Bond Covenants and Institutional Blockholding
Cinder Xinde Zhang
Simiao Zhou∗
First draft: January 5, 2012
This draft: November 11, 2013
Abstract
The literature is unclear about whether the bond contract is designed ex ante to
enhance the ex post efficiency. Using a sample of 10,513 public bonds issued between
1979 and 2008, we find that institutional shareholders’ blockholding has a significant and
positive effect on the number of restrictive covenants in the bond contract. This effect is
robust to different measures of blockholding and alternative regression models, and cannot
be explained away by the endogeneity of institutional blockholding. We identify a stronger
effect of blockholding for active blockholders and short-term blockholders, suggesting that
covenants are employed ex ante to mitigate potential conflicts between shareholders and
bondholders. Moreover, we find that bond covenants also benefit shareholders by reducing
their borrowing costs that are related to agency conflicts. Our findings suggest that the
bond contract is preemptively designed to capture the ex-post efficiency.
∗
Both authors are affiliated with the School of Finance, Shanghai University of Finance and Economics; 777
Guoding Road, Shanghai 200433, China. Zhang can be reached at [email protected], and Zhou
can be reached at [email protected]. We thank Patrick Bolton, Jiaping Qiu, Jay Wang, Dolly
King, Kai Li, and seminar and conference participants at Southwestern University of Finance and Economics,
Zhejiang University, the XMU-UNCC 2012 International Symposium on Risk Management and Derivatives,
the 2013 SMU-SUFE Summer Institute of Finance Conference, and the 2013 Northern Finance Association
Meetings for their helpful comments and suggestions. The usual disclaimer applies.
Bond Covenants and Institutional Blockholding
First draft: January 5, 2012
This draft: November 11, 2013
Abstract
The literature is unclear about whether the bond contract is designed ex ante to
enhance the ex post efficiency. Using a sample of 10,513 public bonds issued between
1979 and 2008, we find that institutional shareholders’ blockholding has a significant and
positive effect on the number of restrictive covenants in the bond contract. This effect is
robust to different measures of blockholding and alternative regression models, and cannot
be explained away by the endogeneity of institutional blockholding. We identify a stronger
effect of blockholding for active blockholders and short-term blockholders, suggesting that
covenants are employed ex ante to mitigate potential conflicts between shareholders and
bondholders. Moreover, we find that bond covenants also benefit shareholders by reducing
their borrowing costs that are related to agency conflicts. Our findings suggest that the
bond contract is preemptively designed to capture the ex-post efficiency.
1
1
Introduction
Recent empirical studies on financial contracting reveal that covenants strengthen creditor
rights and significantly affect corporate investment (Chava and Roberts, 2008; Nini, Smith,
and Sufi, 2009; Demiroglu and James, 2010), capital structure (Roberts and Sufi, 2009; Nini,
Smith, and Sufi, 2012), and operating efficiency and stock price performance (Nini, Smith,
and Sufi, 2012). These results suggest that covenants in corporate debt contracts are binding. They also highlight the importance of carefully allocating creditor control in financial
contracts. While those studies find that debt covenants are effective ex post, it is important
to determine whether the allocation of control rights is efficient ex ante. In this paper, we
attempt to answer the question: Are covenants strategically designed to define an efficient
allocation of control rights ex ante?
Financial contracting theories indicate that the rationale for covenants is controlling
agency conflicts and defining control rights. Smith and Warner (1979) argue that covenants
are used to control the bondholder-shareholder conflict to lower the agency costs of debt.
(Aghion and Bolton, 1992) theoretically demonstrate that when firm owner is rent seeking,
debt can be the optimal contract. Their theory explains that state contingency control right
allocation is optimal in their setting. Bond covenant is therefore proposed in their paper to
allocate control right in debt contracts. In this paper, we investigate empirically how agency
conflicts shape the structure of bond covenants. We find robust evidence that heightened
agency conflicts increase the stringency of covenants in bond contracts. Our results suggest
that the potential conflicts between bondholders and shareholders are preemptively considered
in bond covenants.
To measure the restrictiveness of covenants, we follow Billett, King, and Mauer (2007)
and divide bond covenants into 15 categories. These categories of covenants fall into four
major groups, including payout restrictions, investment restrictions, financing restrictions,
and event-driven covenants. We then add up the total number of categories to construct a
covenant index approximating the level of protection for bondholders. As the value of the
covenant index increases, bond covenants become more restrictive.
To approximate the agency conflicts between shareholders and bondholders, we create two
measures of institutional blockholding: NL5 is the percentage of a company’s shares owned by
2
all institutional investors who hold no less than 5% of total shares outstanding, and Top5 is
the percentage of shares held by the five largest institutional owners of the firm. Concentrated
institutional ownership strengthens shareholder control, because large institutional investors
are incentivized and better able to monitor the management. Active monitoring aligns the
interest of the management with those of shareholders at large or with those of blockholders
in particular. Enhanced shareholder rights would then exacerbate bondholder concerns over
the agency conflicts. Influenced by blockholders, firms may undertake high-risk projects that
maximize blockholders’ interest instead of firm value (Jensen and Meckling, 1976), or forgo
profitable investment opportunities due to the debt-overhang problem (Myers, 1977). To
mitigate these agency costs, firms may find it beneficial to enclose protective covenants in
the bond contract, because giving creditors more control rights facilitates debt financing and
helps to lower the borrowing cost. Empirical studies also find that firm operating and stock
price performance improve following violations of covenants (Nini, Smith, and Sufi, 2012).
Thus, the agency view predicts that more concentrated institutional ownership increases the
restrictiveness of bond covenants.
While concentrated outside ownership strengthens shareholder rights, it could also bring
about shared benefits to bondholders. With effective monitoring, blockholders can force management to take value-enhancing actions (Barclay and Holderness, 1992; Huddart, 1993). Consequently, there would be less need for bond covenants when efficient blockholder-monitoring
provides shared benefits to bondholders. Thus, the shared-benefit hypothesis predicts a negative effect of blockholding on the tightness of covenants. The foregoing analysis indicates
bondholder and shareholder interests can either align or diverge, as reflected in a statement
by Moody’s Investor Service: “While there is substantial overlap between creditor and shareholder interests, there also are important potential conflicts.” Ultimately, the net effect of
institutional blockholding on bond covenants remains an empirical question.
Our empirical investigation shows that the agency channel plays a dominant role in determining the relation between covenants and institutional blockholding. Using 10,513 bonds
issued from 1979 to 2008, we find that institutional blockholding has a significant and positive
effect on covenant protection. This effect persists after we control for firm characteristics, bond
attributes, and credit market conditions. These findings suggest that stronger shareholder
3
control escalates the potential shareholder-bondholder conflicts and, consequently, leads to
more stringent covenant protection.
To test the robustness of our baseline results, we employ a different blockholding measure,
based on the Herfindahl index, to account for the effect of total institutional ownership, and
an alternative estimation equation, the ordered probit model, to fit the ordinal nature of the
covenant index. To address concerns that may arise when a firm issues multiple bonds over
the sample period, we conduct regressions of covenants on blockholding at the firm level. In
all above robustness checks, we find similar results to those obtained in baseline regressions.
To identify the causal effect of block ownership and mitigate the endogeneity concern,
our main approach is to employ instrumental-variable regressions. Our instrument for institutional blockholding is stock liquidity. This can be justified on the basis of the following
arguments. First, liquidity has a direct effect on institutional blockholding because it reduces
transaction costs, making it easier for blockholders to trade shares. Second, liquidity is unlikely to have a direct effect on the covenants of debt contract. The bond security is relatively
illiquid, so that bond contracting terms are unlikely determined by stock liquidity. Existing
theories indicate that liquidity facilitates shareholder governance either through blockholder
intervention (Maug, 1998) or through the threat of blockholder exit (Edmans, 2009; Edmans
and Manso, 2011). Thus, liquidity only has an indirect effect on covenants through enhanced
shareholder control. Two-stage least square regressions show that institutional blockholding
still has a significantly positive effect on covenant protection, and this effect is even stronger
than that in the baseline regressions.
To further identify the channel through which institutional blockholding affects bond
covenants, we carry out regression analysis for sub groups of covenants and for different types
of institutional investors. In our first test, we examine the effect of blockholding on different
groups of covenants. Although the 15 covenant categories are all interrelated, there is some
heterogeneity in the intended purposes across different sub groups of covenant categories. We
break down the covenant index and focus on two major groups of covenant categories. The
first group includes restrictions on payouts and investment activities, and the second group
comprises restrictions on financing activities. We expect that the two groups of covenants are
related in different ways to agency conflicts between bondholders and shareholders. Payout
4
and investment restrictions are designed to directly address shareholder-bondholder conflicts,
while financing restrictions are mainly designed to mitigate claim dilution. Thus, covenants
restricting payouts and investment activities should depend more on institutional blockholding, while financing-related covenants are less affected by concentrated institutional ownership. Our regression results on different types of covenant categories are consistent with this
conjecture.
In our second test, we investigate the effect of institutional blockholding among different
types of investors. We hypothesize that the agency problem is more acute when institutional
shareholding is concentrated among active investors. If the positive effect of institutional
blockholding on covenant protection is mainly driven by agency concerns, then we expect
the effect to become stronger for active investors. Consistent with this hypothesis, we find
that the positive effect of institutional blockholding on covenant protection is much larger for
active institutional investors than for passive institutional shareholders.
Similar to investor activism, the investment horizon also has differential implications on the
bondholder-shareholder relationship. On the one hand, concentrated ownership allows shortterm blockholders to exercise undue influence over management to engage in opportunistic
activities, such as investing in short-run and high-risk projects, and thus elevates bondholdershareholder conflicts. On the other hand, long-term institutional blockholders may protect the
interest of bondholders by considering the firm’s future financing needs (Shleifer and Vishny,
1997) and long-term value (Anderson, Mansi, and Reeb, 2003). Anderson, Mansi, and Reeb
(2003) find that shareholding by founding families reduces bond yields, which suggests that
long-term investors’ interest may be aligned with bondholders’ interests. We find empirical
results that are consistent with the foregoing analysis. Institutional blockholding by shortterm investors is associated with tighter bond covenant restrictions, while the positive effect
of ownership concentration among long-term institutional blockholders is greatly attenuated
and statistically insignificant.
In the last part of our empirical investigation, we consider whether covenants help to reduce
the offering yields of bonds. We find that, exposed to intensified agency conflicts, bondholders
charge a higher yield for firms with more institutional blockholding. This confirms the finding
in Bhojraj and Sengupta (2003) that blockholding was associated with higher offering yield
5
spreads. Our additional analysis shows that covenants help to alleviate the impact of agency
conflicts on yield spreads, and the beneficial effect of covenants is stronger for investmentgrade bonds.
Our paper is closely related to the literature that examines the determinants of bond
covenants. Nash, Netter, and Poulsen (2003) argue that bond issuers’ use of covenants weighs
the benefit of mitigating agency problems and the cost of reducing flexibility. The authors
show that high-growth firms are less likely to include dividend or debt issuance restrictions
in their bond contracts to preserve future flexibility. Malitz (1986) finds that small firms and
highly levered issuers are more likely to include restrictive covenants in their bond contracts
to reduce the financing cost. Billett, King, and Mauer (2007) find that covenant protection
increases with growth option, debt maturity, and leverage. Additionally, covenant protection
significantly reduces the negative relation between leverage and growth opportunities, and thus
it can mitigate the agency costs of debt for high-growth firms. Bengtsson (2011) finds that
covenants are in place to overcome a conflict of interest that arises from debt-like contractual
features of a venture capitalist’s preferred stock. Closet to our paper is Chava, Kumar, and
Warga (2010), who investigate the effects of managerial entrenchment and fraud on different
types of covenants and find that managerial agency risk influences the use of covenants.
