What is this thing called “corporate governance” that is occupying more and more time in the boardroom? Does having more of it create value for stakeholders in the company? Are institutional activists, politicians, regulators and media pundits justified in their efforts to reform it? And are the costly changes they are proposing likely to be socially desirable?
Corporate governance clearly matters to shareholders, customers, employees and parties that interact with the company. Moreover, there is little doubt that governance mechanisms are a necessary and vital part of economic growth and the functioning of a liquid capital market.
The separation of ownership and control that characterises the structure of the corporation means that managers can potentially make self-serving decisions at the expense of stakeholders. It is therefore important to put in place a set of mechanisms to constrain such managerial decisions in a way that maximises the net benefits to stakeholders. As well as giving general protection to stakeholders, an appropriate governance structure should lead to better performance.
In order to understand whether there are indeed net benefits from corporate governance, it is necessary to take an agnostic and scientific look at the data and assess the merits of the claims made by the business press, ratings agencies and many others. For example, are companies with more independent directors on the board or a separate CEO and chair priced at a premium in the market? And do managers of these companies make higher-quality reporting, investing and financing decisions?
There is substantial variation in governance structures across different countries and even across companies within a country. One interpretation of this observation is that each company faces a unique set of problems and costs to institute various mechanisms, and thus a different governance structure is optimal. In other words, governance should not be considered a “one size fits all” proposition.
By contrast, there is a widely held belief among consultants, politicians and ratings agencies that there is a single, optimal governance structure, and that any company that deviates from this has a governance problem. This type of “boiler-plate” governance benchmark is obviously simple to apply. At face value, recommendations such as the appointment of a majority of independent directors to the board and the separation of the roles of CEO and chair seem reasonable. However, adopting these “boiler-plate” recommendations is a costly process and it is far from clear whether these changes will produce better-managed companies as well as satisfy stakeholder objectives.
Measuring corporate governance
Before attempting to assess the impact of governance or managerial decisions on company performance, it is necessary to come to grips with how to measure corporate governance. Can a complex structure of governance be reduced to a summary statistic? How does one go about identifying which of the myriad governance mechanisms should be included?
As might be expected, researchers and ratings services use a multitude of measures. For example, the Institutional Shareholder Services Corporate Governance Quotient is based on 61 variables and the Governance Metrics International Corporate Governance Score is based on 450 data points. Regardless of the precise computation, most of the ratings examine factors related to board and committee structure, executive pay, anti-takeover provisions, concentration of equity ownership and other measures.
Any external assessment of the quality of corporate governance is limited to what can be observed. This means that the majority of measures captured by the various ratings agencies and academic research is based on characteristics, such as board structure and processes, disclosed executive compensation, distribution of ownership, various anti-takeover provisions and so on.
However, these simple measures do not include insights into the inner workings of the corporate board. There are no detailed interviews with management and board members to assess whether their objectives are consistent with stakeholder goals, and no assessment of whether various constituencies are directing the right questions to top management. As a result, ratings based on observable governance characteristics do not answer questions such as: “Are the number of board meetings, the composition of the board and sub-committees, the age of the directors and other structural measures sufficient to capture the complex nature of how an effective board should work?” Yet, these are the types of measures that are the focus of the business press, customers, employees, investors and regulators.
Analysis of governance measures
We obtained data from four major intermediaries that specialise in rating corporate governance practices of companies: Governance Metrics International (GMI), Investor Responsibility Research Center (IRRC), Institutional Shareholder Services (ISS) and The Corporate Library (TCL). For each of the agencies, we analysed the association between their ratings, and subsequent company operating performance and stock returns. A potential benefit of these measures is that they presumably reflect not only formal structural measures of governance but also the expertise of the analysts working for the ratings agencies.
