the rate that equates expected cash flows over a specified period to the current stock price. Our implied cost-of-equity model relies on analysts consensus earnings forecasts from IBES. We collected the mean 1-year ahead and 2-year ahead earnings forecasts to derive cash flow expectations. The model also requires that we collect dividend and price information, which we obtain from Compustat. In our investigation into the relation between corporate governance and the cost of equity, we control for other firm characteristics correlated with the dependent variables. Our cost of equity model controls for traditional risk proxies: U.S. stock market sensitivity (beta) measured over previous year s trading days, where the Fama- French (1993) market proxy is used to describe market return variation, the book value of leverage (debt / assets), firm size measured by the market value of equity after lognormal transformation, and the price-to-book ratio truncated at the 1% level. Returns data are from the CRSP U.S. Stocks database. All other data that are necessary to construct our controls are from the Compustat library. 3

5 EMPIRICAL ANALYSIS We examine the association between GMI s overall corporate governance score and firms cost of equity capital using the abnormal earnings growth valuation model of Easton (2004). The Easton (2004) implied cost-of-equity model takes the form: P t = (x t+2 + r peg * d t+1 x t+1 )/ r 2 peg, (1) where P t is a firm s market price in year t, x is the expected future earnings per share, d is the expected future net dividends per share, and r peg is the implied cost of equity capital. The intuition behind Easton s price-earnings-growth ratio model is fairly straightforward and the model requires only a limited number of inputs. By using analysts forecast for the expected earnings and dividends, we solve for the internal rate of return, which is the implied cost of equity capital. To some extent, the preference for this model is arbitrary. Hail and Leuz (2006) describe this model but also other models that enable computation of the implied cost of capital, such as the residual income models of Claus and Thomas (2001) and Gebhardt et al. (2001). Because Hail and Leuz (2006) find that all four models yield comparable results, we choose the computationally most efficient specification. Moreover, Botosan and Plumlee (2005) compare alternative cost of equity estimators derived from valuation models and conclude that the implied premium from the Easton (2004) model is superior to most other expected return estimates, based on the correlation between the implied rate of return and well-known risk proxies. The advantage of implied cost of equity estimates is that they reflect ex ante expected returns and do not explicitly depend on conventional pricing models using realized returns, which are potentially prone to misspecification. The misspecification problem is a reason for us to eschew existing expected return models that make use of realized returns, especially because none of these models explicitly incorporates factors to describe premiums for the corporate governance attributes of stocks. One critique towards implied cost of capital models concerns the validity of the proxies as a consequence of, for instance, analyst forecasts that are sluggish with respect to information in past stock returns (Guay et al. (2004)). Although we are not primarily interested in the absolute cost of capital estimates but in cross-sectional differences in these estimates across firms, limitations to implied models brought forward in earlier work suggest that our results should be handled with appropriate caution. Measuring firms cost of equity is an unavoidably difficult exercise. After having solved for firms implied cost of equity, we estimate the following pooled regression model: 4

6 J K rpeg, i = α + β jgovscore j, it + γ kck, it + ε it, (2) j= 1 k= 1 where r peg is the implied expected return for firm i in year t, GOVSCORE it is the overall corporate governance score from GMI, and C it is a vector of (financial) control variables. The controls we use are common to studies on expected returns (e.g., Fama and French (1992), Ashbaugh-Skaife et al. (2006), Hail and Leuz (2006)). There are several reasons why governance might be priced in the market. The most widely held idea is that governance has clear risk implications. Before we can discuss those risk implications in greater detail, it is important to define alternative sources of risk. Traditional financial markets theory separates systematic risk from idiosyncratic risk (firm-specific risk) and posits that investors command a premium only for bearing non-diversifiable risk. In the standard capital asset pricing literatures, the degree to which a stock s return co-varies with the return on the market as a whole (i.e, the firm s beta ) is viewed as the single dimension of systematic, non-diversifiable risk that matters for explaining expected returns. However, there now is growing evidence that due to capital market imperfections, firm-specific risk cannot be fully eliminated by means of diversification, which implies that (to some extent) firm-specific risk is priced as well. 3 In addition, there are several studies, such as Fama and French (1993) suggesting that non-diversifiable risk factors other than beta account for much of firm s cost of equity capital. We make no claim about the role of systematic versus idiosyncratic risk in explaining firms cost of equity but we do investigate the relation between corporate governance and these two risk dimensions. We test the association between governance and Beta using the following model: J K Betai = α + β jgovscore j, it + γ kck, it + ε it (3) j= 1 k= 1 where Beta of firm i is measured as the slope coefficient from a regression of daily stock returns for firm i on the daily returns of a market portfolio comprising all NYSE- AMEX-NASDAQ stocks. Each calendar year, we re-estimated the betas. The financial control variables we use in this model are firm size, leverage, and the book-to-market ratio. Our models also control for industry effects. 3 See, for example, Malkiel and Xu (2006). 5

