Who Benefits from Bond Market Modernization?
DAVID MUSTO
University of Pennsylvania,
The Wharton School
& Securities and Exchange Commission
∗
JILLIAN POPADAK
Duke University,
Fuqua School of Business
ABSTRACT
We analyze the role modernization plays in shaping corporate bond market transactions. To identify
the effects of technological change, we use SEC eligibility cut-off rules that allow “Well-Known Seasoned Issuers” (WKSI) to accelerate issuance speeds. We find widespread adoption of incurrencebased covenants, which serve the economic purpose of substituting for investor diligence. Evidence
on covenant pricing and bondholding suggests the transformative power of technology mostly benefits those intended. Learning externalities from transitioning to covenant-heavy bonds coincide with
a contract uncertainty premium, which benefits sophisticated investors. Issuers bear some costs in
the transition but benefit from an expanded set of bondholders and lower volatility. Competition
prevents underwriters from profiting.
JEL classification: G3, G32, G2
Keywords: Raising Capital, Financial Innovation, Financial Intermediation, Fixed Income, Corporate Bonds, Contracting, Covenants, Primary Offerings, Sophisticated Investors, WKSI.
∗
Musto: The Wharton School, University of Pennsylvania, 3620 Locust Walk, Suite 3000, Philadelphia, PA
19104 (e-mail: musto@wharton.upenn.edu) and the Securities and Exchange Commission (SEC). Popadak: Duke
University, Fuqua School of Business, 100 Fuqua Drive, Durham, NC 27708 (e-mail: jillian.popadak@duke.edu). For
helpful comments, we thank Sudheer Chava, Adam Cohen, Kathleen Hanley, Campbell Harvey, Gregor Matvos,
Adriano Rampini, and seminar participants at the Napa Conference on Financial Markets, the Ohio State University,
and Washington University in St. Louis. Joshua Aronoff and Jacob Ledbetter provided research assistance. The
Securities and Exchange Commission disclaims responsibility for any private publication or statement by any of its
employees. This study expresses the authors’ views and does not necessarily reflect those of the Commission, the
Commissioners or other members of the staff.
The corporate bond market is the world’s largest and deepest source of capital for firms; it
facilitates capital provision for a diverse set of issuers and investment opportunities for a broad array
of investors. Because bond markets play such an important role in the financial landscape, enabling
their efficiency is critical. While the objective of a well-functioning bond market is clear, little is
known about how the transformative power of technology affects bond market transactions and
participants. Recent comments by the U.S. Securities and Exchange Commission (SEC) Chairman
suggest investors would widely benefit from adopting new technologies in the bond markets (White
(2014)). Yet theory suggests the benefits may not accrue to all investors because intermediaries
may leverage such innovations to their benefit at the expense of other market participants. We shed
light on this debate by analyzing a regulatory reform that modernized the bond issuance process
and allowed issuers to quickly bring new deals to the market. Using novel data that captures many
complexities of the bond market, we analyze the benefits and costs of transformation.
A rich theoretical literature provides intuition for which market frictions may interact with
changes to the bond issuance process as well as how such frictions affect efficiency. We focus
on the theoretical trade-offs influenced by accelerating the speed of issuance for each of the participants involved in the deal (i.e., issuers, investors, and underwriters). First, we consider the
issuers’ incentives. It is well-known that higher leverage distorts incentives, pushing management
from value maximization toward shareholder value maximization (e.g., Myers (1977), Smith and
Warner (1979)). Introducing debt contract terms such as restrictive covenants may ameliorate
bondholders’ concerns but only at the expense of issuers’ freedom to manage. Second, we consider
the underwriter’s incentives. Conducting due diligence on the issuers’ situation takes time. Legal
risk encourages underwriters to perform due diligence, but at the same time, demands for timely
delivery and pressures from competing underwriters may compress the time allotted to diligence in
order to preserve market share. Third, we consider the investors’ incentives. The bond market has
a free-rider problem where each buyer benefits from other buyers’ diligence. A potential remedy is
a boilerplate contract (Kahan and Klausner (1997)) which simplifies the necessary task, but rigid
contract terms impose costs when the optimal contract deviates substantially from the boilerplate.
The many frictions faced by bond market participants make it difficult to disentangle the costs
1
and benefits of modernization. Moreover, empirically establishing the effects of modernization on
bond market participants is difficult because modernization is not randomly assigned. We overcome
these challenges to identification by examining a particular type of modernization with eligibility
cut-off rules that deliver a quasi-natural experiment. So instead of using a randomization rule to
assign firms to treatment, we use the eligibility cut-offs to naturally assign firms to treatment as
if by randomization. Thus, we have a differences-in-difference estimator. The first difference is
time, pre-modernization vs. post, and the second difference is above the cut-off vs. below, which
differences out the time trends that affected both treated and control firms.
The modernization we exploit involves the SEC defining a special type of securities issuer, a
Well-Known Seasoned Issuer, or WKSI. By definition, WKSI eligibility requires a worldwide market
value of outstanding common equity of $700 million or more. Whether an issuer meets this cutoff is
precisely observable, and what the issuer gets for meeting it is immediate effectiveness of its registration statements. This immediacy from the new SEC rules allows an issuer to proceed as quickly
as it likes from filing its registration to completing its offering, and thus to reduce its investors’
opportunity for diligence substantially. Embracing the fact that technology has enabled high-speed
information sharing, under the new SEC rules, “access equals delivery” meaning traditional marketing documents such as a Prospectus that details the complex contractual arrangements need
only to be accessible online. This rule is therefore not only convenient for strong identification but
also potentially consequential for efficient contracting.
First, we address the effect of modernization on bond issuances. Because accelerated issuance
compresses investors’ time for scrutiny, and scrutiny helps investors gauge a bond’s risk and present
value, acceleration may endanger subscription. To test this logic, we examine the reforms effect on
offering quantities. Because deal volume affects a firm’s ability to finance operations and undertake
new investments, any impact of modernization on issuances is critical to the growth and stability
of the corporate sector. Second, we examine the reform’s effect on contract design. In particular,
we test whether issuers modify contracts to simplify their values across future states, and thus
make them easier to scrutinize on a tighter schedule. Using a new database assembled from SEC
filings, we find the reform facilitates faster issuance speeds, has no effect on the offering amount,
2
and meaningfully affects covenant use.
In particular, WKSI-eligible issuers adopt 2.8 more covenants, on average, relative to ineligible
issuers in the two years following the reform. The new covenants are incurrence-based covenants
that are traditionally thought to protect investors from events such as major asset sales or takeovers.
For example, the change-of-control covenant allows investors in bonds threatened by acquisition
to sell those bonds back to the issuer at 101% of par value. Our point estimates suggest there
was 42 percentage point increase in the use of change-of-control covenants among WKSI-eligible
issuers. Why incurrence-based covenants? Our evidence suggests issuers strengthened the terms of
their bond contracts because investors did not have time to evaluate the likelihood of such events
occurring when faster transactions began to occur. Put another way, the additional covenants serve
the significant economic purpose of substituting for investor diligence.
Second, we examine the pricing effects of the modernization-induced change in contractual
terms. The rapid onset of covenant use isolates changes in valuation with unusual precision. Because
many issuers floated bonds both shortly after and before the transition, it provides us with a set
of bonds from the same issuer, with similar maturity, with and without the contracting change.
The market’s appraisal of the effect of the change on bond value is apparent in the yield difference.
We focus our analysis on the change-of-control covenant, which is economically relevant (Billet et
al. (2007)) with well-established price effects (?) and substantial heterogeneity in the quality of
its underwriting (Credit Roundtable (2008)). We find the average effect of including a change-ofcontrol covenant is 23 basis points less yield, which is in the neighborhood of others’ findings and
indicative of a large impact on bond value. Closer examination, however, reveals that introducing
the covenant tends to increase the yield: we find a significant negative effect on value when an
issuer uses the covenant for the first time, when covenant-inclusion represents a large deviation
from recent contracting by other issuers, and when the issuance terms differ meaningfully from the
base prospectus. These findings suggest initially issuers bear some costs from the modernization.
Who invests in the bonds that are priced with the bargain covenant? Our evidence suggest
mutual funds are buying these bonds. And then, over the subsequent two quarters through secondary market trades, the bargain covenant becomes priced. The initial investors realize gains from
3
trade by selling these bonds, via small volume transactions, to subsequent investors. Overall, this
evolution increases the number of bondholders relative to bonds issued prior to the modernization
reform. It appears mutual funds were able to find the bargain investment, because of prior due diligence on the same issuer. This suggests other investors were free-riding off of the mutual funds’ due
diligence rather than underwriters forewarning their favorite clients about the bargain covenant.
Finally, volatility analyses suggest the modernization is associated with less volatile pricing once
investors learn about the change in contract design. These findings suggest in the longer-term
issuers benefit from the modernization.
Last, we address the effect of modernization on underwriters, as represented by their market
share and fees. On the one hand, underwriters face increasing legal danger when a deal goes south,
but at the same time, they are being encouraged by issuers and regulators to bring deals to market
at lightning speed – and if they turn down the business, they know another, less rigorous firm will
simply step in and execute it anyway. We find no evidence that underwriters are able to charge
higher fees for the timelier issuances. Further, we find no evidence that a single underwriter was
able to gain market share either by transaction count of transaction amount from introducing the
covenant-heavy contract. In fact within two quarters of the SEC reform, all underwriters began to
include the incurrence-based covenants. This suggests the underwriter market is highly competitive
and underwriters were not able to extract rents from the modernization.
Overall, our empirical findings related to the effect of modernization on the bond issuance
process highlight the importance of learning externalities. Our evidence is consistent with the
notion that repetition reduces the cost that investors must expend in learning the meaning of
contract terms. Initial investors after the modernization reforms may not have been aware of the
contract terms because the accelerated issuance speeds constrained their ability to scrutinize them.
Investors’ exposure to this contract uncertainty suggests investors should be compensated in terms
of risk and reward. From this viewpoint, the first issuers bear a type of compliance cost when
issuing post reform, but once the terms have become familiar through use over longer periods of
time, the standardization of terms suggest investors deserve less of an uncertainty premium. This
implies for all to benefit from modernization the costs of contract analysis must be reduced. One
4
potential solution is to use increasingly standardized contract terms as our analysis reveals, but the
externalities such a solution imposes should be carefully considered (e.g., less effective markets for
corporate control).
Our findings contribute several important empirical results relevant to both academic research
and policymakers. Our work complements work examining the effects of modernization on investors’ ability to hedge new equity issuance, issuer’s information disclosure and placement channel
(Fenn (2000); Hanley and Hoberg (2013); Gao and Ritter (2010)). By highlighting how learning externalities shape the pricing and inclusion of covenants, we contribute to the prior literature which
considers variations in agency costs (Kahan and Yermack (1998), Bradley and Roberts (2004)), path
dependence (Kahan and Klausner (1997); De Franco et al. (2014)), and the supply-side (Murfin
(2012)) as reasons for covenant restrictiveness. By emphasizing the time-series variation in covenant
pricing, we contribute to evidence on the price of covenants (Billet et al. (2007)). Our finding that
context-dependent decision-marking affects sophisticated investors shows recent work focusing on
unsophisticated investors generalizes (Barber and Odean (2008); Hastings and Shapiro (2013)).
Our evidence on bondholders’ due diligence efforts complements research examining bond market
clienteles’ preferences and portfolios having a real effect on prices (Becker and Ivashina (2014)).
More broadly, our research provides empirical evidence for competing theories of the economic
consequences of financial innovation (Lerner and Tufano (2011); Zingales (2015)).
Section I reviews the theory and evidence on bond market modernization and details the main
hypotheses. Section II outlines the SEC regulatory reforms. Section III details how the regulatory
reform is used to construct estimators that isolate the impact of bond market modernization.
Section IV presents the data. Section V describes the results. Section VI places the results in
context, and Section VII concludes.
I.
Theoretical Framework
Financial innovation, including both product and process innovation, has been one of the most
influential trends prevailing in international financial markets since the 1980s. There are a number
of surveys which document the emergence and growth of new financial markets and products such
5
as derivatives and securitized assets (e.g., Tufano (2003)). To a lesser extent, there exists research
covering financial process innovation such as the implementation of information technology into
financial transactions; the financial process innovation research, however, is primarily theoretical
rather than empirical. Hence, our empirical application provides an important testing ground for
competing theories.