Our paper contributes to the literature on debt covenants in two important ways. First,
we find that the potential conflicts between shareholders and bondholders are preemptively
considered in designing the structure of debt covenants. Although an increasing number of
studies show that the debt contract is ex post effective following a violation of covenants, it
is less known empirically whether the use of tight covenants is ex ante efficient. Our paper
fills the gap by showing that covenants are used as a bondholder governance mechanism in
response to aggravated agency conflicts. Our results connect the ex ante design of covenants
with the ex post effectiveness of control allocation and suggest that the bond contract is
preemptively designed to capture the ex-post efficiency. Second, our paper’s results shed some
light on the interaction between bondholder and shareholder governance. As a bondholder
governance mechanism, covenants can benefit shareholders by reducing their borrowing cost,
which increases with stronger shareholder control.
Our paper is also related the literature that studies the relation between bond yields
6
and corporate governance (see, among others, Bhojraj and Sengupta, 2003; Cremers, Nair,
and Wei, 2007; Klock, Mansi, and Maxwell, 2005; Chava, Livdan, and Purnanandam, 2009).
The literature shows that the bond yield generally increases (falls) when the bondholdershareholder relationship deteriorates (ameliorates). Our results complement these findings
and indicate that the use of covenants helps to reduce bond yields when the agency conflicts
between shareholders and bondholders become more acute. We use a comprehensive covenant
protection index rather than individual covenants. Cremers, Nair, and Wei (2007) show that
individual categories of covenants help to mitigate the potential conflict between the firm’s
shareholders and creditors in the event of mergers and acquisitions. Our paper stresses the
importance of using an overall covenant index that covers 15 categories, as theoretical and
empirical evidence suggests that different types of bond covenants are in fact interrelated
(Smith and Warner, 1979; Billett, King, and Mauer, 2007).
The rest of the paper proceeds as follows. Section 2 describes data and summary statistics.
Section 3 presents our baseline results, and their robustness is studied in Section 4. Section 5
discusses the endogeneity issue and presents results that mitigate the concern. Section 6
examines the effect of blockholding on different types of covenants, and Section 7 discusses
the effect of investor type on covenant protection. Section 8 investigates the effect of covenants
on bond yields. Section 9 concludes the paper.
2
Data and Sample Characteristics
2.1
The Bond Issue Sample
Our primary data sources are the Fixed Income Security Database (FISD) for public debt
issuance and covenant usage information, the Thomson 13-F database for institutional ownership data, and Compustat for firm-level accounting variables. Because of data availability
in the Thomson 13-F database, we restrict sample years to the period between 1979 and
2008. We start with 73,797 corporate bonds in FISD, which are issued by non-financial U.S.
firms.We then eliminate 39,950 corporate medium-term notes (MTN), since they have little information in FISD on covenant use. For the remaining 33,847 bond issues, we obtain
covenant information from FISD for 19,351 bonds.
We next match the issuer CUSIP’s of the 19,351 bond issues with records in the Thomson
7
13-F database. Specifically, for each issuer CUSIP in FISD, we trace the entire history of
CUSIP changes, if any, via the master file provided by the CUSIP Service Bureau. We then
use the CUSIP history to find the matched firm in the Thomson 13-F database to retrieve
institutional share holding in the most recent quarter, prior to the offering date. The matching
procedure leaves a final sample of 10,513 bond issues between 1979 and 2008.
Figure 1 shows that the number of issues remains at low to moderate levels in 1980’s, it
increases sharply in early 1990’s, and it stays at relatively high levels throughout the rest of
the sample period. For a bond issue, we group all covenants into 15 broad categories and
present the annual average number of covenants in Figure 2.1 . The solid line represents the
average number of covenants for the 10,513 bond issues in our sample, and the dashed line is
for the entire 19,351 bond issues in FISD that have covenant information. Figure 2 shows that
the average number of bond covenants increases in early years and stabilizes in mid-1990’s.
Additionally, the time series pattern in our sample is almost identical to that in the FISD
universe. Thus, bond issues in our sample is quite representative of those in the FISD universe
in term of covenant information.
Table 1 describes the bond issues for our initial sample. The average offering yield is
7.27%. Comparing the yield of a bond issue to the yield of U.S. Treasury bonds of similar
maturity, we find that the average yield spread is 1.75%. The average bond issue has an
offering amount of $378.52 million and a maturity of 12.66 years, and more than 96% of
the issues are senior bonds. We create a set of issue-level dummy variables, which indicate
whether the bond is callable, puttable, or convertible. We find that more than half of the
bond issues are callable, while the percentage of puttable and convertible bonds is relatively
small. We also extract from FISD a bond’s credit ratings on the offering date. For issues that
do not have ratings at issuance, we search for ratings issued within 1 year of the offering date.
Among those ratings, we first choose the rating immediately prior to issuance, and, if no prior
ratings exist, we then use the earliest post-issuance rating. Table 1 shows that about 73%
of the issues are rated by Standard & Poor’s; of those, 46.70% are investment-grade issues.
Similar statistics are observed in Moody’s ratings. We generate an investment-grade dummy
variable that equals 1 if both Standard & Poor’s and Moody’s ratings for the bond issue is
above the investment grade. Table 1 shows that, among the 8,280 rated bond issues, 61.4%
1
We explain the details of constructing the covenant index in Section 2.2
8
are investment grade.
2.2
2.2.1
Regression Variables
Measures of Covenant Protection and Institutional Blockholding
Central to our empirical analysis is the relationship between bondholder protection and institutional blockholding. Following Billett, King, and Mauer (2007), we construct a covenant
index by dichotomizing bondholder protection in 15 broad categories of covenants.2 If a bond
issue has no covenants in any of the 15 categories and FISD indicates that the covenant information is available for the issue, then we assign a value of zero to the covenant index, which
represents the least protection for bondholders. On the other hand, if an issue has covenants
in all categories, then the index equals 15, which represent the most protection.
To approximate outside shareholder control, we construct two measures related to institutional blockholding. The first measure, NL5, is the percentage of shares owned by all
institutional equity holders who hold no less than 5% of total shares outstanding. The second
measure, TOP5, is the percentage of shares held by the five largest institutional investors.
We use institutional blockholding, rather than total institutional ownership, because it better
reflects the effective monitoring by institutional investors. While concentrated institutional
holding represents greater incentives and ability to influence the firm’s decisions and thus better aligns the shareholder-manager interest, aggregate institutional ownership may not result
in such an alignment. Institutions with minor stakes have little incentive to be involved in
firm-specific decisions, a situation which leads to the free-rider problem pointed out Grossman
and Hart (1980). Table 2 provides summary statistics for the three key variables. As shown
in Table 2, half of the bond issues have covenants in 3–6 categories, while the median number
of covenants is 4. Top-five institutions own an average of 21.3% of outstanding shares, and
the average holding is 12% for blockholders who own at least 5% of the firm.
2
These 15 categories of covenants are grouped as follows. Two categories of payout restrictions limit a firm’s
ability to pay dividends and repurchase shares. Three categories of investment restrictions include clauses on
the investment policy, asset sale, and merger. Seven categories of financing restrictions are concerned with
funded debt, subordinated debt, senior debt, secured debt, total leverage, sale & leaseback, and stock issuance.
The last three categories are event-driven covenants, including rating and net worth triggers, poison put, and
cross-default clauses.
9
2.2.2
Firm-Level Accounting Variables and Other Controls
We match bond issues to the issuers’ accounting data in Compustat via the historical CUSIP
table. For each issue, we first find the fiscal year in which the bond offering date falls and label
it as Year 0. We then extract firm-level accounting information at the end of (fiscal) Year −1.
Previous studies document that leverage, firm size, growth options, and debt maturity are
important determinants of bond covenants. We construct empirical measures for these firm
characteristics. To measure a firm’s leverage, we compute the ratio of the book value of longterm debt to the sum of the book value of long-term debt and the market value of equity,
where the market value of equity is equal to the product of the fiscal year-end share price and
shares outstanding. To account for changes in the long-term debt that take place between
the end of Year −1 and the offering date, we adjust the leverage ratio in the following way.
First, we add the amount of new issues in FISD to the level of long-term debt at the end
of Year −1. We then subtract any existing bond issues that were long-term at the end of
Year -1 and would have become shot-term at the offering date. Finally, we deduct the amount
of existing issues that are redeemed between the end of Year −1 and the offering date. We
measure growth options by the market-to-book asset ratio, where the market value of assets
equals the book value of total assets minus the book value of equity plus the market value
of equity. Firm size is measured by the natural logarithm of net sales, and short-term debt
is approximated by debt in current liabilities, divided by total assets. The debt in current
liabilities includes short-term notes and the current portion of long-term debt. Additionally,
our empirical regressions control for firm profitability, as captured by the EBIT-to-assets ratio,
and asset tangibility, as defined by the ratio of fixed assets to total assets. Table 2 displays
the summary statistics of these accounting variables. To mitigate the impact of outliers, we
winsorize all firm-level accounting variables and the blockholding measures at the top and
bottom 1 percentiles.
Also presented in Table 2 is a credit premium variable, which captures the interest rate
conditions at the time of issuance.3 We define the credit premium as the difference in average
yields of BBB and AAA corporate bonds. Bond yields are retrieved from the Federal Reserve
3
Instead of using the credit premium variable, we control for a full set of year dummy variables and find
qualitatively similar results in our baseline regressions. It is also worth noting that adding industry (based on
two-digit SIC codes) fixed effects does not change our main findings in the paper.
10
Economic Database. We match each bond issue with the credit premium in the month prior
to the offering date. Table 2 shows that, on average, BBB bonds yield 89.3 base points higher
than AAA bonds in the sample period.
3
Baseline Results
In this section, we empirically examine the relationship between the level of covenant protection and institutional blockholding. We see a clear positive relation between covenant protection and blockholding in Figures 3 and 4, where we plot the average value of the covenant
index against different deciles of the blockholding measures. Except for the first decile of the
Top5 blocking measure, we observe that the index of covenant protection increases when institutional blockholding increases. The preceding analysis suggests that bondholders demand
more covenant protection as the firm’s ownership becomes more concentrated among outside
institutional blockholders. To further explore the relationship between covenant protection
and shareholder rights, we turn now to the multivariate regression analysis.
In Tables 3 and 4, we investigate the effect of institutional blockholding on the level
of covenant protection, where the institutional blockholding is measured by NL5 and Top5
variables respectively. In all empirical regressions, we adjust standard errors by the firm-level
clustering and report t-statistics accordingly. Our empirical results for all regression models
are very similar in both tables. The model in column 1 is a simple linear regression with NL5
as the only explanatory variable. The model reinforces the preceding descriptive results that
institutional blockholding has a significantly positive effect on the covenant protection. Take
the blockholding measure NL5 as an example. Its coefficient is 3.113, with a t-statistic equal
to 9.00. A one standard deviation increase in blockholdings raises the covenant index by 0.4,
and this increase accounts for 10% of the median covenant index, given the sample median
value of 4.
