What if Regulating Executive Pay Doesn’t Work? A Brief Primer on the Limits to Knowledge in Corporate Governance Research

Below is a brief missive on the topic of this Thursday’s World Growth Panel Discussion, entitled “Corporate Governance Rating Agencies and Conflicts of Interest: Harming Pension Funds, Individual Investors, and Company Employees, Investor Harm and Future Policy Implications.” The flyer for the event is included at the end of this email.

After WorldCom, Enron, and the other turn-of-the-millennium financial scandals, the loose structure of federal, state, securities exchange and self-regulation of corporate governance that had evolved up to that point was regarded by many to inadequately limit the fundamental principal-agent problems between investors and management within U.S. firms. The Sarbanes-Oxley Act of 2002 was one response to this perceived failure.

But another – less well-known – response is what has become known as the “corporate governance industry.” The corporate governance industry – composed of governance advisers, governance rating firms, and proxy advisers, sometimes operating as business units of a single company – plays a major role in corporate governance policymaking, and, because of its influence with institutional investors, exerts great control over corporations and their organizational structures.

The corporate governance industry is built upon the notion that certain corporate governance characteristics are systematically related to firm performance. The idea is that since principal-agent problems – including high executive pay – are “bad,” structures that remove those problems should broadly be favored. While that notion seems intuitive – so much so that it is the basis for restricting compensation in the banking industry lately – in reality the idea is anything but accepted or established beyond a theoretical ideal.

Of course, the key problems are how to measure “corporate governance” and how to measure “firm performance.” In the real world, it is difficult to measure the effects of singular measures or corporate governance. Hence, corporate governance firms use proprietary models to distill myriad soft behavioral factors into “corporate governance ratings,” which are thought to relate to “form performance.”

Early empirical academic studies found some relationships between corporate governance ratings and various measures of “firm performance.” For instance, Brown and Caylor (2004) suggested that high RiskMetrics Group’s Institutional Shareholder Services (ISS) CGQ scores were associated with higher current stock returns, higher accounting returns, lower stock return volatility, and higher dividends. Brown and Caylor’s (2006) follow-up study suggested there existed a favorable relationship between Tobin’s Q and an index created from 51 governance variables collected by ISS (and identified as important elements of ISS ratings).

Since that period, however, important research has contradicted those findings and raised serious questions about both the reliability and utility of corporate governance models and assumptions. Moreover, Brown and Caylor’s 2004 article was commissioned directly by ISS, suggesting a possible conflict of interest.

To give just a short sampling of the research that does not support the corporate governance hypothesis, Daines, Gow, and Larcker (2008) conduct statistical analyses of four ratings: ISS’s CGQ, GovernanceMetrics’s GMI, The Corporate Library’s TCL, and Audit Integrity’s Accounting and Governance Risk (AGR) metrics and find that, with the possible exception of the AGR, “governance ratings have either limited or no success in predicting firm performance or other outcomes of interest to shareholders.” They also find little correlation among the ratings, a result they suggest indicates either that the ratings measure different corporate governance metrics or significant measurement error in the different metrics.

Bhagat, Bolton, and Romano (2007) similarly conclude that existing corporate governance ratings do not accurately predict firm performance.  They examine academic corporate governance indices that form the basis of commercial ones in widespread use and criticize what they see as commercial misuse of academic methodologies.  Moreover, Bhagat, Bolton, and Romano also find that there is no single best measure of performance that is adequate to make informed decisions regarding firm quality.  In fact, they find that one variable – outside directors’ stock ownership – by itself outperforms leading academic indices.

Other authors find that, rather than simply being ineffective, corporate governance ratings may even have adverse effects on firm performance. Rose (2007) finds that “the corporate governance industry may have similarly harmful effects on the competition for capital by pressuring companies to adopt a homogenized set of governance rules which may be ill-suited to the companies’ respective situations.”  He also argues that one-size-fits all governance ratings that are often unproven can have adverse impacts on significant shareholder decisions.

In a similar vein, Koehn and Ueng (2005) argue that firms are often pressured to obtain corporate governance ratings from high-profile firms such as ISS and GovernanceMetrics International.  However, they find that the governance rating metrics championed by these firms are “not good indicators of either the quality of a firm’s earnings or of its ethics,”  and may in fact be negatively correlated with annual stock appreciation and ethics scores.  In this regard, corporate governance ratings harm both the firms pressured to obtain them and pension fund and individual investors who rely on them.

