Convertibles and hedge funds as distributors of equity

exposure

Stephen J. Brown, Bruce D. Grundy, Craig M. Lewis and Patrick Verwijmeren*

First draft: 11 November 2009

This draft: 15 February 2010

*Stephen Brown (sbrown@stern.nyu.edu) is from New York University Stern School of

Business, Bruce Grundy (bruceg@unimelb.edu.au) and Patrick Verwijmeren
(patrickv@unimelb.edu.au) are from the University of Melbourne, and Craig Lewis
(craig.lewis@owen.vanderbilt.edu) is from Vanderbilt University’s Owen Graduate
School of Management. This paper has benefited from comments by Nick Bollen,
Sudipto Dasgupta, Robert Faff, Joseph Fan, Doug Foster, Andrew Hertzberg, Mike
Lemmon, Bryan Lim, Mark Mitchell, Phong Ngo, Garry Twite and seminar participants
at Vrije Universiteit Amsterdam, RSM Erasmus University, and the 2009 ANU Summer
Camp. We gratefully acknowledge funding from a Faculty Research Grant at the
University of Melbourne and thank Melissa Roodzant and Daniel Selioutine for research
assistance.
 Convertibles and hedge funds as distributors of equity

exposure

Abstract
Hedge funds are major players in the convertible securities market. In this paper we
directly observe their involvement in the market for newly-issued convertibles and find
that hedge funds provide 73.4% of convertible financing in the period 2000-2008. Hedge
funds are subject to less regulation than other investors and their business model is based
on expertise in short-selling. When buying convertibles, hedge funds typically combine a
long position in a convertible with a short position in the underlying stock. By buying
newly issued convertibles and simultaneously creating short stock positions in the open
market, hedge funds can effectively distribute equity exposure to well-diversified
shareholders. We hypothesize that firms use the distributive ability of hedge funds as a
substitute for a seasoned equity offering. We find strong evidence that firms choosing to
issue convertibles have characteristics that make a seasoned equity offering expensive,
and also have characteristics that reduce the cost for hedge funds to establish and
maintain short positions. A higher fraction of a convertible issue is privately-placed with
hedge funds when the issuer’s stock return is more volatile and when the issuer has a
higher probability of financial distress, which are characteristics increasing the cost of
seasoned equity offerings. Hedge fund involvement also increases when institutional
ownership, stock liquidity, issue size, limitations on callability and concurrent stock
repurchases suggest that shorting costs will be lower. We further observe that virtually all
convertibles in our sample are very equity-like, consistent with the convertibles being
issued by ‘would-be’ equity issuers. We find that discounts on convertibles issued to
hedge funds are not higher than the discounts on convertibles issued to other buyers,
which is in line with hedge funds serving as relatively low-cost distributors rather than as
investors of last resort. We further observe that hedge funds do not bear the costs of
maintaining short positions for extremely long periods—on average no more than 13.38%
of convertibles are still outstanding five years after they were originally issued.

JEL classification: G2, G32

Keywords: Convertible securities, convertible arbitrage, hedge funds

2
1. Introduction

Convertible arbitrage hedge funds combine long positions in convertible securities

with short positions in the convertible issuer’s stock. These funds are subject to less

regulation than other investors and their business model is based on expertise in short-

selling.1 Anecdotal evidence suggests that hedge funds are in fact a major player in the

convertible security market. This paper directly examines the participation of hedge funds

in convertible issues and establishes that the majority of the US convertibles issued

during the 2000 – 2008 period are initially purchased by hedge funds.

We construct a database of the initial buyers of 803 privately placed convertibles

issued under Rule 144A.2 Securities issued under Rule 144A do not require registration

with the SEC, but they can only be sold in the secondary market with no lock-up period

by the set of qualified institutional buyers listed in a registration statement issued after the

original prospectus of the convertible.3 The registration statement therefore provides a list

of the original buyers in convertible security offerings. The data show that 73.4% of the

financing of newly-issued convertibles in our data set is provided by hedge funds.4

Virtually all convertibles in our sample are ‘equity-like’ in that the average

convertible delta at the time of issue is 0.867. Only three of 803 issues have deltas below

one half and only ten percent (twenty-five percent) of issues have deltas below 0.725
1
See Fung and Hsieh (1999) for a description of the regulatory environment for US hedge funds and
explanations of how and why the trading behavior of hedge funds differs from other investors.
2
Private placements under Rule 144A represent the majority of convertible issues in recent years. For
example, De Jong, Dutordoir, and Verwijmeren (2009) report that 95% of convertible issues during the
2003 – 2007 period are privately placed.
3
Qualified institutional buyers (QIBs) are institutions with over $100 million in assets.
4
This confirms anecdotal evidence that about 70%-80% of convertibles are bought by hedge funds. A
Financial Times article (Skorecki, 2004) states that in 2003 “some [convertible] bonds have been issued
exclusively to hedge funds and on average they have been responsible for buying about 70 percent of new
issues.” A 2004 Wall Street Journal article (Pulliam, 2004) reports that “hedge funds play[ed] a big role in
the roughly $600 billion convertible-bond market, which saw $97 billion in new issues last year. As much
as 80% of those issues were bought by hedge funds, according to brokers who work on convertible-bond
trading desks.”

3
(0.816). A similar observation is made by Lewis and Verwijmeren (2009) and is

consistent with firms issuing convertibles as a substitute for equity. The observation

suggests the following possibility. Instead of issuing high cost equity, firms privately

place convertibles with hedge funds. By shorting stock to hedge against changes in the

stock price, these funds distribute the equity exposure to diversified investors in the open

market.

By placing a convertible with hedge funds, a firm can receive financing today while

avoiding the discounts and underwriter fees associated with a secondary equity offering.

Where those costs are higher than the costs associated with the private placement and the

security is eventually converted, the firm will have issued equity at a lower cost. We find

that hedge fund involvement in convertible issues is greater when the issuer is more

financially distressed, when the issuer’s return volatility is higher, and when the firm is

listed on NASDAQ. Altinkilic and Hansen (2003), Corwin (2003), and Eckbo, Masulis,

and Norli (2007) argue that these characteristics are indicative of relatively high costs of

issuing seasoned equity.

The attractiveness of convertibles to hedge funds depends on both the cost of

shorting the issuer’s stock and on the design of the security itself. We find that hedge

funds purchase a smaller fraction of a particular issue and comprise a smaller fraction of

the buyers of the issue if the issue is callable. Call features complicate hedging since the

decision to call is in the hands of the firm. A call will redistribute wealth between

convertible-holders and stock-holders. This redistribution is not a hedgeable co-

movement of the bond and the stock and makes it more difficult to determine the optimal

number of shares to short.

4
 Hedge funds are also more involved in issues that are small relative to the market

value of the firm’s equity; i.e., when hedging the issues requires borrowing only a small

fraction of the shares outstanding. This finding is also consistent with the ability to hedge

being a determinant of involvement by hedge funds. We also find evidence that hedge

fund involvement is positively related to concurrent stock repurchases. De Jong,

Dutordoir, and Verwijmeren (2009) argue that these concurrent stock repurchases

facilitate hedge funds in obtaining their short positions.

To further investigate the role of convertible hedge funds as distributors of equity

exposure, we compare firms that issue convertibles to hedge funds to firms that issue

seasoned equity. We find that firms selling convertibles are more financially distressed

and have more volatile returns relative to seasoned equity issuers. The issuers of

convertibles would therefore have faced high costs of issuing seasoned equity compared

to those firms that chose to issue equity. In addition, convertible issuers have a higher

average level of institutional ownership and have more liquid stock than firms that make

seasoned equity offers. Institutional ownership facilitates the borrowing of stock to set up

a short position (D’Avolio, 2002), while stock liquidity also reduces the costs of

establishing and maintaining a short position.

The paper makes two main contributions to the literature. Primarily, we document

the role that hedge funds play as distributors of equity exposure. Brophy, Ouimet, and

Sialm (2009) study PIPE (Private Investments in Public Equity) issues and conclude that

hedge funds serve as investors of last resort for the issuing firms in that hedge funds

“provide capital for companies that are otherwise constrained from raising equity capital”

5
(Brophy, Ouimet, and Sialm, 2009, p. 541).5 PIPE issuers are in general very small and

distressed and provide large discounts to investors. The firms in our sample are

substantially larger and are not as distressed, and issuing convertibles to hedge funds is

unlikely to be their only financing option. Further, the convertible issuers in our sample

do not issue bonds at higher discounts to hedge fund investors than to other investors.

Second, our paper investigates the matching of particular securities and a targeted

group of buyers. The literature on this topic is scarce as the original buyers of new

security offerings cannot typically be observed. One of the first major contributions in

this field is Wruck (1989), who focuses on private equity sales. Our hand-collected

database allows us to study the original buyers of 144A privately placed convertible

securities. We are able to show that both investor and investee identity matters in the

convertible debt market in the sense that hedge fund involvement in an issue depends on

characteristics of both the issuing firm and the security design. Prior studies have

examined security design choice from the perspective of the suppliers of convertible

securities. For example, Lewis and Verwijmeren (2009) argue that important motivations

for design choices are earnings management, credit rating concerns, and reductions of

taxes, refinancing costs, and managerial discretion costs. We add to these studies by

viewing security design choice from the perspective of the suppliers of capital. We

document a relation between the involvement of hedge funds and the incorporation of

call features, cash settlements, call spread overlays, and put features into the convertibles’

design.