When is financial process innovation beneficial and when is it not? To answer this question,
theory separates the rent-seeking components of financial innovation from those that benefit society. We categorize the economic consequences of financial innovation as follows: enhanced efficiency; exploitation of unsophisticated investors; exacerbation of managerial agency problems; and
rent-capture by financial intermediaries. We borrow from the economic theories underlying these
consequences to derive a set of null and alternative hypotheses that result from bond market modernization via accelerated issuance speeds. As emphasized by Lerner and Tufano (2011), we are
careful to consider the complexity of externality when evaluating the financial innovation’s welfare
implications.
Enhanced efficiency resulting from bond market modernization is our null hypothesis. In the
traditional view, efficiency gains from financial innovation are realized either through transaction
costs savings or its risk-reducing properties (Allen and Gale (1994)). In our setting, if before the
regulation to accelerate bond issuance speeds, bond contracts are too customized, and modernization leads to a more standardized contract with a higher usage and lower transaction costs, this
would be evidence of enhanced efficiency. The efficiency gains may also occur indirectly through
financial intermediaries. For example, if financial process innovation increases competition among
financial intermediaries, the economic surplus from such competition could accrue to issuers and/or
bondholders through firm’s ability to raise capital in larger amounts and at a lower cost than they
could otherwise and/or superior underwriting as a result of intermediaries competing for market
share. The timing component of our setting suggests a third type of efficiency gain – improved
market timing. In capital markets that are inefficient or segmented, market timing benefits firm’s
investors at the expense of entering and exiting investors in the market. In a survey of CFOs,
Graham and Harvey (2001) found a market-timing factor ranked above four traditional factors in
6
the decision to issue debt. In contrast to these traditional efficiency arguments, some theorists
demonstrate how with the additional assumption of belief disagreement among investors, financial innovation increases transaction costs, promotes excessive trading, and/or increases risk (e.g.,
Simsek (2013a); French (2008); Simsek (2013b)).
Exploitation of unsophisticated investors, primarily retail investors, is a popular perception of
a harmful consequence of financial innovation (Zingales (2015)) and is an alternative hypotheses to
efficiency gains. Such exploitation comes in both direct and indirect forms. First, investors may be
directly sold a bond they do not understand and would have never wanted had they understood
it. In our setting, if after modernization, investors buy a bond with contractual covenants that
they would not have wanted had they understood or had the time to understand, this would be
evidence of a detrimental effect of modernization. Second, if investors are attracted to bonds
that are very convenient for sophisticated investors, who buy the cheap version and disregard the
expensive version, but turn out to be costly for the unsophisticated investors. In our setting, if after
modernization, sophisticated investors buy bonds with “free” or mispriced contractual features but
sell them to retail investors at higher prices, this would be evidence of another detrimental effect
of modernization.
Free-riding by investors potentially amplifies any of the harmful consequences to unsophisticated
and/or time-constrained investors brought about by financial innovation. The free-rider problem
is simple: a bond is placed with many investors, each of whom benefits if others make the costly
effort to ensure that the bond is worth the offering price. If financial innovation leads to the
development of a new contractual term, the issuer can achieve a degree of customization that may
not be available with terms that recent issuers have used. In contrast, the use of a contractual term
which has been commonly used in the past can offer other benefits, because the effort potential
investors need to exert is more costly when the bond’s contractual features make it hard to compare.
This implies that as more issuers employ the same contract term, it becomes significantly more
attractive for comparative reasons. Such intuition has been formalized and documented by Kahan
and Klausner (1997). In our context, free-riding may reduce and delay the efficiency gains from
innovation, because at first, the additional requirement on investors to learn any new contract terms
7
may make them unwilling to subscribe to the bond offering or require issuers to yield a greater
return to potential investors in the form of a contract uncertainty premium. This is an important
alternative hypothesis that we test.
Because financial engineering provides an extremely flexible tool to exploit contractual relationships, managerial agency problems may be exacerbated by financial innovation and reduce the
aggregate efficiency gains. In our setting, the firm’s investors may be unaware that their agents, the
managers, are putting in place covenants such as anti-takeover provisions which insulate managers
from the market for corporate control. Billet et al. (2007) show the inclusion of a change-of-control
covenant, which is the dominant anti-takeover covenant in bonds, reduces the probability of becoming a takeover target by 50%. Because of the sizeable economic consequences of such entrenchment
for investors, this is a critical alternative hypotheses.
Rent-capture by financial intermediaries has both theoretical reasons and empirical evidence
to claim that its’ an important component of financial innovation that needs to be considered as
an alternative hypothesis. Gennaioli, Shleifer, and Vishny (2015) theorize that financial intermediaries cater to investor preferences by engineering securities perceived to be safe but exposed to
neglected risks. Because the risks are neglected, issuance volume is excessively high and intermediaries profit from such transactions. Similarly, Dang, Gorton, Holmstrom, and Ordonez (2014)
suggest intermediaries may leverage the timely information to their benefit at the expense of other
market participants. These rent-capture theories compete with more traditional reputation arguments which suggest financial firms’ value their reputations and intend to be infinite-lived, so any
innovation in underwriting or transactions will be value-added.
II.
SEC Regulatory Reform
In 2005, the SEC adopted a series of new rules that reformed the registration, communication
and public offering processes under the Securities Act of 1933. These new rules, which became
effective December 1, 2005, were meant to improve and modernize the securities issuance process,
and they serve as our source of identification. The regulatory change offers a unique opportunity
to study the implications of financial innovation, to the extent that the change had a large, sudden,
8
and direct effect on the speed at which new bonds were issued for a segment of issuers in the
corporate bond market.
The SEC reforms focus on modernizing the communications regulatory scheme applicable to
registered securities offerings. Eckbo, Masulis, and Norli (2007) provide an overview of these
reforms, which they describe as “sweeping changes” and “major innovation.” While the new rules
defined four types of issuers, our focus is on the creation of a new category of issuers called “well
known seasoned issuers” (WKSI). We focus on the WKSIs because they receive treatment from the
perspective of a statistical experiment meant to isolate the effect of innovation on the bond issuance
process. Before detailing the reforms aimed at modernization or treatment that the WKSIs are
subject to, it is important to be clear about what issuers meet the SEC definition of WKSI-eligible.
While the definition involves several components, the requirement of a worldwide market value of
outstanding common equity of $700 million or more is the binding constraint for the vast majority
of issuers.
A WKSI-eligible issuer is one that: (1) meets the registrant requirements of Form S-3 or Form F3, which include being current and timely in its reporting obligations under the Securities Exchange
Act of 1934; (2) as of a date within 60 days of the determination date1 , has either a worldwide
market value of outstanding voting and non-voting common equity held by non-affiliates of $700
million or more; or both (i) registered and issued at least $1 billion in aggregate principal amount of
non-convertible debt or preferred stock for cash, not exchange, during the past three years, and (ii)
will only offer non-convertible debt or preferred stock (unless the issuer is also eligible to register
a primary offering of its securities on Form S-3 or Form F-3); and (3) is not an “ineligible issuer2 ”
1
The determination date is the later of (i) the filing date of the issuers most recent shelf registration statement
(S-3); (ii) the date of the most recent amendment to the issuers shelf registration for purposes of satisfying Section
10(a)(3) of the Securities Act; and (iii) the filing date of the issuers most recent 10-K or 20-F (if the issuer has not
filed a shelf registration statement for the prior 16 months).
2
“Ineligible issue” includes issuers that (i) are not current in their Exchange Act reporting requirements; (ii)
are blank check companies, shell companies, penny stock issuers or limited partnerships offering other than through
a firm commitment underwriting; (iii) within the past three years, have filed for bankruptcy; (iv) within the past
three years, have been convicted of any felony or misdemeanor under certain provisions of the Exchange Act; (v)
within the past three years, were made the subject of any judicial or administrative decree or order arising out
of a governmental action that prohibits certain conduct regarding (including future violations of) the U.S. federal
securities laws, requires them to cease and desist from violating the anti-fraud provisions of the U.S. federal securities
laws or determines that they have violated those anti-fraud provisions; or (vi) within the past three years, have had
any registration statement subject to an SEC refusal order or stop order.
9
or “asset backed issuer.”
The treatment the WKSIs receive under the new rules is the flexibility to access the capital
markets at will relative to other issuers and reductions in required information about the issuance
prior to launching. In particular, WKSIs qualify for on-demand automatic shelf registration (ASR),
which means WKSIs register securities offerings on significantly simplified registration statement
that automatically and immediately becomes effective. Under the new Rule 462, an ASR and
any post-effective amendment to an ASR also becomes effective immediately upon filing with the
SEC. Moreover, revised Rule 415(a)(1)(x) permits primary offerings to occur immediately after
effectiveness of an ASR. Under new Rules 456 and 457, the WKSI need not pay filing fees upon
filing of the ASR, but rather can wait until the securities are “taken off of the shelf” and the actual
offering commences. As a result, WKSIs face neither the time constraints, the financial constraints,
nor the information risks associated with the SEC staffs review of a registration statement.
In addition, under the new Rule 430B, the ASR and any Prospectus filed in connection with
the ASR may omit a significant amount of information. Specifically, the WKSI may omit the
description of the securities including the amount offered, but not including the type of securities,
the distribution plan, any information that is unknown or not reasonably available (Rule 409), any
potential covenants, whether the issuance is a primary or secondary offering, the identity of other
issuers and the names and respective amounts of any selling securities holders. Rule 430B and Rule
512(a) allow the WKSI to include any omitted information in either a post-effective amendment
to the ASR, a prospectus supplement filed under Rule 424(b) or by incorporation by reference to
the WKSI’s other Exchange Act reports. Finally, Rule 413 permits WKSIs to add new classes
of securities and additional issuers to its ASR through a post-effective amendment. Together,
these changes eliminated the post-effective amendment requirements that obligated issuers to file
amendments related to any fundamental disclosure change. As a result, the reforms streamline the
issuance process for WKSIs.
Once the ASR is effective, which occurs immediately, the WKSI can sell its securities whenever
it wants. Depending on the WKSI’s view of the market, this could be immediately after filing, or it
could be months later. At the time of the actual offering when the WKSI launches the deal, bringing
10
the securities to the market, the WKSI must simply file a post-effective amendment or a Rule 424(b)
Prospectus supplement. The deadline for filing a Rule 424(b) Prospectuses is the second business
day following the earlier of the pricing date and the date first used after effectiveness for public
offerings or sales. In practice, the Preliminary Prospectus Supplement is typically filed the same day
that the takedown launches, while the Final Prospectus is filed the morning after pricing. Typically,
new issues trace the timeline outlined in Figure 1. As the figure shows potential investors, even if
they may hear rumors of a prospective offering, do not get the Preliminary Prospectus Supplement,
which is likely to include significantly more information such as the restrictive covenants, until just
fifteen minutes or so before the deadline for expressing interest to the bookbuilder. Diligence in so
little time would be significantly strained.
How different is a WKSI offering from what those issuers did previously? While many commentators have stated that the reforms were a major innovation and the timeline provides a sense
of the compressed timeframe investors’ face, an example helps solidify the differences. Figure 2
illustrates the issuance of a new bond by General Mills prior to the reform and Figure 3 illustrates
the issuance of a new bond by General Mills after the reform. The figures describe the information
contained in each SEC filing, what may change from one document to the next, as well as the length
of the document, the number of times a covenant is mentioned and the textual overlap between
the base Prospectus and the Final Prospectus Supplement. What is striking to note is that in the
post-reform era there are only minutes to review a document that is 87 pages long that has changed
more than 25% of the content in it including the covenants.
Specifically, what we see is that after the WKSI reform, the issuer registered its potential new
securities issuances using Form S-3/ASR, which does not require SEC review. The S-3/ASR usually
includes some type of base Prospectus, but this does not include covenant terms. The S-3/ASR
provides prospective buyers with basic firm information, risk factors, use of proceeds, and the plan
of distribution. Then, at any point thereafter even when the risk factors or business outlook in the
base Prospectus may have changed substantially, the WKSI issuer can go to market. When the
issuer goes to market that would be the first time potential buyers learn about the issuance. In
practice, rumors usually get out a few days before launch, or large bond investors have in-house
11
models to predict when a popular credit may go to market. On that offering date, under rule
424(b), the issuer submits a Preliminary Prospectus Supplement, which will have a description
of the covenants, updated risk factors, and firm financials, but the supplement often becomes
available at virtually the same time the new issuance is actually priced as indicated in Figure 1.