7 We examine the relation between governance and idiosyncratic risk in a similar manner. We estimated the following pooled regression model: J K ISYNC _ RISKi = α + β jgovscore j, it + γ kck, it + ε it (4) j= 1 k= 1 where idiosyncratic risk (ISYNC_RISK) is defined as the standard deviation of the daily residual returns from the regressions we perform to estimate beta. The control set in this model comprises firm size, leverage, the price-to-book ratio, beta and industry dummy variables. We inspected summary statistics, which are not tabulated here, on some basic financial data and the governance data used in this research note. For our sample, the average implied cost of equity capital is in the order of 11 percent. The implied estimate does not produce extreme outliers, as our inspection of the respective standard deviation points out. To describe the association between the cost of equity and governance, we present pooled regressions of the implied cost of equity on the traditional risk factors (size, beta, leverage, and price-to-book) and on GMI s overall rating. Table 1 shows the results. The coefficients on the control variables are significant and carry the expected sign in our models. The models in Table 1 vary in the choice of industry control variables. While the first model includes a dummy for financial firms, the second model includes dummies that account for variation in expected return across the 48 industries identified by Fama and French (1997). As for the coefficient on the governance rating, both models produce a negative loading that is significant at the 10% level (and almost significant at the 5% level under the second regression model). Hence, firms with better governance enjoy a lower cost of equity capital. Also from an economic perspective, the coefficient on the governance variable is plausible: a 10-point increase, i.e., from the lowest rating of 0 to the highest rating of 10, decreases the cost of equity by approximately 1 full percent (e.g., from 10 to 9 percent). In Table 2, we relate corporate governance to a firm s beta and to idiosyncratic risk. We report regression results for both idiosyncratic risk according to an estimated CAPM and idiosyncratic risk observed under the three-factor model. The models almost consistently suggest that leaders in corporate governance have lower systematic risk and lower idiosyncratic risk than do laggards. The negative relation between the governance score and the dependent variables in Table 2 is highly statistically significant, independent of the model we estimate. 6

8 SUMMARY AND DISCUSSION OF IMPLICATIONS We find evidence that better governance is associated with lower firm-specific risk, lower systematic risk, and a lower implied cost of equity capital. These relationships are likely to carry several important implications for company managers, investors and scholars. When markets are assumed efficient then long-run returns of stocks should equal the expected return. The implication of our finding, then, is that weakly governed firms should earn a higher absolute return compared to better-governed firms. This return difference does not reflect an investment opportunity for investors in search of underpriced stocks. Instead, it represents a risk premium associated with holding fundamentally riskier stocks. Moreover, because the relation we observe is not subsumed by the usual risk factors related to market sensitivity (beta), distress (price-book) and firm size, it is an open question whether current models that use realized portfolio returns to estimate the cost of equity, such as the widely employed CAPM, accurately measure a securities expected return. Especially our finding that weaker governance is associated with higher firm-specific risk is interesting in this context because studies are increasingly suggesting that firm-specific risk cannot be fully diversified away as a consequence of capital market imperfections. These imperfections could cause investors to command a premium for bearing firm-specific risk, which then factors into the cost of equity capital. Moreover, since the CAPM rests on the assumption of perfect capital markets it is reasonable to state that a separate governance risk factor is needed in cost of equity models. Ashbaugh-Skaife et al. (2006) is a recent attempt to develop such a model. In addition, because most investors are judged using performance evaluation models that do not explicitly incorporate governance risk premiums, the possibility emerges that investors holding weakly governed firms earn a positive risk-adjusted returns (i.e, positive alpha) and outperform investors holding better governed firms because not all risks are adequately accounted for by current models. When the omitted governance risk premium would be introduced in those performance evaluation techniques, this abnormal performance should disappear. Thus, a potential extension of current performance attributions models would be to add a governance risk premium that captures non-diversifiable risk associated with weaker governance. Possibly, there are implications for firms capital structure decisions. Our finding that investors price protect against weak governance by lowering the equity prices (raising the cost of equity) of weakly governed companies could suggest that those companies are more exposed to the market for debt. Implicit in that assumption is the view that, unlike equity investors, bondholders do not respond unfavorably to weaker overall governance. Because there is now some evidence in the literature that 7

