Chairman Reed, Ranking Member Bunning, and members of the Subcommittee, I want to thank you for inviting me to testify. Effective corporate governance is a crucial foundation for economic growth, and I am honored to have been asked to participate.
A. Are There Any General Lessons for Corporate Governance from the Financial Crisis?
Some have described the ongoing financial crisis as reflecting poorly on US corporate governance, as with the accounting scandals and stock market bubbles of the late 1990s and early 2000s that led to the Sarbanes-Oxley Act. Unlike those episodes, however, the ongoing financial crisis has not exposed new and widespread problems with the basic governance of most US publicly held corporations. Outside the financial and automotive sectors, most companies have suffered only as a result of the crisis, and did not contribute to or cause it. Stock prices have fallen across the board, but most price declines have more to do with the challenges facing the real economy, and the spillovers from the financial sector on companies in need of new capital, and little to do with any general problem with corporate governance. As a result, we have learned relatively little about many long-standing concerns and debates surrounding the governance of publicly held corporations – and there are few if any easy lessons that can be drawn from the crisis for corporate governance generally.
I do not mean to minimize those concerns and debates, or suggest lawmakers should remain passive in the field of corporate governance. To the contrary, the crisis makes reform more important and urgent than ever, because well-governed companies recover and adapt more readily than poorly governed firms. But the best reform path will need to attend to differences between governance across industries, and ways that corporate governance interacts with industry-based regulation – and in particular, financial industry regulation – if legal changes are not to make things worse, rather than better. Governance flaws at Citigroup differed dramatically from governance flaws at GM, and attempts to fix the problems at firms like GM through laws directed at all public companies could make things worse at firms like Citigroup.
One important problem at financial firms was excessive risk-taking, stemming from a so-called “bonus culture” of compensation practices strongly linked to share prices. But the risks that financial firms took on were harmful for the nation as a whole because the financial firms were so important (and complex) and existing resolution authority so weak and poorly designed that those financial firms could not generally be allowed to fail. As a result, in economic terms, financial firms’ compensation practices did not take into account the external effects on taxpayers in the event of insolvency. In effect, financial firms were allowed to gamble with taxpayer money. This would have been true even if managers of those firms had been perfect stewards of shareholder wealth. The suggestion of my colleagues Holger Spamann and Lucian Bebchuk (2009) – praised by the New York Times editors earlier this week – that financial firms be required to link compensation to returns on their bonds as well as their common stocks reflects this point. Shareholders are not the only important corporate constituency to consider in setting corporate governance rules for banks.
At most public companies, the diagnosis has not been the same. If anything, the conventional critique of the governance of non-financial companies is that boards and managers have tended (from the shareholder perspective) to be excessively resistant to change, and to have tied executive compensation too weakly with performance. When commentators attempt to link compensation at firms like AIG and claims about excessive executive compensation at public companies generally, they fail to acknowledge that most shareholders do not mind if executives make an enormous amount of money, as long as shareholders also gain. Efforts to increase shareholder power to encourage managers more strongly to pursue shareholder wealth could – at financial firms – undermine efforts by bank regulators to restrain risk-taking by those same firms. The most important practical lesson of the financial crisis is, then, this: whatever form general corporate governance reform takes, careful thought should be given to exempting – or at least allowing relevant financial regulatory authorities to exempt or override – financial firms from those reforms.
B. Evidence on Policy Options
Turning from the general lessons of the financial crisis to some of the specific governance reforms that have been discussed or proposed in the last few years, it is important to bear in mind that corporate governance is not rocket science – in fact, it is much more complicated than rocket science. Corporations are in their simplest sense large groups of people coordinating their activities for profit. Science has a hard enough task tracking inert matter moving through space; it has a harder time predicting the behavior of a single actual or typical human; and it has the hardest time of all attempting to describe or predict how large groups of people will act – if for no other reason than researchers cannot experiment on large groups of people in realistic settings. As a result, there are few consensus views among researchers about any non-trivial topic in corporate governance, and evidence tends to emerge slowly, is rarely uncontested, and is subject to constant (and often dramatic reevaluation). As a result, everything that you do in setting rules for corporate governance should keep the fragility of the evidence in mind: set rules that can be changed by delegating to regulatory agencies; direct those agencies to review and reassess their own rules regularly; and provide “opt outs” and “sunsets” to governance mandates that are expected to last indefinitely, as at many corporations.