5
PIPEs are private placements in which a public company issues equity securities to a group of private
investors without registering the shares. PIPEs differ from traditional private placements because a
registration statement is filed soon after signing of the purchase agreement, which allows for the resale of
shares sold to PIPE investors (Chaplinsky and Haushalter, 2009).

6
 The remainder of this paper is organized as follows. Section 2 discusses prior

studies on convertible arbitrage and constructs testable predictions on the role that hedge

funds play. Section 3 describes our data set and provides information on the proportion of

convertible securities that are bought by hedge funds. Section 4 presents our empirical

tests on the relation between hedge fund involvement and firm and issue characteristics.

Section 5 investigates the differences between firms with secondary equity offerings and

firms issuing convertibles to hedge funds. Section 6 examines convertible underpricing.

We report additional analyses of hedge fund shorting in Section 7 and of the effective life

of convertible bond issues in Section 8. Section 9 contains our conclusions.

2. Hedge funds and convertible arbitrage

2.1. Prior evidence

Hedge funds try to achieve an absolute return irrespective of the return on broad

stock or bond indices, which makes short-selling a natural part of their strategy (Fung and

Hsieh, 1999).6 Calamos (2003) discusses the typical buy-and-hedge strategy of buying

convertibles and shorting the stock of the issuing firm with the short position being

determined by the convertible’s delta. Delta is the sensitivity of a derivative’s price to

small changes in the price of the underlying, and is between zero and one. A delta-neutral

position attempts to exploit underpricing of the convertible security while hedging

against changes in the stock price.

A first way in which prior studies have examined the prevalence of hedge fund

purchases of newly issued convertibles is by focusing on short interest in the stock of the

6
Other investors are less likely to use short-selling. Fung and Hsieh (1997) for example find that mutual
fund returns are highly correlated with standard asset classes, while hedge funds generate returns that have
low correlation with the returns of mutual funds and standard asset classes.

7
issuer. Brent, Morse, and Stice (1990) document that US firms with convertible debt

outstanding report higher monthly short interest than other companies.7 Choi,

Getmansky, and Tookes (2009) document the effects of convertible arbitrage activity on

market liquidity by using the change in monthly short interest around convertible issue

dates as a proxy for arbitrage activity. They conclude that arbitrage activity increases the

liquidity of the stock because arbitrageurs trade in the opposite direction to the market’s

movements: arbitrageurs add to their short position when the stock price increases (as an

increase of the stock price increases the delta) and close out part of their short position

when the stock price decreases.

The main downside of using short interest data to examine convertible arbitrage is

that changes in short interest may be due to a belief that the share is overvalued. Thus it

might be that issue announcements are associated with valuation shorting by those who

take a relatively more pessimistic view of an announcement’s implications for firm value.

Moreover, monthly short interest data are noisy as confounding events can occur in the

month of a convertible issue. An alternative is to use daily short sale information. De

Jong, Dutordoir, and Verwijmeren (2009) examine daily short sale data from NYSE TAQ

database’s REG SHO file (which covers short-selling transactions from January 2005)

and document that daily short sales peak at convertible issue dates. Unfortunately, the

daily TAQ short sale data are only available for a limited number of NYSE firms and the

database has been discontinued in July 2007. Using daily short sale data also does not

solve for the effect of valuation shorting on announcement dates, as De Jong, Dutordoir,

7
Ackert and Athanassakos (2005) and Loncarski, Ter Horst, and Veld (2009) report similar findings for
Canadian firms using bi-monthly short interest data.

8
and Verwijmeren document that 92.58% of the convertible debt offerings in their sample

are announced on the issue date or the day prior.

A second way of examining convertible arbitrage is to examine fund flows into

convertible arbitrage hedge funds. Choi, Getmansky, Henderson, and Tookes (2009) and

De Jong, Duca, and Dutordoir (2009) show that aggregate convertible bond issuance is

positively related to the flow of funds into convertible arbitrage hedge funds. Mitchell,

Pedersen, and Pulvino (2007) find that convertible prices will drop relative to their

fundamental values when convertible arbitrageurs are more capital constrained. Agarwal,

Fung, Loon, and Naik (2009) show that the returns of convertible arbitrage funds are

positively related to the supply of convertible bonds and argue that an increase in the

supply of bonds means an increase in the availability of mispricing opportunities to be

exploited. Agarwal, Fung, Loon, and Naik also show that the returns of convertible

arbitrage hedge funds can be largely explained as the return to a buy-and-hedge strategy.

Our data allow us to directly examine the involvement of specific hedge funds in

specific convertible issues. For each convertible in our sample we are able to observe

how much of the issue is purchased by hedge funds. In this particular sense, our paper is

most closely related to Brophy, Ouimet, and Sialm (2009). Brophy, Ouimet, and Sialm

study traditional and structured PIPEs between 1995 and 2002 using data from the

Sagient database, which reports the involvement of hedge funds in PIPE issues.8 They

find that hedge funds account for 15.6% of investments in traditional PIPEs and 71.9% of

investments in structured PIPEs. They further show that the larger the involvement of

8
Traditional PIPEs are common stock PIPEs, with or without warrants, and convertible security PIPEs,
with or without warrants, while structured PIPEs are common stock reset PIPEs, structured equity lines,
floating convertibles, and reset convertibles.

9
hedge funds in an issue, the worse the post-issue performance of the issuing firm. This

negative relation leads them to conclude that hedge funds are investors of last resort.9

Our paper differs from Brophy, Ouimet, and Sialm (2009) in multiple ways. First,

Brophy, Ouimet, and Sialm do not distinguish within the set of traditional PIPEs between

convertible PIPEs and common stock PIPEs.10 Second, firms that issue PIPEs are in

general very small: Chaplinsky and Haushalter (2009) report that over the 1995 to 2000

period the mean (median) total asset value of PIPE issuers was $54 million ($18 million),

and Brophy, Ouimet, and Sialm report that over the 1995 to 2002 period the mean total

asset value of PIPE issuers was $178 million. In contrast, our sample consists of

substantially larger firms: the average value of total assets of the firms issuing the

convertibles in our sample is over $3 billion.11 Our data allow us to examine the role that

hedge funds play when issuers are likely to have a choice of financing methods open to

them. Third, we have detailed information on the contractual features of the convertibles

in our sample. This allows us to examine the relation between hedge fund involvement

and convertible security design as well as the relation between the decision to issue a

convertible and characteristics of the issuing firm.

9
Dai (2007) also uses the Sagient database to focus on PIPEs and compares PIPE investments by venture
capitalists with PIPE investments by hedge funds. Dai finds that venture capital-led PIPEs (issues in which
the lead investor is a venture capital fund) outperform hedge fund-led PIPEs. Dai concludes that venture
capitalists provide a certification (rather than monitoring) role.
10
Ellis and Twite (2008) estimate that during the 2002 to 2005 period, traditional PIPEs accounted for 78%
of PIPE issues and 95.1% of PIPE capital raised. Of the 1,169 traditional PIPEs in their sample only 282
are convertible PIPEs. Ellis and Twite argue that PIPEs are not simply securities issued to hedge funds as
financiers of last resort, but that PIPEs are more likely to be issued when firms have profitable growth
options and face a severe information asymmetry.
11
Brophy, Ouimet, and Sialm exclude 144A offerings as 144A private placements are “issued by larger and
more mature companies and are not considered PIPEs due to different regulatory treatments” (Brophy,
Ouimet, and Sialm (2009, p.547)).

10
2.2. Empirical predictions

Consider a firm that seeks to issue equity when an underwriter would be unwilling

to bear a large exposure to the firm’s equity unless the firm pays a substantial fee and/or

offers a large discount on its shares. Such a firm may be able to use the comparative

advantage of hedge funds to distribute equity exposure. Hedge funds engaged in

convertible arbitrage are able to use their knowledge of the borrowing and short-sale

market to hedge themselves while distributing risk to a large number of well-diversified

investors. Issuing convertibles may allow ‘would-be’ equity issuers to raise capital at

lower costs. Although an underwriter could also hedge by shorting stock, they may not

have a comparative advantage at doing so and their marketing claims may not be seen as

credible while they maintain a large short position in the shares they seek to place.

2.2A. Empirical predictions about the decision to issue a convertible

We predict that firms will issue convertibles to hedge funds when the costs of

issuing seasoned equity are high relative to the costs of establishing and maintaining a

short position. We use three main proxies for the costs of issuing seasoned equity. The

first proxy is the firm’s probability of financial distress, as the firm’s financial condition

is a strong predictor of the costs of issuing securities (Eckbo, Masulis, and Norli, 2007).

Kim, Palia, and Saunders (2005) show that underwriter spreads are higher for firms with

higher leverage and lower profitability. Provided that the costs of establishing and

maintaining a short position do not rise in a manner similar to the increase in the costs of

issuing seasoned equity, firms with a higher probability of financial distress are more

likely to choose to issue convertibles to hedge funds. Our second proxy is the return

volatility of the firm’s stock. Altinkilic and Hansen (2003) and Corwin (2003) report that

11
the costs of seasoned equity offers are significantly higher for issuers with more volatile

stock returns. Provided that the costs of establishing and maintaining a short position are

not comparably higher for more volatile firms, volatile firms are predicted to be more

likely to issue convertibles to hedge funds. Altinkilic and Hansen (2003) and Corwin

(2003) also find that NASDAQ firms have substantially higher costs of issuing seasoned

equity, even after controlling for variables like return volatility. We use a NASDAQ-

listing as a third proxy for the costs of issuing seasoned equity.

