Board Structure and Capital Structure

Daniel Ferreira, London School of Economics

Miguel A. Ferreira, NOVA School of Business and Economics
Beatriz Mariano, University Carlos III

This Version: March 2012

Abstract
Board structure has been linked to firm environment and characteristics, but the evidence
is based on correlations rather than causal estimates. In particular, there is evidence of a
positive correlation between board independence and financial leverage. Using a
regression discontinuity design, we show that the number and fraction of independent
directors increases following a debt covenant violation, as creditors can use the threat of
accelerating loan payments to demand a more independent board of directors. The effect
is economically important as a covenant violation implies up to one additional
independent director in the board. Our findings establish a causal effect from firm
characteristics leverage to board structure through a transfer of control from
shareholders to creditors.

JEL Classification: G32; G34

Keywords: Corporate boards, Board independence, Capital structure, Covenant violations

E-mail: d.ferreira@lse.ac.uk
E-mail: miguel.ferreira@novasbe.pt

E-mail: bmariano@emp.uc3m.es

1. Introduction
While previous research has shown that board structure and firm characteristics are related, it has
been much less clear on how they are related and what are the channels and mechanisms behind
these correlations. Thus, we know that firm characteristics and board structure are correlated but
a casual effect has not been established yet. The goal of this paper is to study this issue by
identifying a specific channel through which firm characteristics affect board structure.
The determinants of board structure can be classified into three broadly defined hypotheses:
the scope of operations hypothesis, the monitoring hypothesis, and the negotiation hypothesis.
The scope of operations hypothesis suggests that the size and complexity of a firms operations
affect its board structure (Fama and Jensen (1983)). As a firm grows and diversifies, it faces an
increasing demand for specialized board members who can perform tasks such as auditing and
setting managerial compensation. The monitoring hypothesis emphasizes the importance of a
firms business environment for the optimal design of its board structure (e.g., Demsetz and
Lehn (1985), Raheja (2005), and Adams and Ferreira (2007)). The negotiation hypothesis
emphasizes the role of the negotiation between the CEO and outside directors as an important
determinant of board composition (Hermalin and Weisbach (1998)).
Empirical evidence by Boone, Field, Karpoff and Raheja (2007), Coles, Daniel, and Naveen
(2008), and Linck, Netter, and Yang (2008) finds elements of these theories to explain board size
and independence. In particular, they find that board structure and capital structure are related.
The findings in these papers indicate that there is a positive and significant correlation between
financial leverage and board independence and board size. They interpret this finding as
indicating that firms with high leverage, which depend on external resources to a greater extent,
could have greater advising requirements (Pfeffer (1972) and Klein (1998)). Thus, leverage (and
2

also diversification and firm size) are interpreted as proxies for complexity and the CEOs need
for advice. These papers, however, do not show that the leverage-board structure link is robust to
omitted variable concerns and endogeneity concerns. Moreover, they do not identify the channel
through which debt impacts board structure.
To achieve this goal, we examine the impact of debt covenant violations on board structure.
Covenant violations are a natural experimental ground to investigate the influence of creditors on
corporate boards. There are several reasons why financial covenant violations provide an
excellent opportunity to study the influence of creditors on board structure. First, when a firm
violates a covenant, control rights shift to creditors, who can use the threat of accelerating the
loan payments to exert considerable influence over shareholders. Covenant violations provide
creditors with the same rights as a payment default, including the ability to accelerate any
outstanding principal and to terminate any unused revolving credit facility capacity. Although
creditors almost always waive the violation, the threat associated with these rights enables
creditors to exert considerable influence over managerial decision-making and shareholders.
Second, covenant violations are common, and their discrete nature enables us to employ a
regression discontinuity design to address the concern that board structure and firm
characteristics are jointly determined. The discrete nature of the covenant violation generates a
potentially exogenous source of variation in the distance to the covenant threshold that can be
used to estimate the effect of firm characteristics on board structure. Therefore, the goal of our
empirical strategy is to show that shareholders would not have altered their board structure in the
same manner had the covenant violation not occurred.
In this paper, we investigate the link between leverage and board structure and identify a
specific mechanism through which leverage affects board structure. We extend previous
3

empirical papers on board structure determinants by showing that the relation between board
structure and leverage is robust in several ways. We show that there is a positive and significant
relation between leverage and the number of independent directors, fraction of independent
directors, and total number of directors in the boardroom. This relation is robust to a long list of
possible covariates and after the inclusion of firm fixed effects, suggesting that time-invariant
unobserved firm characteristics cannot explain our empirical findings.
We then empirically identify a specific channel (debt covenants) and mechanism (transfer of
control rights from shareholders to debtholders) behind the link between leverage and board
structure. Using a regression discontinuity design, we find a positive and significant relation
between debt covenant violations and both the number and fraction of independent directors in
the boardroom. The magnitude of the effect is economically important. We find that debt
covenant violations imply that the number of independent directors in the boardroom increases
by up to one director. This effect is significantly stronger than those found in other studies of the
determinants of board structure. Indeed, our research design allows us to pin down the direct
effect of a change in firm characteristics on board structure.
Debt covenant violations are not similarly related to the total number of directors in the
boardroom, which indicates that creditors use their influence to pressure for a more independent
board but at the same time they do not increase board size. So there seems to occur a restructure
of the board of directors following a covenant violation with inside directors being replaced by
independent directors. The interpretation is that debtholders see an increase in the boards
monitoring role as an important mechanism to maximize the chance that the firm repays its debt
through future cash flows.

The remainder of the paper is organized as follows. Section 2 describes the sample and the
data. Section 3 presents our evidence on the relation between board structure and leverage using
panel regressions. In Section 4 we study the relation between debt covenant violations and board
structure using a regression discontinuity empirical framework. Section 5 concludes.

2. Data and Methodology

We start with nonfinancial firms in the Investor Responsibility Research Center (IRRC) database
between 1990 and 2007. We obtain board data from Compact Disclosure for the 1990-1995
period and from IRCC for the 1996-2007 period. The governance index of Gompers, Ishii, and
Metrick (2003) (GIM) is from IRRC and CEO compensation and tenure data are from
ExecuComp. We obtain accounting and segment data from Compustat and stock returns from
CRSP. This sample contains 2,554 firms and 19,603 firm-year observations. Table 1 presents
descriptive statistics of the variables used in our study. Table A.1 in the Appendix presents
detailed definition of the variables. The descriptive statistics in Table 1 are comparable to those
of other studies in the board of directors literature (e.g., Ferreira, Ferreira, and Raposo (2011)).
This sample is merged with covenant violations data obtained from Amir Sufis website and
from the Reuters Loan Pricing Corporations (LPC) DealScan database. The dataset which we
obtain from Amir Sufis website was prepared using information from the 10-Q and 10-K filings
on the SEC Edgar website.1 Using an algorithm described in Nini, Smith, and Sufi (2011), they
identify financial covenant violations for a large number of publicly traded firms. They construct
an indicator variable of whether or not the firm reports a violation of a financial covenant during

This dataset is available at http://faculty.chicagobooth.edu/amir.sufi/data.htm.

the corresponding quarter. When using this dataset in Section 3 (Tables (2)-(4)), the sample
period is restricted to the years between 1996 and 2007.2
The database of covenant violations reported in SEC filings is compiled at quarterly
frequency as opposed to board data which is annual. This is because changes in the composition
of the board of directors typically occur no more than once a year after their approval at the
annual shareholder meeting. This difference in data frequency requires having to convert the
quarterly covenant violation indicator variable from this dataset into an annual covenant
violation indicator variable. Specifically, we define a violating firm as one that violates a
covenant in at least one quarter at a given year. The sample that results from merging the
covenant violation data with our initial sample contains 2,264 firms and 12,173 firm-year
observations. From this sample, we calculate that 23% of the firms violate a covenant at least
once during the sample period (526 firms), and 7% of the firm-year observations correspond to a
covenant violation event (896 firm-year observations).
As a robustness test, we provide additional results in Section 4 (Tables (6)-(7)) using debt
covenant information extracted from DealScan for the 1996-2007 period. The DealScan database
contains information on syndicated loans made by banks to typically large firms around the
world. It provides information about loan amounts, maturity, type of loan, syndication, covenants,
pricing, among others. We eliminate all the firms for which we do not have any loan with
information on covenants or which are not bound by either a current ratio, a net worth, tangible
net worth or a debt-to-EBITDA ratio covenant at least once during our sample period.3 We

This corresponds to the first dataset compiled by Nini, Smith, and Sufi (2011). It also covers the first quarter of
2008 which we do not consider given that the frequency of our remaining data is annual.
3

