Corporate Governance Indices Roberta Romano Yale Law School, NBER and ECGI International Conference on Institutional Quality Madrid, Jan. 22, 2009.

Corporate Governance Indices

Roberta RomanoYale Law School, NBER and ECGI

International Conference on Institutional Quality

Madrid, Jan. 22, 2009

Outline of Remarks

• Review evidence on the relation between key institutions of corporate governance and performance

• Review the prominent corporate governance indices and examine more closely whether there is a relation between governance indices and performance

• Consider policy implications of review’s bottom line (that there is no systematic relation between indices and performance)

Corporate Governance and Performance

• Research on governance mechanisms in U.S. has not systematically identified positive effects– Independent boards (no effect; may be negative)– Shareholder activism – proposals (no effect)– Shareholder activism - proxy fights (positive)– Inside ownership (nonlinear)– Outside blocks (evidence mixed)– Outside director ownership (no effect or positive,

depending on measure)– Executive compensation (evidence mixed)

• Concern: governance mechanisms are numerous and interaction effects probable

Corporate Governance Indices

• Index idea: because governance operates on many dimensions, desirable to combine elements into one number (index) to represent the quality of firms’ corporate governance

Academic Indices for U.S. Firms

• G-Index (Gompers, Ishii & Metrick 2003)– First and most prominent (2001 w.p.)– Equally weighted sum of 24 governance components

collected by IRRC (primarily takeover defenses)• E-Index (Bebchuk, Cohen & Ferrell 2005)

– Subset of 6 G-index components thought most effective at blocking hostile bids

• Gov-Score & Gov-7(Caylor & Brown 2006)– Eq. wt sum of 51 governance components collected

by ISS (board, compensation; few defenses)– Subset of 7 components (only 2 defenses); identified

empirically (correlated with performance)

Proprietary Governance Indices

• Vendors: RiskMetrics (ISS), Glass-Lewis, Egan-Jones, Governance Metrics Int’l, The Corporate Library

• Differences in construction from academic indices (marketing governance expertise):– Component weights vary– Takeover defenses deemphasized– Relative vs. absolute rankings

• Marketed to institutional investors for proxy voting and investment decisions

Governance Indices and Performance: Initial Research

• GIM: trading strategy of selling firms with worst governance and buying those with best produces abnormal returns (8.5% per year) in 1990s; good governance positively related to Tobin’s Q and stock returns

• BCF: E-index did as well as G-index; only components associated with performance are those in E-index

• Caylor & Brown: Relation between index and performance due mainly to compensation and board components, not defenses; Gov-7 does better than E-index in explaining performance

GIM’s Explanation

• Investor misperception of impact of governance on performance in 1990 (start of trading strategy period)

• Managers expect poor performance and adopt defenses

• Poor governance is correlated with something else that causes poor performance

Robustness of Findings

• Several studies find GIM results are not robust:– Example 1: Controlling for performance before

defenses adopted (1980s), relation in 1990s disappears (Lehn et al. 2006)

– Example 2: Analyst forecasts predict poor performance of poor governance firms; no difference in stock returns surrounding earnings announcements of poor and good governance firms; no trading gains in 2000s (Core et al. 2006)

– Example 3: Relation not independently significant but depends on interrelation of governance and block institutional ownership (high quality governance on both measures); those firms also appear to have greater risk (Cremers & Nair 2005)

Single Governance Mechanism vs. Governance Index

• Can a single mechanism ever be as effective a measure of quality as an index?– Theory: if board has right incentives it can provide

effective oversight, given pivotal position • Proxies: independence, outsider stock ownership

– Empirical: identification and measurement error should be lower with one variable, as governance institutions’ interactions are complex

• E.g., Gillan, et al. (2007): firms with more independent boards have more defenses (good governance components are substitutes not complements)

Note: few formal models of governance and no adequate theory of when or whether governance components are complements or substitutes)

Econometric Issues

• Governance quality and performance are not independent– Example: some governance features are

motivated by incentive-based models of managerial behavior, which also affect performance (e.g., ownership)

• Simultaneous equation estimation with instrumental variables a solution – But finding proper instruments is difficult

Comparing indices’ relative performance controlling for

endogeneity• Accounting performance (ROA)

– G and E indices significantly correlated; Gov-score, TCL unrelated

• Stock returns– No relation

• Disciplinary CEO turnover after poor performance – Better governed firms under G and E indices less likely to experience

such turnover after poor performance– Gov-Score, TCL unrelated

• Single governance measures do as well or better:– Outside directors’ median stock holdings positively related to ROA – Higher outside directors’ median stock holdings and higher % board

independent increase disciplinary turnover after poor performance

(Results from Bhagat & Bolton 2008)

Do Commercial Indices Do Better?

• Daines et al. (2008) find no systematic relation between leading commercial indices (ISS, TCL, GovernanceMetrics), and performance (ROA, excess stock returns, Tobin’s Q)– Note: does not employ simultaneous equation

methodology

• Bhagat & Bolton (2008) findings indicate more complex measures (e.g., Gov-Score), which are more like commercial indices, do not outperform simpler ones (e.g., G- index)

Policy Implications for Investors: Choice of an Index

• There is no one best governance index to identify a firm’s quality

• Appropriateness of index depends on use to be made (“best” is specific to context) – None useful if objective is to predict stock

performance – G or E index useful to predict accounting

performance

Policy Implications for Regulators: Choice of Regulatory Regime

• “One-size-fits-all” regulation should be rejected

• Preferred ordering of regulatory approach: (i) Disclosure (ii) Comply or Explain (iii) Mandates

– Best-practices equivalent to an equal-weighted index and therefore suffers from same problems: does not permit appropriate matching of governance to firms

– Burden on non-complier to explain inconsistent with implication that same governance system does not fit all firms or all contexts

– Disclosure – no such inconsistency (no benchmark), byproduct, some informational costs imposed

Speculation on Implication for Emerging Markets

• World Bank uses governance ratings• “Radical relativism” sensible for developed

markets may need adaptation– Background components of good governance

for legal/political system (e.g., “rule of law”, independent judiciary) may be more universal than firm’s corporate governance

– But local organization, culture differ across nations, which may affect relation between law and economic development

Conclusion

• There is no consistent relation between performance and governance indices, or individual governance devices

• There is, accordingly, no “best” measure of governance: it depends on context and firm circumstances

• Regulatory regime should therefore permit flexible variation in governance across firms