Previous studies identify firm size, leverage, growth options, and debt maturity as important determinants of bond covenants. In column 2, we control for these covenant determinants,
as well as additional firm characteristics and macroeconomic conditions. For large firms, where
information is readily available and reputation is both established and reviewed often, the cost
of explicit covenants exceeds any possible benefits (Malitz, 1986). Using log sales as a proxy
11
for firm size, we find that larger firms tend to use fewer restrictive covenants. We also find that
a few other previously identified determinants significantly affect the covenant index in our
sample. For instance, firms with a higher market-leverage ratio have more covenant restrictions in their bond contracts, suggesting that bondholders demand more covenant protection
because of default concerns. Short-term debt significantly reduces the usage of covenants,
implying that short-term bonds could ease agency conflicts between shareholders and bondholders (Myers, 1977; Billett, King, and Mauer, 2007).
We also control for firm profitability, asset tangibility, and the credit premium. Higher
profitability may escalate agency concerns over free cash flows. In such an situation, bond
covenants help to mitigate the agency costs by disciplining the manager. Consistent with
the agency view, our results in column 2 show that profitability has a significantly positive
effect on the use of covenants. Unlike leverage, tangible assets could mitigate default concerns
on the firm’s bond issues. Indeed, the results in column 2 demonstrate that fewer covenant
restrictions are observed for issuers with more fixed assets. The credit premium is used to
control aggregate economic conditions during the bond issuance. The significantly negative
coefficient on the credit premium suggests that higher returns on the credit market lower the
level of covenant protection. Controlling for all above firm characteristics and the measure
for macroeconomic conditions, we find that institutional blockholding is still positive and
significant at the 1% level.
In column 3, we further control for issue-level characteristics: the natural log of issue size,
maturity, and whether the bond has an embedded call, put, or conversion option. With other
things equal, bond issues with larger size and shorter maturity tend to have more covenants.
Callable bonds tend to have more protection, while the opposite holds for puttable and convertible bonds. As we sequentially increase control variables in the regressions in Tables 3
and 4, the coefficient on institutional blockholding always remains positive and significant
at the 1% level, suggesting that the level of covenant protection increases as institutional
blockholders accumulate more shares.
In column 4 of Tables 3 and 4, we further control for credit ratings of the bond issues.
The results show that the effect of blockholding on covenant protection is largely reduced
both in terms of magnitude and the level of statistical significance when we control for the
12
investment-grade dummy variable. The attenuated effect of institutional blockholding is expected, because such covenant determinants as agency conflicts would already be factored in
by rating agencies in their evaluation of the credit risk of a bond issue. Once ratings are controlled for, the incremental effect of institutional blockholding is thus weakened. We confirmed
our conjecture by running a probit regression of the investment-grade dummy. The explanatory variables include institutional blockholding and the same set of issue-level, firm-level, and
macro-level control variables as in Tables 3 and 4. Untabulated results show that, for both
blockholding measures, the coefficient is negative and significant at the 1% level. In addition,
the pseudo R-squares of the probit regressions are 0.48, and the model correctly classifies
the bonds’ ratings for about 84% of the observations. These results suggest that the agency
conflict is an important determinant of bond ratings. As institutional blockholding increases,
a proxy for increasing shareholder control and, thus, more severe bondholder-shareholder
conflicts, the bond is less likely to be rated above the investment grade.
In the last column of Tables 3 and 4, we add an interaction term between institutional
blockholding and the investment-grade dummy, which allows us to differentiate the effect of
block ownership between bonds with different credit ratings. Take the result in Table 3 as an
example. The coefficient on NL5 is −0.068 and statistically insignificant. This implies that,
for non-investment-grade bonds, institutional blockholding does not have a significant effect
on covenant protection. The coefficient on the interaction term is 1.333 and significant at
the 5% level. Combining the preceding two estimates, we find that the effect of blockholding
on covenant protection is equal to 1.265 for investment-grade bonds. In addition, the joint
hypothesis test shows that this positive effect is significant at the 1% level. The result in
column 5 of Table 3 demonstrates that institutional blockholding has a significantly positive
effect on covenant protection for investment-grade bonds, but the effect becomes insignificant
for non-investment-grade bonds. The same conclusion can be drawn in column 5 of Table 4,
where we use Top5 as a measure for institutional blockholding. Thus, the results in column
5 of both Table 3 and Table 4 shows that the attenuated effect of institutional blockholding
in column 4 is mainly driven by the fact that block ownership has an insignificant effect for
non-investment-grade bonds. Bond ratings certainly incorporate the information on agency
conflicts inherent in a bond issuer. Institutional blockholding is unlikely to have a great
13
impact on bond covenants in the presence of ratings, especially for junk bonds where default
concerns aggravate the potential agency conflicts. Hence, we do not use the bond rating
dummy in subsequent regressions.
4
Robustness Checks
In this section, we present a variety of robustness checks for our baseline results. In Table 5, we first examine whether our results are robust to alternative model specifications. We
argued that institutional blockholding, rather than total institutional shareholding, better
reflects the effective monitoring by institutional investors. Our baseline results in Tables 3
and 4 showed that, as institutional blockholding increases, the enhanced shareholder control
intensifies bondholder concerns, and thus more covenant protection is observed in the debt
contract. As a robustness check, we take into account the possibility that total institutional
ownership also could capture agency conflicts indirectly and thus influence the covenant design. Consider two firms, H and L, with the same level of block ownership. Yet, Firm L
has a much lower total institutional shareholding than Firm H. If blockholders are perceived
as having superior or insider information, then the fact that Firm L has much lower institutional ownership by other minority shareholders may indicate that it is opaque and has
more powerful blockholders. Thus, bondholders are more concerned about lending to Firm L
than to Firm H. Our primary measures of institutional blockholding do not distinguish these
two types of firms, as they have the same level of block ownership. To differentiate agency
conflicts inherent in these two types of firms, we construct an alternative measure of ownership concentration, based on the Herfindahl index of institutional shareholding. In particular,
our Herfindahl index variable is equal to the sum of squares of all institutional investors’
shareholding. This variable better captures ownership concentration among all institutiona
investors and will have a larger value for Firm L compared to Firm H.
Using the alternative measure of ownership concentration, we present the regression results in column 1 of Table 5. The coefficient on Herfindahl index is positive and significant at
the 1% level. This result is similar to those we obtained using the two primary blockholding
measures, and reinforces our conclusion that covenant protection increases with more concentrated institutional ownership. In untabulated regressions, we also augment Model 3 in both
14
Table 3 and Table 4 by including total institutional ownership in addition to blockholding.
Conditional on NL5 and other control variables, the coefficient on total institutional shareholding is negative but insignificant. This coefficient is very close to zero and insignificant in
the regression involving Top5.
In columns 2 and 3 of Table 5, we employ ordered probit models to accommodate the
categorical and ordered nature of our dependent variable, the covenant index. The ordered
probit models better capture the ordinal nature of the dependent variable, but at the extra
cost of assuming a normally distributed error term. The results show that the estimated
coefficients on all regressors are qualitatively the same as the baseline results for Model 3 in
Tables 3 and 4. Both measures of blockholdings are positive and significant at the 1% level
after controlling for the full set of bond-, firm-, and macro-level variables.
In Table 6, we investigate whether our main results are robust to different data structure.
Our baseline regressions are at the issue level. In our sample, it is common for firms to offer
multiple bond issues, even in the same year. Over the entire sample period, the average
(median) number of issues for a firm is 9.86 (5). Among the firms issuing multiple bonds,
the average (median) number of days between two consecutive issues is 657.80 (330) days.
Within a year, the average (median) number of issues offered by a firm is 1.78 (1), and,
among multiple-bond issuers, the average (median) number of days between two consecutive
offerings is 97.37 (83) days. To examine whether our main results hold at the firm level,
we first collapse all bond issues in the same fiscal year into one observation by calculating
the weighted averages for the covenant index and the institutional blockholding variables.
The weight is an issue’s offering size as a percentage of the total amount of bonds issued in
the same fiscal year. We then regress the weighted-average covenant index on the weightedaverage blockholding, controlling for firm accounting variables at the end of fiscal Year −1
and the average credit premium over fiscal Year 0. The first two columns of Table 6 report
the firm-year regressions. We find that both measures of institutional blockholding have a
positive and highly significant effect on the covenant index.
The results in columns 1 and 2 are consistent with the interpretation that higher ownership
concentration aggravates agency conflicts between shareholders and bondholders, and thus
leads to more covenant protection in the bond contract. Alternatively, one may interpret
15
the results as being driven by the persistence in bond covenants. Suppose that there are
two groups of firms initially. One group has more covenants, and the other group has less.
If covenants help to reduce the cost of financing, blockholders may prefer to hold stocks of
those firms that use more covenants. If the use of covenants is persistent, firms with more
covenants initially tend to have more covenants in their subsequent bond issues. Thus, firms
with a large (small) number of covenants will attract more (fewer) blockholders in subsequent
years when they offer new bonds. Pooling all years together for the two groups of firms, we
are likely to find a positive effect of blockholding on covenants. However, this positive effect
is mainly due to the persistence of covenants, rather than to the agency conflicts. To alleviate
this concern, we re-run the firm-year regressions by including only the first year’s information
for each firm, and present the estimation results in columns 3 and 4. The results of the pure
cross-sectional regressions confirm a significantly positive effect of blockholding on covenant
protection, lending additional support for the interpretation of agency conflicts.
5
Endogeneity
In this section, we examine whether the endogeneity of institutional blockholding would explain away our main results. The endogeneity issue arises if institutional blockholding is
simultaneously determined by bond covenants. This occurs when blockholders can anticipate
the covenant structure of a bond issue and make share-trading decisions by incorporating their
anticipation. The endogeneity concern is also present due to an omitted-variable problem.
Demiroglu and James (2010) show that the tightness of financial covenants is related to ex
post improvement in the borrower’s covenant variable and declines in investment spending and
net debt issuance. Their results suggest that firms use tight covenants to signal private information concerning their future performance. In the meantime, institutional blockholders, as
sophisticated investors, may also possess insider information regarding firms’ prospects. Thus,
unobserved information is an omitted variable that affects both blockholding and the tightness
of covenants, rendering institutional blockholding an endogenous explanatory variable.
We address the endogeneity problem in two ways. First, we examine whether the anticipation issue renders institutional blockholding endogenous. If blockholding is affected by
the anticipation of the structure of bond covenants, we should observe notable changes in
16
blockholding prior to bond offerings and expect the changes to explain the covenant structure
of the bond issues. In our sample, however, the change in blockholding from the second to
the first quarter before bond offerings has an average of only 0.21% for NL5 and 0.26% for
Top5. Comparing the sample average of 12% for NL5 and 21.3% for Top5, we see that the
changes in blockholding are insubstantial. In Table 7, we investigate whether the changes in
blockholding prior to bond offerings explain the restrictiveness of covenants. In all regressions,
the change in blockholding has an insignificant coefficient. The preceding results suggest that
blockholders are unlikely to trade shares in anticipation of the covenant structure of a future
bond issue.
Next, we run instrumental-variable regressions in Table 8 to further mitigate the endogeneity concern. We use stock liquidity as an instrument variable for institutional blockholding.
This can be justified on the basis of the following arguments. First, previous theoretical work
and empirical evidence show that liquidity has a direct effect on institutional blockholding.
Liquidity reduces transaction cost, making it easier for blockholders to accumulate or sell
large blocks of shares. In theoretical models of Maug (1998) and Edmans (2009), liquidity
induces the initial formation of block ownership. The empirical finding of Edmans, Fang, and
Zur (2013) shows that liquidity increases the probability of activist hedge funds acquiring a
large block of shares in a firm.