Chava and Roberts (2008) focus on covenants involving the current ratio and the (tangible) net worth. Demiroglu
and James (2010) on those involving the current ratio and the debt-to-EBITDA ratio. These seem to be covenants

merge these sample of loans to our initial sample using company name and loan origination date.
This is the sample we use to identify whether a firm violates any of the covenants or how far
away it is from violation.
The covenant variable from DealScan gives a covenant threshold which we compare to the
corresponding accounting variable calculated for a given point in time during the lifetime of the
loan. If the accounting variable exceeds the covenant threshold there is no covenant violation; if
it does not exceed (or is equal to) the covenant threshold there is a covenant violation. We apply
this decision rule to three out of the four covenants that we consider. For the case of the debt-toEBITDA, the accounting ratio should not exceed the covenant threshold for a non-violating firm.
Specifically, we take each of the previous loans with covenant information and consider that the
firm is bound by this covenant (or covenants) at every quarter until the loan maturity date. We
follow Chava and Roberts (2008) and choose to use quarterly frequency in this part of the
analysis. This is because most borrowers report their financial statements to creditors every
quarter to be monitored on the compliance with the covenant requirements. We use Compustat
data at quarterly frequency to compute the relevant accounting variables. When we compare the
covenant threshold to the value of the corresponding accounting we are effectively calculating a
measure of distance to default at a given quarter. Since a firm might have more than one active
loan at a given quarter, we calculate distance to default using the minimum covenant threshold
(maximum for the debt-to-EBITDA ratio) at a given quarter. Since these values are measured at
a quarterly basis we characterize the behavior of the firm at a given year by calculating the
minimum value across the four covenants within one year. This sample contains 613 firms and

which appear frequently in loan contracts in general (e.g., Chava and Roberts (2008) and Dichev and Skinner
(2002)).

1,879 firm-year observations. From this sample, we calculate that 17% of the firms violate a
covenant at least once during the sample period (103 firms), and 14% of the firm-year
observations correspond to a covenant violation event (260 firm-year observations).

3. Board Structure, Leverage and Covenant Violations

In this section, we study the relation between leverage, debt covenant violations and board
structure using OLS and firm fixed effects panel regressions. We consider three measures of
board structure: number of independent directors, board independence and board size. The
explanatory variables of interest are financial leverage and debt covenant violations.
We first examine the relation between debt covenant violations and board structure using
graphical analysis. We first show the means of our three board structure variables for violators
from four years before the violation through four years after the violation.
Panel A of Figure 1 shows the evolution of the number of independent directors around a
covenant violation. The number of independent directors is nearly constant during the four years
before the violation. However, when the covenant violation occurs at year 0 the number of
independent directors increases and keeps increasing in the years following the violation
(between 0 and +2). Panel B of Figure 1 shows a similar pattern for the number of directors as a
fraction of board size. Panel C of Figure 1 shows how the trend in board size changes before and
after a covenant violation.
3.1. Number of Independent Directors
Table 2 presents the outcome of ordinary least squares (OLS) and firm fixed effects panel
regressions, where the dependent variable is the number of independent directors. Our
8

explanatory variable of interest is book leverage, which is defined as the ratio of total debt (i.e.,
long term debt and debt in current liabilities) to total assets. Later we will consider alternative
definitions of leverage. All explanatory variables, including leverage, are measured with a twoyear lag relative to board structure variables since the effect of firm characteristics on board
structure is likely to take time to happen. OLS regressions include industry (two-digit SIC) and
year fixed effects. All reported t-statistics are adjusted for heteroskedasticity and within-firm
correlation using clustered standard errors.
Column (1) presents the coefficients of an OLS regression of the number of independent
directors on leverage and other variables that have been shown to be correlated with board
structure. There is strong evidence of a positive and statistically significant relation between the
number of independent directors and leverage. The leverage coefficient is 1.15 with a t-statistic
of 4.12. The effect of leverage on the number of independent directors is economically
significant if compared with the effects of other important board structure determinants. Using
the specification in column (1), a one standard deviation increase in leverage increases the
number of independent directors by roughly 0.26 directors.
With respect to the other explanatory variables, we find that firm size, firm age and the
number of business segments are all positively and significantly related to board independence.
These findings are consistent with the scope of the operations hypothesis that more complex
firms require more independent boards. Consistent with the monitoring hypothesis, we find a
statistically negative relation between the number of independent directors and the market-tobook ratio and stock return volatility.
Endogeneity problems are a crucial concern in empirical research on corporate governance.
In our setting, there could be many reasons for board structure and leverage to be jointly
9

determined. We first address the potential endogeneity problems using firm fixed effects
methods that control for unobserved sources of firm heterogeneity. Fixed effects methods solve
joint determination problems in which an unobserved time-invariant variable simultaneously
determines both leverage and board structure. It is also equivalent to looking only at within-firm
changes in leverage.
Columns (2) and (3) of Table 2 present firm fixed effects estimates, as well as year fixed
effects and t-statistics adjusted for firm-level clustering. There is still evidence of a positive
relation between the number of independent directors and leverage. In Column (2), the estimate
of the leverage coefficient is 0.68 with a significant t-statistic of 2.75. Column (3) presents
estimates for a specification that includes additional explanatory variables which are the
governance index (GIM) and CEO ownership and tenure. The estimate and t-statistics of the
leverage coefficient remain very similar to those in column (2). Moreover, the governance index
(GIM) coefficient is positive and statistically significant, which is consistent with the idea that
the number of independent directors is higher in firms that are insulated from the market for
corporate control (Gillan, Hartzell, and Starks (2006)). The coefficient of CEO ownership is
negative and statistically significant, which is consistent with the suggestion of Hermalin and
Weisbach (1998) that board structure is influenced by the negotiations between CEOs and
outside directors. The evidence indicates that the presence of independent directors is negatively
related to the degree of CEO influence.
The fixed effects results go a long way toward dismissing omitted variables explanations as
sources of endogeneity. Because only the effects of within-firm changes in board structure are
taken into account, firm-specific omitted variables cannot explain the observed relation between
leverage and the number of independent directors. Firm fixed effects estimates, however, do not
10

address reverse causality concerns, i.e., that changes in board structure could affect leverage. In
the analysis that follows and in the next section, we will use financial covenants violations and a
regression discontinuity design to establish a casual effect from leverage to board structure. In
particular, we will establish a link between leverage and the presence of independent directors in
the boardroom.
We start by examining the effect of debt covenant violations on the number of independent
directors in the boardroom. Column (4) of Table 2 presents estimation results where we add the
covenant violation dummy as reported in the SEC filings as an explanatory variable to the
specification in column (3). The results show that the number of independent directors increases
following the covenant violation. The covenant violation dummy coefficient is 0.2, with a tstatistic of 2.53. The coefficient on leverage is still positive and significant in column (4).
The model reported in column (5) adds the covenant control variables: debt to assets ratio,
return on assets, interest expenses to assets ratio, net worth to assets ratio, and cash to assets ratio.
The choice of these controls is based on the most common financial covenants employed in
private credit agreements. The covenant violation dummy coefficient is 0.18, with a t-statistic of
2.36, which is statistically significant at the 5% level. Furthermore, the leverage coefficient is no
longer statistically significant. This is a first indication that covenant violations is the channel by
which leverage impacts board structure.
To disentangle the effect of the covenant violation from changes in board structure that
would have otherwise occurred, we estimate the impact of the covenant violation by focusing
only on discontinuous changes in board structure occurring at the covenant threshold. We apply
this discontinuity approach by including as right-hand side variables a covenant violation
indicator variable along with linear, nonlinear, and step functions of the underlying variables on
11

which covenants are written. With the inclusion of these functions, the point estimate on the
covenant violation indicator variable is identified under the assumption that shareholder
preferences over board structure are not discontinuous exactly at the covenant threshold. This
assumption is valid as long as shareholders, in the absence of financial covenants, would not
have chosen the exact same ratios and levels of the ratios as creditors to determine financial
policy. This quasi-discontinuity approach is similar to the one in Roberts and Sufi (2009).
In this way, columns (6)-(7) present additional specifications of the number of independent
directors regression. The specification reported in column (6) includes the covenant control
variables and high-order polynomial terms (squared and cubic) for each of the covenant control
variables. Finally, column (7) includes the covenant control variables, high-order polynomial
terms (squared and cubic) for each of the covenant control variables, and quintile indicator
variables for each of the covenant control variables. Even with this extensive set of covenant
control variables, the covenant violation dummy coefficient is largely unaffected, remaining at
0.2 with a t-statistic of 2.51. The leverage coefficients is also insignificant in these specifications.
3.2. Board Independence
The number of independent directors could be higher because the board includes more directors.
We next examine the relation between leverage and board independence, i.e., the number of
independent directors as a fraction of board size. Table 3 presents the outcome of OLS and firm
fixed effects panel regressions, where the dependent variable y is a logistic transformation (or the
log odds ratio) of the fraction of independent directors z (i.e., y = ln(z/(1-z)). We use a logistic
transformation because the fraction of independent directors is bounded between zero and one.
The models in Table 3 are similar to those in Table 2.
12

