Topic of Research: Securities Lending and Corporate Governance

 

Tentative Title of Paper: Stock Lending and Proxy Voting: Does the Tail Wag the Dog?

 

 

Introduction

 

In the December, 2007 issue of the Journal of Finance, a team of academic researchers alleged that the securities lending market serves as a secondary market for corporate proxy votes, in that some share borrowers regularly abuse their proxy voting privileges.  Savvy investors, the academics claimed, often borrowed shares to ‘buy votes’ in defiance of US regulations and accepted practices in other countries. Some in the financial media took these studies as fact and ran articles explaining “How Borrowed Shares Swing Company Votes ”, leading to a “Corporate Voting Charade.”  In part, as a consequence, concerns were voiced by legislators, regulators, trade groups, and beneficial owners. Researchers at the Center for the Study of Financial Market Evolution were tasked to determine if these charges were true.

 

Careful analysis of the published study raised questions about quality of the data, the knowledge that the authors had of market functions, the quality of the analysis, and the accuracy of the findings.  These criticisms suggest the need for a fresh and unbiased approach.

 

 

Background

 

In addition to the argument for proxy vote manipulation, certain academics also allege that there are frequent instances of distortion in the compilation of, and accounting for proxy votes as a result of inefficient procedures in the securities finance sector of the capital markets. Indeed, recent press reports have created the mistaken perception that the stock loan market is responsible for multiple voting of the same shares (Drummond, 2006), thus violating the fundamental governance principle of ‘one-share, one-vote’. It has also been alleged that even if there is no multiple voting, the market for stock lending allows vote buying by informed market participants at zero cost from uninformed atomistic shareholders in order to influence the outcome of important proxy contests in a way that may not be in the best interest of shareholders.

 

A recent working paper by Christoffersen, Geczy, Musto, and Reed (2007), which is scheduled for publication in the Journal of Finance, claims to furnish empirical evidence that there is an active market for vote trading within the US corporate stock loan market. Based on a proprietary data set consisting of loans of U.S. shares by a custodian bank in 1999 and by a broker-dealer between 1996 and 2001, they report that stock loans spike on record dates, increasing on average from 0.21% to 0.26% of outstanding shares, and claim that “the spike in borrowing on the record date strongly supports the existence of some record date capture.”  In their analysis, the spike is higher for firms with poorer performance, for votes that turn out to be close, and for votes that elicit greater support for shareholder proposals or greater opposition to management proposals.

Practitioners respond to the academic allegations by asserting that the published studies are misinformed about current procedures in securities lending/finance and that interpretation of the underlying motivation for spiking volume before record-date, if such exists, is at best superficial and, at worst, intentionally misleading in an effort to contrive an attractive thesis for publication.

 

In a related paper, which has been prominently discussed in the Wall Street Journal, Hu and Black (2006) argue that there has been a decoupling of economic ownership and voting ownership, which has potentially far more onerous consequences for corporate governance. They try to analyze what they term the new vote buying and its potential benefits and costs by considering a set of case studies. However, they do not test their hypothesis on a large statistical sample, and thus, their empirical evidence appears limited to a handful of isolated cases. Their study also fails to assess whether the phenomenon they claim to observe is economically significant. For example, transferring votes from uninformed to well-informed investors should improve economic efficiency, as suggested by Christoffersen et al. (2007).

 

Taken together, the allegations in Christoffersen et al. (2007) and Hu and Black (2006) seem to challenge the social value of the securities lending market, which in the U.S. probably totals more than $4 trillion in stocks, corporate bonds and government securities on any given day. This critical stance contradicts the long-standing views of market regulators, as well as practitioners, who regard securities lending as an important contributor to capital market efficiency.

 

Starting with the July 1999 report (http://www.bis.org/publ/cpss32.htm) produced by the Technical Committee of the International Organization of Securities Commissions (IOSCO) and the Committee on Payment and Settlement Systems of the central banks of the Group of Ten countries (CPSS), numerous banking authorities and market regulators have endorsed the securities lending market as essential to a well-functioning capital market.  In their view, lending provides liquidity in direct and indirect ways. Examples of critical market activities supported by securities lending include repo transactions, M&A arbitrage in stock-for-stock deals (where an arbitrageur buys the target company’s stock and sells short the acquiring company’s stock by borrowing these shares), and trading in options and other derivative instruments (both for replicating certain option strategies and for fulfilling certain futures contracts where, for example, the cheapest-to-deliver T-Bond has to be borrowed for immediate delivery).

