Are Busy Boards Effective Monitors?

Cummings and Greg (2000).

Lublin (2001).

We check for whether other news events might explain the observed abnormal returns. However, a Lexis-Nexis search around the announcement date fails to uncover release of other significant corporate news.

See the Report of the National Association of Corporate Directors Blue Ribbon Commission on Director Professionalism (1996), and the Core Policies, Positions and Explanatory Notes from the Council of Institutional Investors (1998).

Ferris, Jagannathan, and Pritchard (2003) also use the average return on assets (ROA) over 1993–1995 as a measure of performance. As with their market-to-book ratio regressions, the ROA specifications do not control for firm-specific effects.

We calculate ROA as operating income before depreciation (Compustat item 13) plus the decrease in receivables (Compustat item 2), the decrease in inventory (Compustat item 3), the increase in current liabilities (Compustat item 72), and the decrease in other current assets (Compustat item 68). We scale this measure by the average of beginning- and ending-year book value of total assets (Compustat item 6).

Total capital adds the market value of the firm’s equity, book value, long-term debt, and an estimated market value of preferred stock. We calculate the market value of preferred stock by dividing preferred dividends over the prevailing yield on Moody’s index of high-grade industrial preferred stocks.

This potential endogeneity, however, works against uncovering the negative relation that we document between firm performance and busy outside directors.

The Poisson model specifies that if λ is defined by log (λ) = Xβ, where X is a vector of independent variables and β is a parameter vector, then the probability of n outside directors obtaining a directorship in a given year is given by: λⁿ e^−λ/λ! The log-likelihood function of this specification is maximized over β to produce maximum likelihood estimates and is given as,where C₁ is a constant that does not change the maximization process, N is the number of firms, T is the number of time periods per firm, and n_it is the number of outside directors obtaining a directorship in firm i in year t.

The statistical significance of a covariate is given by the log-likelihood ratio statistic. Under the null hypothesis that \( \beta = {{\beta }_0} \),\( - 2\log \left[ {\frac{{L({{\beta }_0})}}{{L({{\beta }^{{MLS}}})}}} \right] \) is distributed \( {{\chi }^2} \) with K degrees of freedom, where K is the number of elements of \( \beta \), and \( \frac{{L(\beta )}}{{L({{\beta }^{{MLS}}})}} \) is the likelihood ratio statistic. For the hazard estimation, presented as model (1) of Table VII, we estimate p-values for the test that an individual covariate is zero. Below each covariate, we report its risk ratio in parenthesis, a transformation of the estimated coefficient that is easier to interpret. The risk ratio is defined as the ratio of the hazard function under one set of covariates (\( X^{\prime} \)) to the hazard under a base case set of covariates (\( {{X}^0} \)): \( R(X^{\prime},{{X}^0}) \equiv \frac{{{{h}_i}(t|X^{\prime})}}{{{{h}_i}(t|{{X}^0})}} \). For example, the percentage change in the hazard given a one-unit change in the k’th quantitative covariate is simply estimated as \( \frac{{{{h}_i}(t|{{X}_k} = {X} + 1)}}{{{{h}_i}(t|{{X}_k} = \hat{X})}} - 1 = \exp [{{\beta }_k}(\hat{X} + 1) - {{\beta }_k}(\hat{X})] - 1 = \exp [{{\beta }_k}] - 1 \).

The option value is adjusted for dividend payouts (Merton 1973).

We determine that boards have switched status from busy to nonbusy by focusing on the remaining outside directors on the board. It is possible, of course, that a firm may replace a departing busy director with a busy appointee, leaving the board’s status unchanged. To account for this, we also track changes in board composition for a 6-month period following the director’s departure and consider these changes in determining the board’s busy status. Using this process, we identify 19 firms that switched to nonbusy status and obtain virtually identical results to those reported herein.