As one example, to my knowledge, there is no reliable large-scale empirical evidence – good or bad – on the effects of shareholder access to a company’s proxy statement, along the lines proposed by the SEC and mandated by S. 1074, H.R. 3269 and H.R. 2861, because there has no been no significant observed variation in such a governance system within any modern developed economy. This does not mean that there is no information relevant to evaluating how such a system would operate in practice, or that there is no basis on which such a system could be recommended or adopted. Rather, the absence of observed variation means that there is no general body of data that is capable of revealing whether such a system would consistently have good or bad effects on shareholder welfare – and no such data will exist unless and until a large number of companies voluntarily adopt such a system or are required to by law. That is generally true of many corporate governance proposals, and to require such data before adopting rule changes would effectively freeze laws governing corporate governance in place indefinitely, preventing further inquiry or development of evidence.
Nonetheless, there are some corporate governance topics about which evidence is better than others. Here I set out what is necessarily an abbreviated summary of the evidence on three topics addressed in one or more bills pending in the current Congress, including the Shareholder Bill of Rights Act of 2009 (S. 1074): (a) say on pay, (b) mandatory separation of the chairman and CEO positions, and (c) mandatory annual board elections.
a. Say on Pay
The proposed requirement that shareholders be given an advisory vote on executive pay has the advantage that it is very similar to a requirement adopted in another jurisdiction (the United Kingdom (UK)) that has capital markets and laws that are otherwise similar to those applicable in the United States.
This fact enables a research approach that is otherwise unavailable: a before-and-after test of board and shareholder responses, compensation practices, stock market reactions and shareholder returns, and other items of interest surrounding the adoption of say-on-pay in the UK. Different researchers have conducted several investigations of this kind and the results published at least informally. Those researchers report that say-on-pay’s adoption in the UK:
• improved the link between executive pay and corporate performance (Ferri & Maber 2007);
• led firms (both before and after relatively negative shareholder votes) to adopt better pay practices (id.);
• led activist shareholders to target firms with weak pay-performance links and those with higher-than-expected executive compensation levels (id.; Alissa 2009);
• did not reduce or slow the overall increase in executive compensation levels (Ferri & Maber 2007; Gordon 2008).
Together, these findings suggest that say-on-pay legislation would have a positive impact on corporate governance in the US. While the two legal contexts are not identical, there is no evidence in the existing literature to suggest that the differences would turn what would be a good idea in the UK into a bad one in the US.
Researchers have also exploited the introduction of earlier say-on-pay legislation in the US to examine stock price reactions to the prospect of such a governance reform. Consistent with the UK findings, they report that stock investors appear to have viewed the proposed legislation as good for firms with higher-than-typical executive compensation, firms with weak pay-performance links, and firms with weak corporate governance measured in various ways (Cai & Walkling 2009). They also report data showing that the market reacted positively at most sample firms to the proposed legislation. The same researchers also report that shareholder-sponsored efforts to introduce say-on-pay rules at individual firms – particularly when sponsored by unions with low stock holdings in the targeted firms – were not well-received by the stock market, in part because they were not directed at firms with higher-than-typical executive compensation or firms with weak pay-performance links, but instead simply at companies that happen to be large. The researchers suggest that their findings show that one-size- fits-all say-on-pay legislation may be harmful, but this implication does not in fact follow from their findings. If anything, the UK evidence summarized above suggests that general say-on-pay legislation will weaken the ability of special interest shareholder activists to exploit executive compensation as an issue, and will lower the costs of the broad run of shareholders to use their advisory votes on pay to target firms that are most in need of pressure to improve pay practices.