We obtain similar evidence using board independence as a dependent variable to that using
the number of independent directors. Columns (1) present OLS estimates and columns (2) and (3)
firm fixed effects estimates. There is evidence of a positive and statistically significant relation
between board independence and leverage. In column (1), the estimate of the leverage coefficient
is 0.34 with a significant t-statistic of 3.09. The effect of leverage on board independence is
economically significant if compared with the effects of other important board structure
determinants.
With respect to the other explanatory variables, the OLS estimates are in general consistent
with the scope of the operations and the monitoring hypotheses. However, the evidence in favor
of these hypotheses is significantly weaker when using firm fixed effects.
Columns (4)-(7) presents the outcome of firm fixed effects panel regression models, where
the dependent variable is the number of independent directors as a fraction of board size. In these
models we include both the covenant violation dummy and leverage as explanatory variables.
These models are similar to those in Table 2. In all models, we find a positive and significant
relation between board independence and the covenant violation dummy. For example, even with
an extensive set of covenant control variables as in column (7), the covenant violation dummy
coefficient is largely unaffected, remaining at 0.08 with a t-statistic of 2.33. Furthermore, we
find that the leverage coefficient is in most cases no longer statistically significant when we add
the covenant violation dummy as a explanatory variable.
3.3. Board Size

13

Table 4 presents the outcome of OLS and firm fixed effects panel regressions, where the
dependent variable is board size. The models in Table 4 are similar to those in Tables 2 and 3.
Column (1) presents OLS estimates and columns (2)-(7) firm fixed effects estimates.
Columns (1) and (3) show evidence of a positive and statistically significant relation between
board size and leverage. In column (1), the leverage coefficient is 1.02, with a t-statistic of 3.74.
The firm fixed effects estimates are significantly different. In column (3), there is still a positive
relation between board size and leverage but statistically significant only at the 10% level.
With respect to the other explanatory variables, we find that firm size and firm age are
positively and significantly related to board size. These findings are consistent with the scope of
the operations hypothesis that more complex firms require more directors. Consistent with the
monitoring hypothesis, we find a statistically negative relation between board size and marketto-book ratio and stock return volatility. The governance index coefficient is positive and
statistically significant, which is consistent with the idea that the number of directors is higher in
firms that are insulated from the market for corporate control. The coefficient of CEO ownership
is negative and statistically significant, which is consistent with the hypothesis that board
structure is influenced by the negotiations between CEOs and outside directors.
Columns (4)-(7) presents the outcome of firm fixed effects panel regression models, where
the dependent variable is the number of directors. In these models we include both the covenant
violation dummy and leverage as explanatory variables. These models are similar to those in
Table 2. In all models, we find an insignificant relation between board size and the covenant
violation dummy.
Overall, the evidence in Tables 2-4 show a positive and significant relation between the
presence of independent directors in the boardroom and leverage. Highly levered firms seem to
14

have both a higher number of independent directors and a higher fraction of independent
directors. This finding is not explained by unobserved firm heterogeneity. In contrast, leverage is
not reliably associated with larger boards. These findings are consistent with previous studies
(e.g., Coles, Daniel and Naveen (2008) and Linck, Netter, and Yang (2008)) that also find that
leverage is positively associated with leverage. They interpret these findings as suggesting that
firms with high leverage depend on external resources to a greater extent and could have greater
advising requirements. Therefore, board size and independence increase in firm complexity and
advising benefits. Our findings in this section extend previous studies by showing that the
relation between leverage and board independence is robust to omitted variable concerns by
using a firm fixed effects estimator. This empirical fact is important as other determinants of
board independence are not significantly related to board independence when we use a firm fixed
effects estimator.
Our findings in this section also support a casual effect of leverage on board structure and
provide evidence of the channel by which this effect occurs. We find that that the number and
fraction of independent directors increases following a debt covenant violation. In contrast, there
is no evidence that board size increases following a violation. Taken together these findings
support the idea that following a covenant violation there are important changes in a firms board
structure as there is a transfer of control rights from shareholders to debtholders. Creditors can
advise the firm on how best to manage through the violation to maximize the chance that the
firm repays its debt through future cash flows. We present evidence that one of those actions
taken by creditors is to increase the presence of independent directors in the board and therefore
improve board monitoring. These changes do not seem to take place by increasing the number of
directors, but by replacing inside directors by independent directors.
15

3.4. Alternative Leverage Measures

In this section we further investigate the relation between board structure and leverage. We have
defined leverage as the ratio of total debt to total assetsbook leverage. We now investigate the
relation between board structure variables and alternative measures of leverage.
If leverage affects board structure because of the transfer of control rights to debtholders due
to covenant violations or default then book leverage is an appropriate measure of leverage as
covenants are typically written on ratios calculated using book values. On the other hand, market
leverage captures the market perception of the financial health of the firm and incorporates all
available information about firms future prospects. It may be the case that the optimal board
structure changes by other reasons than violations and therefore market leverage could be a
better proxy. If the channel by which leverage affects board structure is not only the transfer of
control to creditors, then we should also expect market leverage to be positively related to board
structure.
Columns (1)-(3) of Table 5 present estimates of firm fixed effect regressions, where the
dependent variable is the number of independent directors, board independence and board size.
The main explanatory variables of interest are the ratio of total debt to the market value of assets,
which is defined as total debt plus market value of equity, and the covenant violation dummy.
There is no significant relation between the board structure variables and market leverage and the
covenant violation dummy results are fully consistent with the book leverage results. This is
consistent with the fact that covenant violations and technical default seem to be an important
channel by which debt affects board structure.
We next use long term debt over total assets as a measure of leverage, in alternative to
considering total debt (i.e., long term debt and debt in current liabilities). Debt covenants are
16

typically written on total debt to assets ratios rather than long term debt. Thus, if indeed covenant
violations is an important mechanism by which leverage affects board structure, total debt seems
to be a more appropriate measure and we should find stronger results using total debt than using
long term debt.
Columns (4)-(6) of Table 5 present estimates of firm fixed effect regressions, where the
dependent variable is the number of independent directors, board independence and board size.
The main explanatory variables of interest are long-term debt to total assets and the covenant
violation dummy. We find similar results to the market and book leverage cases.

4. Discontinuity Regressions
In this section, we use DealScan information to isolate the analysis to a sample of loans for
which we know the covenant thresholds. Such an analysis alleviates two potential concerns
associated with the results in the previous section: (1) the exact covenant threshold for which the
violation occurs is unknown; and (2) we observe only reported covenant violations. The analysis
and sample are similar to those found in Chava and Roberts (2008).
In Table (6), we estimate firm fixed effects panel regressions of the number of independent
directors, board independence and board size. All explanatory variables are lagged two periods.
All regressions include year fixed effects and t-statistics are adjusted for firm-level clustering.
The explanatory variable of interest is the covenant violation dummy, whose coefficient
represents the impact of a covenant violation of board structure. Because of the inclusion of a
firm-fixed effect, identification of the coefficient of interest comes only from those firms that
experience a covenant violation. Therefore, we restrict our attention to the subsample of firms

17

that experience at least one covenant violation; however, the estimated treatment effect using the
entire sample of firms is qualitatively similar.
Moreover, because of the measurement problems associated with determining distance to
default and because loan agreements contain adjustments to the covenant level over the life of
the loan, we might misclassify or miss some violations using the covenant threshold at the
origination date. This is not a concern if we examine the covenant violations that are reported in
the borrowers financial statements for which we can identify a loan agreement and respective
covenant thresholds in DealScan. As Beneish and Press (1993) explain, violations reported in the
borrowers financial statements are likely to be more serious violations that have not been cured
or waived at the time the financials were prepared. For these reasons, we start by using reported
violations in SEC filings as our main explanatory variable as in Demiroglu and James (2010).
We then check the robustness of our findings using violations identified based only on the
covenant slack relative to the covenant threshold obtained from DealScan, without using the
information reported to the SEC.
Panel A of Table 6 presents the estimation results for the entire sample consisting of loans
containing either a current ratio, debt-to-EBITDA, or net worth-to-assets ratio covenant.
Columns (1), (4) and (7) of Table 6 present estimation results with only the covenant violation
dummy as an explanatory variable. Columns (1) and (4) show that covenant violations are
associated with an increase in the number of independent director and board independence. The
coefficient of the covenant violation dummy is 0.46 in the number of independent directors
regression, with a t-statistic of 1.90. The coefficient of the covenant violation dummy is 0.24 in
the board independence regression, with a t-statistic of 2.20. Column (7) shows that covenant