In our example, the Preliminary Prospectus Supplement was provided the same morning that the
new issuance is priced. Under Rules 164 and 433, the pricing and terms of the deal are recorded
and distributed in the form of a term sheet that is used to market the securities the day of the
launch and is called a Free-writing Prospectus (FWP), which is a short document that contains
no information about covenants. Then, within two business days of the offering date, the Final
Prospectus Supplement must be filed with all the details combined from the Free-writing Prospectus
and any potential changes from the Preliminary Prospectus Supplement if there was negotiations
for changes in contracts terms such as restrictive covenants between potential buyers and the issuer
prior to pricing.
In contrast to the post-reform era, our example WKSI-eligible issuer filed an S-3 shelf-registration
form that had pre-specified details about potential takedowns (i.e., the amount and the types of
issuance such as convertible debt, unsecured debt, etc . . . ). If those issuers need to increase the size
or plan of distribution, they would need to file a post-effective amendment to the registration and
wait for review, which typically takes about 48 hours. In our example, the issuer filed an S-3/A, and
this filing also documented a small change in covenants. Not until the post-effective amendment
is approved would it be a day when they could offer/price the bond. The S-3 shelf registration
statement includes a base Prospectus, but in contrast to the S-3/ASR base Prospectus, the S-3 base
Prospectus provide details about covenants and other terms ultimately contained in the indenture.
There is, therefore, never a need to file a Supplement to the base Prospectus/registration statement
at the time of issuance, because investors have had substantial time to scrutinize the terms. In
our example, investors have had more than one year to familiarize themselves with the terms of a
potential issuance from General Mills. Then, just like in the WKSI-era, in the pre-reform era, our
issuer documented the final version of the indenture including the pricing terms after the offering
is completed and accounting for any changes made as a result of negotiations with investors in a
12
Final Prospectus Supplement filed under Rule 424(b).
Why did the SEC even adopt the reforms in the first place? By integrating disclosure and
process under the Securities Act of 1933 and the Securities Exchange Act of 1934, the new rules
were meant to reduce the procedural aspects of securities offering and enhance capital formation
processes. While the SEC had circulated preliminary reform rules for commentary well before the
adoption, the final rules were adopted with several changes to address a number of points raised by
commenters. Hence, the exact date of the adoption and the details of the adoption are unlikely to
have been anticipated by market participants. Further, it seems plausible to believe that firms that
needed to raise capital in public debt markets in a timely manner would not have delayed raising
debt until the rules were in place. This suggests that the adoption of the rules did have a sudden
and direct effect on those that qualified, and those that qualified vs. those that did not qualify were
subject to similar pre-treatment trends from a statistical viewpoint. In the next section, we detail
how we use this SEC reform to isolate the effect of modernization on bond market participants.
III.
A.
Identification Strategy
Difference-in-differences Analyses
We analyze the role modernization plays in shaping corporate bond market transactions. To
empirically identify the effects of technological change, we use eligibility cut-off rules from the
SEC’s WKSI regulatory reform that allowed bond issuers to significantly accelerate issuance speeds.
Specifically, the WKSI issuers receive special filing exemptions that allow them to quickly go to
market. The thresholds for eligibility provide a unique natural experiment. The firms just below
the $700 million market value of common equity outstanding threshold remain subject to regulatory
delays while those just above the threshold are not. As long as the firms are statistically indistinguishable along other dimensions close to the thresholds, the firms that do not qualify for WKSI
status serve as an ideal counterfactual or control group to those treated firms that are subject to
the financial modernization. Hence, the regulatory reform provides a quasi-natural experiment in
which a difference-in-differences estimator captures the average treatment effect of the SEC reforms
13
aimed at bond market modernization.
In our differences-in-difference design, we compare treated issuers (i.e., those that are eligible for
WKSI) with a control issuers (i.e., those that are ineligible for WKSI) in the quarters pre-reform
and post-reform. Typically, one cannot observe the counterfactual of what the contract design
would have been absent treatment, but in this unique case, we are able to get a very good idea
of what it would have been. To ensure our control firms are similar to our treated firms, we limit
our sample to issuers near the $700 million threshold; doing so reduces the potential bias from
unobservable factors that might be correlated with the threshold and with outcomes of interest
such as pricing. Furthermore, because most firms visit the capital markets infrequently, firms are
not required to be in the sample for all time periods around the event.
We estimate the average treatment effect using the following issuer-panel regression:
Yjit = α + βEligiblei + δ (Eligible × P ost)it + γXjit + g (CEjit , T otjit ) + ft + εjit
(1)
where Yjit represents outcome variables of interest such as pricing, quantity, and risk-sharing for
issuance j by issuer i in quarter t, Eligible is an indicator for issuers that are eligible for WKSI
status, P ost is an indicator for the post-reform period, and Xjit is a vector of issuer and issuance
controls. We include time fixed effects, ft , to ensure that we estimate the impact of the regulatory
reform after controlling for any fixed differences in contracting practices across time. An indicator
for P ost is not included in the specification as it is collinear with the time fixed effects. To
account for potential covariance among issuer outcomes, we cluster the standard errors at the issuer
level. Finally, g (CEjit , T otjit ) is a flexible polynomial that is included in some specifications; the
polynomial controls for how far the treated and control issuers are from Eligible based on the
conditions of common equity market capitalization (CEjit ) and total offerings in the past three
years (T otjit ). The flexible polynomial is a more conservative approach to difference-in-difference
estimation, because it safeguards against comparing firms that are far from the qualifying threshold.
14
B.
Within-Issuer Analyses
The rapid modernization of the bond issuance process isolates changes in valuation and invest-
ment with unusual precision. Because many issuers floated bonds both shortly after and before the
transition, it provides us with a set of bonds from the same issuer, with similar maturity, with and
without any modernization-induced contractual changes. This byproduct of the reform allows us to
control for all issuer-specific variables by comparing the post-onset bond with a same-issuer, similarmaturity pre-onset bond. We use this within-issuer estimator to isolate different mechanisms via
which modernization impacts the market participants.
Our within-issuer sample focuses on WKSI-eligible issuers that changed contractual terms after
the reform. By collecting a sample that spans a 10-year horizon, we started with a sufficiently large
number of issuers that transitioned. The large sample size allows us to use a within estimator (i.e.,
fixed effects estimator) for our analyses. The within estimator measures the association between
individual-specific deviations of dependent and independent variables from their respective average
values over the observations. Because our issuer-specific pairs always involve two observations, the
dataset can also be viewed as a panel dataset with two periods. In this special case of k = 2, the
fixed effects estimator is statistically equivalent to a first difference estimator.
To examine the pricing of contractual changes, we estimate the following two observation issuerpanel regression:
Yjik = α + βCovjik + γXjik + fi + fk + εjik
(2)
where Yjik is the dependent variable for issuance j by issuer i at offering period k. Example outcome
variables include yield less treasury, underpricing, etc . . . Covjik is an indicator for a covenant of
interest. We include issuer fixed effects, offering year fixed effects, as well as a vector of issuance
controls Xjik , which includes credit rating, maturity, and offering amount.
To explore the heterogeneity underlying the subsequent pricing empirically, we return to the
first-difference estimator and focus on salience and novelty as potential sources of variation. Our
exact first-differences specification is as follows:
15
∆Yji = β∆Salientji + γ4Xji + ∆fk + ∆εji
(3)
where ∆Yji is the difference in the dependent variable for issuance j by issuer i with and without
the covenant. Salientji is an indicator for the covenant being the salient attribute relative to
pricing or offering amount. F irstji indicates when the covenant is the first time that issuer has
ever included such a covenant. Dissimilarji indicates the contract terms included at the time
of issuance had changed significantly from the base prospectus (defined as below median textual
similarity). Because of the differencing in this specification, we have effectively controlled for issuer
fixed effects, but we continue to control for offering year fixed effects. Further, we control for
differences in a vector of issuance controls Xji , which includes credit rating, maturity, offering
amount, and offering yield.
C.
Underwriter Analyses
While the difference-in-differences and within-issuer analyses allow us to estimate the effects of
modernization on issuers and investors, we are also interested in the effects of modernization on
underwriters. To examine the effects on underwriters, we estimate the following regression:
Yit = α + βU nderwritingQualityit + γXit + fi + ft + εit
(4)
where Yit is the dependent variable of interest for underwriter i in in quarter t. Example outcomes
include the number of new bond issuances, dollar volume of new bond issuances, market share, and
underwriting fees. U nderwritingQualityit is the key variable of interest and measures aspects of
the quality of the underwriting in the bond contract such as the inclusion of restrictive covenants,
changes from the base Prospectus to the final Prospectus, the quality of the legal language used in
the indenture as well as the time allowed for investor due diligence. We include underwriter fixed
effects, quarter fixed effects, as well as a vector of other issuance controls including credit rating,
maturity, offering amount, offering yield, and underwriter fees when appropriate.
16
IV.
Data
To examine the role modernization plays in shaping corporate bond market transactions, we
collect detailed information on firm characteristics of bond issuers, the contractual design of bond
issuances, the underwriters for the issuances, the quality of the underwriting, the pricing of the
bonds, the investors in the bonds, and the subsequent trading patterns of the bonds.
A.
Bond Issuers
We start with bond-level data from Mergent’s Fixed Income Securities Database (FISD) which
is a comprehensive database of publicly-offered U.S. bonds. FISD provides details on corporate debt
issues and the issuers. Mergent provides its’ own unique identifier for each issuer, the nine-character
Committee on Uniform Securities Identification Procedures (CUSIP) that uniquely identifies an
issuer and the type of security, the name of the issuer as recorded in the prospectus, and the
name of the issuer as reported to the CUSIP. Using the unique CUSIP, we cross-reference this
database with Thomson Reuters’ eMAXX database for U.S. corporate bond issuances as well as
FINRA’s Trade Reporting and Compliance Engine (TRACE) database. Approximately 90% of
bonds are consistently recorded across the three databases; for those with conflicting data, we
search on Bloomberg and SEC’s Edgar database to resolve discrepancies. Based on the available
issuer details such as CUSIP, ticker, and name from the bond-level dataset, we match the issuance
data with Compustat to obtain financial and accounting data for these firms. Further, we match
the data with corporate governance data from FactSet.
B.
Bond Contracts and Underwriting
While Mergent’s FISD is a relatively comprehensive database of corporate bond issuances, its’
coverage of bond contract features is incomplete. For example, less than 10% of the bonds in its
database have information on the covenants recorded. Even when the sample is limited to U.S.
corporate bonds with an investment grade rating issued after 2000, only 70% of the bonds in its
database have information on the covenants recorded. Further, the information provided in FISD
on the covenants is typically only a yes or no indicator for if the covenant was included in the final
17
contract rather than information about the quality of the covenant’s underwriting or key thresholds.
To overcome these data limitations, we web scraped the 424(b), S-3, S-3/A, S-3/ASR, FWP, and
POS AM filings associated with new bond issuances from the SEC Edgar website. The Securities
Exchange Act of 1933 requires issuers to file registration statements with the SEC before making
an issue available for purchase by the public. As is outlined in Section II, the SEC reforms aimed
at modernization altered the required form, timing of the filing, and type of information recorded
in the filings. As such, we analyze the text contained in these filings in multiple ways. First,
we obtain a simple count of the number of words in the file as well as the number of times the
word “covenant” is mentioned as proxies for complexity. Second, we analyze the quality of the
underwriting by examining specific phrases included by the underwriter in the contract. Third,
we compute the linguistic similarity between the information available in the base Prospectus,
which available at the time of pricing, and the Final Prospectus. Appendix B details the natural
language processing techniques used to compute the linguistic similarity. The intuition for the
similarity measure is that a score of 1 means the contracts are identical, a score of 0.75 would
indicate 75% overlap in contracts, and 0 would indicate no overlap in contracts.