Second, liquidity is unlikely to have a direct effect on the covenants of debt contract. The
bond security is relatively illiquid in nature, so that bond contracting terms are unlikely determined by stock liquidity. Therefore, stock liquidity only affects the use of bond covenants
indirectly through its influence on shareholder governance. Maug (1998) demonstrates that
liquidity encourages blockholders to intervene because it enables them to purchase additional
shares at a price that does not reflect the gains from intervention. Thus, liquidity is good
for shareholder governance as it allows blockholders to benefit from monitoring through informed trading. Edmans (2009) and Edmans and Manso (2011) argue that liquidity induces
information acquisition and facilitates blockholder trading. Selling shares by blockholders
can exert downward pressure on the stock price, hurting the manager ex post if he has equity interest in the firm. Ex ante, liquidity is a shareholder governance that controls the
manager through the threat of blockholder exit. Although the specific mechanisms differ in
17
the previous two strands of research, a general conclusion is that liquidity has a beneficial
effect on shareholder governance. Enhanced shareholder control may aggravate bondholder
concerns on agency conflicts, which leads to more covenant protection in the bond contract.
Thus, liquidity affects the use of bond covenants only through the presence of institutional
blockholdings.
The above two properties of liquidity validates its candidacy for the instrument variable.
In this paper, we construct two empirical measures for liquidity. The first measure, turnover,
is defined as the ratio of the daily trading volume to shares outstanding, averaged over all
trading days in the first fiscal year prior to the bond offering date. A stock’s turnover should
be positively related to its liquidity. Following Bekaert, Harvey, and Lundblad (2007) and
Fang, Noe, and Tice (2009), we construct our second liquidity measure, non-zero returns,
as the proportion of non-zero daily returns in the first fiscal year before the offering date.
Lesmond, Ogden, and Trzcinka (1999) argue that if the value of an information signal is insufficient to outweigh transaction costs, market participants will choose not to trade, resulting
in an observed zero return. Thus, the proportion of non-zero returns is negatively related to
transaction costs and positively related to liquidity.
Using the two liquidity measures as instruments, we report the two-stage least square
regressions in Table 8. We present the first-stage regressions in columns 1 and 3, and the
second-stage regressions in columns 2 and 4. Both turnover and non-zero returns variables
have positive and significant effects on institutional blockholding. This suggests that stock
market liquidity reduces transaction costs, making it easier for blockholders to trade more
shares. In the second-stage regressions, we see that the coefficients on both blockholding
measures are still significantly positive. The effects of blockholding are even larger than those
observed in our baseline OLS regressions. Thus, endogeneity does not seem to explain away
our results, and the instrument-variable estimations even strengthen our main findings.
6
The Effect of Blockholding on Different Types of Covenants
In this section, we refine the covenant index variable and further examine the effect of blockholding on different types of covenants. The covenant index covers 15 different categories of
covenants, among which we focus further on two groups. The first group of covenant categories
18
include restrictions on payouts and investment activities, and the second group comprises restrictions on financing activities. Although the 15 covenant categories are all interrelated, the
correlations vary among different groups of covenant categories. For instance, Billett, King,
and Mauer (2007) find that positive correlations tend to be the largest within the broader
categories of payout restrictions, investment restrictions, and financing restrictions.
Payout restrictions contain two categories of covenants. They limit dividend payments of
the issuer or a subsidiary of the issuer, and confine the issuer’s freedom to repurchase shares.
Thus, covenants restricting payouts are designed to protect bondholders from possible wealth
transfer, and to help mitigate the underinvestment problem since a dividend restriction prohibits the distribution of free cash to shareholders (Myers, 1977; Nash, Netter, and Poulsen,
2003). Investment restrictions include three covenant categories. Investment policy restrictions prevent risky investments made by the issuer and/or subsidiary. Asset sale clauses limit
the issuer’s ability to sell assets or confine the issuer’s use of the proceeds from the sale of
assets. A merger restriction specifies that a consolidation or merger of the issuer with another firm is restricted. Thus restrictions on investment activities are put in place to alleviate
bondholder concerns with the asset substitution problem (Jensen and Meckling, 1976; Nash,
Netter, and Poulsen, 2003). In summary, by restricting payouts and investment activities,
the first group of covenant categories directly prevent bondholders from being expropriated
by shareholders.
The second group has seven covenant categories, restricting financing activities. The first
four categories limit the issuance of additional debt. They include restrictions on issuance of
any debt with initial maturity of one year or longer and restrictions on issuing additional subordinate, senior, and secured debt. The total leverage test clauses restrict total indebtedness
of the issuer and/or require the issuer to maintain a specified minimum net worth or ratio of
earning to fixed charges. A sale-leaseback covenant restricts the issuer and/or its subsidiary
from selling and simultaneously leasing back assets, where the lease agreement represents
senior liabilities to the firm. Finally, the stock issue restriction prohibits the issuer and/or
its subsidiary from issuing additional common or preferred stock. To summarize, financing
restrictions are mainly designed to mitigate claim dilution.
The preceding analysis shows that the two groups of covenants are related in different
19
ways to agency conflicts between bondholders and shareholders. Unlike financing restrictions,
payout and investment restrictions are designed to directly address shareholder-bondholder
conflicts. Thus, covenants restricting payouts and investment activities should depend more on
institutional blockholding, while financing-related covenants are less affected by concentrated
institutional ownership. To test this conjecture, we regress different groups of covenants on
institutional blockholding and firm-, issue-, and macro-level control variables. Table 9 present
the regression results. The dependent variables in the first two columns are the number of
covenants restricting payouts and investment, the dependent variables in columns 3 and 4 are
the number of financing-related covenants. As expected, both blockholding measures have
positive and significant effects on the number of covenants restricting payouts and investment
activities, while neither of them has a significant impact on the number of financing-related
covenants.
7
The Differential Effects of Institutional Blockholding
In the preceding baseline models, we investigated the effect of institutional blockholding on
the level of covenant protection for bondholders. In theory, two conflicting effects are at work.
On the one hand, institutional blockholders can effectively monitor the management and force
them to take value-enhancing actions (Barclay and Holderness, 1992; Huddart, 1993). Thus,
there is less need for bond covenants, because efficient monitoring provides shared benefits to
bondholders. On the other hand, concentrated outside ownership (blockholding) better aligns
the management to shareholders, which can aggravate the agency conflicts between shareholders and bondholders (Jensen and Meckling, 1976). Intensified agency concerns cause
bondholders to demand more protection via covenants. Our results so far show that institutional blockholding has a positive impact on the restrictiveness of bond covenants, suggesting
that ex ante, agency concerns dominate the shared-benefits effect. In this section, we further
examine whether agency concerns are the main channel through which institutional blockholding heightens the covenant protection. To this end, we classify institutional investors into
different types and investigate whether the positive effect of blockholding is more pronounced
among the type of institutions for whom the shareholder-bondholder conflicts are more likely
to occur ex-post.
20
7.1
The Effects of Institutional Blockholder Activism
If institutional blockholding affects the covenant protection mainly through the channel of
agency conflicts, then this effect should be stronger for more active institutional investors.
This is because active institutional blockholders are expected to be more engaged and involved in firms’ decision-making, presumably leading to more severe bondholder-shareholder
conflicts. To empirically test this hypothesis, we follow Brickley, Lease, and Smith (1988, 1994)
and Almazan, Hartzell, and Starks (2005) and classify institutions into two groups: potentially active investors and potentially passive investors. The active group includes investment
companies, independent investment advisors, corporate pension funds, public pension funds,
and university and foundation endowments; the passive group includes banks and insurance
companies. Detailed information on institutional investor type is provided by Professor Brian
Bushee.4
We then calculate our institutional blockholding measures separately among active and
passive institutional investors. Specifically, for all investors whose holdings are used to calculate the NL5 and Top5 blockholding measures, we split them into active and passive groups
and re-calculate the two blockholding measures among each group. Finally, we jointly examine the effects of blockholding among active and passive institutional investors and report
estimation results in Table 10.5 All specifications indicate that the effect of institutional
blockholding among active investors is positive and significant, while the effect of institutional blockholding among passive investors is either negative or insignificantly positive. The
signs of coefficients on all control variables remain the same as they are in the baseline regression models. These results suggest that bondholders are more concerned about lending
to firms with concentrated ownership among active blockholders, and thus, in equilibrium,
bondholders retain more control rights to protect their own interests.
4
More information on institution types is available at http://acct3.wharton.upenn.edu/faculty/bushee/.
We exclude the miscellaneous-group investors without any identifiable types. However, our results are not
sensitive to the treatment of the miscellaneous group. If we include blockholding by miscellaneous investors
as an additional control variable, our results are similar to those in Table 10. It is blockholding among active
investors that has a significantly positive and larger effect on covenant protection.
5
21
7.2
The Effects of Investors’ Investment Horizons
While institutional blockholding aligns the management to shareholders, we hypothesize that
the severity of potential conflicts between shareholders and bondholders depends on the investment horizons of institutional blockholders. To verify our hypothesis, we examine whether
institutional investors’ investment horizons have different effects on the severity of the agency
conflicts and, in turn, on the extent of covenant protection. Following Bushee (2001) and
Bushee and Noe (2000), we divide institutional investors into three groups: dedicated investors, quasi-indexers, and transient investors. We expect dedicated (transient) investors to
have long-term (short-term) investment horizons and label them as long-term (short-term)
investors.
Similar to what we did in Section 7.1, we calculate institutional blockholding measures
separately for long-term and short-term institutional investors and jointly examine their effects
on covenant protection. The estimation results are reported in Table 11.6 Across all regression
models, the effect of institutional blockholding among short-term investors is positive and
significant at the 1% level, while the effect of blockholding among long-term institutional
investors is mostly insignificant. Comparing the effects of both types of investors, we see that
the positive effect on the covenant protection is much larger for short-term investors than for
long-term investors. Thus, our findings imply that bondholders demand more control rights
when lending to firms with more concentrated short-term institutional investors. This may
result from the fact that short-term shareholders are more likely to engage in opportunistic
activities, such as taking on short-run and high-risk projects, which could be detrimental to
bondholders (Leland, 1994; Ju and Ou-Yang, 2006). Our findings also suggest that longterm relationships help to mitigate agency conflicts between bondholders and shareholders
(Anderson, Mansi, and Reeb, 2003; Shleifer and Vishny, 1997).
6
We exclude quasi-indexers in our analysis, because they are less likely to make independent holding decisions. The results are qualitatively the same if we include blockholding by quasi-indexers as an additional
control variable. We see that the effect of blockholding among short-term investors is significantly positive and
is also the largest if we further control for blockholding by quasi-indexers.
22
8
The Effects of Covenants on Yield Spreads
We have shown that more concentrated ownership intensifies bondholder concerns on agency
conflicts, leading to more covenants in bond contracts. Exposed to the potential agency
problem, bondholders can also charge a higher yield to compensate themselves when lending
to a firm. Since bond covenants are a mechanism to control the inherent agency problem in
lending, we would expect that the proper use of covenants can help attenuate the effect of
agency conflicts on bond yields. We examine the effect of covenants on yield spreads in the
last part of our empirical analysis.