9 several governance mechanisms not welcomed by equity investors are not eschewed by bond investors, the capital structure effects of governance are theoretically sensible. 4 However, we leave this question a challenge for future research. REFERENCES Ashbaugh-Skaife, H., D. Collins and R. Lafond (2006), Corporate Governance and the Cost of Equity Capital, Working Paper. Ashbaugh-Skaife, H., D.W. Collins and R. Lafond (2006b), The Effects of Corporate Governance on Firms Credit Ratings, Journal of Accounting and Economics, forthcoming. Bauer, R., B. Frijns, R. Otten and A. Tourani-Rad, (2005), The impact of Corporate Governance on Corporate Performance: Evidence from Japan, Working Paper. Botosan, C. and M. Plumlee (2005), Assessing Alternative Risk Premium, The Accounting Review, forthcoming. Claus, J. and J. Thomas (2001), Equity Premia as Low as Three Percent? Evidence from Analysts' Earnings Forecasts for Domestic and International Stock Markets, Journal of Finance, vol. 56, pp Easton, P. (2004), PE Ratios, PEG Ratios, and Estimating the Implied Expected Rate of Return on Equity Capital, The Accounting Review, vol. 79, pp Fama, E.F. and K.R. French (1993), Common Risk Factors in the Returns on Stocks and Bonds, Journal of Financial Economics, vol. 33, pp Fama, E.F. and K.R. French (1997), Industry Costs of Equity, Journal of Financial Economics, vol. 43, pp Gebhardt, W., C. Lee and B. Swaminathan (2001), Toward an Implied Cost of Capital, Journal of Accounting Research, vol. 39, pp For example, Ashbaugh et al. (2006b) find that firms with stronger shareholder rights have a lower credit rating, consistent with the idea that more shareholder rights can be detrimental to bondholder wealth because shareholders might encourage managers to undertake riskier projects at the expense of increased bondholder risk. 8

10 Guay, W., S. P. Kothari and S. Shu (2004), Properties of Implied Cost of Capital Using Analysts Forecasts, Working paper, The Wharton School, MIT, and Boston College. Hail, L. and C Leuz (2006), International Differences in the Cost of Equity Capital: Do Legal Institutions and Securities Regulation Matter?, Journal of Accounting Research, forthcoming. Malkiel, B.G. and Y. Xu (2006), Idiosyncratic Risk and Security Returns, Working Paper. White, H. (1980), A Heteroskedasticity-consistent Covariance Matrix Estimator and a Direct Test for Heteroskedasticity, Econometrica, vol. 48 pp

11 Table 1. Governance and the cost of equity: pooled regression results Table 1 shows the outcome of estimating models for the implied cost of equity, where the cost of equity for U.S. firms is computed using the Easton (2004) computation, and where the independent variables are the market value of equity after log transformation, the firm s beta based on daily stock returns over the last year, the book debt-to-assets ratio, the price-to-book ratio (truncated at the 1% level), year fixed effects, and the aggregate corporate governance index. The first model additionally controls for whether the firm is a financial services company according to the industry classification scheme of Fama and French (1997). The second model contains all industry dummy variables based on Fama and French s 48 industry classifications. The table reports pooled OLS coefficients with t-statistics (in parentheses) based on White (1980) standard errors. Pooled sample period: Implied Cost of Equity Intercept 0.10 *** 0.09 *** (11.88) (10.55) GOVSCORE -9.6 E-4 * E-3 * (-1.70) (-1.95) Firm Size E-3 ** E-3 (-2.07) (-1.41) Beta 0.02 *** 0.01 *** (6.23) (4.57) Debt / Assets 0.04 *** 0.04 *** (5.77) (5.33) Price / Book Value E-3 *** E-3 *** (-4.91) (-4.32) Year Fixed Effects Y Y Financial Industries Controlled Y Y All Industries Controlled N Y # Observations Adj. R-squared *** Significant at 1% level, ** at 5% level, * at 10% level.

12 Table 2. Governance, Systematic Risk and Idiosyncratic risk: pooled regressions Table 2 shows the outcome of estimating models for firms beta and for idiosyncratic risk, where beta and idiosyncratic risk are computed using daily stock returns observed during the previous calendar year, The independent variables are the market value of equity after log transformation (firm size), beta (only in regressions for idiosyncratic risk), the book debt-toassets ratio, the price-to-book ratio (truncated at the 1% level), year fixed effects, and the overall corporate governance rating. Idiosyncratic risk is computed using, respectively, the single-index CAPM and the three-factor model of Fama and French (1993), and is the defined as the standard deviation of the residual daily returns from those models. The table reports pooled OLS coefficients with t-statistics (in parentheses) based on White (1980) standard errors. Pooled sample period: Stock Beta Idiosyncratic Risk CAPM Fama-French Intercept 2.06 *** 2.00 *** 0.03 *** 0.03 *** (30.83) (30.62) (14.35) (14.54) GOVSCORE *** *** E-2 *** E-2 ** (-2.65) (-3.57) (-2.45) (-2.31) Firm Size *** *** E-2 *** E-2 *** (-9.74) (-8.83) (-8.78) (-8.66) Beta 0.53 E-2 *** E-2 *** (9.95) (8.74) Debt / Assets *** E-2 *** 0.53 E-2 *** (-5.21) (-1.62) (3.61) (3.47) Price / Book Value -6.17E E E-4 *** 1. 1 E-4 ** (-0.17) (1.36) (2.70) (2.25) Year Fixed Effects Y Y Y Y Financial Industries Controlled Y Y Y Y All Industries Controlled N Y Y Y # Observations Adj. R-squared *** Significant at 1% level, ** at 5% level. 11