 

In these and other cases, securities lending not only provides essential liquidity for the smooth and efficient functioning of the broader capital markets, and helps to avoid market squeezes and the establishment of corners. Hu and Black (2006) cite a few examples where M&A arbitrage seems to lead to an economically inefficient economic outcome, but these limited examples do not invalidate the generally efficient outcome of M&A arbitrage. In M&A deals, there are other examples of potentially-inefficient economic outcomes, such as two-tier tender offers, so that risk arbitrage should not be singled out as the sole cause of such outcomes. In any event, a careful statistical test is needed to evaluate the seriousness of the Hu and Black (2006) allegations. The examples they cite might prove to be simply a few isolated anomalies.

 

The Christoffersen et al. (2007) study is potentially quite troubling. It reaches the paradoxical conclusion that while these votes are potentially very valuable in contentious proxy contests, the average vote sells at a zero price. The study uses a limited and aged sample to test its central thesis. It also suffers from at least three other important deficiencies: (a) an imperfect generalization of its empirical results to the broader issue of how the entire securities lending market operates, and in particular, its failure to acknowledge the many different perfectly valid reasons for securities lending in general and stock lending in particular; (b) a failure to take into account properly the cost and risks of carrying a large stock loan position; and (c) a failure to consider fully how the nature of the corporate governance event can affect the demand for stock loans.

 

As Hu and Black (2006) point out, Christoffersen et al. (2007) seem to conclude that ill-informed lenders consciously yield their shares to better-informed borrowers in order to benefit from this information aggregation without ever learning the foolishness of their ways. However, even casual empiricism reveals that it is the borrowers who are the active agents in the market, and it must be demonstrated that borrowers have a positive economic interest in stock borrowing. Stock borrowers are not always outside agents, which is contrary to one of the assumptions in Christoffersen et al. (2007). A more thoughtful study that correctly reflects the way the stock loan market currently operates, especially if that study uses a more comprehensive and up-to-date data set, could very well come to a very different conclusion and resolve the apparent zero-cost paradox.

 

 

RESEARCH OBJECTIVES

 

The research findings of Christoffersen et al. (2007) and Hu and Black (2006) and all reasonable alternative explanations for their findings should be very carefully considered before accepting the policy implications of any new regulations that might be imposed on the entire securities lending industry. Professor Hu’s argument that the securities lending market not only enables, but actively facilitates abuses in corporate governance by share borrowers could precipitate a calamitous termination of institutional securities lending programs resulting in a loss of needed revenue for shareholders and beneficiaries, as well as a reduction in liquidity and operational efficiency within the global capital markets.

 

1. How strong is the empirical support for the Hu citation of Christoffersen's allegation that "the spike in borrowing on the record-date strongly supports the existence of some record-date capture"? Specifically, for a representative sample of stocks with material or contentious votes, how frequently does volume in the securities lending markets "spike", as defined in the Christoffersen article? Is the incidence of volume spikes in contentious issues significantly greater than in a representative sample of non-contentious issues?

 

2. Is an attempt to accumulate votes by borrowing shares necessarily abusive, as alleged in the Hu article on empty voting in the securities lending markets, as they truly exist? That is, does an increase in share lending volume before a proxy record-date necessarily indicative of an attempt to accumulate votes?  Is an accumulation of votes necessarily for the benefit of new holders? Alternatively, how often are the pre-proxy borrowed votes intended for long-term holders whose shares have been placed on loan by their broker-dealers?  Similarly, how often are pre-proxy borrowed votes intended for delivery to long-term short-sellers whose brokers are substituting newly borrowed shares for older borrowed shares being recalled by lenders, who may be acting in order to vote their positions?

 

3. Do market forces in securities lending serve to limit the potential for market abuses with respect to the proxy voting and corporate governance process?  For example, at what point does the rising cost to borrow shares render impractical the possible manipulation of corporate governance through the use of derivative, as defined in the Hu article?  In other words, does the securities lending market create a form of economic “self-regulation”, thus acting as a deterrent to would-be derivatives-market manipulators?

 

4. CalPERS will not lend shares of certain companies around voting record dates and will only lend shares, according to Hu, to "those who have a legitimate right to the proxy as a benefit of true ownership." How is the legitimacy of the proxy "right" defined?  How is the legitimacy of "true ownership" defined?  To what extent, if any, should the voting franchise in the corporate governance process be redefined (or restructured) as a "benefit" in the cross-border, derivatives-hedged portfolios of institutional investors?  Should lending agents and broker-dealers modify their practices to protect the legitimacy of proxy rights for true owners?