A potential fourth proxy for the costs of issuing equity is firm size as measured by

the market value of equity. Evidence on the effect of firm size on the costs of issuing

securities has however been mixed. Hansen and Torregrosa (1992) and Corwin (2003)

provide some evidence that larger firms have lower costs of issuing equity, while

Gompers and Lerner (1999) find that issuing costs are positively related to firm size for

venture capital-backed IPOs. Many other studies find no strong effect of firm size on

underwriter spreads and security underpricing; see Eckbo, Masulis, and Norli (2007) for

an overview of studies on the relation between various firm characteristics and the costs

of issuing seasoned equity.

Just as certain firm characteristics may make a secondary equity offering more

expensive, other characteristics of the issuing firm affect the cost of establishing and

maintaining a short position. Since short sellers must pay cash in lieu of the dividend to

the lenders of stock, the management of a hedge fund’s cash flows may be more

expensive when the convertible is issued by a dividend-paying company. Further, the

difficulty of shorting shares depends on the ease with which shares can be borrowed.

D’Avolio (2002) finds that institutional ownership explains about 55% of the variability

12
in loan supply across stocks. Since issuers with higher institutional ownership will have a

higher availability of shares to be borrowed (see also Asquith, Pathak, and Ritter, 2005),

such issuers are more likely to find that issuing convertibles is cheaper than issuing

seasoned equity.

All else equal, a would-be arbitrageur is more likely to be able to short the desired

number of shares when that number of shares is small relative to the number of shares

outstanding. When investigating the choice between issuing seasoned equity and

distributing exposure via hedge funds we take as our measure of relative size the ratio of

the issue proceeds to the market value of the equity outstanding. Similarly, the costs of

establishing and maintaining a short position will be lower when the issuer’s stock has

greater liquidity and again the likelihood of a convertible issue in preference to a

seasoned equity offering should be higher.

Conditional on the firm issuing a convertible, the attractiveness of the convertible

to hedge fund investors versus other potential buyers is a second question.

2.2B. Empirical predictions about hedge fund purchases of particular convertible issues

When the cost of issuing seasoned equity is relatively high and the firm does decide

to issue a convertible, it does not necessarily follow that the bond is more attractive to

hedge fund buyers than to other investors in convertibles. But if the decision to issue the

convertible was in part driven by an investigation of potential demand from hedge funds,

then the same factors that predict an increased likelihood of a convertible issue relative to

an equity issue will be associated with a higher fraction of the convertible being

purchased by hedge funds. This will imply that the fraction of a convertible issue

purchased by hedge funds should be higher when the probability of financial distress is

13
higher, when the issuer’s return volatility is higher, and when the issuer’s stock has a

NASDAQ-listing.

There is a separate reason why hedge fund involvement may be higher when the

issuer’s return volatility is higher. Part of a hedge fund’s profits can come from playing

the role of a market-maker willing to buy stock after a stock price decline and sell after a

stock price increase. When stock price movements are liquidity-induced and

subsequently reverse, the fund’s dynamic hedging strategy can be inherently profitable

and hedging a more volatile stock involves more opportunities to trade.

Following the reasoning in subsection 2.2A certain issuer characteristics are

predicted to make a convertible bond more or less attractive to hedge funds. Issues by

dividend-paying companies are predicted to be less attractive. Higher institutional

ownership and greater liquidity of the issuer’s stock and a smaller relative size for the

issue are all predicted to make a convertible more attractive to hedge fund investors.

When investigating the attractiveness of different convertible issues to hedge fund buyers

we take as our measure of relative size the ratio of the delta-neutral short stock position to

the number of shares outstanding.

Not only do particular characteristics of the issuer affect the relative costs of

secondary offerings versus distribution via hedge funds, firms that do decide to issue

convertibles should structure the issue so as to reduce the cost of establishing and

maintaining a short position. There are two ways to do this: first, by restricting or

eliminating the bond’s call features, and second, by undertaking a concurrent share

repurchase. Limits on callability increase the attractiveness of convertibles to hedge funds

because funds need to constantly balance their long bond position with a short stock

14
position. Call features complicate this balancing since the decision to call remains in the

hands of the issuing firm. Whenever the occurrence of a call is not a deterministic

function of the stock price and time, hedging will not be perfect.

A concurrent stock repurchase by the issuing firm will cater to the needs of hedge

funds by allowing funds to short-sell borrowed stock at a predetermined price, namely the

repurchase price. De Jong, Dutordoir, and Verwijmeren (2009) examine instances where

firms combine convertible issues with stock repurchases (the combination is known as a

Happy Meal), and conclude that convertible arbitrage explains both the size and speed of

execution of these stock repurchases.

We predict that hedge funds will purchase a larger fraction of convertible bond

issues with limited callability and a larger fraction of convertible issues accompanied by

a stock repurchase.

3. Data and summary statistics

Most recent convertible issues in the US have been privately placed under Rule

144A: Marquardt and Wiedman (2005) report that 84% of convertibles issued during

2000 – 2002 are privately placed, and De Jong, Dutordoir, and Verwijmeren (2009)

report that for the period 2003 – 2007 the percentage of privately-placed convertible

issues is about 95%. For our sample period, January 2000 to March 2008, the percentage

of privately-placed convertibles is 85% and 84% of the total proceeds raised by all the

convertible bond issues in the SDC database come from privately-placed issues.12

12
Convertible issues have been relatively popular since 2000. The aggregated convertible proceeds are on
average $68 billion per year for the period 2000 to 2008, while in the 1990s this average was $26 billion.

15
 We collect 1,142 privately-placed convertible issues from SDC. We require that

firms have an offering prospectus available on the SEC’s Edgar database. This eliminates

201 of the 1,142 issues. We also require a registration statement, which eliminates an

additional 138 issues. The registration statement lists the original buyers of the

convertible issue.13 This leaves 803 offerings with detailed information on the design of

the convertible and the identity of the original buyers.

Many buyers are easily classified as hedge funds since they have the words

“convertible arbitrage” in their name. We use the full lists of hedge fund managers in

Bloomberg and TASS and undertake an extensive internet search on each buyer (various

websites allow for a search of self-registered hedge funds, for example

http://www.hedgeco.net) in an effort to determine which buyers are hedge funds. We can

classify 39.4% of the 4,335 buyers as hedge funds.14

Panel A of Table 1 reports the average number of buyers per privately placed

convertible issue.

[ please insert Table 1 here ]

On average, a convertible issue is bought by 64 different buyers. Note that 64 × 803 is

much larger than 4,335, i.e., many buyers are involved in multiple convertible offerings.

The number of buyers varies considerably over the convertible issues. Some convertibles

13
The registration filings are typically S-3/A filings, although in some cases the selling security-holders can
be found in S-3, S-3ASR, or 424B filings.
14
There are three hedge funds in our sample that TASS classifies as “Global Macro” hedge funds, and an
additional two hedge funds in our sample have the words “Global Macro” in their name. As global macro
funds do not necessarily combine a long position in a convertible with a short position in stock, we have re-
estimated all the results in this paper with global macro hedge funds classified as non-hedge funds. We
have also checked the impact of the classification of brokerage firms on our results, as one practitioner has
suggested that brokerage firms possibly trade on behalf of small hedge fund clients. We find that our
conclusions in the paper are robust to changing the classifications of global macro hedge funds and
brokerage firms.

16
are bought by a single buyer and the maximum number of different buyers of a single

issue in our sample is 320.

That less than half of buyers are hedge funds is not inconsistent with financial press

reports that hedge funds buy 70% to 80% of all convertible issues. A typical hedge fund

is involved in a larger number of different convertible issues than a typical non-hedge

fund buyer of convertibles. Panel A of Table 1 indicates that on average more than half

the buyers of individual convertible bond issues (56.4%) can be classified as hedge funds.

Further, hedge funds often buy a larger percentage of a particular offering than other

buyers.15 Panel A reports that on average hedge funds account for 73.4% of the

investments in individual convertible issues.

There is dispersion in the percentage of each issue purchased by hedge funds: some

issues are entirely purchased by hedge funds, while other issues have no hedge funds

involvement at all. In a first attempt to shed light on this heterogeneity, we report the

fraction of the issue purchased by hedge funds per industry in Panel B of Table 1. Hedge

funds account for the majority of the convertible proceeds in every one of the 12 Fama-

French industry classifications. No simple industry affect is apparent.

[ please insert Figure 1 here ]

Hedge fund involvement has grown through time. Figure 1 reports the average

percentage of convertible issues purchased by hedge funds per year. Hedge funds account

for about 60% of the investments in privately-placed convertibles in 2000. By the first

quarter of 2008, this percentage has grown to 80%.

15
The average investment by an individual hedge fund expressed as a fraction of the offering proceeds is
2.31% in the sample. The average individual non-hedge fund investor buys 0.82% of an issue.

17
 We construct a final sample that we use to examine the relation between hedge

fund involvement, security design, and firm characteristics. Following common practice

we delete financial firms and utilities (112 issues). We also require that the issuing firms

have data available on Compustat, CRSP and the Thomson-Reuters Institutional

Holdings database, which deletes an additional 62 observations.