b. Mandatory Separation of Chairman and CEO Positions
In comparison to research on say-on-pay rules, the evidence on the proposal to mandate the separation of the chair and the CEO of public companies is more extensive and considerably more mixed. At least 34 separate studies of the differences in the performance of companies with split vs. unified chair/CEO positions have been conducted over the last 20 years, including two “meta-studies.” Dalton et al. (1998) (reviewing 31 studies of board leadership structure and finding “little evidence of systematic governance structure/financial performance relationships”) and Rhoades et al. (2001) (meta-analysis of 22 independent samples across 5,271 companies indicates that independent leadership structure has a significant impact on performance, but this impact varies with context). The only clear lesson from these studies is that there has been no long-term trend or convergence on a split chair/CEO structure, and that variation in board leadership structure has persisted for decades, even in the UK, where a split chair/CEO structure is the norm.
One study provides evidence consistent with one explanation of the overall lack of strong findings: optimal board structures may vary by firm size, with smaller firms benefiting from a unified chair/CEO position, with the clarity of leadership that structure provides, and larger firms benefiting from the extra monitoring that an independent chair may provide given the greater risk of “agency costs” at large companies. Palmon et al. (2002) (finding positive stock price reactions for small firms that switch from split to unified chair/CEO structure, and negative reactions for large firms). If valid, this explanation would suggest that it would be a good idea for any legislation on board leadership to (a) limit any mandate to the largest firms and (b) permit even those firms to “opt out” of the requirement through periodic shareholder votes (e.g., once every five years).
c. Mandatory Annual Board Elections
The evidence on the last legislative proposal I will address – mandatory annual board elections (i.e., a ban on staggered boards) – is thinner and at first glance more compelling than that on board leadership structure, but on close review is just as mixed. There have been at least two studies that focus on the specific relationship between annual board elections and firm value (Bebchuk & Cohen 2005; Faleye 2007), and a number of other papers that include annual board elections in studying the relationship between broader governance indices and firm value more generally (e.g., Gompers et al. 2003; Cremers & Ferrell 2009). Most (but not all) conclude that annual board elections (either on their own or in combination with other governance practices) are associated with higher firm value, as measured by the ratio of firms’ stock prices to their book values. The governance-valuation studies, however, generally suffer from a well-known “endogeneity” problem – that is, it is difficult (and given data limitations, sometimes impossible) to know whether annual elections improve firm value, or firm value determines whether a company chooses to hold annual elections. While there are statistical techniques that can address this issue, none of the studies to date have presented compelling evidence that annual elections lead to better performance, at least in the last 20 years, during which time public companies rarely switched from annual to staggered elections. Moreover, the longer a given study of this type has been available for others to attempt to replicate, the more fragile the findings have appeared to be, suggesting that the bottom-line conclusions of more recent studies may not hold up in the face of continued research.
Evidence on annual elections is further complicated by the fact that companies that “go public” for the first time continue to adopt staggered board elections at high rates, as late as 2007. Since the evidence regarding the purported ability of staggered boards to improve firm value has been known for some time, and since shareholders have the ability to adjust the prices they pay for newly issued IPO shares to reflect governance practices, the fact of continued adoption of staggered board elections prior to IPOs suggests that there may be a social advantage to permitting these structures, at least when adopted before a company goes public. Other researchers have made a similar point about “dual class” capital structures, which give low or no votes to public investors, while letting founders or their family members retain high vote stock. SEC rules and stock exchange listing standards have for a long time permitted such structures to be adopted in the US only prior to a company going public, and not once a company has gone public. Such structures, as with staggered board elections, have long been thought to reduce firm value, measured by reference to public stock prices. Yet, as with staggered boards, some companies continue to adopt dual class structures – and some have done quite well by their shareholders (e.g., Google Inc. – still up over 300% since its IPO despite the recent market meltdown).