To determine the quality of the underwriting, we follow legal precedents, which highlight two
fundamental issues in the design of a change-of-control covenant. The first is the extent to which the
covenant’s effectiveness depends on a credit rating change. Almost all investment grade bonds have
some ratings condition. There is variance as to whether the bond must (1) cross from investment
grade to high yield; (2) be high-yield; or (3) be lowered from investment grade to high-yield or
from high-yield to an even lower rated high-yield. The second fundamental design issue is what
defines a change-of-control event. There are five common triggers: (1) a person acquires more than
50% of the voting stock of the issuer; (2) a sale of substantially all assets; (3) a merger where
the issuer’s shareholders don’t continue to own a majority of the merged firm; (4) a continuing
directors test; and (5) adoption of a plan of liquidation or dissolution. Finally, there is also a
“holding company” exception that can affect the operation of these triggers. The holding company
exception nullifies some of the five common triggers, because in an acquisition of the issuer by
another public company, the merger transaction does not result in a surviving corporation that is
18
more than 50 percent owned by any single “person” or “group” then the trigger event legally did
not occur.
We focus on the change-of-control covenant’s underwriting because this covenant has important
economic consequences but also because it’s underwriting has been scrutinized at the highest level
of courts. Based on the court findings and best practices published by Credit Roundtable (2008),
there is solid legal precedent for determining the quality of the underwriting for this particular
covenant. For the covenant to have strong underwriting, it cannot be subject to the defective
holding company clause, it must have all five of the triggers, it must be dual triggered, and it
must not require rating verification. For the covenant to have moderate underwriting, it cannot
be subject to the defective holding company clause, it must have at least four of the triggers, it
must be dual triggered, but it can require rating verification. For the covenant to have moderate
underwriting, it can be subject to the defective holding company clause, it must have at least three
of the triggers, it must be dual triggered, but it can require rating verification. For the covenant
to have weak underwriting, it either has two or fewer triggers or it is not dual triggered.
C.
Bondholders and Bond Trades
The bondholding data for our analysis comes from Thomson Reuters’ eMAXX database. The
data provides comprehensive coverage of quarterly fixed income holdings by individual insurance
companies, mutual funds, pension funds, annuity funds, hedge funds, and other institution-like
investors (e.g., governments and hospitals). This allows us to construct a panel dataset with the
total par value held by these sophisticated investors; we also know the number of holding portfolios,
buying portfolios, and selling portfolios in any time period. With details on the other holdings by
these investors, we know their experience holding covenant-light vs. covenant-heavy bonds, holding
bonds when a firm becomes the target of a leveraged buyout (LBO), etc We combine the bondholder
information with trade information from FINRA’s TRACE database. This database provides intraday transaction level trades. We use this data to determine the liquidity, trading volume, return
volatility, and price paths for our sample of bonds.
19
V.
A.
Results
Difference-in-Differences Results
A causal interpretation for our difference-in-differences estimator is predicated on the assump-
tion that absent the SEC reform, the WKSI-eligible issuers would, on average, have had the same
changes in bond issuance outcomes such as changes in pricing, quantity issued, and contractual
terms as ineligible issuers. Hence, we begin by examining the actions of eligible and ineligible
issuers prior to the reform. Firms eligible for the WKSI status are very similar to our sample of
ineligible firms in the years before the SEC reform became effective. Table I reports the ex-ante
characteristics of WKSI-ineligible firms in column (1) and the ex-ante characteristics of WKSIeligible firms in column (2). The p-value from a t-test of differences in means is reported in column
(3). The two groups are similar in market-to-book ratio, profitability, ROE, investment intensity,
and recent growth in cash, assets, and revenue. Put another way, we are unable to reject the null
hypothesis that WKSI-eligible issuers are similar to ineligible issuers in all of these dimensions. This
evidence supports the interpretation that the effects of the SEC reforms aimed at modernization
are caused by changes in the nature of the bond issuance process, rather than other unobservable
contemporaneous changes differentially affecting control and treatment issuers. As expected, the
issuers are slightly different in terms of firm size. For this reason, we include additional controls
for distance from the qualifying size-based threshold to ensure issuers far from the threshold and
thereby the likely counterfactuals are not driving the result.
To estimate issuers responses to the SEC reforms aimed at modernization, we compare changes
in the WKSI-ineligible and WKSI-eligible issuers’ outcomes around the time of the reform. We
use the difference-in-differences estimation strategy described in Section III. Panel A of Table II
presents the effects of the SEC reform on the price and quantity of newly issued bonds. Because
accelerated issuance speeds shorten investors’ time for scrutiny, and scrutiny tends to make investors
more comfortable subscribing to an offering, the faster schedule may come at a cost of endangering
subscription. To test this logic, we first examine the reforms effect on offering prices and quantities.
Because deal volume and higher borrowing costs affect a firm’s ability to finance ongoing operations
20
and undertake new investments, these factors are critical to the growth and stability of the U.S.
corporate sector of the economy. Column (1) and (2) of Panel A show no evidence of an effect on
the offering amount or pricing of an issuance. Column (3) and (4) repeat the estimation but using
a four-year window on each side of the reform. If the effect of the reform took time to manifest
in the pricing and quantity, the longer would capture these results. Again, we find no evidence to
support pricing and quantity as the margins of adjustment for issuers after the reform.
Where we find evidence of adjustment after the reform is in contract design and risk-sharing
specifically. Panel B of Table II presents the effects of the SEC reform on contract design. We
test contract design, because a possible remedy for the threat to the price and quantity involves
changing the contract design to reduce the risk borne by the investor and thereby incentivize
the investor to be more comfortable with subscribing to the offering even with the faster schedule.
This hypothesis is consistent with the traditional view of financial innovation which proves efficiency
gains are realized either through risk-reducing properties (Allen and Gale (1994)). We find that,
on average, treated issuers statistically significantly increase covenant usage as a result of the SEC
regulatory reform. The estimates reported in Column (1) indicate that, after the SEC reform,
WKSI-qualifying issuers increased their use of restrictive covenants by adding, on average, 2.8
additional covenants relative to the issuers that did not qualify for WKSI status. Column (2)
through (5) show the specific covenants for which issuers increased their use. All of the covenants
are incurrence-based, restrictive covenants such as asset sale, change-of-control, fixed charge, and
negative pledge covenants.
Our point estimates in Panel B and C of Table II suggest the change-of-control covenant is
the most common covenant to be adopted as a result of the modernization event. We find WKSIeligible issuers, on average, increased change-of-control covenant use by 29% to 42% as a result of
the reform. The 29% point estimate is based on a four-year window around the reform whereas
the 42% estimate is based on the two-year window around the reform. Both point estimates are
significantly different from 0 at a 1% level. Such an increase in covenant usage is very economically
meaningful. In fact, prior to the reform, less than 10% of new issuances included the restrictive
change-of-control covenant but after the reform around 65% of new issuances included the covenant.
21
While the reform cannot explain all of the increase in restrictive covenant usage, it does explain a
meaningful portion of it.
Figure 4 illustrates visually the evidence supporting the point estimates reported in Table II
related to risk-sharing for the WKSI-eligible issuers after the SEC regulatory reform. Plot (1), the
upper left plot, illustrates an increase in total number of covenants included in a bond indenture
for the treated issuers. Plot (2) illustrates the percentage point increase in use of asset sale clauses.
Plot (3) illustrates the percentage point increase in change-of-control covenants. Plot (4) illustrates
the percentage point increase in fixed charge covenants. In each plot, the navy side represents the
8 quarters before the reform, the reform is demarcated by a black vertical line, the maroon side
represents the 8 quarters after the reform, and ninety-percent confidence intervals are shaded.
What the plots reveal is that the number of covenants increased by 2 covenants, an additional
25% of bonds had asset sale covenants, an additional 30% had change-of-control covenants, and an
addition 12% had fixed charge covenants.
Overall, the plots in Figure 4 provide suggestive evidence of the mechanism driving the effect
of the SEC reform on covenants. Because bonds are issued quickly under the reform, issuers
and their investment bank advisors may want to offset the effect of the shorter schedule on the
difficulty of selling the bonds by adding contractual terms that make the bond’s future payoffs
simpler. Arguably, the change-of-control covenant makes the future payoffs simpler because a
bond that pays 101 in specific future states is simpler than a bond with unspecified value in those
states. Similarly, asset sale covenants protect bondholders from potential sharp changes in price
attributable to major asset sales.
As a final check of the difference-in-differences results, we examine the identifying assumption
that the treatment and control group would have evolved along parallel paths had it not been for
the SEC reform. One way to assess the validity of such an assumption is to examine the data
to determine if the issuers with varying status post reform displayed similar trends prior to the
reform. Figure 5 illustrates that the issuers did indeed display parallel trends prior to the reform,
but post reform the trends diverged. Because there does not appear to be a violation of the
parallel trends assumption and our results are robust to the inclusion of a flexible polynomial that
22
enables comparison of the most similar issuers, it is plausible the identifying assumptions for our
difference-in-differences research design are satisfied.
B.
Within-Issuer Results
Covenant inclusion should be priced by the market. Prior research suggests that covenants that
restrict managements’ actions by protecting bondholders’ interests should benefit management in
terms of lowering borrowing costs. For this reason, we study a set of issuers that floated bonds both
shortly after and before the reform. Doing so provides us with a set of bonds from the same issuer,
with similar maturity, with and without the contractual change. Because we want to benchmark
any price changes to prior pricing studies, we focus on the change-of-control covenant, which first
became popular in the 1980s and consequently has been studied.
Table III summarizes the characteristics of our bond issuers and their issuances that had similar
maturities but one bond had the contractual change and one did not. We observe that bonds with
the restrictive covenant have similar offering amounts and yields, ratings at issuance, and gross
spread. As expected, the time to evaluate the contract is significantly shorter for the issuances
with the restrictive covenants at only 1.4 hours on average or 0.6 hours at the median. The average
issuance with a change of control covenant has 3.3 of the 5 triggers, but 48% of issuances include the
holding company exception that can prohibit the effective operation of the triggers. The majority of
issuances require at least two agencies to publicly verify the cause of the downgrade is attributable
to the change-of-control event. Our text-based analysis of the bond contracts show that the bonds
with the covenants have slightly longer contracts and unsurprisingly, the word covenant occurs
more frequently.
Table IV shows the results from examining the initial pricing for the issuer-specific bond pairs, in
which issuers floated bonds both before and after the change, and thus had concurrently-outstanding
bonds both with and without the covenant. Column (1) of Panel A reveals that on average, all
else equal, including the covenant reduces the issuers’ yield by 23 basis points. This finding is
statistically significant and robust to the inclusion of issuance controls, issuer fixed effects, and
year fixed effects. While this suggests issuers are being compensated from the new contractual
23
arrangement, the constraints on investors time to evaluate the offering terms at issuance suggest
this average may belie substantial heterogeneity.
To explore the heterogeneity in bond pricing, we explore price-heterogeneity that stems from
context-dependent decision-making (McFadden (2001)). We examine three measures that capture
different features of context-dependent decision-making: (1) boilerplates (i.e., the information-based
case), (2) saliency (i.e., the behavioral-based case), and (3) dissimilarity (i.e., time-based case). We
want to account for the potential costs borne by issuers and investors when an issuance breaks
the boilerplate contract to adopt the change-of-control covenant, its’ use of a change-of-control
covenant is particularly salient relative to recent issuances, or the contractual details of its issuance
differ substantially from the base prospectus provided to investors in the initial S-3 filing.
The logic for the boilerplate test of context-dependent decision-making is as follows. Because any
one investor gains little from evaluating the precise terms of an offering, the disincentive for investor
due diligence promotes free-riding and the reuse of previous deals’ terms. Underwriters, therefore,
are unlikely to benefit either in terms of profit or market-share from customizing bond contracts for
the exact risks of an issuer. This practice of boilerplate contracting or sticky covenants (Kahan and
Klausner (1997), De Franco et al. (2014)) is largely borne out in the summary statistics, where we
find 80% of the content from the most recent bond issuance is retained in the next bond issuance.
This is even more remarkable when one considers the fact that issuers often change underwriters
from one issuance to the next. We specifically examine the within-issuer pairs and see if when an
issuers’ issuance first includes the change-of-control covenant, if the pricing is correct. Further, if
the pricing seems out-of-line with the risk-return profile it generates, we examine whether issuers
or investors are amassing the benefits of the mispricing.