4. Determinants of the fraction of a convertible bond issue acquired by hedge funds

In this section we examine the relation between the fraction of a convertible issue

privately-placed with hedge funds and proxies for the costs of directly issuing seasoned

equity and the costs of indirectly distributing equity exposure via hedge funds. The set of

issuer characteristics examined include the likelihood of financial distress as measured

by a firm’s Altman Z-score (Z-scores are higher for firms with a lower chance of

bankruptcy); the return volatility of the issuer’s stock; a dummy for whether the issuer’s

stock have a NASDAQ-listing; firm size as measured by the market value of equity (in

regression analyses we employ the natural logarithm of this variable); a dummy for

whether the bond is issued by a company with dividend-paying stock; the percentage

institutional ownership of the issuer’s stock; the relative size16 of the issue as a proxy for

the fraction of the outstanding stock that would have to be sold short in order to hedge

ownership of the entire bond issue; and the Amihud liquidity measure (a high Amihud-

score denotes illiquidity).

The set of issue characteristics examined are those predicted in subsection 2.2B to

be related to hedge fund involvement (i.e., callability and concurrent stock repurchases)

plus other issue characteristics studied in Lewis and Verwijmeren (2009). Issue

16
For a given issue size, relative size is an issuer characteristic. Of course for a given issuer size, relative
size becomes a characteristic of the issue.

18
characteristics involve both dummy and continuous variables. The set of dummies denote

whether an issue is callable; whether an issue is accompanied by a stock repurchase;

whether an issue occurs in conjunction with a call spread overlay; whether an issue has a

cash settlement feature; and whether an issue is accompanied by put rights. The

continuous variables measure the size of the convertible’s delta and the size of the issue

proceeds. We make no prediction about how the issue characteristics other than

callability and concurrent stock repurchases will be related to hedge fund involvement,

but include these variables simply as controls. Exact definitions of the issue and issuer

characteristics are reported in Appendix A.

[ please insert Table 2 here ]

Table 2 provides a univariate analysis of the relation between various issuer and

issue characteristics and the fraction of a convertible issue sold to hedge funds. For our

overall sample, the average equity value of the issuing firms is in excess of $4 billion

dollars, and the average issue proceeds are $313 million. A substantial fraction (53.6%)

of the issuing firms is listed on NASDAQ. Of the 629 convertible issues in the sample,

8.1% involve a call spread overlay, 40.4% have a cash settlement feature, and 42.4%

have attached put rights. Interestingly, only 72.3% of the convertible issues are callable

some time prior to their maturity; Korkeamaki and Moore (2004) report that this

percentage is 98.2% over the period 1980 – 1996.

Table 2 also reports average values of issuer and issue characteristics for

subsamples. The first two subsamples contain equal numbers of issues and consists of

convertible securities with above-median purchases by hedge funds (meaning that hedge

19
funds buy more than 75.3% of the issue), and issues with below-median hedge fund

participation. We also construct two subsamples based on whether hedge funds buy the

majority of the convertible offering (584 of the 629 issues) or only a minority of the issue

(45 issues).

We find that differences in issuer characteristics are significant at the 1% level in

three instances. Convertibles with above-median hedge fund involvement are

significantly more likely to have been issued by firms with higher return volatility,17 and

by firms whose stock are more likely to be listed on NASDAQ. Both these characteristics

are identified by Altinkilic and Hansen (2003) and Corwin (2003) as likely to increase the

cost of a seasoned equity offering. The third significant univariate relation between issuer

characteristic and hedge fund involvement is that hedge funds are more likely to buy only

a minority of a convertible issue when the issuing firm is smaller.

When focusing on issue characteristics, we find that hedge funds have a

significantly lower involvement in convertibles with a call feature and a significantly

higher involvement in convertible issues accompanied by a stock repurchase. Both of

these findings are in line with our predictions.18 We further find that hedge fund

involvement is significantly higher in issues with call spread overlays and issues with

cash settlement features.

17
We measure return volatility as the annualized stock return volatility, estimated with ten years of
monthly stock return data. In robustness tests we have measured volatility as the annualized daily stock
return volatility over trading days [240, 40] relative to the issue date, as in Lewis, Rogalski, and Seward
(1999). The results throughout our paper are qualitatively unchanged.
18
We have collected more detailed call information for the callable convertible issues with an above-
median hedge fund involvement. We find that only 4.9% of these issues have a time to first call of zero.
The average (median) time to first call for these issues is 4.69 (5.00) years. As hedge finds can be expected
to hold their positions for shorter than 5 years, many callable convertibles can still be treated as non-
callable by hedge funds.

20
 We draw our main conclusions from the multivariate analysis reported in Table 3.

For ease of interpretation, Table 3 also sets out the predictions for the signs of the

coefficients developed in subsection 2.2B. Some firms in our sample are responsible for

multiple convertible offerings.19 In line with recommendations of Petersen (2009), we

therefore cluster standard errors at the firm level.

[ please insert Table 3 here ]

In Model 1, the dependent variable is the percentage of proceeds purchased by

hedge funds. Firms that issue convertibles because they face high costs of directly issuing

equity are likely to have first confirmed that demand from hedge funds is likely to be

high. Thus a higher fraction of convertibles issued by relatively distressed, more volatile

and/or NASDAQ-listed firms is predicted to be purchased by hedge funds. We find that

hedge fund involvement is positively related to the likelihood of financial distress (low

values for the Altman Z-score), and is also significantly higher for firms with high return

volatility and a NASDAQ-listing. The attraction to convertibles issued by companies with

highly volatile stock may also reflect an attraction to the trading profits from market-

making while dynamically hedging. Note that a significantly smaller percentage of a

convertible issue is purchased by hedge funds when the convertible is issued by a smaller

firm.

Hedge fund involvement is predicted to be higher when the costs of establishing

and maintaining a short position are lower. The significant relation between relative size

and hedge fund involvement is in the predicted direction and, also as predicted,

19
The 629 convertibles in the sample are issued by 474 different firms: 364 firms issue a single
convertible, 76 firms issue two convertibles, 26 firms issue three convertibles, 5 firms issue four
convertibles, and 3 firms are responsible for five different convertible issues.

21
convertibles with call features are of less interest to hedge fund investors. We find some

evidence (at the 10% level) that hedge fund involvement in an issue is positively related

to the concurrent repurchase of stock. We find no evidence that hedge fund involvement

is significantly higher when the firm pays no dividends, when more of the issuer’s stock

is held by institutions, or when the stock is more liquid.

Regarding the control variables, we find that cash settlements, call spread overlays,

and put features are all positively related to the fraction of a convertible issue that is

purchased by hedge funds. Lewis and Verwijmeren (2009) argue that cash settlement

features and call spread overlays facilitate earnings management and that this may be

important to managers whose bonuses are tied to reported earnings and/or to shareholders

unable to see through reported earnings.20 A potential explanation for the relation

between these design characteristics and hedge fund involvement is that firms with

relatively high costs of issuing seasoned equity are more likely to try to make their

financial reporting look as favorable as possible. Put features provide investors with more

downside protection than is typically associated with a convertible security. The positive

relation with hedge fund involvement can possibly be explained by put features being

issued by risky firms with high informational asymmetry, as these are firm characteristics

that are likely to make a seasoned equity offering relatively expensive.

20
Cash settlement allows firms to decide whether payment will be in common stock or in cash (for the
value of common stock). At the time of our study, accounting rules were such that the potential dilution
associated with the offer is not reflected in fully diluted earnings for most cash settlements (see Lewis and
Verwijmeren, 2009). Call spread overlays require a firm to use part of the issue proceeds to purchase call
options on its own shares, struck at the conversion price, and write call options on its own shares at a higher
strike price. The net effect is an increase of the strike price in the conversion option. Had the issuer simply
offered the convertible bond with a higher conversion price, the interest offered on the convertible would
have been higher and the firm would report higher interest expense.

22
 In Model 2 we use the percentage of buyers classified as hedge funds as the

dependent variable. This variable focuses on the number of hedge funds serving as

distributors, instead of the size of their involvement. The results are qualitatively similar

to those obtained from Model 1.

5. Determinants of the decision to distribute equity exposure via hedge funds

In this section we will compare seasoned equity issuers to firms that issue

convertibles to hedge funds. This analysis can shed more light on whether selling equity-

like convertibles to hedge funds is indeed a substitute for ‘would-be’ equity issuers with

relatively high costs of directly issuing seasoned equity.21 We obtain seasoned equity

offerings from the SDC database from January 2000 to March 2008. We impose the same

restrictions as on our sample of convertibles; i.e., we delete financial institutions, utilities,

and firms with missing data in Compustat, CRSP, or the Thomson-Reuters Institutional

Holdings database. Our seasoned equity sample consists of 2,198 offerings.22

We use a binary logit model to compare the firm characteristics of seasoned equity

issuers to the characteristics of the 584 convertible issues in which hedge funds purchase

the majority of the issue. Table 4 reports the results.