18

violations do not seem to have a significant effect on board size. These findings are consistent
with those from the quasi-discontinuity regression approach in the previous section.
Columns (2)-(3), (5)-(6), and (8)-(9) include the set of control variables used in Tables 2-4
and smooth functions of order five of the distance from the default boundary defined in terms of
current ratio, debt-to-EBITDA or net worth-to-assets ratio. The goal of including these functions
is to isolate the treatment effect to the point of discontinuity and to address the concern that the
distance to the covenant threshold contains information about board structure not captured by the
other determinants. With this extensive set of controls, the covenant violation dummy coefficient
is even stronger in the number of directors and board independence regressions. The coefficient
of the covenant violation dummy is 0.58 (t-statistic is 2.42) in the number of independent
directors regression in column (3), and 0.27 (t-statistic is 2.48) in the board independence
regression in column (6).
Panel B of Table 6 presents the estimation results using the discontinuity sample. This
sample contains only those firm-years observations that fall within a narrow range ( 0.3) around
the covenant threshold. The results in this table confirm the findings obtained with the entire
sample that both the number of independent directors and board independence increases
significantly following a covenant violation. The magnitude of the effects is stronger using the
discontinuity sample. We estimate that the number of independent directors in the boardroom
increases between 1 and 1.3 directors following a covenant violation. The results in this panel of
Table 6 also show evidence that board size significantly increases in response to covenant
violations. The effects of covenant violations on the number of independent directors and board
independence are consistent with our previous estimates, while the effect on board size is not.

19

In panel A and B of Tables 7, we follow a similar exercise focusing on the violations that we
can identify using DealScan information alone. The results remain consistent with our main
findings.
In sum, board structure changes significantly in response to covenant violations. The results
indicate that the presence of independent directors increases significantly following covenant
violations. These findings suggest that covenant violations impact board structure via the transfer
of control rights from shareholders to debtholders. Moreover, these findings establish a channel
that explain the link between board structure and leverage.

5. Conclusion
We investigate the link between leverage and board structure and identify a specific mechanism
through which leverage affect board structure. We empirically identify a specific channel (debt
covenants) and mechanism (transfer of control rights from shareholders to debtholders) behind
the link between leverage and board structure. Using a regression discontinuity design, we find a
positive and significant relation between debt covenant violations and both the number and
fraction of independent directors in the boardroom. The magnitude of the effect is economically
important. We find that debt covenant violations imply that the number of independent directors
in the boardroom increases by more than one director.
Debt covenant violations are not similarly related to the total number of directors in the
boardroom, which indicates that creditors use their influence to pressure for a more independent
board but at the same time they do not increase board size. The interpretation is that debtholders
see an increase in the boards monitoring role as an important mechanism to maximize the
chance that the firm repays its debt through future cash flows.
20

21

References
Adams, R. and D. Ferreira, 2007. A theory of friendly boards. Journal of Finance, 62, 217-250.
Boone, A., L. Field, J. Karpoff, C. Raheja, 2007. The determinants of corporate board size and
composition: an empirical analysis. Journal of Financial Economics 85, 66101.
Chava, S., and M. Roberts, 2008. How does financing impact investment? The role of debt
covenants. Journal of Finance 63, 2085-2121.
Coles, J. L., Daniel, N. D., Naveen, L., 2008. Boards: Does one size fit all?. Journal of Financial
Economics 87, 329356.
Demiroglu, C., and C. James, 2010, The information content of bank loan covenants, Review of
Financial Studies 23, 3700-3737.
Demsetz, H. and K. Lehn, 1985. The structure of corporate ownership: Causes and consequences.
Journal of Political Economy 93, 1155-77.
Dichev, I., and D. Skinner, 2002. Large-sample evidence on the debt covenant hypothesis.
Journal of Accounting Research 40, 1091-1123.
Fama, E. and M. Jensen, 1983. Separation of ownership and control. Journal of Law and
Economics 26, 301-325.
Ferreira, D., M. Ferreira, and C. Raposo, 2011. Board structure and price informativeness.
Journal of Financial Economics 99, 523-545.

22

Gillan, S.L., Hartzell, J.C., Starks, L.T., 2006. Tradeoffs in corporate governance: evidence from
board structures and charter provisions. Unpublished working paper. Texas Tech University.
Gompers, P., J. Ishii, and A. Metrick, 2003. Corporate governance and equity prices. Quarterly
Journal of Economics 118, 107-155.
Hermalin, B. and M. Weisbach, 1998. Endogenously chosen boards of directors and their
monitoring of the CEO. American Economic Review 88, 96-118.
Linck, J., Netter J., and T. Yang, 2008. The determinants of board structure. Journal of Financial
Economics 87, 308-328.
Klein, A., 1998. Firm performance and board committee structure. Journal of Law and
Economics 41, 275-303.
Nini, G., D. Smith, and A. Sufi, 2011. Creditor control rights, corporate governance, and firm
value. Working paper, University of Chicago.
Pfeffer, J., 1972. Size and composition of corporate board of directors. Administrative Science
Quarterly 21 , 218-228.
Raheja, C., 2005. Determinants of board size and composition: a theory of corporate boards.
Journal of Financial and Quantitative Analysis 40, 283306.
Roberts M, Sufi A. 2009. Control rights and capital structure: an empirical investigation. Journal
of Finance 64, 16571695

23

Table 1
Summary Statistics
This table reports the mean, standard deviation, minimum, 10th percentile, median, 90th percentile, maximum, and
number of observations for each variables. The sample consists of observations on Investor Responsibility Research
Center (IRRC) firms from 1990 to 2007. Financial industries are omitted (SIC codes 6000-6999). Board data are
taken from IRRC and Compact Disclosure. Governance data are taken from IRRC. Executive compensation data are
from ExecuComp. Accounting and segment data are from Compustat. Stock return data are from the Center for
Research in Security Prices. Refer to Table A1 in the Appendix for variable definitions. Financial ratios are
winsorized at the bottom and top 1% level.
Mean
Number of independent directors
Fraction of independent directors
Number of directors
Firm size
Leverage
Firm age
Number of segments
Market-to-book
R&D
Stock return volatility
Free cash flow
Return on assets
Governance index (GIM)
CEO ownership
CEO tenure
Dividend dummy
Positive net income dummy
Interest expenses
Net worth-to-assets
Cash
Current ratio
Tangible net worth-to-assets
Debt-to-EBITDA

6.083
0.666
9.031
4684
0.223
22.3
2.347
2.050
0.035
0.399
0.072
0.139
9.2
0.025
7.4
0.540
0.817
0.018
0.492
0.136
2.437
0.347
1.808

Standard Minimum
10th
Deviation
percentile
2.515
1.000
3.000
0.171
0.077
0.429
2.636
2.000
6.000
16729
0
136
0.175
0.000
0.000
18.7
1.0
4.0
1.750
1.000
1.000
1.635
0.532
1.020
0.065
0.000
0.000
0.237
0.107
0.181
0.115
-1.903
-0.035
0.111
-1.877
0.045
2.7
1.0
6.0
0.058
0.000
0.000
7.4
0.0
1.0
0.498
0.000
0.000
0.387
0.000
0.000
0.015
0.000
0.001
0.212
-1.053
0.246
0.172
0.000
0.006
2.165
0.052
0.915
0.275
-1.089
-0.007
4.259
-3.105
0.000

24

Median
6.000
0.700
9.000
947
0.215
17.0
1.000
1.577
0.000
0.340
0.081
0.141
9.0
0.003
5.0
1.000
1.000
0.015
0.481
0.061
1.907
0.355
1.326

90th
Maximum
percentile
9.000
22.000
0.875
0.944
12.000
26.000
9111
467093
0.444
0.892
51.0
82.0
5.000
10.000
3.504
63.426
0.112
0.659
0.680
2.235
0.184
0.363
0.253
0.434
13.0
19.0
0.079
0.340
17.0
56.0
1.000
1.000
1.000
1.000
0.036
0.162
0.783
0.964
0.384
0.963
4.257
23.863
0.703
0.950
4.502
31.781

