Corporate bond pricing and ownership heterogeneity

    Corporate Bond Pricing and Ownership Heterogeneity Kershen Huang, Alex Petkevich PII: DOI: Reference:

S0929-1199(15)00131-5 doi: 10.1016/j.jcorpfin.2015.11.001 CORFIN 976

To appear in:

Journal of Corporate Finance

Received date: Revised date: Accepted date:

31 January 2015 30 October 2015 3 November 2015

Please cite this article as: Huang, Kershen, Petkevich, Alex, Corporate Bond Pricing and Ownership Heterogeneity, Journal of Corporate Finance (2015), doi: 10.1016/j.jcorpfin.2015.11.001

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT Corporate Bond Pricing and Ownership Heterogeneity∗ Kershen Huang

RI P

Alex Petkevich∗

T

McCallum School of Business, Bentley University, Waltham, MA 02452

SC

Department of Finance, College of Business and Innovation, University of Toledo, Toledo, OH 43606

NU

Abstract

We examine how heterogeneity in institutional equity ownership affects bondholders. Firms

MA

with larger short-term (long-term) institutional ownership are associated with higher (lower) future bond yield spreads. The adverse effect of short-term ownership on bond pricing is driven by issuing firms that have larger financial distress risk and larger equity volatility.

ED

The favorable effect of long-term ownership appears to be more systematic. Further, this bond pricing effect is stronger in cases where shareholder rights are relatively weak. Finally,

CE

PT

the effect of short (long) horizons is driven by concentrated (diffused) institutional holdings.

JEL Classification: G12, G32, G34

AC

Keywords: Agency cost of debt, Information, Institutional investors, Investment horizon

Corresponding author. Tel.: +1 419 530 2548; fax: +1 419 530 2873. E-mail address: [email protected]. ∗ We thank Colin Campbell, Doina Chichernea, Stephen Christophe, Len Rosenthal, Tommy Thompson, the seminar participants at Bentley University and the University of Toledo, and the conference participants at the 2015 Financial Management Association (FMA) European Conference in Venice, Italy and the 2015 Southwestern Finance Association (SWFA) Annual Meeting in Houston, TX for insightful comments and suggestions. We also thank Jeffry Netter (the editor) and an anonymous referee for their helpful comments. Any remaining errors or omissions remain the responsibility of the authors. ∗

ACCEPTED MANUSCRIPT 1. Introduction The debt and equity holders of a firm are claimants to the same aggregate value. The

T

two parties, however, differ in their payoff structures. Creditors are protected by the abso-

RI P

lute priority rule, but unlike shareholders, they do not fully enjoy the upside potential of firm value; this fundamental difference leads to discrepancies in their attitudes toward risk.

SC

Effectively, shareholders hold a call option on firm assets and, therefore, prefer the higher risk (Merton, 1974). If shareholders have the power to impact management decisions and

NU

firm policies in ways that are consistent with their own incentives, the higher risk that they introduce would at some point hurt creditor wealth. Therefore, the characteristics of the

MA

shareholders for the debt are essential.

Within the composition of the shareholder base at the firm level, institutions most likely

ED

fit the profile of “large minority shareholders” that have the ability to mitigate information asymmetry problems, and they are thus suitable for our study (Grossman and Hart, 1980;

PT

Shleifer and Vishny, 1986). The impact of institutional investors on creditor wealth can be

CE

mixed. On the one hand, they may improve the information environment (Sengupta, 1998; Healy et al., 1999) and mitigate shareholder-manager conflicts such as shirking (Jensen and Meckling,

AC

1976), empire building (Jensen, 1986), and inefficient compensation (Murphy, 1985). When institutions alleviate problems that hurt all stakeholders, creditors are better off. On the other hand, by aligning management incentives with those of their own, they may create shareholder-creditor conflicts due to different payoff structures (Jensen and Meckling, 1976; Myers, 1977). Under these situations, creditors can be worse off. In either case, creditors should take into consideration the presence of institutions when pricing their debt investments. Prior research has provided aggregate evidence that institutional investors are overall beneficial to bond pricing, but the benefits are reversed in cases where institutional holdings are concentrated (Bhojraj and Sengupta, 2003). In this paper, we examine this bond pricing effect further by introducing heterogeneity 1

ACCEPTED MANUSCRIPT among institutions. Specifically, we study how the equity investment horizons of institutions affect the corporate bond pricing of investee firms. We assert that when institutions are

T

short-term oriented, they may pressure management to adopt such a focus, even at the cost

RI P

of long-term value (Bushee, 2001; von Thadden, 1995). The mere focus on short-term profits for shareholders more likely benefits them at the expense of creditors. For instance, firms may

SC

undertake sufficiently risky investment projects that have negative NPVs or cut important expenses to boost earnings despite profitable future growth opportunities – in both examples,

NU

conflict between shareholders and creditors is exacerbated.2 Shareholders in these conflicts benefit at the expense of existing creditors, but eventually they bear the costs. All else

MA

equal, higher yields lead to higher future borrowing costs that hurt shareholders. Therefore, when institutions plan to hold their equity investments for a longer period, they may want