The logic for the saliency test of context-dependent decision-making is as follows. A buyer exhibits a context effect if his choice between two alternatives systematically depends on the presence
of other options. In a recent study, Hastings and Shapiro (2013) provide large-scale evidence of
context-dependent decision-making among retail consumers in gasoline markets while Barber and
Odean (2008) find evidence of context-dependent decision-making among retail traders in equity
markets. In contrast to the boilerplate example, context-dependent choice arises in this case when
24
a buyer’s attention is drawn to salient attributes of goods, such as quality or price (Bordalo, Gennaioli, Shleifer (2012); Bordalo, Gennaioli, Shleifer (2013)). An attribute is defined as salient for a
good when it stands out among the good’s attributes relative to that attribute’s average level in
the choice context. The main prediction is that buyers attach disproportionately high weight to
salient attributes, and their choices are tilted toward goods with higher quality/price ratios; they
are bargain-hunting based on the most salient attribute. We specifically examine the within-issuer
pairs and see if when an issuers’ issuance includes a covenant that is salient relative to offering yield
and quantity, if the initial secondary market pricing is correct. Further, if the pricing seems out-ofline with the risk-return profile it generates, we examine whether issuers or investors are amassing
the benefits of the mispricing. The bargaining-hunting motive supported by theory suggests the
primary alternative hypothesis is that investors benefit.
Columns (2) – (4) of Table IV present tests for price heterogeneity in the initial pricing for the
issuer-specific bond pairs. The specification is the same as in Eq.(3) and includes indicator variables
for if the issuance is the first by that issuer with the covenant or if the covenant is the salient
attribute when the covenant is included. We find that significant heterogeneity underlies the initial
pricing of the covenant. Our evidence suggests initial investors benefit from the modernization
in terms of a free or bargain covenant and issuers suffer because they are not fully compensated
in terms of yield reduction. In particular, the observed heterogeneity suggests even sophisticated
institutional bond investors do not take into account fully all the information available to them in
the bond contract, but overemphasize the information their minds focus on. Columns (2) – (4)
provide statistically significant evidence to suggest when the covenant is the most salient attribute
relevant to offering amount and pricing, issuers receive less of a reduction in yield. Similarly, when
an issuer first includes such a covenant and breaks the boilerplate, the issuer receives less of a
reduction in yield. Finally, when the final contract terms are dissimilar to the base contract terms
the covenants are priced at bargain levels.
How long does it take for bond traders to arbitrage away the price inconsistency? Figure 6
examines the persistence of the pricing heterogeneity over time. The y-axis plots the reduction
in yield accrued to the issuer and the x-axis plots the quarters since issuance. The left-hand side
25
plot illustrates the pricing differential when covenants are the salient attribute; it reveals that
those issuances trade with a sizeable 20 basis point yield differential for at least a year following
the offering slowly mitigating three years later. The right-hand side plot illustrates the pricing
differential when an issuer first includes such a covenant; it reveals that those issuances do not
trade with a sizeable differential, which indicates they more quickly arbitrage back to a yield
indistinguishable from other issuances. This suggests that while the modernization reform initially
brought benefits to investors, those benefits quickly dissipated.
C.
Bondholder and Underwriter Results
Who were the investors that initially benefited from the bond market modernization? In par-
ticular, was there one group of bond investors performing their due diligence while others free-ride?
Were underwriters forewarning their favorite institutional clients in hopes of gaining future business or market share? Or were the benefits to investors spread evenly across different groups of
investors? We explore the answers to these questions both graphically and statistically.
Table V provides descriptive statistics of the bondholding data for our study. The data comes
from Thomson Reuters’ eMAXX database and provides comprehensive coverage of quarterly fixed
income holdings by individual insurance companies, mutual funds, pension funds, annuity funds,
hedge funds, and other institution-like investors (e.g., governments and hospitals). This allows us
to construct a panel dataset with the total par value held by these sophisticated investors; we also
know the number of holding portfolios, buying portfolios, and selling portfolios in any time period.
With details on the other holdings by these investors, we know their experience holding covenantlight vs. covenant-heavy bonds, holding bonds when a firm becomes the target of a leveraged
buyout (LBO), etc . . . . Using the bondholder data, Figure 7 examines which investors initially
invested in the covenant-heavy bond contracts with the favorable covenant pricing. The figure
suggests institutional investors, and mutual funds in particular, buy the bonds that are priced as
though they have a free covenant.
Table VI provides statistical evidence about the bondholders and bond trades that eventually
lead to the covenant pricing. Turnover in ownership shows the initial mutual fund investors realize
26
gains from trade by selling these bonds. Overall, this evolution increases the number of bondholders
relative to the covenant-light bonds. Volatility analyses suggest the covenant usage and dispersion
in ownership are associated with less volatile pricing once learning occurs and this price stability
benefits the bond issuers. Panel B, C, and D of Table VI focus specifically on the issuances with
“bargain” covenants. There do appear to be some differences across these attributes. What we
consistently see is that mutual funds are the ones who initial bought the bargain covenants and
sold the bonds as the covenant became priced in the market.
Table VII delves into why a difference exists in trading and bondholding behavior between the
bargain and priced covenants. Specifically, we examine the role of bondholder characteristics in
terms of ability to perform due diligence and prior issuer-specific and LBO experience as potential
explanations. Tests of differences across the two types of pricing reveals that despite the shorter
time for diligence, many of the mutual fund investors had prior issuer experience which helped
facilitate evaluation of the contractual terms. Surprisingly, prior LBO experience does not seem
to affect subsequent bondholding suggesting that the specific form of the covenant is not what
mattered but rather that the covenant serves to simplify due diligence by protecting downside risk
in highly idiosyncratic events.
Were underwriters faced with increasing pressure to deliver bonds at fast speeds reducing the
quality of their underwriter? We explore this question both visually and statistically. Figure 8
examines which underwriters use covenant-heavy bond contracts after the modernization reform.
The graph shows that all underwriters quickly adopted the new covenant-heavy contracts. Additional statistical analyses of the underwriters’ market share pre and post the reform show suggest
the underwriting market for corporate bonds is quite competitive as no firm was able to achieve a
statistically significant gain in market share by three quarters after the reform.
Table VIII explores the extent to which underwriters benefited from the SEC reforms that
modernized the bond issuance process. The specification is as in Eq.(4). The dependent variable
looks at various dimensions on which underwriters could profit, including new issuance count and
dollar volume, as well as market share and fees. We find no evidence that underwriters benefit
from the modernization. The underwriters that protect themselves from legal liability by writing
27
stronger contracts with more restrictive covenants appear to lose a small amount of business and
see no changes in their fees. When focuses on the bond issuances with restrictive covenant packages,
underwriters that provide investors with greater time for scrutiny appear to receive smaller fees and
less market share. While strong underwriting in terms of legal language leads to some additional
business, it does not appear to be enough to offset the reduction in business from simply including
the restrictive covenant. Overall, these results suggest that underwriters do not benefit from bond
market modernization.
VI.
Discussion
Our results converge to support the traditional economic efficiency hypothesis that financial innovation generates gains through transaction costs savings (e.g., improved market-timing for firms
issuing public debt) and risk-reducing properties (e.g., improved underwriting of bond contracts).
Because the U.S. corporate bond market operates within a structure” of rules, technology, market
practices, and other constraints, it is important to recognize that any incremental innovation produces a complex interaction among regulation and other factors like competition. And that these
incremental changes can produce significant, sometimes unintended economic consequences that
may not become evident for a period of years or even decades.
The complex interactions resulting from the U.S. corporate bond market structure create the
unique and surprising findings from our analyses. First, we find widespread adoption of incurrence
covenants such as change-of-control covenants. While at first glance, this could appear to be the
traditional risk-reduction property, we show these covenants serve a new purpose of simplifying due
diligence in addition to the traditional shifting of risk purpose. Second, we delve into the theories
that distinguish the rent-seeking components of financial innovation from those that benefit society.
We find the costs and benefits of corporate bond market modernization are not evenly distributed
across market participants. Learning externalities from transitioning to covenant-heavy bonds coincide with a contract uncertainty premium, which benefits sophisticated investors at the expense
of subsequent investors. Issuers bear some costs, because they are not fully compensated for the
managerial control relinquished to bondholders through the covenants. Issuers, however, benefit
28
from an expanded set of bondholders and lower bond return volatility from the more standardized contract. We find no evidence to support rent-seeking activities by financial intermediaries.
One interpretation of this finding is competition prevents underwriters from profiting; a second
interpretation is reputational concerns prohibit rent-seeking by underwriters.
As with any natural empirical experiment, there could be concerns that the results may not
extrapolate to other contexts. Accelerating the speed of bond issuances is a very specific type of
financial innovation. It is plausible that other types of financial innovation proposed by the SEC
that focus on transparency, trading speed, and changes in the number and nature of trading venues
would impact market participants differently. What does extrapolate across all types of financial
innovation is the finding that modernization, security design, the standardization of contract terms,
and pricing are intertwined. This dynamic reality means that a one-size fits all approach to market
modernization without consideration of the externalities and what should be standardized may not
be appropriate.
To understand the economics consequences of financial innovation for issuers and bondholders,
we examine a sample of firms that exhibit variation in issuance practices by undergoing a transition
from issuing covenant-light bonds to covenant-heavy bonds. While this approach has the benefit
of removing confounding factors idiosyncratic to an individual issuer, one could be concerns that
the results do not generalize because this is selected sample of issuers. To test the veracity of
this generalizability claim, we investigate a second mutually exclusive sample from the first. We
construct a matched sample based on a propensity score estimator. While this sample cannot be
considered as randomly assigning contract design to issuers, because it takes a significantly different
approach to testing the underlying hypotheses, when the results from the two samples coincide that
suggests the original results do generalize.
Specifically, this second propensity-score matched approach enhances the previous results for
several reasons. Unlike the previous sample, where we limited our sample size by requiring the
match to be homogenous by issuer, maturity, liquidity, and covenant transition status, the propensity score sample is drawn from the universe of all investment grade bond issuers. Thus, this
constructed sample allows us to generalize our results and also because it’s a mutually exclusive
29
sample it provides a robustness check for the previous method. Further, previous studies have
confirmed that propensity score matching methods allow for more accurate inferences in an observational study. Finally, since the pool of potential matches is large, the data is well suited for using
this method.
Appendix C shows that the fundamental results from the within-issuer sample do not change
when using the larger more representative sample of issuers. Table CI shows the characteristics
of the issuers and issuances for the matched sample. Table CII shows the results from examining
the initial pricing for the propensity-score matched bond pairs. Column (1) of Panel A reveals
that on average, all else equal, including the covenant reduces the issuers’ yield by 27 basis points.
This finding is statistically significant and robust to the inclusion of issuance controls, issuer fixed
effects, and year fixed effects. It is also qualitatively similar to the 23 basis points found using the
issuer-specific sample. Column (1) of Panel B provides suggestive evidence that covenant inclusion
is associated with greater underpricing, which is again similar to the findings in the issuer-specific
sample. Table CIII shows that the subsequent trading behavior also coincides fairly closely with
the previous findings. Comparing the inferences from the two samples suggests that the results
from the smaller transition sample generalize to the universe of investment-grade bonds. Further,
because the two approaches had different identifying assumptions, this suggests our inferences are
not fragile to a single identifying assumption.
VII.
Conclusion
In this paper, we study the role modernization plays in shaping U.S. corporate bond market
transactions. In 2005, the SEC adopted a series of rules that allowed large issuers to significantly
accelerate issuance speeds. Under the new rules, potential investors have 36 minutes, at the median, to evaluate Prospectus terms, the issuers credit story, and pricing expectations. For the large
majority of new issue transactions, there is no conference call with the management of the issuer.
At the same time, the new issue market has grown substantially and multiple issuers often come to
market at the same time, which exacerbates the constraints on investors time to review each transaction. To identify the effects of this modernization event, we exploit eligibility cut-off thresholds
30
that are part of the SEC rules.
