How Much Do Directors Influence Firm Value?

Aaron Burt is Assistant Professor of Finance at the University of Oklahoma Price College of Business; Christopher Hrdlicka is Assistant Professor of Finance and Business Economics at the University of Washington Foster School of Business; and Jarrad Harford is Professor of Finance and Chair of the Department of Finance and Business Economics at the University of Washington Foster School of Business. This post is based on their recent article, forthcoming in the Review of Financial Studies.

Every company has a board of directors. Debates rage over whether they do their job; what is the ideal mix of insiders and outsiders, men and women, management and labor; and whether directors are too busy or whether busyness is an outcome of quality. But until now, we have not even been able to answer the most basic question: How much do directors actually influence the value of companies?

To be fair, this is a hard question to answer. Much about boards is unobservable. Board meetings are private, with the minutes only rarely being made public. Directors work with management outside of these meetings, making it impossible to see all the relevant interactions. Worse, we cannot simply look at arrivals and departures of directors because of the endogenous matching between a director and company.

Narrowing the question to particular events like mergers and acquisition have allowed a glimpse of director influence. Natural experiments such as mandates on the female fraction of boards or the unexpected death of a director also show boards matter. The effects of directors have even been traced to commonality in events such as switching exchange listings, or commonality in practices, such as similar tax minimization strategies. Taken together, prior studies have shown a vast array of director influence in specific instances, but they cannot answer how much do directors influence firms overall. Their answers are limited both because they utilize only a subset of events and because there is no obvious way of aggregating over the events studied.

In our paper How Much Do Directors Influence Firm Value? which is forthcoming in the Review of Financial Studies, we utilize the fact that directors sit on multiple boards to provide an unconditional measure of the value influence of a director. We find that, on average, a director’s influence causes variation of 1% of aggregate market capitalization per year. For perspective, our sample’s average annual variation in market value is 14.5%. This 1% swing from a single director’s influence accounts for 6.5% of the total observed volatility. Knowing that directors wield such influence makes the debates over optimal board structure and regulation even more important.

Our measure is based on a new method that exploits the commonality in news across firms which share directors. If directors matter in the sense of influencing their firms, then when a director is seated on two boards, the comovement of those firm’s stock should increase. We document this increased comovement using the network of shared directorships between 1996 and 2015. To eliminate other potential sources of comovement (such as common risk factor exposures), we use the idiosyncratic returns of one firm in a linked pair as a signal of a director’s actions or influence and study the subsequent price responses at the other firm in the pair.

If this signal is value-relevant, then the price at the other firm should respond in the same direction. This assumes the director influences firms the same way on average, making this measure an understatement of the true influence if directors provide more nuanced influence on firms. This common response can operate through three channels: (1) linked firms announce similar outcomes; (2) the market learns of news at the first firm and the price of the second firm reacts in anticipation of a similar outcome; (3) the market learns about the director’s general quality and the price of the second firm responds. (Figure 3 of the paper illustrates two examples of the first channel.) In all three cases we use the market’s processing of information unavailable to us as econometricians and policy makers to glean the importance of the shared director.

We translate this signal into the value of a director by forming a trading strategy. Each month we sort firms into 5 portfolios by the ranking of the previous month’s idiosyncratic returns for the other firms which their directors oversee. We buy firms with directors whose other firms did well the previous month and short firms with directors whose other firms did poorly. Value-weighting this strategy gives a long-short alpha of 9% per year.

By averaging across firms in this strategy, unrelated noise averages out to zero, leaving only the value impact of the common event—the shared director. Thus, the alpha of this long-short strategy is equivalent to an event study’s average abnormal return of a set of firms having the most extreme director events in a given month. Because we know the probability of being included in the long or short portfolio (approximately 10% each), we can generalize beyond a typical event study and convert this conditional value measure to the unconditional value measure of 1% of market cap variation per year.

The strategy’s positive alpha implies that investors do not fully monitor directors in real time. If investors monitor the real-time performance of all the firms a director advises, any cross-firm information should be simultaneously reflected in the prices of all the connected firms, eliminating any trading profit. In particular, the limited attention appears concentrated in the director’s influence at small firms. This limited attention to directors has implications for how directors build reputation and improve their labor market outcomes.

For our measure to have the interpretation of value influence, we establish exogeneity of the link in several ways. We show the results are robust to other known economic links between firms such as common industry, customer-supplier relationships, strategic alliances and geographic location of firms. We show the common comovement is limited to the linkage period and that the delay in the comovement is stronger when the director linkage is harder to discern (e.g., is not in machine-readable datasets). One would not expect predictability to vary with this publicity under other alternative hypotheses for the comovement.

The complete article is available for download here.

Both comments and trackbacks are currently closed.