Overall, the results of our analyses were similar regardless of the source of the ratings. We found no evidence that the summary ratings were associated with stock returns. Alternatively stated, an investment strategy based on the governance ratings was not a profitable one. We did, however, find some evidence that governance ratings are associated with the level of future operating performance. The results for operating performance were strongest with data from TCL, so we explain the analysis of these specific ratings in more detail below.
The analysis for the TCL data was drawn from 2,012 companies for the period 2002-2004. The overall TCL “board effectiveness” rating is based on the following components: board composition, CEO compensation, shareholder responsiveness, accounting quality, strategic decision-making, litigation problems and takeover defences. Thus, the TCL rating appears to be a relatively comprehensive and broad-based measure of corporate governance.
Each of the components is given a rating between A and F, and the combination of the categories also yields an A-F overall score for board effectiveness. Companies receiving an A, B or C grade are considered to be have “good” governance and those with a D or F rating to have “bad” governance.
We utilised the discrete (good versus bad) distinction in the TCL ratings and examined stock price performance after issuance of their reports. The analyst reports are compiled after the end of each proxy season. To ensure that the information in the reports is available to investors, we tracked the stock performance from July 1 in the year that the rating is disseminated through to June 30 the following year. TCL’s analysts conduct their analysis in conjunction with other sources during the proxy season period (up to the middle of 2003 for the 2002 fiscal year). Therefore, examining stock returns for the period of July 1 2003 through to June 30 2004 will capture the 12-month period following the issuance of the TCL rating. We repeated this process for the three years for which we have data and rebalanced the portfolios every year on July 1.
In Figure 1, we illustrate the average returns that would have been earned for $1 invested in companies with “good” ratings for board effectiveness and those with “bad” ratings. As is evident from our analysis, there is very little difference in the returns to an equity investor based on the TCL board effectiveness rating. Indeed, statistical tests revealed no difference between the two portfolios. We found a similar lack of stock return predictability for the composite ratings from GMI and ISS, as well as the shareholder rights provisions collected by IRRC.
A potential explanation for the stock return results is that the market correctly anticipates the future performance of “good” and “bad” governance companies, and thus there should be no differential stock price returns for these two groups. However, if governance is important to performance, we should be able to detect performance differences using measures of operating performance. Thus, a more direct way to assess performance is to see whether “good” governance companies have higher return on assets (that is, the ratio of operating income to total assets).
In our research, we examined the ability of the TCL board effectiveness ratings to predict both the change in and the level of one year-ahead return on assets. Consistent with our stock return results, we found no evidence that the TCL board effectiveness rating is associated with change in operating performance. We did, however, find some evidence that companies with “good” TCL ratings have a higher level of operating performance in the year following the release of the rating relative to those with “bad” ratings. Companies receiving a “good” TCL rating experienced higher return on assets in the order of 1.5 per cent in the year following the release of the board effectiveness rating.
Collectively, the evidence suggests that there may be a modest association between governance scores and future operating performance, but this association does not translate into future company value (as measured by changes in stock price).
While not necessarily damning of measures of governance, the lack of a robust pattern between popular governance ratings and future performance should raise questions about the assertion that governance changes will improve shareholder returns. Of course, the time period we examined for these ratings is quite limited (July 2002 through to December 2004), and it may take a longer period for the valuation effects of governance to materialise. Nevertheless, our empirical analysis is based on the most current data and collectively it suggests that there is not much evidence supporting recent claims that governance is associated with performance.
A limitation of our analysis is that we have only used overall scores. Furthermore, these governance ratings are somewhat unsophisticated aggregates of multiple measures of governance structures. To address this limitation, we conducted our own study on the relation (or lack thereof) between corporate governance and company value. This involved a detailed statistical analysis of 2,106 US companies with fiscal years ending between June 2002 and May 2003. Our sample spans many sectors of the economy and represents over 70 per cent of the market capitalisation of the Russell 3000 as of the end of 2003.