We find modernization meaningfully affects risk-sharing with widespread adoption of covenants,
which we argue simplifies the due diligence required to value the bond. We find a set of incurrencebased covenants gained popularity after the regulatory change. Because incurrence-based covenants
provide value to bondholders in idiosyncratic, highly speculative situations such as future business
combinations, spin-offs, or asset sales, we contend the inclusion of such covenants serves as a laborreducing device when time for contract scrutiny is compressed. Put another way, the additional
covenants serve the significant economic purpose of substituting for investor diligence.
In contrast to the critics that suggest financial innovation largely benefits the financial intermediaries who capture rents from exploiting new rules to their advantage, our evidence from bond
placement and secondary market trades suggests the modernization of the bond issuance process
mostly benefits those intended issuers and investors. We find underwriters operate in a competitive landscape and no underwriter captured market share or additional fees as a result of the
modernization. Learning externalities from transitioning to covenant-heavy bonds coincide with a
contract uncertainty premium, which benefits sophisticated investors. In particular, mutual funds
are rewarded for the superior information they gather in their due diligence process. We find issuers
bear some costs in terms of initial underpricing but benefit from an expanded set of bondholders
and reduced volatility once the new contract terms are learned.
Overall, our evidence highlights the importance of standardization in conjunction with modernization as regulators assess ways to improve current bond market practices. Our evidence is
consistent with the notion that repetition reduces the cost that investors must expend in learning
the meaning of contract terms. Hence, standardization of corporate bond issuances provides issuers
and investors a channel for improvement. Primary among these improvements is the prospect of
a more stable, less volatile issuance environment with improved transparency and lower new issue
concessions. Of course, in any setting, the features included in the standardized contract need to
be carefully assessed. In particular, our findings that change-of-control covenants are included in
over 60% of new issuances and the potential externality that places on the market for corporate
control should be carefully considered.
31
REFERENCES
Allen, F. and D. Gale, 1994, Financial Innovation and Risk Sharing, Cambridge, MA: MIT Press.
Asquith, P. and T. Wizman, 1990, Event Risk, Covenants and Bondholder Risk in Leveraged
Buyouts, Journal of Financial Economics 27, 195–214.
Barber, B. and T. Odean, 2008, All That Glitters: The Effect of Attention and News on the Buying
Behavior of Individual and Institutional Investors, Review of Financial Studies 21, 785–817.
Bebchuk, L., A. Cohen, and A. Ferrell, 2009, What Matters in Corporate Governance?, The Review
of Financial Studies 22, 783–827.
Becker, B., and V. Ivashina, 2014, Reaching for Yield in the Bond Market, The Journal of Finance
forthcoming.
Benveniste, L.M., A. Ljungqvist, W.J. Wilhelm, and X. Yu, 2003, Evidence of Information Spillovers
in the Production of Investment Banking Services, Journal of Finance 58, 577–608.
Billet, M., T. King, and D. Mauer, 2007, Growth Opportunities and the Choice of Leverage, Debt
Maturity, and Covenants, The Journal of Finance LXII, 697–730.
Bolton, P., and A. Faure-Grimaud, 2010, Satisficing Contracts, The Review of Economic Studies
77, 937–971.
Bordalo, P., N. Gennaioli, and A. Shleifer, 2012, Salience Theory of Choice Under Risk, Quarterly
Journal of Economics, 1243–1285.
Bordalo, P., N. Gennaioli, and A. Shleifer, 2013, Salience and Consumer Choice, Journal of Political
Economy 121, 803–843.
Bradley, M., and M. R. Roberts, 2004, The Structure and Pricing of Bond Covenants, Working
Paper.
Bratton, W., 2006, Bond Covenants and Creditor Protection: Economics and Law, Theory and
Practice, Substance and Process, European Business Organization Law Review 39.
Cheng, A. and King, T., 2009, Corporate Governance and Financial Contracting: Bondholder
Takeover Defenses in Poison Puts, Corporate Ownership and Control, 7, 9–20.
Cook, D. and Easterwood, J., 1994, Poison Put Bonds: An Analysis of their Economic Role, The
Journal of Finance XLIX, 1905–1920.
Crabbe, L., 1991, Event Risk: An Analysis of Losses to Bondholders and “Super Poison Put” Bond
Covenants, The Journal of Finance XLVI, 689–706
The Credit Roundtable, 2008, Improving Covenant Protections in the Investment Grade Bond
Market, available at http://www.iimemberships.com/dl/creditroundtable/.
Dang, T., G. Gorton, B. Holmstrom, and G. Ordonez, 2014, Banks as Secret Keepers, NBER
Working Paper No. 20255.
32
De Franco, G., F. Vasvari, D. Vyas, and R. Wittenberg-Moerman, 2014, Sticky Covenants, Working
Paper.
Dichev, I. and D. Skinner, 2002, Large-Sample Evidence on the Debt Covenant Hypothesis, Journal
of Accounting Research 40, 1091–1123.
Eckbo, E., R. Masulis, and O. Norli, 2007, Security Offerings, Handbook of Corporate Finance:
Empirical Corporate Finance, E. Eckbo, editor, North-Holland/Elsevier, Chapter 13.
Fenn, G., Speed of Issuance and the Adequacy of Disclosure inthe 144A High-yield Debt Market,
Journal of Financial Economics 56, 383–405.
Fields, J., D. Kidwell, and L. Klein, RJR/Nabisco and Event Risk Protection, Financial Services
Review, 143–156.
French, 2008, The Cost of Active Investing, AFA Presidential Address.
Gao, X., and J. Ritter, The Marketing of Seasoned Equity Offerings, Journal of Financial Economics 97, 33–52.
Gennaioli, N., A. Shleifer, and R. Vishny, 2015, Neglected Risks: The Psychology of Financial
Crises, American Economic Review Papers and Proceedings, forthcoming.
Graham, J., and C. Harvey, 2001, The Theory and Practice of Corporate Finance: Evidence from
the Field, Journal of FInancial Economics 60, 187–243.
Gulati, M., and R. Scott, 2013, The 3 1/2 Minute Transaction: Boilerplate and the Limits of
Contract Design, University of Chicago Press.
Hanley, K. and G. Hoberg, 2012, Litigation Risk, Strategic Disclosure and the Underpricing of the
Initial Public Offering, Journal of Financial Economics 103, 235–254.
Hastings, J. and J. Shapiro, 2013, Fungibility and Consumer Choice- Evidence from Commodity
Price Shocks, Quarterly Journal of Economics 128, 1449–1498.
Kahan, M., and M. Klausner, 1993, Anti-Takeover Provisions in Bonds: Bondholder Protection or
Management Entrenchment, UCLA Law Review 40, 931–954.
Kahan, M., and M. Klausner, 1997, Standardization and Innovation in Corporate Contracting,
Virginia Law Review 83, 713–771.
Kahan, M., and D. Yermack, 1998, Investment Opportunities and the Design of Debt Securities,
Journal of Law, Economics, and Organization 14, 136–151.
Kamenica, E., 2008, Contextual Inference in Markets: On the Informational Content of Product
Lines, American Economic Review 98, 2127–49.
Lerner, J., and P. Tufano, 2011, The Consequences of Financial Innovation: A Counterfactual
Research Agenda, NBER Working Paper No. 16780.
McFadden, D., 2001, Economic Choices, American Economic Review 91, 351–378.
33
Murfin, J., 2012, The Supply-side Determinants of Loan Contract Strictness, The Journal of Finance LXVII, 1565–1601.
Myers, S., 1977, Determinants of Corporate Borrowing, Journal of Financial Economics 5, 147–145.
Noe, T., M. Rebello, and J. Wang, The Evolution of Security Design, Journal of Finance 61,
2103–2135.
Simsek, A., 2013, Speculation and Risk Sharing with New Financial Assets. Quarterly Journal of
Economics 128, 1365–1396.
Simsek, A., 2013, Financial Innovation and Portfolio Risks, American Economic Review Papers
and Proceedings 103, 398–401.
Smith, C., Warner, J., 1979, On Financial Contracting: an Analysis of Bond Covenants, Journal
of Financial Economics 7, 117–161.
Tufano, P., 2003, Financial Innovation. In Handbook of the Economics of Finance (Volume 1a:
Corporate Finance), eds. Constantinides G, Harris M, and Stulz R. New York: Elsevier, 307–36
White, M., 2014, Intermediation in the Modern Securities Markets: Putting Technology and Competition to Work for Investors, Speech before Economic Club of New York, June 20, 2014.
Zingales, L., 2015, Does Finance Benefit Society? AFA Presidential Address.
34
Figures and Tables
Time
9:00 AM
Step
Underwriters announce a new issue and alert investors via Bloomberg and phone.
Sales force begins discussions regarding investors’ interest and pricing expectations.
Investors attempt to evaluate the issuer’s credit quality and potential terms.
Investors submit an order without finalized pricing or terms.
9:15 AM
Preliminary prospectus is provided to investors.
Offering goes subject and books are closed (i.e. orders of interest stop being accepted).
10:00 AM
Price guidance is announced.
11:00 AM
Deal is launched with final pricing terms which can differ meaningfully from guidance.
3:00 PM
Underwriters set and release allocations to investors.
3:30 PM
Deal is priced near market close.
Figure 1. Example bond offering timeline post reform.
35
Pre-SEC Reform, WKSI-eligible Issuer (General Mills) 2.625% Notes due October 24, 2006, CUSIP 370334AX2
Date
December 21, 2001
February 8, 2002
Time
Similarity
Page Covenant
Info in File
Count Word Count Measure
Company information
(financials), intended use of
proceeds, description of debt
The issuer registers $8 billion of
securities the company may offer
Shelf Registration debt securities that are to be
including restrictive covenants
25
22
NA
and Prospectus
offered on a delayed or continuous
(e.g., asset sale), plan of
basis.
distribution, and information
about what terms of the indenture
can be changed.
SEC Filing File Description Description
S-3
S-3/A
36
September 17, 2003
(Offering Date)
September 19, 2003 12:09 PM 424B5
September 24, 2003 9:44 AM 8-K
The issuer amends the registration
statement, which will become
effective when the SEC, acting
pursuant to Section 8(a),
Amendment No. 1
determines. The information in the
to S-3 Shelf
Preliminary Prospectus
Registration and
Supplement and the accompanying
Preliminary
Preliminary Prospectus is not
Prospectus
complete and may be changed. The
issuer cannot sell securities until
the registration statement has been
declared effective by the SEC.
Prospectus Supplement dated
September 17, 2003, but filed
September 19, 2003.
This document is a Prospectus
Supplement and supplements the
Prospectus
attached Prospectus, which is part
Supplement (Final of a registration statement that the
Prospectus)
issuer filed with the SEC. This
Prospectus Supplement contains
the terms of this offering of notes.
Current report
Agreements related to issuance
Company information (risk
factors, recent developments,
financials), description of the
notes including restrictive
65
covenants (e.g., asset sale),
underwriters, use of proceeds, and
firm capitalization.
25
NA
Description of the notes including
pricing information, restrictive
36
covenants, underwriters, use of
proceeds, and firm capitalization.
25
0.906
Underwriting agreement dated
September 17, 2003.
NA
NA
3
Figure 2. Example of bond issuance filings before the SEC reform: This figure describes the SEC filings and information
contained in the filings for a WKSI-eligible issuer prior to the reform. To quantify the due diligence required to understand the
terms of the contract, a page count, covenant word count, and a text-based similarity measure are included. A similarity measure
of 1 indicates the Final Prospectus is identical to the terms available at the time of issuance, whereas a similarity measure of 0
indicates no overlap in content.
Post-SEC Reform, WKSI-eligible Issuer (General Mills) 1.500% Bond due April 27, 2027, CUSIP 370334BX1
Date
Time
SEC Filing File Description
February 20, 2015 4:48 PM S-3ASR
Page
Count
Covenant
Similarity
Word Count Measure
Description
Info in File
Automatic Shelf
Registration
General registration of
unidentified amount of debt
securities. Statement becomes
effective automatically without
SEC review. Each time the
issuer sells debt securities, the
issuer will provide a Prospectus
Supplement that will contain
specific information about the
terms of that offering, and the
issuer may add, update or
change information from what
is contained in this Prospectus.