Obs.
19603
19603
19603
19603
19603
19603
19603
19603
19603
19603
19603
19603
15706
14147
13764
19603
19603
18620
19602
19598
19326
19324
19603

Table 2
Regression of Number of Independent Directors, Leverage and Covenant Violations
Estimates of OLS and firm fixed effects panel regressions of the number of independent board directors are shown.
Covenant controls are debt to assets, return on assets, interest expenses to assets, net worth to assets, and cash to
assets. Specifications include covenant controls raised to the second and third power and covenant controls quintile
indicators. All regressions include year fixed effects. Refer to Table A1 in the Appendix for variable definitions. The
sample consists of observations on Investor Responsibility Research Center firms from 1990 to 2007. Financial
industries are omitted (SIC codes 6000-6999). Robust t-statistics adjusted for firm-level clustering are in parentheses.
*, **, *** indicate significance at the 10%, 5%, and 1% levels.
(1)

(2)

(3)

0.67***
(2.81)
0.34***
(4.41)
0.22**
(1.98)
0.08
(1.24)
-0.49***
(-3.50)
1.45
(1.48)
-0.10
(-0.92)
1.04*
(1.66)
-0.24
(-0.33)
0.08***
(2.75)
-3.72***
(-3.63)
-0.00
(-0.56)

(4)
0.20**
(2.53)
0.66***
(2.74)
0.35***
(4.55)
0.21*
(1.96)
0.08
(1.24)
-0.50***
(-3.57)
1.48
(1.52)
-0.13
(-1.14)
1.01
(1.61)
-0.15
(-0.21)
0.08***
(2.75)
-3.66***
(-3.58)
-0.00
(-0.56)

(5)
0.18**
(2.36)
0.32
(0.66)
0.38***
(4.66)
0.21*
(1.92)
0.08
(1.23)
-0.54***
(-3.66)
1.55
(1.58)
-0.14
(-1.28)
0.82
(1.31)
0.01
(0.01)
0.08***
(2.74)
-3.64***
(-3.54)
-0.00
(-0.52)
5.30
(1.61)
-0.14
(-0.29)
0.28
(1.04)

(6)
0.19**
(2.47)
0.94
(0.83)
0.38***
(4.57)
0.21*
(1.93)
0.08
(1.19)
-0.55***
(-3.62)
1.28
(1.29)
-0.15
(-1.33)
0.89
(1.40)
-0.04
(-0.04)
0.08***
(2.75)
-3.61***
(-3.52)
-0.00
(-0.54)
5.65
(0.60)
0.17
(0.23)
-0.23
(-0.25)

(7)
0.20**
(2.51)
1.69
(1.17)
0.36***
(4.39)
0.19*
(1.80)
0.08
(1.19)
-0.53***
(-3.58)
1.13
(1.13)
-0.15
(-1.38)
0.93
(1.46)
-0.64
(-0.60)
0.08***
(2.68)
-3.50***
(-3.38)
-0.00
(-0.57)
17.30
(1.51)
0.46
(0.57)
0.41
(0.31)

1.15***
(4.12)
0.61***
(16.75)
0.48***
(8.61)
0.26***
(4.12)
-1.05***
(-9.05)
2.23**
(2.40)
-0.87***
(-4.86)
0.37
(0.44)
0.83
(0.89)

0.68***
(2.75)
0.37***
(4.71)
0.26**
(2.26)
0.09
(1.30)
-0.53***
(-3.74)
1.49
(1.52)
-0.12
(-1.04)
1.11*
(1.76)
-0.33
(-0.46)

No
No
No
No
No
9,312
0.398

Yes
No
No
No
No
9312
0.184

Yes
No
No
No
No
9312
0.191

Yes
No
No
No
No
9312
0.191

Yes
Yes
No
No
No
9312
0.192

Yes
Yes
Yes
Yes
No
9312
0.193

Yes
Yes
Yes
Yes
Yes
9312
0.195

Covenant violation dummy - SEC

Leverage
Firm size (log)
Firm age (log)
Number of segments (log)
Market-to-book (log)
R&D
Stock return volatility
Free cash flow
Return on assets
Governance index (GIM)
CEO ownership
CEO tenure
Interest expenses
Net worth
Cash

Fixed Effects
Covenant controls
Covenant controls squared
Covenant controls third power
Covenant controls quintile indicators
Observations
Adjusted R-squared

25

Table 3
Regression of Board Independence, Leverage and Covenant Violations
Estimates of OLS and firm fixed effects panel regressions of the logistic transformed fraction of independent board
directors are shown. Covenant controls are debt to assets, return on assets, interest expenses to assets, net worth to
assets, and cash to assets. Specifications include covenant controls raised to the second and third power and
covenant controls quintile indicators. All regressions include year fixed effects. Refer to Table A1 in the Appendix
for variable definitions. The sample consists of observations on Investor Responsibility Research Center firms from
1990 to 2007. Financial industries are omitted (SIC codes 6000-6999). Robust t-statistics adjusted for firm-level
clustering are in parentheses. *, **, *** indicate significance at the 10%, 5%, and 1% levels.
(1)

(2)

(3)

0.21**
(2.09)
0.01
(0.49)
-0.02
(-0.59)
0.01
(0.44)
-0.11**
(-2.23)
0.48
(1.19)
-0.01
(-0.28)
0.54**
(1.98)
-0.18
(-0.59)
0.01
(0.56)
-0.59
(-1.45)
-0.00*
(-1.86)

(4)
0.08**
(2.23)
0.20**
(2.03)
0.02
(0.65)
-0.03
(-0.62)
0.01
(0.43)
-0.12**
(-2.32)
0.50
(1.22)
-0.02
(-0.48)
0.53*
(1.94)
-0.14
(-0.48)
0.01
(0.55)
-0.57
(-1.40)
-0.00*
(-1.86)

(5)
0.08**
(2.15)
0.15
(0.83)
0.02
(0.76)
-0.03
(-0.64)
0.01
(0.44)
-0.12**
(-2.30)
0.52
(1.28)
-0.03
(-0.55)
0.49*
(1.76)
-0.11
(-0.35)
0.01
(0.55)
-0.56
(-1.39)
-0.00*
(-1.84)
1.30
(0.91)
0.01
(0.03)
0.06
(0.55)

(6)
0.08**
(2.23)
0.96**
(1.97)
0.02
(0.70)
-0.02
(-0.59)
0.01
(0.43)
-0.12**
(-2.19)
0.52
(1.24)
-0.03
(-0.56)
0.51*
(1.81)
-0.09
(-0.21)
0.01
(0.58)
-0.54
(-1.33)
-0.00*
(-1.85)
3.20
(0.76)
0.11
(0.42)
-0.06
(-0.16)

(7)
0.08**
(2.33)
0.88
(1.47)
0.02
(0.65)
-0.03
(-0.64)
0.01
(0.41)
-0.12**
(-2.18)
0.46
(1.11)
-0.03
(-0.58)
0.50*
(1.80)
-0.45
(-1.01)
0.01
(0.54)
-0.55
(-1.34)
-0.00*
(-1.85)
6.42
(1.24)
0.16
(0.55)
-0.21
(-0.40)

0.34***
(3.09)
0.07***
(4.58)
0.10***
(4.28)
0.07***
(2.78)
-0.24***
(-5.03)
1.56***
(4.35)
-0.09
(-1.15)
0.32
(0.93)
0.34
(0.90)

0.21**
(2.08)
0.02
(0.52)
-0.02
(-0.57)
0.01
(0.43)
-0.12**
(-2.28)
0.47
(1.17)
-0.01
(-0.28)
0.57**
(2.05)
-0.21
(-0.69)

No
No
No
No
No
9282
0.245

Yes
No
No
No
No
9282
0.281

Yes
No
No
No
No
9282
0.283

Yes
No
No
No
No
9282
0.284

Yes
Yes
No
No
No
9282
0.284

Yes
Yes
Yes
Yes
No
9282
0.285

Yes
Yes
Yes
Yes
Yes
9282
0.287

Covenant violation dummy - SEC

Leverage
Firm size (log)
Firm age (log)
Number of segments (log)
Market-to-book (log)
R&D
Stock return volatility
Free cash flow
Return on assets
Governance index (GIM)
CEO ownership
CEO tenure
Interest expenses
Net worth
Cash