 

 

Current Literature

 

D’Avolio (2002) is the first paper to describe the US stock loan market in any detail and to investigate how it operates. He obtained eighteen months of data on stock loan transactions, loan fees, and loan recalls from a large institutional investor that functions as a regular stock lending intermediary. He concludes that stock loan fees are nominal and that stock loan recalls are rare.  Geczy, Musto and Reed (2002) use one year of data on stock loans extended by a major stock lender in the US to study the effect of short-selling costs and the institutional and legal constraints on short-selling strategies. They checked the effect of these restrictions on various types of firms, including IPO firms, Dot Coms, large cap firms, growth firms, and firms with low-momentum stocks. The proposed research would build on this earlier work.

 

Directly relevant to the proposed research topic are the papers by Christoffersen et al. (2007), which uses the same data set as Geczy, Musto and Reed (2002), and Hu and Black (2006), which were discussed previously. We expect that correcting the deficiencies found in these two papers could lead to different conclusions. For example, Hu and Black (2006) describe various strategies market participants might use to decouple share voting from shareholders’ economic interests, and they discuss short-term and long-term regulatory implications of their findings. None of these recommendations make any sense if the strategies they describe are limited to a few isolated cases, especially if they result in the transfer of shares from poorly informed to well informed shareholders.

 

 

Testable Hypotheses

 

 

Research Methodology

 

The first and most important task was to collect an up-to-date sample of equity loan transactions that document the identity and reasons for the loan where such data are available, subsequent proxy-voting data with details of proposals being voted upon, pricing of such loans, reversals of these transactions, and market trading data, such as volume and stock price.

 

The plan is to employ standard statistical techniques to investigate the six testable hypotheses. The research methodology is similar to those employed in the D’Avolio (2002), Geczy, Musto and Reed (2002), and Christoffersen et al. (2007) papers. This research will avoid certain inaccuracies in Christoffersen et al. (2007), by insuring that our research methodology is faithful to the manner in which the stock loan market currently operates. Having the ability to work with stock loan market professionals will be invaluable in that regard.

 

 

Data Required to Perform the Study

 

The data required for the academic study are identified below. One record per security, per trading day, will be provided to include the following statistics:

 

 

Research Output

 

The research output will be a written report that (a) specifies the issues researched, (b) describes how the stock loan market works (and identifies the features Christoffersen et al. (2007) appear to have overlooked or misunderstood), (c) explains the research methodology, (d) presents the research findings, (e) explains the conclusions drawn from the research findings, and (6) distinguishes our conclusions form those of Christoffersen et al. (2007) and Hu and Black (2006). Interim results will be presented to CSFME and the industry sponsors no later than six months from the date the research data is provided.

 

 

Publication of Research Results

 

An academic paper based on the research report will be submitted for publication to a top refereed journal. The academic paper will acknowledge the financial and other support furnished by CSFME and the other sponsors of the research study.  It will be presented in standard scientific style.

 

 

Access to Lending Agent Data Base

 

The academic researchers need access to CSFME’s robust stock loan data base, including assistance from CSFME staff in pulling the data items needed from that data base. From this, we plan to create a separate, secure database for our study incorporating all the data items from the various sources mentioned in this proposal. We would use this database to perform the hypothesis testing. All members using the database will sign a standard confidentiality agreement.

 

 

Proposed Research Funding

 

To be determined.

 

 

Possible Follow-on Studies

 

The possibility exists for conducting follow-on research concerning the role of the stock loan market in promoting stock market efficiency.  However, the CSFME database can be used only for the purposes of this research, exclusively within the confines of this project. All data and copies of working papers will be placed in academic escrow at the end of the study. 

 

Going forward, as a result of this study, the academic research team will gain a familiarity with the CSFME database that will enable them to perform the additional queuing and short selling impact studies that have been considered. The team will be given the opportunity to submit additional proposals for follow-on studies.

 

 

References

 

- Christoffersen, Susan Kerr Christopher Geczy, David K. Musto, and Adam V. Reed, 2007, “Vote Trading and Information Aggregation,” Working Paper, McGill University (January) http://ssrn.com/abstract=686026.

 

- D’Avolio, Gene, 2002, “The Market for Borrowing Stock,” Journal of Financial Economics, 66 (2-3), 271-306.

 

- Drummond, Bob, 2006, “Corporate Voting Charade, April, Bloomberg Markets,” http://www.bloomberg.com/media/markets/apr_ft_proxy.pdf

 

- Geczy, Christopher, David K. Musto, and Adam V. Reed, 2002, “Stocks are Special Too: An Analysis of the Equity Lending Market,” Journal of Financial Economics, 66 (2-3), 241-269.

 

- Hu, Henry T. C., and Bernard Black, 2006, “The New Vote Buying: Empty Voting and Hidden (Morphable) Ownership,” Southern California Law Review, 79 (May), 811- 908.

 

- Smith, Randall, 2006, “Moving the Market: Short-Selling Inquiry Is Widened to Cover Stock- Lending Business,” Wall Street Journal.