Another potential explanation for corporate governance ratings’ failure to accurately predict firm performance is the conflicts of interest between the firms being evaluated and the rating agencies themselves. Such conflicts of interest were the subject of recent a policy briefing sponsored by the Millstein Center for Corporate Governance and Performance at the Yale School of Management. The Briefing focused on firms such as the RiskMetrics Group, which provide voting advice to institutional investors while also providing structural governance advice to the firms. Investors at the briefing suggested that the dual role for corporate governance rating agencies may allow companies purchasing governance guidance “to ‘game’ the system, whereby less potentially disruptive voting recommendations are given to investors if the company of interest is also a client of the corporate governance rating agency’s consulting services.”

Like rating shopping in the credit rating industry, seemingly “optimal” corporate governance ratings are only optimal from the perspective of the models used to produce them. The depth of today’s recession is in part a result of the need not just to report losses on structured financial instruments, but to restructure those financial instruments to fit an improved model of credit performance that accounts for what we now acknowledge as faulty financial engineering. Corporate governance structures are far more complex than even the most complicated financially-engineered security and will take considerably longer to restructure toward stability if the corporate governance rating models used to engineer corporate governance structures prove wrong.

Hence, not only are today’s “innovative” regulatory policies at risk of misapplying results from a thin body of empirical work, but in “getting it wrong,” such policies may be putting in place populist restrictions on optimal performance that will be very difficult to change should they prove unhelpful or even destructive to U.S. economic performance and competitiveness.

† Hermann Moyse, Jr./Louisiana Bankers Association Professor of Finance, Louisiana State University, Senior Fellow at the Wharton School, and Partner, Empiris LLC. Contact information: joseph.r.mason@gmail.com; (202) 683-8909 office. Copyright Joseph R. Mason, 2009. All rights reserved. Past commentaries and testimony are blogged on http://www.rgemonitor.com/financemarkets-monitor/bio/626/joseph_mason.

4 Responses to "What if Regulating Executive Pay Doesn’t Work? A Brief Primer on the Limits to Knowledge in Corporate Governance Research"

  1. Ric Marshall   October 30, 2009 at 9:20 am

    Your ‘brief missive’ may be well reasoned, but it rests on a base of half-truths and misconceptions. You highlighted one of the key limitations of your argument when you wrote of the Daines study, “They also find little correlation among the ratings, a result they suggest indicates either that the ratings measure different corporate governance metrics or significant measurement error in the different metrics.” Had Daines, et al, studied the various ratings systems more carefully they would not only have concluded that they do indeed “measure different corporate governance metrics”, they would have also discovered that The Corporate Library’s ratings suffer from none of the shortcomings and limitations cited. But the Daines study is flawed in other ways as well. For example, we have never been able to determine where or how the authors obtained our ratings, why they chose to test them in ways that differ considerably from their intended and application, or why they limited their efforts to a single one year period, given the long-term focus of most corporate governance research. The other studies cited focus primarily on the RiskMetrics Group’s “CGQ” ratings, which are distributed free of charge in support of that company’s proxy advisory and corporate consulting business. Our ratings, on the other hand, are available by subscription only, are fully independent and cannot be gamed by the companies that we rate. We do not under any circumstances provide consulting services to these companies. We rate companies by identifying certain key indicators of governance risk, using a rigorous methodology that completely avoids the “one size fits all” problems that plague the RiskMetrics Group’s ratings. They are based entirely on factual data collected from public filings. Independent third party testing has proven these ratings to be strongly linked with lowered investment risk, and susceptible of significantly enhancing investment returns. A specialized version of our ratings developed for D&O liability underwriters have proven consistently effective at predicting future securities class action suits. In most cases our ratings accurately predicted not only the steep decline of the financial services industry as a whole but also identified which companies would be most seriously impacted, with far greater accuracy than the credit rating firms. Such outcomes are hardly the result of a “thin body of empirical research”. They are the result of our long history of involvement with corporate governance.Ric MarshallChief Analyst, The Corporate Libraryhttp://blog.thecorporatelibrary.com/blog/

  2. Cheri Gaudet   May 12, 2010 at 9:11 am

    Here is The Corporate Library’s updated response to the “Rating the Raters” study and information about our corporate governance ratings.http://blog.thecorporatelibrary.com/blog/2010/05/corporate-governance-ratings-rating-the-raters.html