We first investigate how blockholding affects the yield spread of a bond, where yield spread
is computed as the offering yield minus the yield of U.S. Treasury bonds with similar maturity. We regress yield spreads on institutional blockholding plus other firm-, issue-, and
macro-level control variables, and present the results in columns 1 and 4 of Table 12. The coefficients on both blockholding measures are positive, suggesting that the number of covenants
increases with intensified agency conflicts. We augment the first set of regressions with the
covenant index and its interaction with blockholding measures. The parameter of interest is
the coefficient on the interaction term, which captures the differential effect of blockholding
on covenants for bond issues with different number of covenants. We see in columns 2 and
5 that, for both blockholding measures, the coefficients on the interaction term are negative,
and the coefficient on Top5 ×Covenant is significant at the 5% level. These results suggest
that although bondholder concerns on agency conflicts increase the yield spreads, this effect
is smaller if more covenants are present in the bond contracts. In columns 3 and 6, we further
interact the product of blockholding and covenants with the investment-grade dummy. We
find that the effect of covenants on yields differs for bonds with different credit ratings. The
positive effect of blockholding on yield spreads is reduced by covenants, significantly more so
for investment-grade bonds than for non-investment-grade bonds. Overall, our results in Table 12 suggest that covenants help to alleviate the impact of agency conflicts on yield spreads,
and the beneficial effect of covenants is stronger for investment-grade bonds.
23
9
Conclusion
Bond covenants are designed ex ante to address ex post agency conflicts. In this paper,
we show empirically that the agency concern is significant in shaping the structure of bond
covenants. Larger institutional blockholdings align the management to shareholders, aggravating bondholder concerns on ex post agency conflicts. Thus, bondholders retain more contingent control rights by including more restrictive covenants in the bond contracts. Consistent
with this hypothesis, this paper’s empirical results show that more covenants are incorporated
in corporate bonds issued by firms with higher institutional blockholdings. This positive effect
is robust to different measures of blockholding and alternative regression models. Using stock
liquidity as an instrument, we find that the effect of blockholding is still positive and becomes
even stronger in two-stage least square regressions. We also find that the positive effect of
institutional blockholding on covenant protection is more pronounced for bond issuers whose
ownership is more concentrated among active and short-term investors. This bolsters our
conjecture that institutional blockholding increases covenant protection through the channel
of agency conflicts. While covenants enhance bondholder governance, they can also benefit
shareholders by reducing the firm’s borrowing cost. Consistent with this view, we find that
covenants help to reduce yield spreads that are related to bondholder concerns on agency
conflicts.
24
References
Aghion, P. and Bolton, P., 1992. An incomplete contracts approach to financial contracting.
The Review of Economic Studies 59 (3), pp. 473–494.
Almazan, A., Hartzell, J., and Starks, L., 2005. Active institutional shareholders and costs of
monitoring: Evidence from executive compensation. Financial Management 34 (4), 5–34.
Anderson, R.C., Mansi, S.A., and Reeb, D.M., 2003. Founding family ownership and the
agency cost of debt. Journal of Financial Economics 68 (2), 263 – 285.
Barclay, M.J. and Holderness, C.G., 1992. The law and large-block trades. Journal of Law
and Economics 35 (2), 265–294.
Bekaert, G., Harvey, C.R., and Lundblad, C., 2007. Liquidity and expected returns: Lessons
from emerging markets. Review of Financial Studies 20 (6), 1783–1831.
Bengtsson, O., 2011. Covenants in venture capital contracts. Management Science 57 (11),
1926–1943.
Bhojraj, S. and Sengupta, P., 2003. Effect of corporate governance on bond ratings and yields:
The role of institutional investors and outside directors. The Journal of Business 76 (3),
pp. 455–475.
Billett, M.T., King, T.H.D., and Mauer, D.C., 2007. Growth opportunities and the choice of
leverage, debt maturity, and covenants. The Journal of Finance 62 (2), pp. 697–730.
Brickley, J.A., Lease, R.C., and Smith, C.W., 1988. Ownership structure and voting on
antitakeover amendments. Journal of financial economics 20, 267–291.
Brickley, J.A., Lease, R.C., and Smith, C.W., 1994. Corporate voting: Evidence from charter
amendment proposals. Journal of Corporate Finance 1 (1), 5–31.
Bushee, B., 2001. Do institutional investors prefer near-term earnings over long-run value?
Contemporary Accounting Research 18 (2), 207–246.
Bushee, B. and Noe, C., 2000. Corporate disclosure practices, institutional investors, and
stock return volatility. Journal of Accounting Research 38, 171–202.
Chava, S., Kumar, P., and Warga, A., 2010. Managerial agency and bond covenants. Review
of Financial Studies 23 (3), 1120–1148.
Chava, S., Livdan, D., and Purnanandam, A., 2009. Do shareholder rights affect the cost of
bank loans? Review of Financial Studies 22 (8), 2973–3004.
Chava, S. and Roberts, M.R., 2008. How does financing impact investment? the role of debt
covenants. The Journal of Finance 63 (5), pp. 2085–2121.
Cremers, K.M., Nair, V.B., and Wei, C., 2007. Governance mechanisms and bond prices. The
Review of Financial Studies 20 (5), 1359–1388.
Demiroglu, C. and James, C.M., 2010. The information content of bank loan covenants.
Review of Financial Studies 23 (10), 3700–3737.
Edmans, A., 2009. Blockholder trading, market efficiency, and managerial myopia. The
Journal of Finance 64 (6), 2481–2513.
25
Edmans, A., Fang, V.W., and Zur, E., 2013. The effect of liquidity on governance. Review of
Financial Studies 26 (6), 1443–1482.
Edmans, A. and Manso, G., 2011. Governance through trading and intervention: A theory
of multiple blockholders. Review of Financial Studies 24 (7), 2395–2428.
Fang, V.W., Noe, T.H., and Tice, S., 2009. Stock market liquidity and firm value. Journal of
financial Economics 94 (1), 150–169.
Grossman, S. and Hart, O., 1980. Takeover bids, the free-rider problem and the theory of the
corporation. Bell Journal of Economics 11 (1), 42–64.
Huddart, S., 1993. The effect of a large shareholder on corporate value. Management Science
39 (11), 1407–1421.
Jensen, M. and Meckling, W., 1976. Theory of the firm: Managerial behavior, agency costs
and ownership structure. Journal of Financial Economics 3, 305–360.
Ju, N. and Ou-Yang, H., 2006. Asset Substitution and Underinvestment: A Dynamic View.
Working paper series, Duke University.
Klock, M.S., Mansi, S.A., and Maxwell, W.F., 2005. Does corporate governance matter to
bondholders? Journal of Financial and Quantitative Analysis 40 (04), 693–719.
Leland, H.E., 1994. Corporate debt value, bond covenants, and optimal capital structure.
The Journal of Finance 49 (4), pp. 1213–1252.
Lesmond, D.A., Ogden, J.P., and Trzcinka, C.A., 1999. A new estimate of transaction costs.
Review of Financial Studies 12 (5), 1113–1141.
Malitz, I., 1986. On financial contracting: The determinants of bond covenants. Financial
Management 15 (2), pp. 18–25.
Maug, E., 1998. Large shareholders as monitors: Is there a trade-off between liquidity and
control? The Journal of Finance 53 (1), 65–98.
Myers, S.C., 1977. Determinants of corporate borrowing. Journal of Financial Economics 5,
147–175.
Nash, R.C., Netter, J.M., and Poulsen, A.B., 2003. Determinants of contractual relations
between shareholders and bondholders: investment opportunities and restrictive covenants.
Journal of Corporate Finance 9, 201–232.
Nini, G., Smith, D.C., and Sufi, A., 2009. Creditor control rights and firm investment policy.
Journal of Financial Economics 92 (3), 400 – 420.
Nini, G., Smith, D.C., and Sufi, A., 2012. Creditor control rights, corporate governance, and
firm value. Review of Financial Studies 25 (6), 1713–1761.
Roberts, M.R. and Sufi, A., 2009. Control rights and capital structure: An empirical investigation. Journal of Finance 64 (4), 1657–1695.
Shleifer, A. and Vishny, R., 1997. A survey of corporate governance. Journal of Finance 42,
737–783.
Smith, C.W. and Warner, J.B., 1979. On financial contracting: An analysis of bond covenants.
Journal of Financial Economics 7 (2), 117 – 161.
26
600
400
200
0
Number of Bond Issues
800
Figure 1: The Number of Bond Issues across Time: 1979–2008
1980
1990
2000
2010
Calendar Year
5
4
3
2
Mean of Covnents Index
6
Figure 2: The Average Number of Covenants across Time: 1979–2008
1980
1990
2000
Calendar Year
27
2010
4.5
Mean Number of Covenants
5
5.5
Figure 3: The Number of Covenants and Institutional Blockholdings (NL5 )
0
2
4
6
8
Deciles of Percentage of Shares for All Investors
Holding More Than 5 Percent
10
4
Mean Number of Covenants
4.5
5
5.5
Figure 4: The Number of Covenants and Institutional Blockholdings (Top5 )
0
2
4
6
8
Deciles of Percentage of Shares Held by Top 5 Investors
28
10
Table 1: Bond Characteristics at Issuance
The sample consists of public bonds issued in the period 1979–2008 by U.S. firms that have information
on institutional ownership. The yield spread equals the offering yield of a corporate bond issue minus
the yield of U.S Treasury bonds of similar maturity. The investment-grade dummy variable equals 1
if the both Standard & Poor’s and Moody’s ratings for the bond are above the investment grade.
Mean
Median
P25
P75
7.27
1.75
378.52
12.66
0.609
0.089
0.172
7.14
1.15
200.00
10.00
1.000
0.000
0.000
6.03
0.66
125.00
6.75
0.000
0.000
0.000
8.60
2.10
377.38
15.00
1.000
0.000
0.000
Freq.
Pct.
Cum. Pct.
543
7386
2188
33
243
120
10513
5.17
70.26
20.81
0.31
2.31
1.14
100.00
5.17
75.42
96.23
96.55
98.86
100.00
Freq.
Pct.
Cum. Pct.
S&P non-investment grade
S&P investment grade
S&P not rated
2758
4910
2845
26.23
46.70
27.06
26.23
72.94
100.00
Moody’s non-investment grade
Moody’s investment grade
Moody’s not rated
2684
4863
2966
25.53
46.26
28.21
25.53
71.79
100.00
N
Mean
SD
8280
0.614
0.487
Offering yield (%)
Yield spread (%)
Offering amount (million)
Years to maturity
Callable bond dummy
Puttable bond dummy
Convertible bond dummy
Senior secured
Senior
Senior subordinate
Junior subordinate
Subordinate
None
Total
Investment grade dummy
29
Table 2: Summary statistics for variables in regressions
The sample consists of public bonds issued in the period 1979–2008 by U.S. firms that have information
on institutional ownership. Issuers’ accounting variables are measured at the end of first fiscal year
prior to the issuance date. The covenant index variable summarizes protection for bondholders in 15
categories, and The total possible number of covenants for a bond ranges from 0 to 15. The NL5
investors variable is the percentage of shares owned by all institutional blockholders, who hold no less
than 5% of total shares outstanding. The Top5 investors variable is the percentage of shares held
by the five largest institutional investors. Leverage is the book value of long-term debt divided by
the sum of the book value of long-term debt and the market value of equity, where the book value of
long-term debt is adjusted for changes in the long-term debt that take place between the end of Fiscal
Year −1 and the offering date. Credit premium is the difference in average yields of BBB and AAA
corporate bonds. The two blockholding measures and all accounting variables are winsorized at the
top and bottom 1 percentiles.