Fixed Effects
Covenant controls
Covenant controls squared
Covenant controls third power
Covenant controls quintile indicators
Observations
Adjusted R-squared

26

Table 4
Regression of Board Size, Leverage and Covenant Violations
Estimates of OLS and firm fixed effects panel regressions of the number of board directors are shown. Covenant
controls are debt to assets, return on assets, interest expenses to assets, net worth to assets, and cash to assets.
Specifications include covenant controls raised to the second and third power and covenant controls quintile
indicators. All regressions include year fixed effects. Refer to Table A1 in the Appendix for variable definitions. The
sample consists of observations on Investor Responsibility Research Center firms from 1990 to 2007. Financial
industries are omitted (SIC codes 6000-6999). Robust t- tatistics adjusted for firm-level clustering are in parentheses.
*, **, *** indicate significance at the 10%, 5%, and 1% levels.
(1)

(2)

(3)

0.43*
(1.70)
0.43***
(5.11)
0.43***
(3.53)
0.09
(1.32)
-0.35**
(-2.22)
0.48
(0.57)
-0.19
(-1.62)
0.91
(1.39)
-0.90
(-1.16)
0.10***
(3.45)
-2.85***
(-3.38)
0.01
(1.19)

(4)
0.06
(0.81)
0.43*
(1.68)
0.43***
(5.13)
0.42***
(3.52)
0.09
(1.32)
-0.35**
(-2.24)
0.49
(0.58)
-0.20*
(-1.68)
0.90
(1.38)
-0.88
(-1.12)
0.10***
(3.45)
-2.83***
(-3.37)
0.01
(1.19)

(5)
0.05
(0.73)
0.36
(0.74)
0.45***
(5.17)
0.42***
(3.51)
0.10
(1.35)
-0.39**
(-2.40)
0.68
(0.79)
-0.20*
(-1.73)
0.71
(1.07)
-0.66
(-0.84)
0.10***
(3.47)
-2.82***
(-3.37)
0.01
(1.23)
1.97
(0.56)
-0.04
(-0.08)
0.50*
(1.75)

(6)
0.06
(0.75)
-0.62
(-0.55)
0.45***
(5.21)
0.42***
(3.51)
0.09
(1.33)
-0.41**
(-2.48)
0.45
(0.51)
-0.21*
(-1.79)
0.73
(1.10)
-0.85
(-0.88)
0.10***
(3.45)
-2.88***
(-3.46)
0.01
(1.24)
-2.38
(-0.24)
-0.01
(-0.01)
0.04
(0.04)

(7)
0.05
(0.70)
0.37
(0.26)
0.44***
(5.04)
0.40***
(3.41)
0.10
(1.36)
-0.40**
(-2.41)
0.35
(0.40)
-0.21*
(-1.83)
0.78
(1.18)
-0.54
(-0.47)
0.10***
(3.41)
-2.74***
(-3.32)
0.01
(1.24)
5.74
(0.50)
0.32
(0.40)
1.11
(0.77)

1.02***
(3.74)
0.70***
(18.01)
0.43***
(7.84)
0.20***
(2.93)
-0.88***
(-7.37)
-1.44
(-1.43)
-1.09***
(-5.51)
0.40
(0.47)
-0.56
(-0.59)

0.44*
(1.72)
0.46***
(5.42)
0.48***
(3.78)
0.10
(1.41)
-0.39**
(-2.46)
0.54
(0.63)
-0.21*
(-1.77)
0.93
(1.42)
-0.93
(-1.19)

No
No
No
No
No
9312
0.406

Yes
No
No
No
No
9312
0.045

Yes
No
No
No
No
9312
0.051

Yes
No
No
No
No
9312
0.051

Yes
Yes
No
No
No
9312
0.052

Yes
Yes
Yes
Yes
No
9312
0.053

Yes
Yes
Yes
Yes
Yes
9312
0.057

Covenant violation dummy - SEC

Leverage
Firm size (log)
Firm age (log)
Number of segments (log)
Market-to-book (log)
R&D
Stock return volatility
Free cash flow
Return on assets
Governance index (GIM)
CEO ownership
CEO tenure
Interest expenses
Net worth
Cash

Fixed Effects
Covenant controls
Covenant controls squared
Covenant controls third power
Covenant controls quintile indicators
Observations
Adjusted R-squared

27

Table 5
Regression of Board Structure and Alternative Leverage Measures
Estimates of firm fixed effects panel regressions of the number of independent board directors, logistic transformed
fraction of independent board directors, and number of board directors are shown. All regressions include year fixed
effects. Refer to Table A1 in the Appendix for variable definitions. The sample consists of observations on Investor
Responsibility Research Center firms from 1990 to 2007. Financial industries are omitted (SIC codes 6000-6999).
Robust t-statistics adjusted for firm-level clustering are in parentheses. *, **, *** indicate significance at the 10%,
5%, and 1% levels.
(1)

(2)
Market Leverage
Nr. of
Fraction of
Dependent variable
Independent Independent
Directors
Directors
Covenant violation dummy - SEC
0.19**
0.08**
(2.48)
(2.34)
Leverage
0.18
-0.20
(0.27)
(-0.80)
Firm size (log)
0.37***
0.02
(4.40)
(0.58)
Firm age (log)
0.19*
-0.03
(1.80)
(-0.60)
Number of segments (log)
0.08
0.01
(1.16)
(0.44)
Market-to-book (log)
-0.53***
-0.14**
(-3.48)
(-2.42)
R&D
1.09
0.44
(1.10)
(1.07)
Stock return volatility
-0.16
-0.02
(-1.40)
(-0.47)
Free cash flow
0.90
0.50*
(1.43)
(1.80)
Return on assets
-0.61
-0.47
(-0.58)
(-1.04)
Governance index (GIM)
0.08***
0.01
(-3.40)
(-1.38)
CEO tenure
-0.00
-0.00*
(-0.58)
(-1.84)
Interest expenses
20.97*
8.64*
(1.89)
(1.72)
Net worth
0.39
0.19
(0.47)
(0.62)
Cash
0.28
-0.31
(0.21)
(-0.58)
Fixed Effects
Yes
Covenant controls
Yes
Covenant controls squared
Yes
Covenant controls third power
Yes
Covenant controls quintile indicators Yes
Observations
9,312
Adjusted R-squared
0.195

Yes
Yes
Yes
Yes
Yes
9,282
0.286

(3)

Nr. of
Directors
0.05
(0.65)
0.72
(1.00)
0.45***
(5.07)
0.40***
(3.39)
0.09
(1.30)
-0.35**
(-2.13)
0.34
(0.38)
-0.23**
(-1.96)
0.74
(1.13)
-0.46
(-0.41)
0.10***
(-3.32)
0.01
(1.22)
5.89
(0.51)
0.15
(0.18)
1.16
(0.80)
Yes
Yes
Yes
Yes
Yes
9,312
0.057

28

(4)

(5)
(6)
Long-term Leverage
Nr. of
Fraction of
Independent Independent
Nr. of
Directors
Directors
Directors
0.20**
0.08**
0.04
(2.50)
(2.37)
(0.46)
0.10
0.03
-0.01
(1.09)
(0.82)
(-0.10)
0.33***
0.02
0.37***
(3.87)
(0.72)
(4.08)
0.20*
-0.01
0.36***
(1.81)
(-0.33)
(3.06)
0.10
0.01
0.11
(1.37)
(0.55)
(1.41)
-0.52***
-0.12**
-0.38**
(-3.23)
(-2.00)
(-2.13)
0.22
0.27
-0.50
(0.23)
(0.66)
(-0.54)
-0.09
-0.01
-0.15
(-0.79)
(-0.12)
(-1.20)
0.83
0.47
0.70
(1.25)
(1.61)
(0.99)
-0.80
-0.34
-1.04
(-0.70)
(-0.71)
(-0.84)
0.09***
0.01
0.10***
(-3.32)
(-1.34)
(-3.19)
-0.00
-0.00*
0.01
(-0.46)
(-1.78)
(1.37)
21.25*
8.29
5.85
(1.86)
(1.63)
(0.48)
0.65
0.16
0.50
(0.75)
(0.53)
(0.59)
0.68
-0.27
1.79
(0.48)
(-0.49)
(1.14)
Yes
Yes
Yes
Yes
Yes
8,592
0.192