The best explanation offered by academic researchers to explain the continued use of dual class structures and staggered board elections is that they provide founders assurance of continued control, which they value more than the stock price of their companies might reflect. Such private value may arise because of particular attachments the founders have towards the companies they have helped build from scratch, or because they hope to pass control of their companies to their children, or because they have developed “firm-specific capital” that they would lose if the company were acquired (and which would be hard to value by outsiders). Some evidence has been developed consistent with these explanations (see Coates 2004, reviewing prior research). This evidence is worth considering not only because dual class structures are analogous to staggered board elections – and interfere with hostile takeovers and shareholder voting rights even more than do staggered board elections – but also because any to mandate annual board elections would also require a ban on dual class structures, or else it w ould simply push companies to adopt the more restrictive dual class structure in lieu of staggered boards.
My recommendations flow from my review of the implications of the financial crisis and my review of evidence above:
First, any corporate governance reform that attempts to shift power from boards or managers to shareholders should either not include financial firms, or should include a clear delegation of authority to financial regulators to exempt financial firms from these power shifts by regulation. Simply directing financial regulators to regulate the same governance practices (as in H.R. 3269) may not suffice to prevent shareholder pressure from encouraging firms to craft ways around those regulations. It would be better more generally to moderate the pressure of shareholders on financial firms to maximize short- term profit at the potential expense of the financial system and taxpayers.
Second, “say on pay” legislation is likely to be a good idea. By enabling shareholders across the board to provide feedback in the form of advisory votes to boards on executive compensation, such a requirement would be likely to increase board scrutiny on one element of corporate governance that has the greatest potential for improving incentives and firm performance in the long run. At the same time, it should be recognized that “say on pay” is not likely to achieve general distributive goals – wealthy CEOs will continue to earn outsize compensation, as long as their shareholders benefit. If the goal of Congress is to reduce wealth or income disparities, say on pay is not the right mechanism, and executive compensation is only a relatively minor part of the picture. For that reason, efforts to use corporate governance practices –- which after all only affect a subset of all US companies, those that have dispersed shareholders – to force a linkage between CEO and employee pay seem to me misguided. It would be better to address pay disparities in the tax code.
Third, while mandating a split between the chair and the CEO is not clearly a good idea for all public companies, it may well be a good idea for larger companies. Because shareholders of those same companies may find it difficult to initiate such a change, given the difficulties of collective action, a legislative change requiring a split leadership structure but permitting shareholder-approved opt outs may improve governance for many companies while imposing relatively minor costs on companies generally. Requiring that companies give shareholders a vote on such a choice episodically (e.g., every five years) would also be a way to help solve shareholders’ inevitable collective action problems without forcing a one-size-fits-all solution on companies generally.
Fourth, mandating that all public companies hold annual elections for all directors is not clearly supported by evidence or theory. It perhaps bears mentioning that other important institutions (the SEC, the Fed, the Senate) permit staggered elections for good reason, and that any rule mandating annual elections would ride roughshod over state law – in Massachusetts, for example, companies are required to have staggered board elections unless they affirmatively opt out of the requirement. In prior writing, I have suggested it be left to the courts to review director conduct with a more skeptical eye at companies that adopted staggered boards prior to the development of the poison pill (Bebchuk, Coates & Subramanian 2001), and I have also suggested elsewhere reasons to consider “re-opening” corporate governance practices put in place long ago (Coates 2004). Both approaches would be better than an across-the-board annual election mandate, which would be likely to lead new companies to adopt even more draconian governance practices without any clear net benefit.
Finally, precisely because there is no good evidence on the potential effects of shareholder proxy access, it would seem to be the best course to move cautiously in adopting rules permitting or requiring such access. For that reason, the most that would seem warranted for a hard-to-change statute to achieve is to mandate that the SEC adopt a rule providing for such access, and thereby to clarify the SEC’s authority to do so. Any shareholder access rule will need to address not only the length of the holding period and ownership threshold required to obtain such access, the ability of shareholders to aggregate holdings to obtain eligibility, rules for independence of nominees and shareholders using the rule, and the availability of the rule to those seeking control or influence of a company. Efforts to specify rules for such access at a greater level of detail will probably miss the mark, and be difficult to correct if experience shows that the access has either provided too much or too little access to accomplish the presumed goal of enhancing shareholder welfare.