N
Mean
Median
SD
P25
P75
Covenants index
NL5 investors
Top5 investors
10513
10513
10513
4.778
0.120
0.213
4.000
0.083
0.205
2.561
0.129
0.112
3.000
0.000
0.137
6.000
0.189
0.283
Log sales
Leverage
Market-to-book assets
Short-term debt/total assets
EBIT/total assets
Fixed assets/total assets
7702
7522
7528
7697
7702
7667
7.607
0.328
1.840
0.047
0.077
0.399
7.818
0.296
1.458
0.025
0.084
0.371
1.980
0.215
1.150
0.062
0.093
0.250
6.331
0.156
1.175
0.005
0.048
0.183
9.088
0.469
1.997
0.063
0.123
0.594
Credit premium
10513
0.893
0.840
0.292
0.690
0.990
30
Table 3: The Effect of Blockholding by Investors Owning No Less Than 5%
The dependent variable is the bond covenant index summarizing protection for bondholders in 15
categories. The total possible number of covenants for a bond ranges from 0, representing the least
protection, to 15, representing the greatest protection. The NL5 investors variable is the percentage of
shares owned by all institutional blockholders, who hold no less than 5% of total shares outstanding.
Other explanatory variables are defined in Section 2. Robust standard errors are adjusted for firmlevel clustering, and the corresponding t-statistics are presented in parentheses. ∗∗∗ , ∗∗ and ∗ indicate
significance at 1%, 5%, and 10% levels, respectively.
(1)
NL5 investors
∗∗∗
3.113
(9.00)
(2)
∗∗∗
1.327
(4.03)
(3)
∗∗∗
1.436
(4.86)
(4)
(5)
∗
0.557
(1.87)
-0.068
(-0.17)
1.333∗∗
(2.18)
NL5×Investment grade
Log sales
-0.242∗∗∗
(-7.36)
-0.489∗∗∗
(-14.26)
-0.413∗∗∗
(-8.35)
-0.408∗∗∗
(-8.03)
Leverage
4.428∗∗∗
(16.29)
2.864∗∗∗
(11.27)
1.090∗∗∗
(3.77)
1.099∗∗∗
(3.79)
0.036
(0.89)
-0.042
(-1.12)
-0.062
(-1.42)
-0.061
(-1.38)
Short-term debt/total assets
-1.841∗∗∗
(-2.84)
-2.127∗∗∗
(-3.42)
-0.778
(-1.31)
-0.727
(-1.21)
EBIT/total assets
3.938∗∗∗
(6.58)
2.573∗∗∗
(5.74)
3.008∗∗∗
(4.59)
3.005∗∗∗
(4.57)
Fixed assets/total assets
-1.039∗∗∗
(-3.70)
-1.311∗∗∗
(-5.88)
-1.241∗∗∗
(-6.10)
-1.233∗∗∗
(-6.03)
Credit premium
-1.311∗∗∗
(-10.21)
-1.079∗∗∗
(-9.23)
-0.504∗∗∗
(-3.98)
-0.497∗∗∗
(-3.91)
Callable bond dummy
0.856∗∗∗
(10.23)
0.515∗∗∗
(5.48)
0.502∗∗∗
(5.27)
Puttable bond dummy
-0.286∗∗∗
(-3.04)
0.096
(0.78)
0.100
(0.81)
Convertible bond dummy
-2.692∗∗∗
(-29.40)
-3.582∗∗∗
(-28.69)
-3.580∗∗∗
(-28.76)
Log issue size
0.439∗∗∗
(8.16)
0.456∗∗∗
(5.17)
0.457∗∗∗
(4.99)
Maturity (in years)
-0.020∗∗∗
(-6.42)
-0.004
(-1.38)
-0.003
(-1.29)
-2.307∗∗∗
(-15.31)
-2.507∗∗∗
(-12.79)
Market-to-book assets
Investment grade dummy
Constant
4.405∗∗∗
(54.91)
6.433∗∗∗
(18.69)
3.875∗∗∗
(6.55)
4.728∗∗∗
(5.23)
4.780∗∗∗
(5.10)
Adj. R2
N obs.
0.02
10513
0.20
7473
0.39
7472
0.52
5865
0.52
5865
31
Table 4: The Effect of Blockholding by Top-5 Investors
The dependent variable is the bond covenant index summarizing protection for bondholders in 15
categories. The total possible number of covenants for a bond ranges from 0, representing the least
protection, to 15, representing the greatest protection. The Top5 investors variable is the percentage
of shares held by the five largest institutional investors. Other explanatory variables are defined in
Section 2. Robust standard errors are adjusted for firm-level clustering, and the corresponding tstatistics are presented in parentheses. ∗∗∗ , ∗∗ and ∗ indicate significance at 1%, 5%, and 10% levels,
respectively.
(1)
Top5 investors
2.880∗∗∗
(6.93)
(2)
(3)
(4)
(5)
1.687∗∗∗
(4.10)
1.815∗∗∗
(4.85)
0.632∗
(1.70)
-0.555
(-1.10)
2.418∗∗∗
(3.15)
Top5×Investment grade
Log sales
-0.247∗∗∗
(-7.55)
-0.494∗∗∗
(-14.50)
-0.414∗∗∗
(-8.39)
-0.406∗∗∗
(-7.86)
Leverage
4.449∗∗∗
(16.33)
2.889∗∗∗
(11.35)
1.097∗∗∗
(3.79)
1.098∗∗∗
(3.79)
0.038
(0.94)
-0.038
(-1.00)
-0.061
(-1.40)
-0.062
(-1.39)
Short-term debt/total assets
-1.742∗∗∗
(-2.68)
-2.030∗∗∗
(-3.25)
-0.749
(-1.27)
-0.665
(-1.11)
EBIT/total assets
3.928∗∗∗
(6.56)
2.553∗∗∗
(5.69)
3.001∗∗∗
(4.57)
3.030∗∗∗
(4.59)
Fixed assets/total assets
-1.025∗∗∗
(-3.68)
-1.302∗∗∗
(-5.88)
-1.239∗∗∗
(-6.10)
-1.214∗∗∗
(-5.94)
Credit premium
-1.311∗∗∗
(-10.26)
-1.078∗∗∗
(-9.27)
-0.508∗∗∗
(-4.02)
-0.501∗∗∗
(-3.95)
Callable bond dummy
0.858∗∗∗
(10.24)
0.516∗∗∗
(5.49)
0.496∗∗∗
(5.20)
Puttable bond dummy
-0.290∗∗∗
(-3.10)
0.096
(0.78)
0.101
(0.82)
Convertible bond dummy
-2.699∗∗∗
(-29.48)
-3.584∗∗∗
(-28.67)
-3.575∗∗∗
(-28.76)
Log issue size
0.433∗∗∗
(8.18)
0.456∗∗∗
(5.17)
0.458∗∗∗
(4.89)
Maturity (in years)
-0.020∗∗∗
(-6.42)
-0.003
(-1.36)
-0.003
(-1.20)
-2.313∗∗∗
(-15.36)
-2.913∗∗∗
(-10.67)
Market-to-book assets
Investment grade dummy
Constant
4.164∗∗∗
(35.28)
6.242∗∗∗
(17.72)
3.733∗∗∗
(6.34)
4.669∗∗∗
(5.15)
4.882∗∗∗
(5.04)
Adj. R2
N obs.
0.02
10513
0.20
7473
0.39
7472
0.52
5865
0.53
5865
32
Table 5: Results for Alternative Regression Models
The dependent variable is the bond covenant index summarizing protection for bondholders in 15
categories. The total possible number of covenants for a bond ranges from 0 to 15. The Herfindahl
index variable is equal to the sum of squares of institutional investors’ shareholding. The NL5 investors
variable is the percentage of shares owned by all institutional blockholders who hold no less than 5%
of total shares outstanding. The Top5 investors variable is the percentage of shares held by the five
largest institutional investors. Other explanatory variables are defined in Section 2. Robust standard
errors are adjusted for firm-level clustering. ∗∗∗ , ∗∗ and ∗ indicate significance at 1%, 5%, and 10%
levels, respectively.
OLS
(1)
Herfindahl index
Ordered Probit
(2)
(3)
10.231∗∗∗
(4.86)
0.759∗∗∗
(5.15)
NL5 investors
0.993∗∗∗
(5.29)
Top5 investors
Log sales
-0.491∗∗∗
(-14.18)
-0.217∗∗∗
(-12.61)
-0.219∗∗∗
(-12.80)
Leverage
2.892∗∗∗
(11.38)
1.154∗∗∗
(8.61)
1.166∗∗∗
(8.68)
-0.041
(-1.10)
-0.023
(-1.16)
-0.020
(-1.03)
Short-term debt/total assets
-2.136∗∗∗
(-3.43)
-0.999∗∗∗
(-3.15)
-0.943∗∗∗
(-2.96)
EBIT/total assets
2.523∗∗∗
(5.61)
1.281∗∗∗
(5.66)
1.270∗∗∗
(5.58)
Fixed assets/total assets
-1.315∗∗∗
(-5.92)
-0.660∗∗∗
(-5.75)
-0.654∗∗∗
(-5.73)
Credit premium
-1.088∗∗∗
(-9.32)
-0.577∗∗∗
(-9.06)
-0.577∗∗∗
(-9.08)
Callable bond dummy
0.861∗∗∗
(10.30)
0.374∗∗∗
(8.63)
0.375∗∗∗
(8.66)
Puttable bond dummy
-0.286∗∗∗
(-3.05)
-0.160∗∗∗
(-3.05)
-0.164∗∗∗
(-3.14)
Convertible bond dummy
-2.692∗∗∗
(-29.34)
-1.321∗∗∗
(-24.93)
-1.326∗∗∗
(-25.01)
Log issue size
0.436∗∗∗
(7.80)
0.223∗∗∗
(11.37)
0.219∗∗∗
(11.31)
Maturity (in years)
-0.020∗∗∗
(-6.43)
-0.009∗∗∗
(-5.22)
-0.009∗∗∗
(-5.19)
0.39
7472
7472
7472
Market-to-book assets
Adj. R2
N obs.
33
Table 6: Results for Firm-Level Regressions
Firm-level regressions are presented in the table. The dependent variable is the bond covenant index
summarizing protection for bondholders in 15 categories. The total possible number of covenants for
a bond ranges from 0 to 15. The NL5 investors variable is the percentage of shares owned by all
institutional blockholders who hold no less than 5% of total shares outstanding. The Top5 investors
variable is the percentage of shares held by the five largest institutional investors. The preceding three
variables are weight-averaged at the firm level for all bond issues corresponding to the same fiscal
year. The weight for a bond issue is its offering amount as a proportion of total amount offered in the
same fiscal year. The credit premium is the difference in average yields of BBB and AAA corporate
bonds, which is observed one month prior to the offering date. For all bonds issued in the same fiscal
year, the simple average of the credit premium is used in the firm-level regressions. Other firm-level
accounting variables are measured at the end of the first fiscal year prior to the offering date. Robust
standard errors are adjusted for firm-level clustering, and the corresponding t-statistics are presented
in parentheses. ∗∗∗ , ∗∗ and ∗ indicate significance at 1%, 5%, and 10% levels, respectively.