Company information, risk
factors, use of proceeds, plan of
distribution, information about
what terms of the indenture can
be changed, and a notice that
the Prospectus Supplement
relating to any debt securities 37
that the issuer offers using this
Prospectus will describe the
amount, price, and other
specific terms including any
restrictive covenants, if
applicable.
18
NA
Description of bond terms,
including restrictive covenants
(e.g., change-of-control
87
covenant), underwriters, use of
proceeds, and firm
capitalization.
22
NA
Bond terms
NA
NA
22
0.737
NA
NA
6:12 AM 424B5
Preliminary
Prospectus
Supplement
April 20, 2015
4:57 PM FWP
Free-writing
Prospectus
Distributed for marketing on
offering date
April 22, 2015
3:48 PM 424B5
Final Prospectus
Supplement
Substantively identical to
preliminary prospectus
April 24, 2015
4:57 PM 8-K
Current report
37
April 20, 2015
(Offering Date)
The information in this
Preliminary Prospectus
Supplement is not complete and
may be changed.
6
Finalized bond terms including
pricing and any negotiated
89
changes such as covenants.
Underwriting agreement dated
Agreements related to offering
4
April 20, 2015
Figure 3. Example of bond issuance filings after the SEC reform: This figure describes the SEC filings and information contained
in the filings for a WKSI-eligible issuer after the reform. To quantify the due diligence required to understand the terms of the
contract, a page count, covenant word count, and a text-based similarity measure are included. A similarity measure of 1 indicates
the Final Prospectus is identical to the terms available at the time of issuance, whereas a similarity measure of 0 indicates no
overlap in content.
(2) Asset Sale Covenant
-20
0
0
% with Covenant
20
40
# of Covenants Added
1
2
3
4
60
5
6
80
(1) Total Covenants
-8
-4
0
4
Quarters Relative to SEC Reform
-8
8
-4
0
4
Quarters Relative to SEC Reform
8
(3) Change-of-control Covenant
15
% with Covenant
5
10
% with Covenant
25
50
0
0
-5
-25
38
75
20
100
(4) Fixed Charge Covenant
-8
-4
0
4
Quarters Relative to SEC Reform
8
-8
-4
0
4
Quarters Relative to SEC Reform
8
Figure 4. The effect of SEC modernization reform: The four plots illustrate the change in risk-sharing associated with the
SECmodernization reform for treated issuers (i.e., WKSI-eligible issuers). Plot (1), the upper left plot, illustrates an increase in
total number of covenants included in a bond indenture for the treated issuers. Plot (2) illustrates the percentage point increase
in use of asset sale clauses. Plot (3) illustrates the percentage point increase in change-of-control covenants. Plot (4) illustrates
the percentage point increase in fixed charge covenants. In each plot, the navy side represents the 8 quarters before the reform,
the reform is demarcated by a black vertical line, the maroon side represents the 8 quarters after the reform, and ninety-percent
confidence intervals are shaded.
(2) Offering Amount
2
0
100
4
Offering Yield
6
8
Offering Amount
200
300
400
10
500
12
600
(1) Offering Yield
-20
-16
-12
-8
-4
0
4
Quarters Relative to SEC Reform
Eligible for WKSI
8
12
-20
16
-16
-12
-8
-4
0
4
Quarters Relative to SEC Reform
Eligible for WKSI
Ineligible for WKSI
8
12
16
Ineligible for WKSI
(4) Change-of-control Covenant
% with Covenant
40
60
20
Total Covenants
7
8
0
6
5
39
9
80
10
100
(3) Total Covenants
-20
-16
-12
-8
-4
0
4
Quarters Relative to SEC Reform
Eligible for WKSI
8
12
-20
16
-16
-12
-8
-4
0
4
Quarters Relative to SEC Reform
Eligible for WKSI
Ineligible for WKSI
8
12
16
Ineligible for WKSI
Figure 5. Parallel paths prior to the SEC modernization reform: In each plot, the treated issuers (i.e., the WKSI-eligible
issuers) are depicted with a maroon line and control issuers (i.e., the WKSI-ineligible issuers) with a navy, dashed line. Plot (1),
the upper left plot, depicts offering yield. Plot (2) depicts offering amount. Plot (3) depicts total covenants. Plot (4) depicts
change-of-control covenants. The plots show the issuers follow parallel paths pre-reform but sharply diverge post-reform in the
two plots related to risk-sharing. The observed “parallel trends” pre-reform suggest difference-in-differences estimates are unbiased
estimates of the effect of the SEC reform, because absent the reform the paths are likely to have been the same for the treatment
and control issuers.
-20
Yield Reduction from Covenant
-10
10
0
20
(1) Boilerplate Test of Pricing
0
1
2
3
4
5
6
7
Quarter Post Issuance
First Issuance with Covenant
8
9
10
11
Other Issuance with Covenant
-20
Yield Reduction from Covenant
-10
0
10
20
(2) Saliency Test of Pricing
0
1
2
3
4
5
6
7
Quarter Post Issuance
Covenant is Salient
8
9
10
11
Covenant is not Salient
Figure 6. Heterogeneity in covenant pricing: The plot examines the initial pricing heterogeneity
and its persistence over time in two settings. Plot (1) illustrates the pricing differential when
an issuer first includes such a covenant whereas Plot (2) illustrates the pricing differential when
covenants are the salient attribute. The two plots differentiate between behavioral and learning
explanations for the covenant pricing. In each plot, the y-axis represents the reduction in yield
accrued to the issuer and the x-axis represents the quarters since issuance. The observations are
drawn from the within-issuer sample.
40
0
20
% New Issuances with Covenant
40
60
80
100
Underwriter Effects
-4
-2
0
2
4
6
Quarters Relative to SEC Reform
Bank of America
J.P. Morgan
8
10
Morgan Stanley
Wells Fargo
Figure 7. Underwriter effects: This graph shows the difference in new issuances with covenants
among four major investment banks. After the SEC reform, J.P. Morgan and Bank of America
were the fastest adopters of the covenant with inclusion of the covenant in a large percentage of
new issuances starting in the first quarter. Morgan Stanley and Wells Fargo soon followed, with
each of the four banks exhibiting an initial spike in usage within a year after the reform took effect.
41
0
% New Holdings with Covenant
40
60
80
20
100
Bondholder Effects
-4
-2
0
2
4
6
Quarters Relative to SEC Reform
Insurance
Pension Funds
8
10
Mutual Funds
Figure 8. Bondholder effects: This graph shows the difference in new holdings with covenants
among three types of bond holders. After the SEC reform, a greater percentage of new holdings of
insurance funds and mutual funds contained the change of control covenant, compared to pension
funds. However, 10 quarters after the reform, pension funds’ new holdings are converging to similar
levels as those of mutual funds.
42
Table I Ex Ante Bond Issuer Characteristics
This table reports summary statistics for bond issuers in the five years before the enactment of the SEC reform
that modernized the bond issuance process. The sample is limited to the bond issuers within $700 million
of common equity eligibility cut-off threshold. The mean and standard deviation (in parentheses) for each
variable are reported separately for two samples of issuers. Column (1) reports estimates for modernization
ineligible issuers and Column (2) reports estimates for modernization-eligible issuers. Column (3) reports
the p-value from a t-test for the difference between the ineligible and eligible issuers. Appendix A provides
the definition of each of the covariates.
Ex ante Issuer Characteristics
Market-to-book
Profitability
ROE
Investment / Assets
Cash Growth
Asset Growth
Sales Growth
Leverage
Firm Size
Observations
Ineligible
(1)
1.229
(0.583)
0.135
(0.058)
0.438
(0.303)
0.221
(0.231)
1.211
(3.534)
0.219
(0.344)
0.213
(0.677)
0.375
(0.112)
7.993
(0.485)
366
43
Eligible
(2)
1.214
(0.586)
0.136
(0.072)
0.445
(1.642)
0.205
(0.220)
1.585
(4.465)
0.198
(0.373)
0.308
(3.253)
0.455
(0.137)
7.216
(0.863)
203
p -value of difference
(3)
(0.774)
(0.758)
(0.955)
(0.397)
(0.305)
(0.499)
(0.680)
(0.000)
(0.000)
Table II Effect of Modernization on Bond Issuance
This table shows difference-in-differences estimates of the average change in bond issuance practices attributable to the SEC reform that modernized the bond issuance process. The sample is limited to the
bond issuers within $700 million of common equity eligibility cut-off threshold. Test-statistics calculated
using robust standard errors, clustered at the issuer level, are in parentheses. The exact specification is as
in Eq.(1): Yjit = α + βEligiblei + δ (Eligible × P ost)it + γXjit + g (CEjit , T otjit ) + ft + εjit , where Yjit
is an outcome variable representing price, quantity, or the inclusion of the listed covenant for issuance j
by issuer i in quarter t, Eligible is an indicator for issuers that are eligible for WKSI status, P ost is an
indicator for the post-reform period, and Xjit is a vector of issuer and issuance controls including issuance
amount, offering yield, and maturity as well as issuer market-to-book, profitability, ROE, investment, cash
growth, asset growth, sales growth, leverage, and firm size, ft are year fixed effects and g (CEjit , T otjit ) is
a flexible polynomial that controls for how far the treated and control issuers are from eligibility based on
the conditions of common equity market capitalization (CEjit ) and total offerings in the past three years
(T otjit ). ***, ** and * indicate p-values of 1%, 5%, and 10%, respectively.
Panel A. The Effect of Modernization on Price and Quantity
2-year Window Around Event
Offering
Offering
Dependent Variable =
Yield
Amount
Difference-in-Differences
2.37
5.96
(1.94)
(41.21)
Yes
Yes
Flexible Polynomial for Cut-offs
Yes
Yes
Issuer & Issuance Controls
Yes
Yes
Year Fixed Effects
Adjusted R2
Observations
4%
306
4-year Window Around Event
Offering
Offering
Yield
Amount
0.74
-22.82
(0.78)
(22.78)
Yes
Yes
Yes
Yes
Yes
Yes
39%
306
19%
697
37%
697
Panel B. The Effect of Modernization on Risk-sharing (2-year Window Around Event)
Dependent Variable = Covenant
Difference-in-Differences
Flexible Polynomial for Cut-offs
Issuer & Issuance Controls
Year Fixed Effects
Adjusted R2
Observations
Total
Covenants
Asset Sale
Clause
Change of
Control
2.80***
(1.04)
Yes
Yes
Yes
8%
306
0.20**
(0.08)
Yes
Yes
Yes
4%
306
0.42***
(0.10)
Yes
Yes
Yes
23%
306
Fixed Charge Negative Pledge
0.07**
(0.03)
Yes
Yes
Yes
11%
306
0.19*
(0.11)
Yes
Yes
Yes
16%
306
Panel C. The Effect of Modernization on Risk-sharing (4-year Window Around Event)
Total
Asset Sale
Change of
Fixed Charge Negative Pledge
Dependent Variable = Covenant
Covenants
Clause
Control
Difference-in-Differences
1.63**
0.04
0.29***
0.09***
0.18**
(0.68)
(0.05)
(0.07)
(0.03)
(0.07)
Yes
Yes
Yes
Yes
Yes
Flexible Polynomial for Cut-offs
Yes
Yes
Yes
Yes
Yes
Issuer & Issuance Controls
Yes
Yes
Yes
Yes
Yes
Year Fixed Effects
2
9%
1%
25%
8%
20%
Adjusted R
Observations
697
697
697
697
697
44
Table III Descriptive Statistics for Within-Issuer Sample
This table provides descriptive statistics for a sample of bond issuers eligible for the SEC modernization reform that issued similar maturity
bonds both before and after the reform. The mean and standard deviation (in parentheses) are reported separately for eligible issuers in Column
(1) and (2) and issuances in Column (3) and (4). Column (1) and (3) report estimates for bonds that include the restrictive covenant (i.e.,
after the reform) and Column (2) and (4) report estimates for bonds that exclude the restrictive covenant (i.e., before the reform). Appendix
A provides the definition of each of the covariates.