[ please insert Table 4 here ]

21
We are not the first to compare firms issuing seasoned equity to firms that issue convertible securities.
For example, Lewis, Rogalski, and Seward (1999) study Stein’s (1992) prediction that convertible debt
issuers have significantly higher adverse selection and financial distress costs than issuers of seasoned
equity. They find evidence in line with Stein’s prediction and conclude that the likelihood of an equity-like
convertible debt issue increases when the costs of a common stock issue are high.
22
A difference with our convertible sample is that not all our seasoned equity offerings are privately
placed. In our sample of 2,198 seasoned equity offerings, only 2 offerings are identified by SDC as 144A
private placements. The private placement information is missing for 1,412 of the seasoned equity
offerings.

23
The dependent variable equals one for convertible issues in which the majority of the

issue is purchased by hedge funds, and zero for seasoned equity offerings. We find that

firms issuing convertibles to hedge funds are more financially distressed than equity

issuers, as the Altman Z-score is negative and statistically significant at the 1% level.

Firms issuing convertibles to hedge funds also have more volatile stock returns than

equity issuers. We find an insignificant increase in the likelihood of issuing a convertible

for firms listed on NASDAQ.

The results in Table 4 show that firms issuing to hedge funds are significantly

larger companies with significantly more liquid stock and significantly higher

institutional ownership than are firms that issue seasoned equity. Greater liquidity and

higher institutional ownership make it easier for hedge funds to set up their desired short

positions. Combined, the results of Table 4 suggest that firms choosing to issue

convertibles to hedge funds rather than to issue seasoned equity are firms that face

relatively high costs of directly issuing equity and also have characteristics that facilitate

the establishment and maintenance of the short positions desired by hedge funds; i.e., that

also have relatively low costs of indirectly distributing equity exposure via hedge funds.

6. Offering discounts

Convertible securities are typically issued at a discount (Ammann, Kim, and Wilde,

2003; Chan and Chen, 2005; Loncarski, ter Horst, and Veld, 2009; De Jong, Dutordoir,

and Verwijmeren, 2009). Potential reasons for convertible debt underpricing include

illiquidity and complexities associated with the valuation of hybrid securities (Lhabitant,

2002). Studying these discounts allows us to distinguish the distribution role of hedge

24
funds from their possible role as a last resort provider of finance. As distributors of equity

exposure, hedge funds require a discount to cover their costs of shorting the underlying

stock. But we do not expect a discount that is substantially larger than any discount on

convertibles issued to other investors. Under the last resort hypothesis, we would expect

higher discounts for convertibles issued to hedge funds. Brophy, Ouimet, and Sialm

(2009) report that when the majority of a PIPE is sold to hedge funds the average

discount is 14.12%, whereas the average discount is significantly lower (9.02%) when the

majority of a PIPE is issued to non-hedge fund investors.

We follow Ammann, Kim, and Wilde (2003), Chan and Chen (2005), Loncarski,

ter Horst, and Veld (2009), and De Jong, Dutordoir, and Verwijmeren (2009) in using a

variant of the Tsiveriotis-Fernandes (1998) model to calculate the theoretical value of

convertibles. This model is argued to be the most popular convertible bond valuation

method among practitioners (Grimwood and Hodges, 2002). We use MATLAB’s

convertible bond pricing algorithm.23 The valuation requires some additional data: the

risk-free rate and credit spread data are taken from Datastream and matched as closely as

possible with the maturity of the convertible bond; the relevant credit rating for the

spread is obtained from Mergent (we assume a BBB rating when the credit rating of the

convertible is unavailable); and call schedules and coupon rates are taken from the issue

prospectuses. We calculate the offering discount by dividing the difference between the

theoretical price and the offer price by the theoretical price. We are able to calculate the

offering discount for 603 issues.24

23
See http://www.mathworks.com/access/helpdesk/help/toolbox/finfixed/cbprice.html.
24
We are not able to calculate the offering discount with MATLAB’s pricing algorithm for our full sample
of observations, as 26 convertibles in our sample have a floating or varying coupon rate.

25
 [ please insert Table 5 here ]

Panel A of Table 5 provides a univariate analysis of the average offering discount.

The average offering discount on the 603 issues is 4.9%. As in Brophy, Ouimet, and

Sialm (2009), we distinguish firms in which hedge funds purchase the majority of the

convertible issue from firms with a minority of hedge fund involvement. Firms with a

majority involvement of hedge funds have an average discount of 4.8% on their

convertible issues, while firms with a minority hedge fund involvement have an average

discount of 6.1%. The difference between the two subsamples is not significantly

different (t-statistic equals 0.510). The finding that the discount is not significantly

higher for issues to hedge funds is evidence against the view that convertibles are sold to

hedge funds as a last resort source of financing.

Panel B reports the results of a multivariate analysis. The dependent variable in

Model 1 is the offering discount and the primary explanatory variable is a dummy

variable that equals one when hedge funds buy the majority of the security offering and

zero otherwise.25 We again find no evidence that hedge funds obtain higher discounts

than other investors. In fact, discounts are insignificantly smaller when the majority of

the issue is sold to hedge fund investors.

The control variables show that the offering discount on the convertibles in our

sample is higher for firms with more volatile stock returns, for smaller firms, and for

firms with more illiquid shares. We find that firms with a listing on NASDAQ provide

25
We follow Brophy, Ouimet, and Sialm (2009) in measuring hedge fund involvement with a dummy
variable that equals one when we observe that hedge funds buy the majority of the security offering, and
that equals zero otherwise. The results are robust to measuring the hedge fund involvement dummy as
either an above-median percentage of the issue purchased by hedge funds or an above-median percentage
of initial buyers of the issue classified as hedge fund buyers.

26
lower discounts on their convertible offerings, all else equal, which is different from

findings in the seasoned equity market. We also find that put rights have a strongly

negative impact on our measure of the offering discount. This finding is sensible as

MATLAB’s convertible bond pricing algorithm does not take put features into account. In

Model 2 we re-estimate our model for only those convertibles with no put features

attached. The results are relatively similar to those from Model 1 in that hedge funds do

not obtain higher discounts than other investors.

Model 3 is a two-stage model. As the offering discount and the involvement of

hedge funds could both be driven by a common unobserved factor, we also include the

inverse Mills ratio. The first stage of the model consists of a probit model regressing the

majority-hedge-fund-involvement dummy variable on the same set of issue and issuer

characteristics examined in Table 3. We calculate the inverse Mills ratio from the first-

stage estimation as in Heckman (1979) and include this ratio as an additional control

variable in the second-stage regression.26

The relation between the offering discount and the involvement of hedge funds is

again negative, and again statistically insignificant. Thus although hedge funds purchase

convertibles at a discount, that discount is not significantly larger than the discount given

to other investors in convertibles. This finding is consistent with convertible bond issues

to hedge funds being a low-cost way of distributing equity exposure, rather than the last

resort financing opportunity open to the issuers.

26
The variables dividend-paying, institutional ownership, and call spread overlay are included in the first
stage of the model, but not in the second stage as these variables do not have a strong predicted impact on
the offering discount. Adding any combination of these variables to the second-stage regression does not
lead to a statistically significant relation between the offering discount and the involvement of hedge funds.

27
7. The establishment of short positions by hedge funds

Choi, Getmansky, and Tookes (2009) examine changes in monthly short interest

data around convertible issue dates and De Jong, Dutordoir, and Verwijmeren (2009)

examine changes in daily short interest data around convertible issue dates. Both sets of

authors document a peak in short interest at the time of the issue. Our data allows us to

directly tie the peak in short interest to the arbitrage activities of hedge fund investors.

We relate the increase in monthly short interest to the increase expected as hedge

funds establish delta-neutral hedge positions. The expected volume of shorting by hedge

funds is the product of the percentage of an issue purchased by hedge funds with the

delta-neutral short position for the entire issue:

face value of entire issue

Expected short sales   delta  % purchased by hedge funds.
conversion price

We use the Compustat mid-month short interest files (available from 2003) to

compute the actual monthly change in short interest.27 We are able to observe the actual

change in monthly short interest at the time of issue for 308 of the convertible bonds in

our sample. The correlation between expected changes and actual changes is 66.4%

(significant at the 1% level). This high correlation coefficient implies that the

contemporaneous jump in short interest is in fact the result of hedge fund hedging.

8. The potential maintenance of short positions by hedge funds

If hedge funds are least-cost distributors of equity exposure a question arises as to

how long hedge funds maintain their ownership of a newly issued bond; i.e., for how long

27
As the cutoff date of the monthly short interest numbers is three trading days prior to the 15th of each
month, we carefully assign the convertible issues to the relevant short interest month (as in Bechmann,
2004).

28
might hedge funds bear the cost of maintaining a short position? If hedge funds were to

never sell convertible bonds on to non-hedge fund investors, they would need to hedge

for the realized life of the bond. While a convertible bond can survive until its notional

maturity, the issuer may go bankrupt.28 Or, the convertible holder may voluntarily

convert or a call may force conversion. Finally, the convertible holder may exercise a put

option inherent in some bonds. Therefore, in order to determine an upper bound on just

how long hedge funds may remain short in an issuer’s stock, we investigate how long

convertible issues survive in practice.

Using balance sheet data (Compustat Item DCVT) we determine the fraction of a

convertible still outstanding in the five years after issue. We examine only instances

where the firm did not already have convertibles outstanding (from an earlier offering) at

the time of issue since we are unable to assign aggregate reductions in a firm’s

convertibles outstanding to particular issues. We assign a missing value code to a

particular year for a given bond if a second convertible bond is issued by that firm. We

also assign a missing value code if the firm no longer exists.