Firm-year OLS
(1)
NL5 investors
(2)
∗∗∗
First-year OLS
(3)
(4)
∗∗∗
2.707
(8.16)
1.420
(2.98)
4.741∗∗∗
(12.38)
Top5 investors
3.008∗∗∗
(5.66)
Log sales
-0.121∗∗∗
(-4.97)
-0.138∗∗∗
(-5.68)
-0.166∗∗∗
(-4.84)
-0.174∗∗∗
(-5.10)
Leverage
2.069∗∗∗
(9.30)
2.308∗∗∗
(10.36)
2.038∗∗∗
(6.93)
2.246∗∗∗
(7.53)
Market-to-book assets
-0.154∗∗∗
(-4.67)
-0.125∗∗∗
(-3.76)
-0.227∗∗∗
(-5.46)
-0.207∗∗∗
(-4.97)
-0.434
(-0.81)
-0.155
(-0.29)
-0.390
(-0.44)
-0.121
(-0.14)
2.438∗∗∗
(6.11)
2.484∗∗∗
(6.20)
1.819∗∗∗
(3.68)
1.802∗∗∗
(3.67)
-0.110
(-0.57)
-0.049
(-0.25)
0.129
(0.54)
0.189
(0.80)
Credit premium
-1.521∗∗∗
(-13.50)
-1.493∗∗∗
(-13.52)
-2.437∗∗∗
(-15.27)
-2.376∗∗∗
(-14.94)
Constant
5.780∗∗∗
(20.60)
5.073∗∗∗
(17.52)
7.079∗∗∗
(20.03)
6.531∗∗∗
(17.60)
Adj. R2
N obs.
0.12
5046
0.14
5046
0.17
1940
0.18
1940
Short-term debt/total assets
EBIT/total assets
Fixed assets/total assets
34
Table 7: Regressions of Covenants on Changes in Blockholding
The dependent variable is the bond covenant index summarizing protection for bondholders in 15
categories. The total possible number of covenants for a bond ranges from 0 to 15. The changes in
NL5 investors variable is quarterly changes prior to the offering date in the percentage of shares owned
by all institutional blockholders who hold no less than 5% of total shares outstanding. The changes
in Top5 investors variable is quarterly changes prior to the offering date in the percentage of shares
held by the five largest institutional investors. Other explanatory variables are defined in Section 2.
Robust standard errors are adjusted for firm-level clustering. ∗∗∗ , ∗∗ and ∗ indicate significance at 1%,
5%, and 10% levels, respectively.
Changes in NL5 investors
(1)
(2)
0.021
(0.05)
0.051
(0.14)
Changes in Top5 investors
Constant
4.762∗∗∗
(74.01)
-0.505∗∗∗
(-14.12)
2.933∗∗∗
(11.64)
-0.068∗
(-1.86)
-2.286∗∗∗
(-3.60)
2.696∗∗∗
(5.94)
-1.431∗∗∗
(-6.23)
-1.092∗∗∗
(-9.29)
0.846∗∗∗
(10.16)
-0.215∗∗
(-2.34)
-2.675∗∗∗
(-29.17)
0.458∗∗∗
(7.65)
-0.021∗∗∗
(-6.66)
4.047∗∗∗
(6.30)
Adj. R2
N obs.
0.00
10426
0.38
7443
Log sales
Adjusted leverage
Market-to-book assets
Short-term debt/total assets
EBIT/total assets
Fixed assets/total assets
Credit premium
Callable bond dummy
Puttable bond dummy
Convertible bond dummy
Log issue size
Maturity (in years)
35
(3)
(4)
0.127
(0.21)
4.761∗∗∗
(73.90)
0.198
(0.35)
-0.505∗∗∗
(-14.12)
2.933∗∗∗
(11.63)
-0.068∗
(-1.86)
-2.285∗∗∗
(-3.60)
2.692∗∗∗
(5.93)
-1.431∗∗∗
(-6.23)
-1.092∗∗∗
(-9.29)
0.846∗∗∗
(10.16)
-0.215∗∗
(-2.35)
-2.676∗∗∗
(-29.15)
0.458∗∗∗
(7.64)
-0.021∗∗∗
(-6.66)
4.049∗∗∗
(6.30)
0.00
10426
0.38
7443
Table 8: Results for Two-Stage Least Square Regressions
The dependent variable is the bond covenant index summarizing protection for bondholders in 15
categories. The total possible number of covenants for a bond ranges from 0 to 15. The NL5 investors
variable is the percentage of shares owned by all institutional blockholders who hold no less than 5% of
total shares outstanding. The Top5 investors variable is the percentage of shares held by the five largest
institutional investors. The two blockholding measures are instrumented by stock liquidity variables.
The turnover variable is calculated as the ratio of the daily trading volume to shares outstanding,
averaged over all trading days in the first fiscal year prior to the offering date. The non-zero returns
variable is defined as the fraction of non-zero-return trading days in the first fiscal year prior to the
offering date. Other explanatory variables are defined in Section 2. Robust standard errors are adjusted
for firm-level clustering, and the corresponding t-statistics are presented in parentheses. ∗∗∗ , ∗∗ and ∗
indicate significance at 1%, 5%, and 10% levels, respectively.
(1)
NL5
(2)
Covenant
(3)
Top5
3.182∗∗∗
(2.62)
NL5 investors
Top5 investors
Log sales
Leverage
Market-to-book assets
Short-term debt/total assets
EBIT/total assets
Fixed assets/total assets
Credit premium
Callable bond dummy
Puttable bond dummy
Convertible bond dummy
Log issue size
Maturity (in years)
Turnover
Non-zero returns
Constant
Adj. R2
N obs.
(4)
Covenant
-0.015∗∗∗
(-7.48)
0.093∗∗∗
(6.65)
-0.015∗∗∗
(-5.96)
-0.043
(-0.89)
0.115∗∗∗
(3.61)
-0.065∗∗∗
(-5.49)
-0.029∗∗∗
(-4.61)
-0.004
(-0.91)
0.030∗∗∗
(4.27)
0.003
(0.53)
0.003
(1.12)
-0.000∗∗∗
(-2.86)
2.949∗∗∗
(7.11)
0.328∗∗∗
(9.75)
-0.060∗
(-1.67)
-0.468∗∗∗
(-14.13)
2.787∗∗∗
(10.52)
-0.036
(-0.95)
-2.125∗∗∗
(-3.35)
2.574∗∗∗
(5.65)
-1.006∗∗∗
(-4.72)
-1.063∗∗∗
(-8.99)
0.841∗∗∗
(9.93)
-0.377∗∗∗
(-3.35)
-2.707∗∗∗
(-29.29)
0.417∗∗∗
(9.47)
-0.017∗∗∗
(-5.49)
3.618∗∗∗
(6.71)
-0.010∗∗∗
(-5.87)
0.068∗∗∗
(5.86)
-0.014∗∗∗
(-6.57)
-0.076∗
(-1.92)
0.111∗∗∗
(4.09)
-0.054∗∗∗
(-5.51)
-0.024∗∗∗
(-4.50)
-0.004
(-1.10)
0.023∗∗∗
(4.32)
0.006
(1.16)
0.004∗
(1.93)
-0.000∗∗∗
(-2.88)
2.624∗∗∗
(7.95)
0.315∗∗∗
(10.35)
0.001
(0.02)
0.14
7335
0.38
7335
0.16
7335
36
3.333∗∗∗
(2.58)
-0.482∗∗∗
(-14.96)
2.859∗∗∗
(11.08)
-0.034
(-0.90)
-2.013∗∗∗
(-3.10)
2.564∗∗∗
(5.61)
-1.033∗∗∗
(-4.91)
-1.072∗∗∗
(-9.15)
0.841∗∗∗
(9.97)
-0.356∗∗∗
(-3.31)
-2.714∗∗∗
(-29.49)
0.413∗∗∗
(9.21)
-0.017∗∗∗
(-5.57)
3.421∗∗∗
(5.97)
0.39
7335
Table 9: The Effect of Institutional Blockholding on Different Types of Covenants
Dependent variables are the number of covenants in different sub groups. The first group includes
covenants restricting payouts and investment activities, and the regression results are presented in
columns 1 and 2. The second group comprises covenants restricting financing activities, and the
regression results are reported in columns 3 and 4. The NL5 investors variable is the percentage of
shares owned by all institutional blockholders who hold no less than 5% of total shares outstanding.
The Top5 investors variable is the percentage of shares held by the five largest institutional investors.
Other explanatory variables are defined in Section 2. Robust standard errors are adjusted for firm-level
clustering. ∗∗∗ , ∗∗ and ∗ indicate significance at 1%, 5%, and 10% levels, respectively.
NL5 investors
Investment & Payout
Financing Covenants
(1)
(3)
(2)
∗∗∗
0.475
(3.48)
0.163
(1.11)
-0.178∗∗∗
(-7.08)
1.380∗∗∗
(12.19)
-0.015
(-0.81)
-0.416
(-1.38)
0.730∗∗∗
(2.85)
-0.523∗∗∗
(-4.74)
-0.390∗∗∗
(-7.12)
0.393∗∗∗
(10.87)
-0.042
(-0.87)
-0.858∗∗∗
(-18.53)
0.107∗
(1.82)
-0.009∗∗∗
(-6.20)
2.361∗∗∗
(4.17)
0.589∗∗∗
(3.31)
-0.180∗∗∗
(-7.19)
1.388∗∗∗
(12.26)
-0.014
(-0.74)
-0.385
(-1.28)
0.723∗∗∗
(2.83)
-0.521∗∗∗
(-4.77)
-0.390∗∗∗
(-7.15)
0.393∗∗∗
(10.87)
-0.042
(-0.89)
-0.860∗∗∗
(-18.55)
0.105∗
(1.80)
-0.009∗∗∗
(-6.26)
2.315∗∗∗
(4.14)
-0.114∗∗∗
(-5.68)
0.758∗∗∗
(6.33)
-0.016
(-0.75)
-0.514∗
(-1.74)
1.006∗∗∗
(4.13)
-0.398∗∗∗
(-3.95)
-0.538∗∗∗
(-9.37)
0.419∗∗∗
(9.59)
-0.297∗∗∗
(-5.80)
-1.951∗∗∗
(-45.23)
0.179∗∗∗
(5.21)
-0.003∗∗
(-2.10)
0.806∗∗
(2.42)
0.210
(1.17)
-0.114∗∗∗
(-5.70)
0.761∗∗∗
(6.35)
-0.015
(-0.73)
-0.502∗
(-1.70)
1.003∗∗∗
(4.11)
-0.396∗∗∗
(-3.92)
-0.538∗∗∗
(-9.39)
0.419∗∗∗
(9.60)
-0.297∗∗∗
(-5.80)
-1.952∗∗∗
(-45.29)
0.178∗∗∗
(5.14)
-0.003∗∗
(-2.08)
0.789∗∗
(2.38)
0.31
7472
0.31
7472
0.43
7472
0.43
7472
Top5 investors
Log sales
Leverage
Market-to-book assets
Short-term debt/total assets
EBIT/total assets
Fixed assets/total assets
Credit premium
Callable bond dummy
Puttable bond dummy
Convertible bond dummy
Log issue size
Maturity (in years)
Constant
Adj. R2
N obs.
(4)
37
Table 10: Investor Activism and the Effect of Blockholding
The dependent variable is the bond covenant index summarizing protection for bondholders in 15
categories. The total possible number of covenants for a bond ranges from 0, representing the least
protection, to 15, representing the greatest protection. The NL5 active (passive) investors variable
is the percentage of shares owned by active (passive) institutional blockholders who hold no less
than 5% of total shares outstanding. The Top5 active (passive) investors variable is the percentage
of shares held by active (passive) institutions who are among the five largest shareholders. Active
institutions include investment companies, independent investment advisors, corporate pension funds,
public pension funds, and university and foundation endowments; passive institutions include banks
and insurance companies. Other explanatory variables are defined in Section 2. Robust standard errors
are adjusted for firm-level clustering, and the corresponding t-statistics are presented in parentheses.
∗∗∗ ∗∗
,
and ∗ indicate significance at 1%, 5%, and 10% levels, respectively.