In order to be included in our sample, we required companies to have available data (much of which was provided by TrueCourse Inc. and Equilar Inc.) on the characteristics of the board of directors, stock ownership by executives and board members, stock ownership by institutions, stock ownership by activist holders, debt and preferred stock holdings, compensation mix variables and anti-takeover provisions. In all, we used 39 different structural measures of corporate governance. There is substantial overlap between the measures that we use and those used by the ratings agencies.
Using a variety of statistical techniques, we assessed whether these 39 measures actually explain differences in managerial behaviour and organisational performance. We found that companies with a lead director experienced higher stock returns. Similar to the analysis with the governance ratings, we did not find any association between our governance measures and changes in operating performance. We did, however, find evidence that companies with activist shareholders, more accounting-based incentive compensation and without poison pills and staggered boards experienced a higher level of operating performance in fiscal 2003.
What is most interesting about our analysis is that, of the 39 governance measures examined, only a small fraction was connected with the level of operating performance and, often, the direction of the associations was unexpected. For example, companies with larger boards and more control by insiders via dual class stock and fewer outsiders on the board exhibit a higher level of operating performance.
Overall, our results imply either that corporate governance is of modest importance (which is difficult to believe) or that the available structural indicators and ratings are not especially useful for measuring it.
Discussions with asset managers
We also interviewed several leading portfolio managers representing large actively managed US and international equity portfolios about their evaluation of governance ratings. We received consistent reports that measures of governance are not effective in generating excess returns (or alpha) for equity portfolios. These responses are likely to reflect several issues. First, as we discussed above, governance is a multi-dimensional construct that is difficult to measure and reduce to a summary statistic. The failure to find an association with stock returns may simply reflect measurement error. Second, there is only a limited period for which governance ratings are available and this makes it impossible to do the type of back-testing that quantitative equity managers typically undertake. Third, the signals that these portfolio managers are already using in their asset allocation decisions may be subsuming any ability of the governance measures to predict future returns. Finally, for measures of governance to be associated with future stock returns, it must be the case that these measures provide predictive information that is not priced by the market in a timely and efficient fashion.
Although governance ratings have been given a very lukewarm reception by some leading actively managed equity funds, there is still a strong interest from equity investors in governance risk of their portfolios. Portfolio managers at some of the larger US pension plans, such as TIAA-CREF and Calpers, have been strong advocates of governance reform. Indeed, their proxy voting has been significantly influenced by perceived governance crises. The ratings agencies themselves are selling ratings and governance advice to a broad base of clients including institutional investors. The fact that there are buyers for these ratings must mean that there is some value to having access to this information. However, the value of these ratings is not related to their ability to predict future company performance.
Conclusions
The relation between corporate governance and organisational performance is of fundamental importance. As might be expected, many conjectures about the features of superior corporate governance have been advanced. In addition, various governance-rating schemes have been proposed as the basis for the design of governance structures. While these ideas make for interesting discussion, there are few compelling results that clearly demonstrate how corporate governance produces the outcomes desired by stockholders or, more broadly, stakeholders. Thus, there is little evidence that the costly governance changes that are currently being imposed on companies will produce expected net benefits in terms of improved performance.
We certainly believe that appropriate governance mechanisms are a necessary and vital part of a capitalistic economy. However, we have considerable concern about whether any of the existing structural measures of governance or governance ratings provide a useful basis for identifying good governance. Before imposing some governance structure on a company, it seems necessary to verify scientifically that the changes are likely to produce the expected outcome. This statement, of course, assumes that we are interested in learning about the value that can be produced from making costly changes to governance mechanisms. If instead, the debate continues to rage based on rhetoric, we should be aware that costly policy decisions are being made without a careful and rigorous analysis of the data.
David Larcker is the Ernst Young professor of accounting at the Wharton School of the University of Pennsylvania.
Scott Richardson is an assistant professor at the Wharton School of the University of Pennsylvania.
Irem Tuna is an assistant professor at the Wharton School of the University of Pennsylvania.