45
Descriptive Statistics for Within-Issuer Sample
Including
Excluding
Covenant
Covenant
Issuer Characteristics
(1)
(2)
Market-to-book
1.241
1.535
(0.554)
(1.417)
Profitability
0.031
(0.004)
(0.019)
(0.084)
ROE
0.131
0.090
(0.266)
(0.300)
Investment / Assets
0.022
0.011
(0.035)
(0.130)
Cash Growth
1.216
1.126
(1.219)
(0.965)
Asset Growth
0.901
0.862
(0.379)
(0.339)
Sales Growth
0.879
0.750
(0.390)
(0.549)
Leverage
0.325
0.238
(0.113)
(0.179)
Firm Size
9.688
6.577
(0.845)
(2.703)
E-index
1.603
2.304
(1.144)
(1.171)
Observations
590
590
Unique Issuers
118
118
Issuance Characteristics
Offering Amount ($mm)
Maturity (Years)
Offering Yield
Rating at Issuance
Gross Spread
Number of Book-running Managers
Time to Evaluate Contract
(Column (3) in hours, Column (4) in days)
Length of Contract (Pages)
Similarity to S-3 Filing
Covenant Quality
Trigger Count
Holding Company Clause
Number of Ratings Needed to Downgrade
Agency Must Verify Downgrade Cause
Covenant Word Count
Change in Covenant Word Count
Including
Covenant
(3)
566,545
(393,102)
6.85
(4.87)
5.15
(1.80)
7.84
(1.54)
5.21
(1.76)
3.24
(1.74)
1.39
(2.63)
59
(23)
0.748
(0.151)
2.60
(1.19)
3.32
(1.46)
0.484
(0.500)
2.37
(0.55)
0.656
(0.48)
19.68
(15.97)
8.39
(15.64)
Excluding
Covenant
(4)
514,385
(428,653)
10.36
(5.35)
5.38
(1.37)
7.25
(1.84)
5.70
(1.50)
2.30
(1.34)
413.49
(438.58)
50
(27)
0.834
(0.169)
16.32
(13.60)
4.66
(13.63)
Table IV Initial Covenant Pricing for Within-Issuer Sample
The table shows the initial price effects from covenant inclusion for the within-issuer sample, which compares
concurrently trading bonds issued by the same issuer with maturities within two years where one issuance
has the restrictive covenant and the other does not. Test-statistics calculated using robust standard errors,
clustered at the issuer level, are in parentheses. The specification for Column (1) is as in Eq.(2): Yjik =
α + βCovjik + γXjik + fi + fk + εjik , where Yjik is the dependent variable for issuance j by issuer i at
offering period k and Covjik is an indicator for the change-of-control. Column (1) shows the average price
effect while Column (2) - (4) examine heterogeneity in the price effect. The specification for Columns (2)
- (4) include indicators for covenant saliency, if the covenant is the first by that issuer, and if the contract
terms included at the time of issuance had changed significantly from the base prospectus (defined as below
median textual similarity). In all specifications, we include issuer fixed effects, year fixed effects, as well as
a vector of issuance controls Xjik , including credit rating, maturity, total covenants, and offering amount.
***, ** and * indicate p-values of 1%, 5%, and 10%, respectively.
Avg. Pricing
(1)
0.227
(1.78)*
Dependent variable = Treasury Less Yield at Offering
Covenant Inclusion
Covenant is Salient Attribute
First Issuance with Covenant
Large Change from Base Prospectus
Issuance & Issuance Controls
Year Fixed Effects
Adjusted R 2
Observations
Yes
Yes
62%
399
46
Heterogeneity in Pricing
(2)
(3)
(4)
0.311
0.459
0.420
(2.18)** (2.93)***
(1.07)
-0.323
(2.34)**
-0.516
(3.46)***
-0.320
(1.78)*
Yes
Yes
Yes
Yes
Yes
Yes
63%
64%
65%
399
399
399
Table V Descriptive Statistics for Bondholders
This table reports summary statistics for bondholders at the time of issuance. The mean and standard
deviation (in parentheses) for each variable are reported separately for two samples of issuers. Column (1)
reports estimates for issuances that included the restrictive covenants and Column (2) reports estimates for
issuances that excluded the restrictive covenant. Appendix A provides the definition of each of the covariates.
Including Covenant
Excluding Covenant
Bondholders at Issuance
(1)
(2)
Number of Bondholders
116.8
99.5
(86.8)
(88.2)
Total Par Held
230,880
248,805
(192,680)
(211,768)
Coverage Ratio
43.6%
39.5%
(18.6%)
(19.6%)
Percent Held by Annunity Funds
4.9%
3.7%
(7.7%)
(7.3%)
Percent Held by Insurance
67.9%
72.1%
(26.0%)
(25.9%)
Percent Held by Mutual Funds
26.1%
20.9%
(22.6%)
(21.9%)
Percent Held by Pension Funds
0.8%
2.7%
(4.3%)
(7.4%)
Percent Held by Others (e.g., Hospital, Government, etc…)
0.3%
0.6%
(0.8%)
(2.4%)
Percent Investing With LBO Experience
48.6%
23.1%
(22.4%)
(24.8%)
Percent of Investors Portfolio in Investment-Grade Bonds
80.5%
80.1%
(10.5%)
(9.9%)
Percent of Investors with Prior Experience with Issuer
41.4%
45.9%
(20.1%)
(20.1%)
Number of Other Issuances Held
1284.6
1155.3
574.2
(667.9)
778
1,554
Observations
284
376
Unique Issuers
47
Table VI Subsequent Bondholdings for Within-Issuer Sample
The table shows the subsequent average price effects and bondholdings from covenant inclusion for the
within-issuer sample, which contains concurrently trading bonds that had maturies within the same two
year window. Test-statistics calculated using robust standard errors, clustered at the issuer level, are in
parentheses. For Panel A, the specification is as in Eq.(2): Yjik = α + βCovjik + γXjik + fi + fk + εjik , where
Yjik is the dependent variable for issuance j by issuer i at offering period k and Covjik is an indicator for
the change-of-control. Each column represents a different time period post issuance. For Panel B, C and D,
the specification is as in Eq.(3): ∆Yji = β∆Salientji + γ4Xji + ∆fk + ∆εji , where ∆Yji is the difference in
the dependent variable for issuance j by issuer i with and without the covenant. Salientji is an indicator for
the covenant being the salient attribute relative to pricing or offering amount. F irstji indicates when the
covenant is the first time that issuer has ever included such a covenant. Dissimilarji indicates the contract
terms included at the time of issuance had changed significantly from the base prospectus (defined as below
median textual similarity). In both specifications, Xji is a vector of controls which includes credit rating,
maturity, offering amount, offering yield, underpricing, and trading volume. ***, ** and * indicate p-values
of 1%, 5%, and 10%, respectively.
Panel A. Dependent Variable =
TREASURY LESS YIELD
INVESTORS HOLDING
RETURN VOLATILITY
Panel B. Dependent Variable =
TREASURY LESS YIELD
MUTUAL FUND HOLDINGS
Panel C. Dependent Variable =
TREASURY LESS YIELD
MUTUAL FUND HOLDINGS
Panel D. Dependent Variable =
TREASURY LESS YIELD
MUTUAL FUND HOLDINGS
Issuance & Issuer Controls
Liquidity/Trading Controls
Year Fixed Effects
Observations
Independent Variable = COVENANT INCLUSION
(Qtr1)
(Qtr2)
(Year1)
(Year2)
0.018
0.679
0.477
0.427
(0.02)
(4.94)***
(1.94)*
(1.34)
24.2
29.7
28.3
50.4
(2.33)**
(2.89)***
(3.00)***
(5.31)***
0.14
0.05
-0.03
-0.08
(1.03)
(0.97)
(1.71)*
(2.11)**
Independent Variable = COVENANT IS SALIENT
(Qtr1)
(Qtr2)
(Year1)
(Year2)
-0.170
-0.112
-0.070
-0.009
(2.56)**
(3.36)***
(1.79)*
(0.22)
11.6%
10.6%
8.6%
8.8%
(2.39)**
(1.82)*
(1.45)
(1.82)*
Independent Variable = FIRST WITH COVENANT
(Qtr1)
(Qtr2)
(Year1)
(Year2)
0.044
0.132
0.057
0.062
(0.75)
(2.01)**
(1.29)
(1.49)
0.7%
-4.8%
-5.1%
-6.0%
(0.14)
(0.90)
(1.20)
(1.46)
Independent Variable = DISSIMILAR TO BASE PROSPECTUS
(Qtr1)
(Qtr2)
(Year1)
(Year2)
-0.133
-0.104
-0.045
-0.037
(2.23)**
(2.15)**
(1.54)
(1.18)
14.7%
12.5%
6.8%
3.4%
(2.45)**
(2.07)**
(1.01)
(0.47)
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
302
298
290
258
48
Table VII Sophisticated, Duped, or Lucky Investors?
This table explores the relationship between investors’ experience evaluating an issuance and the pricing of
the issuances contractual terms. Column (1) examines bonds with bargain covenants and Column (2) bonds
with priced covenants following the modernization to bond issuance process. Column (3) tests if there is a
statistical significant difference in the time available to review the bond, LBO experience and prior issuer
experience for these two pricing outcomes. In both specifications, a vector of issuance controls are used which
include credit rating, maturity, offering amount, offering yield, underpricing, and trading volume. ***, **
and * indicate p-values of 1%, 5%, and 10%, respectively.
Independent Variable =
Dependent Variable =
Time for Due Diligence
LBO Experience
Prior Issuer Experience
Controls and Fixed Effects
Observations
49
Bargain
Covenant
Priced
Covenant
Test of
Difference
(1)
(2)
(3)
-1.788
(2.05)**
0.268
(1.75)*
0.421
(2.26)**
Yes
302
0.882
(1.15)
0.041
(0.35)
-0.415
(2.46)**
Yes
302
(2.41)**
(0.48)
(3.21)***
Yes
302
Table VIII The Benefits Gained by Underwriters from Modernization
This table explores the extent to which underwriters benefited from the SEC reforms that modernized the
bond issuance process. Test-statistics calculated using robust standard errors, clustered at the underwriter
level, are in parentheses. The specification is as in Eq.(4): Yit = α+βU nderwritingQualityit +γXit +fi +ft +
εit where Yit is the dependent variable of interest for underwriter i in in quarter t and U nderwritingQualityit
is the independent variable of interest that measures aspects of the underwriting quality such as the inclusion
of restrictive covenants, changes from the base Prospectus to the final Prospectus, the quality of the legal
language used in the indenture as well as the time allowed for investor due diligence. We include underwriter
fixed effects, quarter fixed effects, as well as a vector of issuance controls including credit rating, maturity,
offering amount, offering yield, and spread when appropriate. Panel A focuses on all bond issuances while
Panel B focuses on the bond issuances that have the restrictive covenant package. The sample is limited to
underwriters with at least 3% market share in a given quarter in the sample. ***, ** and * indicate p-values
of 1%, 5%, and 10%, respectively.
Panel A. All Bond Issuances
New Issuance
(N)
-0.08*
(0.05)
-0.03*
(0.02)
Yes
Yes
55%
753
New Issuance
Volume ($)
-0.13***
(0.05)
-0.04**
(0.02)
Yes
Yes
53%
753
Market Share
(N)
-0.08*
(0.05)
-0.02*
(0.02)
Yes
Yes
21%
735
Market Share
Volume ($)
-0.15***
(0.04)
-0.05**
(0.03)
Yes
Yes
21%
735
Underwriter
Fees
0.016
(0.02)
0.012
(0.01)
Yes
Yes
68%
753
Panel B. Bond Issuances with Restrictive Covenant Packages
New Issuance New Issuance
Dependent Variable =
(N)
Volume ($)
Quality of Underwriting
0.17***
0.17***
(0.05)
(0.05)
-0.07
-0.11*
Time for Diligence
(0.05)
(0.06)
Yes
Yes
Issuance Controls
Quarter & Underwriter Fixed Effects
Yes
Yes
2
57%
51%
Adjusted R
Observations
335
335
Market Share
(N)
0.65***
(0.15)
-0.07***
(0.03)
Yes
Yes
45%
332
Market Share
Volume ($)
0.68***
(0.15)
-0.08***
(0.03)
Yes
Yes
42%
332
Underwriter
Fees
0.70***
(0.05)
-0.07**
(0.03)
Yes
Yes
49%
335
Dependent Variable =
Adding Restrictive Covenants
Change from Base Prospectus
Issuance Controls
Quarter & Underwriter Fixed Effects
Adjusted R2
Observations
50