Yes
Yes
Yes
Yes
Yes
8,562
0.291

Yes
Yes
Yes
Yes
Yes
8,592
0.055

Table 6
Regression of Board Structure and Covenant Violations
Estimates of firm fixed effects panel regressions of the number of independent board directors, logistic transformed
fraction of independent board directors, and number of board directors are shown. All regressions include year fixed
effects. In Panel A regressions include a polynomial function of order five of the distance from the default boundary.
Refer to Table A1 in the Appendix for variable definitions. The sample consists of observations on Investor
Responsibility Research Center firms that are included in the DealScan database with a loan containing covenants
restricting its current ratio, net worth or debt-to-EBITDA ratio to lie above or below a certain threshold. The
discontinuity sample is defined as those firm-year observations in which the absolute value of the relative distance to
the covenant threshold is less than 0.30. The sample period is from 1990 to 2007. Financial industries are omitted
(SIC codes 6000-6999). Robust t-statistics adjusted for firm-level clustering are in parentheses. *, **, *** indicate
significance at the 10%, 5%, and 1% levels.
Panel A: Entire Sample
(3)
(4)
(5)
(6)
Fraction of Independent
Nr. of Independent Directors
Directors
0.46*
0.58**
0.58**
0.24**
0.27**
0.27**
(1.90)
(2.42)
(2.42)
(2.20)
(2.49)
(2.48)
0.41
0.41
0.05
0.05
(0.75)
(0.76)
(0.19)
(0.21)
0.61*** 0.61***
0.08
0.08
(3.67)
(3.63)
(1.16)
(1.12)
0.58
0.59
0.15
0.15
(1.64)
(1.65)
(1.02)
(1.04)
0.11
0.11
0.09
0.09
(0.72)
(0.73)
(1.35)
(1.36)
-0.55*
-0.54*
-0.01
-0.00
(-1.89)
(-1.85)
(-0.06)
(-0.03)
1.84
1.89
0.40
0.40
(0.70)
(0.71)
(0.46)
(0.45)
-0.02
-0.02
-0.02
-0.03
(-0.06)
(-0.07)
(-0.21)
(-0.23)
1.29
1.31
0.17
0.16
(1.07)
(1.10)
(0.27)
(0.26)
-0.77
-0.80
-0.82
-0.82
(-0.53)
(-0.56)
(-1.09)
(-1.09)
0.03
0.01
(0.50)
(0.26)
0.26
-0.21
(0.16)
(-0.25)
-0.00
-0.00
(-0.54)
(-0.47)
-0.01
-0.01
-0.01
-0.01
(-0.60)
(-0.57)
(-1.32)
(-1.29)
(1)

Covenant violation dummy - SEC

Leverage
Firm size (log)
Firm age (log)
Number of segments (log)
Market-to-book (log)
R&D
Stock return volatility
Free cash flow
Return on assets
Governance index (GIM)
CEO ownership
CEO tenure
Default distance

(2)

(7)

0.07
(0.31)

(8)

(9)

Nr. of Directors
0.14
0.14
(0.66)
(0.64)
0.53
0.53
(1.04)
(1.05)
0.64*** 0.65***
(4.72)
(4.76)
0.36
0.39
(1.17)
(1.23)
0.00
0.00
(0.00)
(0.00)
-0.71*** -0.72***
(-2.61)
(-2.66)
0.55
0.54
(0.21)
(0.21)
0.13
0.14
(0.55)
(0.56)
1.65
1.67
(1.54)
(1.57)
0.15
0.14
(0.09)
(0.09)
-0.00
(-0.09)
1.64
(1.22)
-0.01
(-0.71)
0.00
0.00
(0.21)
(0.21)

Default distance

0.00*
(1.85)

0.00*
(1.81)

0.00
(1.36)

0.00
(1.32)

0.00
(1.16)

0.00
(1.15)

Default distance

-0.00**
(-2.05)

-0.00**
(-2.01)

-0.00
(-1.37)

-0.00
(-1.33)

-0.00
(-1.33)

-0.00
(-1.32)

Default distance

0.00**
(2.14)

0.00**
(2.10)

0.00
(1.39)

0.00
(1.35)

0.00
(1.39)

0.00
(1.38)

Default distance

-0.00**
(-2.20)

-0.00**
(-2.15)

-0.00
(-1.40)

-0.00
(-1.37)

-0.00
(-1.42)

-0.00
(-1.41)

1,879
0.193

1,879
0.194

1,871
0.223

1,871
0.224

1,879
0.051

1,879
0.053

Observations
Adjusted R-squared

1,879
0.169

1,871
0.215

29

1,879
0.013

Table 6: continued
Panel B: Discontinuity Sample
(3)
(4)
(5)
(6)
Fraction of Independent
Nr. of Independent Directors
Directors
Covenant violation dummy - SEC 1.31*** 1.21*** 1.17***
0.39*** 0.25***
0.25**
(4.59)
(3.69)
(3.96)
(7.76)
(2.73)
(2.50)
Leverage
-1.23
-0.98
-0.81
-0.69
(-0.48)
(-0.38)
(-0.82)
(-0.71)
Firm size (log)
0.24
0.22
0.03
0.02
(0.37)
(0.34)
(0.14)
(0.07)
Firm age (log)
3.52*** 3.49***
0.89*
0.84*
(3.72)
(3.68)
(1.97)
(1.81)
Number of segments (log)
0.61*** 0.61***
0.16*
0.15
(3.30)
(3.26)
(1.78)
(1.60)
Market-to-book (log)
0.30
0.26
0.03
0.01
(0.25)
(0.20)
(0.07)
(0.01)
R&D
41.65*
42.66*
13.23
13.57
(1.70)
(1.67)
(1.53)
(1.50)
Stock return volatility
1.60**
1.61**
0.95**
0.99***
(2.04)
(2.01)
(2.60)
(2.71)
Free cash flow
-0.08
0.59
0.38
0.77
(-0.02)
(0.19)
(0.25)
(0.54)
Return on assets
-1.87
-2.01
-0.70
-0.81
(-0.51)
(-0.55)
(-0.35)
(-0.40)
Governance index (GIM)
0.01
-0.02
(0.03)
(-0.17)
CEO ownership
6.19***
3.41***
(3.36)
(4.17)
CEO tenure
-0.01
0.00
(-0.40)
(0.07)
(1)

Observations
Adjusted R-squared

269
0.256

(2)

269
0.399

269
0.416

268
0.360

30

268
0.436

268
0.460

(7)

(8)

(9)

Nr. of Directors
0.97*** 1.21*** 1.18***
(2.85)
(3.59)
(3.36)
0.82
1.05
(0.35)
(0.44)
-0.09
-0.13
(-0.17)
(-0.24)
2.91*** 2.89***
(3.66)
(3.69)
0.51*** 0.51***
(2.69)
(2.73)
0.92
0.94
(1.01)
(1.00)
35.92*
37.31*
(1.76)
(1.74)
-0.32
-0.30
(-0.57)
(-0.49)
-2.47
-1.76
(-1.18)
(-0.83)
0.35
0.14
(0.12)
(0.05)
0.05
(0.23)
5.36***
(4.55)
-0.01
(-0.27)
269
0.078

269
0.241

269
0.264

Table 7
Regression of Board Structure and Covenant Violations: Robustness
Estimates of OLS regressions of the number of independent board directors, logistic transformed fraction of
independent board directors, and number of board directors are shown. All regressions include year fixed effects. In
Panel A regressions include a polynomial function of order five of the distance from the default boundary. Refer to
Table A1 in the Appendix for variable definitions. The sample consists of observations on Investor Responsibility
Research Center firms that are included in the DealScan database with a loan containing covenants restricting its
current ratio, net worth or debt-to-EBITDA ratio to lie above or below a certain threshold. The discontinuity sample
is defined as those firm-year observations in which the absolute value of the relative distance to the covenant
threshold is less than 0.30. The sample period is from 1990 to 2007. Financial industries are omitted (SIC codes
6000-6999). Robust t-statistics adjusted for firm-level clustering are in parentheses. *, **, *** indicate significance
at the 10%, 5%, and 1% levels..
Panel A: Entire Sample
(3)
(4)
(5)
(6)
Fraction of Independent
Nr. of Independent Directors
Directors
Covenant violation dummy - Dealscan 0.36*** 0.25***
0.20**
0.12*** 0.11*** 0.09***
(3.69)
(2.81)
(2.38)
(3.25)
(3.04)
(2.63)
Leverage
0.26
0.08
0.21
0.14
(0.65)
(0.22)
(1.25)
(0.86)
Firm size (log)
0.58*** 0.55***
0.04*
0.03
(9.72)
(9.75)
(1.85)
(1.51)
Firm age (log)
0.47*** 0.31***
0.09**
0.04
(5.39)
(3.78)
(2.57)
(1.19)
Number of segments (log)
0.08
0.04
-0.01
-0.01
(0.80)
(0.43)
(-0.20)
(-0.42)
Market-to-book (log)
-0.98*** -0.86***
-0.20**
-0.15**
(-4.86)
(-4.76)
(-2.28)
(-1.99)
R&D
1.23
0.62
0.81
0.54
(0.83)
(0.45)
(1.25)
(0.86)
Stock return volatility
-0.75*** -0.53**
-0.03
0.05
(-3.00)
(-2.27)
(-0.25)
(0.45)
Free cash flow
1.68
0.37
0.62
0.14
(1.46)
(0.35)
(1.18)
(0.28)
Return on assets
1.01
1.42
0.52
0.66
(0.79)
(1.22)
(0.90)
(1.21)
Governance index (GIM)
0.15***
0.05***
(6.19)
(4.80)
CEO ownership
-5.07***
-1.84***
(-3.95)
(-3.27)
CEO tenure
-0.02***
-0.01***
(-2.90)
(-3.70)
Default distance
0.01
0.01
-0.01**
-0.01*
(0.56)
(1.02)
(-2.02)
(-1.70)
(1)