(1)
NL5 active investors
NL5 passive investors
∗∗∗
3.814
(9.77)
0.047
(0.12)
(2)
(3)
∗∗∗
1.613
(4.45)
0.085
(0.23)
Top5 passive investors
-0.241∗∗∗
(-7.25)
4.415∗∗∗
(16.22)
0.030
(0.73)
-1.736∗∗∗
(-2.68)
4.101∗∗∗
(6.85)
-1.065∗∗∗
(-3.76)
-1.311∗∗∗
(-10.21)
Constant
4.423∗∗∗
(56.76)
6.440∗∗∗
(18.53)
-0.487∗∗∗
(-14.28)
2.863∗∗∗
(11.25)
-0.046
(-1.22)
-2.009∗∗∗
(-3.20)
2.702∗∗∗
(5.99)
-1.336∗∗∗
(-5.95)
-1.090∗∗∗
(-9.30)
0.850∗∗∗
(10.11)
-0.291∗∗∗
(-3.08)
-2.677∗∗∗
(-28.91)
0.436∗∗∗
(8.42)
-0.020∗∗∗
(-6.35)
3.911∗∗∗
(6.81)
Adj. R2
N obs.
0.03
10435
0.20
7413
0.39
7412
Leverage
Market-to-book assets
Short-term debt/total assets
EBIT/total assets
Fixed assets/total assets
Credit premium
Callable bond dummy
Puttable bond dummy
Convertible bond dummy
Log issue size
Maturity (in years)
38
(5)
(6)
3.931∗∗∗
(8.70)
-1.676∗∗
(-2.38)
1.850∗∗∗
(4.23)
0.208
(0.42)
-0.242∗∗∗
(-7.03)
4.416∗∗∗
(16.13)
0.037
(0.90)
-1.606∗∗
(-2.50)
4.050∗∗∗
(6.75)
-1.034∗∗∗
(-3.63)
-1.258∗∗∗
(-9.78)
4.262∗∗∗
(38.17)
6.226∗∗∗
(17.18)
2.144∗∗∗
(5.76)
-0.186
(-0.43)
-0.485∗∗∗
(-14.49)
2.844∗∗∗
(11.16)
-0.039
(-1.05)
-1.824∗∗∗
(-2.93)
2.673∗∗∗
(5.96)
-1.315∗∗∗
(-5.91)
-1.041∗∗∗
(-8.90)
0.855∗∗∗
(10.24)
-0.288∗∗∗
(-3.09)
-2.710∗∗∗
(-29.47)
0.441∗∗∗
(9.10)
-0.020∗∗∗
(-6.40)
3.608∗∗∗
(6.55)
0.03
10486
0.20
7458
0.39
7457
1.690
(5.40)
0.045
(0.11)
Top5 active investors
Log sales
(4)
∗∗∗
Table 11: Investment Horizons and the Effect of Blockholding
The dependent variable is the bond covenant index summarizing protection for bondholders in 15
categories. The total possible number of covenants for a bond ranges from 0, representing the least
protection, to 15, representing the greatest protection. The NL5 long-term (short-term) investors
variable is the percentage of shares owned by long-term (short-term) institutional blockholders who
hold no less than 5% of total shares outstanding. The Top5 long-term (short-term) investors variable
is the percentage of shares owned by long-term (short-term) institutional blockholders who are among
the 5 largest shareholders. Long-term (short-term) investors are dedicated (transient) institutions
as defined in Bushee (2001) and Bushee and Noe (2000). Other explanatory variables are defined
in Section 2. Robust standard errors are adjusted for firm-level clustering, and the corresponding
t-statistics are presented in parentheses. ∗∗∗ , ∗∗ and ∗ indicate significance at 1%, 5%, and 10% levels,
respectively.
(1)
NL5 long-term investors
NL5 short-term investors
∗∗∗
1.463
(2.62)
7.724∗∗∗
(9.16)
(2)
(3)
0.607
(1.32)
2.987∗∗∗
(4.04)
0.656
(1.32)
3.158∗∗∗
(4.98)
Top5 long-term investors
Top5 short-term investors
-0.238∗∗∗
(-6.80)
4.611∗∗∗
(15.68)
0.018
(0.42)
-2.405∗∗∗
(-3.36)
3.998∗∗∗
(6.05)
-1.105∗∗∗
(-3.59)
-1.318∗∗∗
(-9.08)
Constant
4.521∗∗∗
(54.96)
6.493∗∗∗
(18.27)
-0.488∗∗∗
(-12.94)
2.991∗∗∗
(10.85)
-0.061
(-1.56)
-2.557∗∗∗
(-3.76)
2.597∗∗∗
(5.25)
-1.377∗∗∗
(-5.63)
-1.100∗∗∗
(-8.42)
0.873∗∗∗
(9.51)
-0.215∗∗
(-2.16)
-2.751∗∗∗
(-27.43)
0.448∗∗∗
(7.07)
-0.021∗∗∗
(-6.38)
3.886∗∗∗
(5.80)
Adj. R2
N obs.
0.03
8763
0.21
6191
0.40
6190
Log sales
Leverage
Market-to-book assets
Short-term debt/total assets
EBIT/total assets
Fixed assets/total assets
Credit premium
Callable bond dummy
Puttable bond dummy
Convertible bond dummy
Log issue size
Maturity (in years)
39
(4)
(5)
(6)
0.876∗
(1.79)
7.204∗∗∗
(8.70)
0.545
(1.14)
2.407∗∗∗
(3.12)
-0.255∗∗∗
(-7.32)
4.493∗∗∗
(16.23)
0.005
(0.13)
-1.891∗∗∗
(-2.71)
4.187∗∗∗
(6.53)
-0.936∗∗∗
(-3.24)
-1.200∗∗∗
(-8.62)
4.460∗∗∗
(49.86)
6.468∗∗∗
(18.78)
0.713
(1.44)
3.131∗∗∗
(4.71)
-0.497∗∗∗
(-14.00)
2.910∗∗∗
(11.22)
-0.068∗
(-1.82)
-2.119∗∗∗
(-3.22)
2.784∗∗∗
(5.81)
-1.221∗∗∗
(-5.42)
-0.979∗∗∗
(-7.61)
0.871∗∗∗
(10.09)
-0.177∗
(-1.92)
-2.814∗∗∗
(-29.62)
0.421∗∗∗
(7.98)
-0.020∗∗∗
(-6.67)
4.069∗∗∗
(7.05)
0.02
9453
0.20
6722
0.39
6721
Table 12: Bond Yields, Institutional Blockholding, and the Effect of Covenants
The dependent variable is the yield spread of a corporate bond issue, where the yield spread is defined
as the difference between the offering yield of the bond issue and the yield of U.S. Treasury bonds of
similar maturity. The NL5 investors variable is the percentage of shares owned by all institutional
blockholders who hold no less than 5% of total shares outstanding. The Top5 investors variable is the
percentage of shares held by the five largest institutional investors. The covenant variable summarizes
protection for bondholders in 15 categories, ranging from 0 to 15. Rating is a dummy variable that
equals 1 if both Standard & Poor’s and Moody’s ratings for the bond issue are above the investment
grade. Other explanatory variables are defined in Section 2. Robust standard errors are adjusted for
firm-level clustering, and the corresponding t-statistics are presented in parentheses. ∗∗∗ , ∗∗ and ∗
indicate significance at 1%, 5%, and 10% levels, respectively.
(1)
NL5 investors
∗∗∗
0.975
(3.21)
NL5×Covenant
(2)
(3)
∗
(4)
(5)
(6)
0.692∗
(1.82)
1.726∗∗
(2.12)
-0.321∗∗
(-1.97)
7.091∗∗∗
(7.40)
-0.837∗∗∗
(-4.35)
-0.813∗∗∗
(-11.35)
0.456∗∗∗
(8.63)
-0.245∗∗∗
(-6.48)
1.649∗∗∗
(6.10)
0.292∗∗∗
(5.17)
0.102
(0.23)
-5.088∗∗∗
(-5.63)
-0.300∗
(-1.90)
1.442∗∗∗
(12.40)
-0.027
(-0.52)
-0.465∗∗∗
(-3.30)
-1.530∗∗∗
(-7.59)
0.223∗∗∗
(2.99)
0.014∗∗∗
(7.42)
-3.019∗∗∗
(-3.99)
0.51
3557
∗∗∗
1.279
(1.88)
-0.152
(-1.16)
5.350
(6.52)
-0.482∗∗∗
(-3.22)
-0.777∗∗∗
(-8.26)
(NL5×Covenant)×Rating
Top5 investors
Top5×Covenant
(Top5×Covenant)×Rating
Covenant
-0.433∗∗∗
(-8.74)
Leverage
2.880∗∗∗
(11.31)
Market-to-book assets
0.310∗∗∗
(5.57)
Short-term debt/total assets -1.075∗∗
(-2.13)
EBIT/total assets
-4.501∗∗∗
(-5.84)
Fixed assets/total assets
-0.870∗∗∗
(-5.33)
Credit premium
1.080∗∗∗
(8.90)
Callable bond dummy
0.236∗∗∗
(4.21)
Puttable bond dummy
-0.753∗∗∗
(-4.92)
Convertible bond dummy
-1.664∗∗∗
(-9.52)
Log issue size
0.292∗∗
(2.54)
Maturity (in years)
0.003
(1.32)
Constant
0.033
(0.03)
Log sales
Adj. R2
N obs.
0.39
4242
0.260∗∗∗
(9.88)
-0.323∗∗∗
(-7.00)
2.335∗∗∗
(9.81)
0.299∗∗∗
(5.83)
-0.293
(-0.63)
-4.862∗∗∗
(-6.85)
-0.451∗∗∗
(-2.94)
1.352∗∗∗
(11.90)
0.157∗∗∗
(3.04)
-0.665∗∗∗
(-4.40)
-1.123∗∗∗
(-6.93)
0.190∗
(1.83)
0.007∗∗∗
(3.14)
-1.051
(-1.10)
0.319∗∗∗
(10.43)
-0.269∗∗∗
(-6.45)
1.939∗∗∗
(7.00)
0.319∗∗∗
(5.41)
-0.152
(-0.33)
-5.586∗∗∗
(-5.86)
-0.332∗∗
(-2.03)
1.462∗∗∗
(11.95)
0.008
(0.15)
-0.491∗∗∗
(-3.51)
-1.415∗∗∗
(-7.05)
0.208∗∗
(2.32)
0.012∗∗∗
(6.21)
-2.021∗∗
(-2.30)
-0.438∗∗∗
(-8.78)
2.904∗∗∗
(11.43)
0.308∗∗∗
(5.48)
-1.075∗∗
(-2.12)
-4.468∗∗∗
(-5.79)
-0.886∗∗∗
(-5.38)
1.076∗∗∗
(8.83)
0.239∗∗∗
(4.25)
-0.745∗∗∗
(-4.86)
-1.656∗∗∗
(-9.48)
0.293∗∗
(2.50)
0.003
(1.24)
0.018
(0.02)
0.313∗∗∗
(7.21)
-0.326∗∗∗
(-7.05)
2.338∗∗∗
(9.90)
0.293∗∗∗
(5.69)
-0.290
(-0.62)
-4.800∗∗∗
(-6.79)
-0.450∗∗∗
(-2.92)
1.354∗∗∗
(11.88)
0.161∗∗∗
(3.08)
-0.665∗∗∗
(-4.37)
-1.114∗∗∗
(-6.90)
0.191∗
(1.82)
0.007∗∗∗
(3.10)
-1.267
(-1.32)
0.46
4242
0.49
3557
0.39
4242
0.46
4242
40