[ please insert Table 6 here ]

Table 6 reports the average (across the non-missing data) of the percentage of the

original issue still outstanding each year after issue. This average is likely to be an

upward biased estimate of the fraction of a convertible issue that actually survives for two

reasons. First, if a firm no longer exists its convertible bonds may have effectively

28
Interestingly, in court-administered reorganizations convertible bondholders have at times been awarded
only the same amount as that awarded to the holder of the number of shares the bond was convertible into;
i.e., bondholders were effectively forced to convert just when they would have most valued down-side
protection.

29
“matured” as a result of bankruptcy. Or a firm may no longer exist because it has been

taken-over, in which case the bondholders may have voluntarily converted. Second,

instances where a firm makes a second convertible bond issue may be cases where the

first bond was called (and the holders were forced to convert). In both of these events, 0%

of the issue remains outstanding. The number of observations used in calculating the

average each year is reported in parentheses. In forming subsamples, we determine

whether a convertible bond issue fell into the below-median or the above-median hedge

fund involvement subsample. We do not consider the minority and majority hedge fund

involvement distinction in this analysis because of the limited number of observations

with minority hedge fund involvement (only 5 observations with minority hedge fund

involvement have a value for “% outstanding after 5 years”).

On average no more than 13.38% of convertibles are still outstanding five years

after issue. For bonds with above-median issuance to hedge funds, the average

survivorship rate after five years declines to 7.82%. Since the mean (median) time to

maturity at issuance of our sample of 629 privately-placed convertible bond issues is

15.26 years (20 years), we can conclude that the effective life of convertible bonds

privately-placed with hedge funds is much less than their notional maturity.29,30

29
The conversion of a high fraction of the convertible bonds issued is consistent with the sample’s mean
(median) conversion premium at the time of issue of 33.81% (30%). The conversion premium is the excess
of the conversion price over the stock price expressed as a percentage of the stock price. Equivalently, the
conversion premium is the face value of the bond relative to its conversion value at the time of issue.
30
We have also examined how short interest changes in the years after a convertible’s issue. This analysis
can potentially provide insights into how long hedge funds maintain their short positions. Although the
results seem to indicate that the initial increase in short interest has evaporated 5 years after the issue,
changes in short interest can be quite volatile. The analysis is also complicated as the expected short
positions of arbitrageurs change when the firm’s stock price changes (due to changes in the delta-neutral
position), and changes in short interest can also be caused by the arrival of new information about the firm.
As mentioned in Section 2.1, another downside of focusing on short interest is that the initial increase in
short interest around the issue date could be partly due to valuation shorting by those who believe their
information is not appropriately reflected in prices.

30
9. Conclusion

Although the majority of buyers in the convertible market are not hedge funds, we

find that hedge funds account on average for 73.4% of the proceeds of privately-placed

convertible issues, and the bulk of capital raised via convertibles is via private-placement.

The typical hedge fund buyer is involved in a larger number of new issues of convertibles

than is the typical non-hedge fund buyer. Also, the typical hedge fund buyer invests more

in any particular new convertible than the typical non-hedge fund buyer invests.

We exploit the variation in hedge fund involvement across issues to show that

investor identity matters in the convertible debt market. We argue that firms with high

costs of seasoned equity offerings privately place convertibles with hedge funds. These

hedge funds simultaneously short stock so as to hedge themselves against changes in the

stock price. In effect, hedge funds are distributing equity exposure into the open market

via their short positions. Firms that have trouble finding investors willing to bear

exposure to large equity positions can instead use hedge funds to distribute this equity

exposure to a larger number of well-diversified investors in the open market.

Although we find that firms attracting hedge fund investors have a higher chance of

bankruptcy and more volatile stock returns, we do not find that these firms offer higher

discounts on their convertibles. This is in line with hedge funds serving as relatively

cheap distributors and not as hedge funds being investors of last resort. The difference

with the findings of Brophy, Ouimet, and Sialm (2009), who study PIPE issuers and do

find evidence for the last resort hypothesis, can be explained by differences in the issuers

of PIPEs and convertibles. PIPE issuers are generally very small and distressed firms,

while the convertible issuers in our sample are much larger. The firms in our sample

31
choose to issue convertibles to hedge funds because this is their least-cost financing

choice, not because it is their only choice.

32
 Appendix A: Description of variables

Amihud Liquidity. The Amihud (2002) measure for liquidity is the daily average of a
firm’s absolute return over trading dollar volume in the year before the offering (× 106).
We use CRSP data to calculate this variable. A high Amihud score denotes illiquidity.

Call spread overlay. A dummy variable equaling one if the convertible is issued with a
call spread overlay, and zero otherwise. In a call spread overlay, the issuer purchases a
call option that mimics the call option embedded in the convertible bond and
simultaneously writes a call option on the same number of underlying shares at a higher
strike price. The net effect is analogous to issuing a convertible bond with a higher
conversion price. We obtain information on call spread overlays from the issue
prospectuses.

Callable. A dummy variable equaling one if the convertible has a call feature, and zero
otherwise. We obtain call information from SDC.

Cash settlement. A dummy variable equaling one if the convertible can potentially be
settled in cash, and zero otherwise. Settlement information is obtained from the issue
prospectuses, as in Lewis and Verwijmeren (2009). Securities that contain cash
settlement features include one of the following conversion choices. Either an issuer 1)
must pay the conversion value (the number of shares a bondholder is entitled to receive
times the stock price at the conversion date) in cash (Instrument A); 2) may choose to
pay either fully in cash or the number of shares a bondholder is entitled to receive
(Instrument B), 3) must pay cash for the accreted value (principal value plus accrued
interest) and may satisfy the conversion spread (the excess of the conversion value over
the accreted value) in either cash or equity (Instrument C or net share settlements), or 4)
may pay any combination of cash and equity (Instrument X), but often there is a stated
policy to settle in cash.

Delta. Delta is the convertible’s sensitivity for small stock price changes and is calculated
as

33
  S 2 
 ln( )  ( r    )T 
T T X 2 
Delta  e N (d1 )  e N  ,
  T 
 

where N(•) is the cumulative probability under a standard normal distribution, S is the
price of the underlying stock measured at day 5 (from CRSP), X is the conversion
price (from SDC), r is the yield on a 10-year US Treasury Bond (from Datastream), δ is
the continuously-compounded dividend yield calculated as Compustat Item DVC
divided by the equity market value (Item PRCC_C x CSHO), σ is the annualized stock
return volatility, estimated with ten years of monthly stock return data (from CRSP),
and T represents the stated maturity of the convertible as of its issuance date (from
SDC). We set the dividend yield to zero if the convertible is protected from dividend
payments.

Delta-neutral short position. The delta-neutral short position is the total number of
common shares that buyers of the convertible issue have to short to obtain a delta-
neutral position. This number of shares can be calculated as (Calamos, 2003):
face value of entire issue
Delta-neutral short position   delta.
conversion price

We obtain information on the face value and the conversion price from SDC.

Dividend-paying. A dummy variable equaling one if Compustat Item DVC exceeds zero
at the beginning of the fiscal year, and zero otherwise.

Financial Distress (Z-score). The Altman Z-score is an estimate of the probability of a

firm’s bankruptcy, and is calculated as 1.2 (Working capital / Total assets) + 1.4
(Retained earnings / Total assets) + 3.3 (EBIT / Total assets) + 0.6 (Market value of
equity / Book value of liabilities) + (Sales / Total assets), which is implemented using
Compustat Items as 1.2 ((ACT  LCT) / AT) + 1.4 (RE / AT) + 3.3 (OIADP / AT) +
0.6 ((PRCC_C  CSHO) / (DLTT + DLC)) + SALE / AT. All variables are measured at
the beginning of the fiscal year. Z-scores above 100 and below 100 are winsorized.

34
Firm size. Firm size is the market value of equity (measured at the beginning of the fiscal
year) calculated with Compustat Items PRCC_C × CSHO. In regression analyses we
employ the natural logarithm of the market value of equity as a measure for firm size.

Institutional ownership. The level of institutional ownership reported in the Thomson-

Reuters Institutional Holdings database scaled by total shares outstanding (both
measured at the fiscal-year end preceding the issue date).

Issue proceeds. Issue proceeds are the gross proceeds of the issue in millions of dollars,
as reported in SDC.

NASDAQ-listing. A dummy variable equal to one if the firm is listed on NASDAQ

(Compustat Stock Exchange Code 14), and zero otherwise.

Put rights. A dummy variable equaling one if the convertible has a put feature, and zero
otherwise. We obtain put information from SDC.

Relative size. A measure of the additional equity exposure being distributed compared to
the initial equity exposure. When comparing convertible bonds with each other we
measure relative size as the ratio of the delta-neutral short position to the shares
outstanding. When comparing convertible issues to hedge funds with seasoned equity
offerings we measure relative size as the ratio of the issue proceeds to the market value
of equity.

Return volatility. Volatility is calculated as the annualized standard deviation of monthly

stock return data (from CRSP) for ten years of data (or for a shorter period if ten years
of data are not available; we set the minimum requirement to twelve months of data).

Stock repurchase. A dummy variable equaling one if the convertible issue is combined
with a stock repurchase. This is the case if either the firm announces that it uses part of
the proceeds of the convertible issue to repurchase stock (in SDC or Factiva), or if both
transactions are announced separately on the same date (in Factiva).