(2)

(7)

(8)

(9)

Nr. of Directors
0.25**
0.11
0.09
(2.48)
(1.14)
(0.93)
-0.01
-0.07
(-0.03)
(-0.21)
0.69*** 0.67***
(10.42)
(10.12)
0.45*** 0.34***
(5.30)
(3.96)
0.13
0.09
(1.24)
(0.89)
-0.79*** -0.70***
(-3.69)
(-3.35)
-1.16
-1.37
(-0.70)
(-0.86)
-1.07*** -0.93***
(-3.62)
(-3.16)
0.95
0.32
(0.78)
(0.26)
-0.28
-0.05
(-0.20)
(-0.03)
0.10***
(3.87)
-3.21**
(-2.39)
0.00
(0.40)
0.03*** 0.03***
(3.17)
(3.43)

Default distance

0.00
(0.92)

-0.00
(-0.10)

0.00**
(2.52)

0.00
(1.50)

-0.00
(-1.36)

-0.00*
(-1.94)

Default distance

-0.00
(-1.24)

-0.00
(-0.06)

-0.00**
(-2.49)

-0.00
(-1.33)

0.00
(0.89)

0.00
(1.50)

Default distance

0.00
(1.36)

0.00
(0.11)

0.00**
(2.45)

0.00
(1.24)

-0.00
(-0.71)

-0.00
(-1.33)

Default distance

-0.00
(-1.41)

-0.00
(-0.13)

-0.00**
(-2.41)

-0.00
(-1.17)

0.00
(0.63)

0.00
(1.25)

2,969
0.335

2,969
0.392

2,956
0.252

2,956
0.306

2969
0.327

2969
0.345

Observations
Adjusted R-squared

2,969
0.196

31

2,956
0.232

2969
0.144

Table 7: continued
Panel B: Discontinuity Sample
(2)
(3)
(4)
(5)
(6)
Fraction of Independent
Nr. of Independent Directors
Directors
0.47
0.62**
0.60**
0.18
0.22**
0.22**
(1.33)
(2.05)
(2.27)
(1.58)
(1.99)
(2.18)
0.22
-0.22
0.44
0.24
(0.28)
(-0.28)
(1.35)
(0.71)
0.66*** 0.59***
0.08
0.06
(4.43)
(4.26)
(1.32)
(0.92)
0.50**
0.35**
0.15*
0.11
(2.54)
(2.03)
(1.88)
(1.50)
-0.07
-0.18
-0.10
-0.14*
(-0.32)
(-0.91)
(-1.27)
(-1.84)
-0.80
-0.66
-0.01
0.03
(-1.52)
(-1.43)
(-0.06)
(0.21)
-1.16
-3.55
-0.94
-2.17
(-0.22)
(-0.84)
(-0.44)
(-1.18)
-0.64
0.08
-0.02
0.23
(-1.16)
(0.17)
(-0.07)
(1.11)
3.46*
1.08
1.58*
0.62
(1.73)
(0.55)
(1.80)
(0.70)
-1.70
-0.25
-1.23
-0.70
(-0.71)
(-0.12)
(-1.25)
(-0.72)
0.21***
0.06***
(4.46)
(2.71)
-3.48
-2.37**
(-1.28)
(-2.02)
-0.05***
-0.02**
(-2.71)
(-2.49)
(1)

Covenant violation dummy - Dealscan

Leverage
Firm size (log)
Firm age (log)
Number of segments (log)
Market-to-book (log)
R&D
Stock return volatility
Free cash flow
Return on assets
Governance index (GIM)
CEO ownership
CEO tenure

Observations
Adjusted R-squared

472
0.253

472
0.353

472
0.454

32

469
0.293

469
0.328

469
0.418

(7)

0.15
(0.43)

472
0.221

(8)

(9)

Nr. of Directors
0.30
0.25
(1.01)
(0.91)
-0.84
-0.99
(-1.12)
(-1.30)
0.71*** 0.67***
(4.39)
(4.29)
0.34
0.21
(1.64)
(1.01)
0.06
-0.02
(0.29)
(-0.11)
-0.98**
-0.88*
(-1.99)
(-1.85)
-1.88
-2.32
(-0.45)
(-0.60)
-0.98*
-0.53
(-1.97)
(-1.08)
1.26
0.13
(0.60)
(0.06)
0.71
1.50
(0.27)
(0.61)
0.17***
(3.53)
0.37
(0.16)
-0.03
(-1.43)
472
0.326

472
0.367

Figure 1
Board Structure around a Covenant Violation
This figure presents means of the number of independent directors, fraction of independent directors and number of
directors in time relative to the first time that a covenant is violated. The violation occurs at year 0. Mean values are
represented by a circle in the center of each vertical line, which corresponds to a 95% confidence interval.

Panel A: Number of Independent Directors

Panel B: Board Independence

33

Panel C: Board Size

34

Appendix
Table A.1
Variable Definitions
Variable
Number of independent directors
Fraction of independent directors
Covenant violation dummy
SEC
Covenant violation dummy
DealScan
Number of directors
Firm size
Leverage
Firm age
Number of segments
Market-to-book
R&D
Stock return volatility
Free cash flow
Return-on-assets
Governance index (GIM)
CEO ownership
CEO tenure
Dividend dummy
Positive net income dummy
Interest expenses
Net worth
Cash

Definition
Number of board members that are independent directors (1990-1995 data from Compact Disclosure and 1996-2001 data
from IRRC).
Ratio of number of independent directors by the number of directors (1990-1995 data from Compact Disclosure and 19962001 data from IRRC).
Dummy variable that takes the value of one if there are a covenant violation reported in SEC filings in a year, and zero
otherwise (SEC 10-Qs and 10-Ks).
Dummy variable that takes the value of one if there are a covenant violation based on covenant slack (current ratio, net
worth, (tangible net worth and debt-to-EBITDA), and zero otherwise (DealScan).
Number of board members (1990-1995 data from Compact Disclosure and 1996-2001 data from IRRC).
Market capitalization in $ millions (Compustat: CSHO x PRCC_F).
Ratio of total debt to total assets (Compustat: (DLTT + DLC) / AT).
Number of years since the stock inclusion in the CRSP database.
Number of business segments in which firm operates (Compustat).
Ratio of market value of assets to book value of total assets (Compustat: (AT + CSHO x PRCC_F - CEQ) / AT).
Ratio of research and development expenditures to book value of assets (Compustat: XRD / AT)
Stock return standard deviation (annualized) estimated with daily stock returns (CRSP).
Ratio of operating income before depreciation minus capital expenditures to total assets (Compustat: (EBITDA - CAPX) /
AT).
Ratio of operating income before depreciation to total assets (Compustat: EBITDA / AT).
Governance index of Gompers, Ishii, and Metrick (2003), which is based on 24 antitakeover provisions (IRRC).
Number of shares held by the CEO divided by number of shares outstanding (EXECUCOMP).
Number of years since the date the director became CEO (EXECUCOMP).
Dummy variable that takes the value of one if the firm pays dividends (Compustat: DVC) and zero otherwise
Dummy variable that takes the value of one if the firm has positive net income (Compustat: NI)
Ratio of interest expenses to total assets (Compustat: XINT / AT).
Ratio of total assets minus total liabilities to total assets (Compustat: (AT - LT) / AT)
Ratio of cash and short-term investments to total assets (Compustat: CHE/ AT)

35