35
 References

Ackert, Lucy F. and George Athanassakos, 2005, “The relationship between short interest
and stock returns in the Canadian market,” Journal of Banking and Finance 29, 1729-
1749.

Agarwal, Vikas, Wiliam H. Fung, Yee Cheng Loon and Narayan Y. Naik, 2009, “Risk
and return in convertible arbitrage: Evidence from the convertible bond market,”
London Business School working paper.

Altinkilic, Oya and Robert S. Hansen, 2003, “Discounting and underpricing in seasoned
equity offers,” Journal of Financial Economics 69, 285-323.

Amihud, Yakov, 2002, “Illiquidity and stock returns: Cross-section and time-series
effects,” Journal of Financial Markets 5, 31-56.

Ammann, Manuel, Axel Kind and Christian Wilde, 2003, “Are convertible bonds
underpriced? An analysis of the French market,” Journal of Banking and Finance 27,
635–653.

Asquith, Paul, Parag A. Pathak and Jay R. Ritter, 2005, “Short interest, institutional
ownership, and stock returns,” Journal of Financial Economics 78, 243-276.

Bechmann, Ken L., 2004, “Short sales, price pressure, and the stock price response to
convertible bond calls,” Journal of Financial Markets 7, 427-451.

Brent, Averil, Dale Morse and E. Kay Stice, 1990, “Short interest: Explanations and
tests,” Journal of Financial and Quantitative Analysis 25, 273-289.

Brophy, David J., Paige P. Ouimet and Clemens Sialm, 2009, “Hedge funds as investors
of last resort,” Review of Financial Studies 22, 541-574.

Calamos, Nick, 2003, Convertible Arbitrage: Insights and Techniques for Successful
Hedging, John Wiley & Sons, Hoboken, New Jersey.

Chan, Alex W. H. and Naifu Chen, 2005, “Convertible bond underpricing: Renegotiable
covenants, seasoning and convergence,” Management Science 52, 1793-1814.

Chaplinsky, Susan and David Haushalter, 2009, “Financing under extreme risk: Contract
terms and returns to private investments in public equities,” Review of Financial
Studies, forthcoming.

36
Choi, Darwin, Mila Getmansky, Brian Henderson and Heather Tookes, 2009,
“Convertible bond arbitrageurs as suppliers of capital,” Review of Financial Studies,
forthcoming.

Choi, Darwin, Mila Getmansky and Heather Tookes, 2009, “Convertible bond arbitrage,
liquidity externalities and stock prices,” Journal of Financial Economics 91, 227-251.

Corwin, Shane A., 2003, “The determinants of underpricing for seasoned equity
offerings,” Journal of Finance 58, 2249-2279.

D’Avolio, Gene, 2002, “The market for borrowing stock,” Journal of Financial
Economics 66, 271-306.

Dai, Na, 2007, “Does investor identity matter? An empirical examination of investments
by venture capital funds and hedge funds in PIPEs,” Journal of Corporate Finance 13,
538-563.

de Jong, Abe, Eric Duca and Marie Dutordoir, 2009, “Do convertible bond issuers cater
to investor demand?” Erasmus University working paper.

de Jong, Abe, Marie Dutordoir and Patrick Verwijmeren, 2009, “Why do convertible
issuers simultaneously repurchase stock? An arbitrage-based explanation,” Erasmus
University working paper.

Eckbo, B. Espen, Ronald W. Masulis and Oyvind Norli, 2007, “Security offerings,” in
Eckbo, B. (ed.), Handbook of Corporate Finance: Empirical Corporate Finance, vol.
1, ch. 4 (Elsevier/North-Holland, Handbooks in Finance Series).

Ellis, Katrina and Garry J. Twite, 2008, “Are PIPEs a bet on growth options?” Australian
National University working paper.

Fung, William and David A. Hsieh, 1997, “Empirical characteristics of dynamic trading
strategies: The case of hedge funds,” Review of Financial Studies 10, 275-302.

Fung, William and David A. Hsieh, 1999, “A primer on hedge funds,” Journal of
Empirical Finance 6, 309-331.

Gompers, Paul and Josh Lerner, 1999, “Conflict of interest in the issuance of public
securities: Evidence from venture capital,” Journal of Law and Economics 42, 53-80.

37
Grimwood, Russell and Stewart Hodges, 2002, “The valuation of convertible bonds: A
study of alternative pricing models,” University of Warwick working paper.

Hansen, Robert S. and Paul Torregrosa, 1992, “Underwriter compensation and corporate
monitoring,” Journal of Finance 47, 1537-1555.

Heckman, James J., 1979, “Sample selection bias as a specification error,” Econometrica
47, 153-161.

Kim, Dongcheol, Darius Palia and Anthony Saunders, 2005, “Are initial returns and
underwriting spreads in equity issues complements or substitutes?” Rutgers University
working paper.

Korkeamaki, Timo P. and William T. Moore, 2004, “Convertible bond design and capital
investment: The role of call provisions,” Journal of Finance 59, 391-405.

Lewis, Craig M., Richard J. Rogalski and James K. Seward, 1999, “Is convertible debt a
substitute for straight debt or for common equity?” Financial Management 28, 5-27.

Lewis, Craig M. and Patrick Verwijmeren, 2009, “Convertible security design and
contract innovation,” Vanderbilt University and University of Melbourne working
paper.

Lhabitant, François-Serge, 2002, Hedge Funds: Myths and Limits, John Wiley & Sons,
London.

Loncarski, Igor, Jenke ter Horst and Chris Veld, 2009, “The rise and demise of the
convertible arbitrage strategy,” Financial Analysts Journal 65, 35-50.

Marquardt, Christine I. and Carol A. Wiedman, 2005, “Earnings management through

transaction structuring: Contingent convertible debt and diluted EPS,” Journal of
Accounting Research 43, 205-243.

Mitchell, Mark, Lasse Heje Pedersen and Todd Pulvino, 2007, “Slow moving capital,”
American Economic Review 97, 215-220.

Petersen, Mitchell A., 2009, “Estimating standard errors in finance panel data sets:
Comparing approaches,” Review of Financial Studies 22, 435-480.

Pulliam, Susan, 2004, “Mixed blessing: How hedge-fund trading sent a company’s stock
on wild ride,” Wall Street Journal (28/12/2004).

38
Skorecki, Alex, 2004, “Convertibles enter era of change: The strategy of selling to hedge
funds was always controversial,” Financial Times (16/01/2004).

Stein, Jeremy C., 1992, “Convertible bonds as backdoor equity financing,” Journal of
Financial Economics 32, 3-21.

Tsiveriotis, Kostas and Chris Fernandes, 1998, “Valuing convertible bonds with credit
risk,” The Journal of Fixed Income 8, 95-102.

Wruck, Karen H., 1989, “Equity ownership concentration and firm value: Evidence from
private equity financings,” Journal of Financial Economics 23, 3–27.

39
 Figure 1

100.0%

90.0%

80.0%

70.0%

60.0%

50.0%

40.0%

30.0%

20.0%

10.0%

0.0%
2000 2001 2002 2003 2004 2005 2006 2007 2008

Figure 1. Percentage of convertible proceeds purchased by hedge funds

This figure shows the average involvement of hedge funds in convertible securities that
are privately placed under Rule 144A in a given year. The year 2008 percentage applies
to convertible issues during the first quarter of the year.

40
Table 1. Hedge fund involvement and industry classifications

The sample period is January 2000 – March 2008. Panel A of this table reports the
number of buyers per privately placed convertible issue in our sample. Panel A also
reports the percentage of buyers in a convertible security that can be classified as a hedge
fund as well as the fraction of the issue purchased by hedge funds. Panel B shows the
fraction of the issue purchased by hedge funds for issues of firms in each of the 12 Fama-
French industry classifications.

Panel A.
N Mean Median St.dev. Min. Max.
Number of buyers 803 64.1 54.0 44.6 1 320
% of buyers classified 803 56.4% 55.6% 18.6% 0.0% 100.0%
as hedge funds
% of issue purchased 803 73.4% 75.3% 15.9% 0.0% 100.0%
by hedge funds

Panel B.
N Mean Median St.dev. Min. Max.
Consumer nondurables 15 66.4% 73.3% 22.9% 3.0% 90.4%
Consumer durables 9 65.0% 65.8% 14.7% 47.7% 82.8%
Manufacturing 50 70.0% 73.7% 19.0% 1.5% 100.0%
Energy 41 72.9% 74.1% 19.3% 0.0% 99.3%
Chemicals 9 78.8% 83.6% 18.0% 38.5% 100.0%
Business equipment 224 72.8% 75.4% 16.1% 10.8% 100.0%
Telecommunications 33 72.8% 71.6% 11.1% 47.4% 100.0%
Utilities 22 72.4% 75.4% 19.3% 7.2% 94.5%
Wholesale & retail 53 72.6% 71.6% 14.4% 40.3% 97.9%
Healthcare 160 75.5% 75.9% 14.4% 34.6% 100.0%
Financial 90 74.3% 77.8% 17.0% 0.0% 100.0%
Other 97 75.2% 75.7% 13.5% 39.2% 100.0%

41
 Table 2. Issuer and issue characteristics and hedge fund involvement
The sample period is January 2000 – March 2008. Issues by utilities and financial
institutions are excluded. This table presents a univariate analysis of the differences in
issuer and issue characteristics for privately placed convertible issues with an above-
median participation by hedge funds (hedge funds purchase more than 75.3% of the
convertible) and a below-median participation of hedge funds (less than 75.3%). The
table also reports the differences for issues in which the majority is purchased by hedge
funds versus issues with only a minority purchased by hedge funds. The difference in
means t-statistics do not assume equal variances for the two samples being compared. See
Appendix A for a description of issuer and issue characteristics. *, ** and *** indicate
significance at the 10%, 5% and 1% levels.

All Subsamples with differing levels of

Issues hedge fund involvement
Above- Below- Diff. of Diff. of
median median means t- Majority Minority means t-
statistic statistic
Issuer characteristics
Financial Distress (Z-score) 16.070 13.813 18.349 1.907* 15.946 17.674 0.367
Return volatility 0.326 0.339 0.312 2.898*** 0.327 0.303 1.393
NASDAQ-listing 0.536 0.595 0.476 3.006*** 0.543 0.444 1.266
Firm size ($ mill) 4143 3860 4429 0.609 4299 2126 3.555***
Dividend-paying 0.197 0.168 0.227 1.865* 0.200 0.156 0.784
Institutional ownership 0.680 0.684 0.676 0.481 0.677 0.721 1.306
Relative size: Delta-neutral 0.134 0.138 0.130 0.715 0.130 0.180 1.323
short position  shares
outstanding
Amihud Liquidity 0.030 0.024 0.037 0.581 0.031 0.018 0.926

Issue characteristics
Callable 0.723 0.646 0.802 4.448*** 0.714 0.844 2.258**
Stock repurchase 0.116 0.152 0.080 2.837*** 0.123 0.022 3.878***
Call spread overlay 0.081 0.127 0.035 4.265*** 0.086 0.022 2.530**
Cash settlement 0.404 0.494 0.313 4.689*** 0.411 0.311 1.373
Put rights 0.424 0.421 0.428 0.183 0.425 0.422 0.032
Delta 0.867 0.860 0.873 1.681* 0.865 0.888 1.834*
Issue proceeds ($ mill) 313 304 322 0.478 319 241 2.537**

42
 Table 3. Hedge fund involvement, issuer characteristics, and issue characteristics

The sample period is January 2000 – March 2008. Issues by utilities and financial
institutions are excluded. This table presents the results of an OLS regression model
estimating the relation between various issuer and issue characteristics and the
involvement of hedge funds. The dependent variable in Model 1 is the percentage of the
issue purchased by hedge funds. The dependent variable in Model 2 is the percentage of
the issue’s buyers classified as hedge funds. See Appendix A for a description of the
issue and issuer characteristics. We report standard errors clustered at the firm level in
parentheses. *, **, and *** indicate significance at the 10%, 5%, and 1% levels.
% of issue
% of purchasers
Prediction purchased by
classified as hedge funds
hedge funds
(1) (2)
Constant 0.874*** 0.769***
(0.061) (0.079)
Financial distress (Z-score)  0.001** 0.001*
(0.000) (0.000)
Return volatility + 0.214*** 0.317***
(0.071) (0.083)
NASDAQ-listing + 0.025** 0.024*
(0.014) (0.014)
Firm size 0.022*** 0.037***
(0.005) (0.007)
Dividend-paying  0.025 0.030
(0.016) (0.020)
Institutional ownership + 0.045 0.069*
(0.030) (0.041)
Relative size: Delta-neutral short  0.200*** 0.204***
position  shares outstanding (0.053) (0.076)
Amihud Liquidity  0.004 0.005
(0.008) (0.011)
Callable  0.055*** 0.040**
(0.016) (0.020)
Stock repurchase + 0.030* 0.019
(0.017) (0.024)
Call spread overlay 0.054** 0.050**
(0.022) (0.031)
Cash settlement 0.034** 0.030*
(0.013) (0.017)
Put rights 0.036*** 0.065***
(0.014) (0.017)
N 629 629
R2 0.132 0.137

43
Table 4. Determinants of the decision to distribute equity exposure via hedge funds
This table compares convertible issues in which the majority of the issue is purchased by
hedge funds with seasoned equity offerings. The sample period is January 2000 – March
2008. Issues by utilities and financial institutions are excluded. The dependent variable of
the binary logit model is a dummy variable equal to one for convertible issues in which
the majority of the issue is purchased by hedge funds and to zero for seasoned equity
offerings. The elasticities are in the form d(lny)/d(lnx). See Appendix A for a description
of the issue and issuer characteristics. * and *** indicate significance at the 10% and 1%
levels.

Convertible issues to hedge funds versus

seasoned equity issues
Coefficients Elasticities
Constant 3.989***
(0.475)
Financial distress (Z-score) 0.008*** 0.151
(0.002)
Return volatility 0.703*** 0.420
(0.207)
NASDAQ-listing 0.122 0.064
(0.138)
Firm size 0.252*** 1.354
(0.059)
Dividend-paying 0.239 0.032
(0.179)
Institutional ownership 1.785*** 0.775
(0.257)
Relative size: proceeds  market 0.045* 0.010
value of equity (0.024)
Amihud Liquidity 11.341*** 0.600
(1.941)

N 2,782
Pseudo R2 0.172

44
Table 5. Hedge fund involvement and offering discounts

The sample period is January 2000 – March 2008. Issues by utilities and financial
institutions are excluded. Panel A reports the offering discount for our overall sample,
issues in which the majority of the convertible is bought by hedge funds, and issues in
which the minority of the convertible is bought by hedge funds. The offering discount is
the difference between the theoretical price and the offer price relative to the theoretical
price. Theoretical prices are calculated using MATLAB’s pricing algorithm:
http://www.mathworks.com/access/helpdesk/help/toolbox/finfixed/cbprice.html. The
difference in means t-statistic does not assume equal variances for the two samples being
compared. Panel B presents the estimated relation between the offering discount and the
involvement of hedge funds. In Model 1 and 2 we estimate an OLS regression model;
Model 2 excludes convertible bonds that have put features attached. In Model 3 we
estimate a two-stage model. The first stage of the two-stage model consists of a probit
model in which the dependant variable is a dummy corresponding to instances where the
majority of the issue is purchased by hedge funds and the explanatory variables are
various issuer and issue characteristics (similar to the regressions reported in Table 3).
We calculate the inverse Mills ratio from the first-stage estimation as in Heckman (1979)
and include this ratio as an additional control variable in the second-stage regression. See
Appendix A for a description of the issue and issuer characteristics. We report standard
errors clustered at the firm level in parentheses. *, ** and *** indicate significance at the
10%, 5% and 1% levels.

Panel A.
Issues with a Issues with a
Difference of
majority of the minority of the
All Issues means t-
issue purchased issue purchased
statistic
by hedge funds by hedge funds

Offering discount 0.049 0.048 0.061 0.510

45
Panel B.
Offering discount
(1) (2) (3)
Constant 0.163*** 0.109* 0.152*
(0.059) (0.060) (0.078)
Majority purchased by hedge funds 0.020 0.010 0.008
(0.026) (0.019) (0.064)
Financial Distress (Z-score) 0.000*** 0.000* 0.000**
(0.000) (0.000) (0.000)
Return volatility 0.236*** 0.366*** 0.234***
(0.056) (0.063) (0.060)
NASDAQ-listing 0.031** 0.034** 0.031**
(0.014) (0.016) (0.013)
Firm size 0.017*** 0.017** 0.017***
(0.006) (0.007) (0.005)
Relative size: Delta-neutral short position  0.044 0.045 0.047
shares outstanding (0.064) (0.084) (0.060)
Amihud Liquidity 0.033*** 0.074 0.033
(0.010) (0.110) (0.021)
Callable 0.020 0.018 0.019
(0.013) (0.016) (0.014)
Stock repurchase 0.025 0.044** 0.024
(0.017) (0.021) (0.018)
Cash settlement 0.011 0.020 0.011
(0.013) (0.019) (0.012)
Put rights 0.084*** 0.084***
(0.014) (0.013)
Inverse Mills ratio 0.006
(0.031)

N 603 353 603

R2 0.227 0.192

46
Table 6. Estimated percentage of convertible bond issue outstanding in years after
issue

The sample period is January 2000 – March 2008. Issues by utilities and financial
institutions are excluded. In this table we determine the fraction of a convertible still
outstanding in the five years after issue. We assign a missing value code to a particular
year for a given bond if a second convertible bond is issued by that firm. We also assign a
missing value code if the firm no longer exists. The number of issues used in calculating
the average is shown in parentheses. An above-median participation by hedge funds
indicates that hedge funds purchase more than 75.3% of the convertible. The difference
in means t-statistics do not assume equal variances for the two samples being compared.
* indicates significance at the 10% level.

Above- Below-
Difference of
median hedge median hedge
% outstanding after n years Total means t-
fund fund
statistic
involvement involvement
94.22% 93.02% 95.50%
1 year 1.12
(280) (144) (136)
86.69% 83.17% 90.17%
2 years 1.73*
(221) (110) (111)
65.40% 64.14% 66.50%
3 years 0.35
(172) (80) (92)
41.42% 33.45% 47.71%
4 years 1.89*
(145) (64) (81)
13.38% 7.82% 18.15%
5 years 1.73*
(104) (48) (56)

47