ED

to incentivize managers to achieve a higher firm value. This can stem from a lower cost of debt that not only increases the amount that the firm can distribute to its shareholders but

PT

also strengthens its ability to raise capital and profit from growth opportunities. As a result, creditors may demand a higher (lower) return on their debt securities when observing short-

CE

term (long-term) equity investors. To capture this characteristic of institutions, we use an

AC

empirically observable proxy based on portfolio turnovers to categorize institutional equity investors into two groups: short-term and long-term institutional owners (Gaspar et al., 2005; Yan and Zhang, 2009). We focus on bondholders for three reasons. First, bonds are an important source of financing. According to data on flows of funds to corporations, public debt financing makes up 32% of external financing for nonfinancial businesses in the US, compared to 11% from new stock issuances (Hackethal and Schmidt, 2004).3 Second, bonds provide us with a cleaner 2

These are related to asset substitution (Jensen and Meckling, 1976) and debt overhang (Myers, 1977) problems, respectively. 3 Considering that debt securities mature, the ability to raise funds through corporate debt is even more important than what these numbers indicate.

2

ACCEPTED MANUSCRIPT setting to examine the impact of institutional equity investors on creditor wealth. Since 1970, more than 95% of newly issued bonds and commercial papers have been sold to fi-

T

nancial institutions such as insurance companies, mutual funds, and pension funds: the

RI P

same types of entities that form the group of institutional equity investors. Third, like the stocks that equity investors hold, bonds are marketable securities. In sum, by examining

SC

the relation between bond pricing and the characteristics of institutional equity investors, we narrow our analysis to parties that are of similar demographics (institutions) and operate

NU

under similar informational environments (securities markets) while playing different roles as claimants to firm value (debt and equity, respectively). Given the importance of marketable

MA

debt financing, variations in bond yield spreads due to institutional characteristics can have economically significant consequences on the market value of existing liabilities, the cost of

ED

future financing for a firm, and the capital allocation among market participants. Using yield spreads dynamically inferred from corporate bond prices and quotes data, we

PT

first show that firms with larger short-term (long-term) institutional ownership are associated with higher (lower) future spreads. The results justify the importance of understanding

AC

tions.

CE

shareholder heterogeneity in determining bond prices and are consistent with our expecta-

We then conduct two sets of subsample analyses to validate our inferences. In the first set, we ensure that the channel of agency cost of debt is at work. We find that the effect of short-term ownership is driven by borrowing firms with higher financial distress risk, where such conflicts are most likely to occur. The effect of long-term ownership, on the other hand, appears to be more systematic. This finding remains qualitatively similar whether we employ measures of financial distress risk that are based on default probability estimations (Merton, 1974; Vassalou and Xing, 2004), credit ratings (Altman, 1989), or leverage (Frank and Goyal, 2009). In the second set of subsample analyses, we verify that the channels through which institu3

ACCEPTED MANUSCRIPT tional investors affect bond pricing are as we describe. For short-term institutions, more frequent trading leads to larger equity volatility (Cremers and Pareek, 2015; Chichernea et al.,

T

2015). In addition, quick profits are more easily realized when there are large fluctuations

RI P

in prices (Derrien et al., 2013). If short-term institutions incentivize managers through their trading behavior (e.g., selling pressure), we expect to see higher spreads because of the ten-

SC

dency of these institutions to trade in stocks with high volatility. For long-term institutions, the emphasis is more likely on the health of the firm, as opposed to temporary jumps and

NU

drops in market prices (Bushee, 1998). If the proxies for investment horizons indeed capture how institutions impact firm policies as we assert, long-term investors should not exhibit

MA

variations in their impact on corporate debt pricing among the low- and high-volatility subsamples. Consistent with what we expect, we find that the effect of short-term investors

ED

on bond pricing is due to firms with higher equity volatility while the effect of long-term investors does not vary across the subsamples.

PT

These findings indicate that bondholders are hurt most when there are more short-term institutional investors and when risk is high. Given the importance of information in the

CE

face of risk, we further explore how the market for corporate control interacts with the

AC

bond pricing effect of institutions’ investment horizons. While stronger shareholder rights can negatively affect creditors due to agency costs (Klock et al., 2005; Cremers et al., 2007), this form of external governance also presents a source of information for bondholders. As opposed to measures of external governance that rely on the effect exerted by all shareholders that have the ability to perform a takeover (Gompers et al., 2003; Bebchuk et al., 2009), our measure for institutional investment horizons captures the characteristic of a subset of existing investors. We test to see how the two effects interact. We find that both the harms from short-term institutions and benefits from long-term institutions are significantly more pronounced when external governance is weak. Therefore, according to our interpretation, when external governance from the market for corporate control is weak, it is more likely that 4

ACCEPTED MANUSCRIPT short-term investors exacerbate agency conflicts of leverage and there is more room for longterm investors to make improvements. The reasons are twofold: First, existing institutional

T

ownership is more powerful in this situation because it is less likely that outsiders can share

RI P

or replace their power. Second, when the possibility of a takeover as a monitoring mechanism is not as effective, creditors rely more on their observations of existing institutional ownership

SC

in the pricing of debt.

Finally, we test whether more concentrated ownership leads to stronger bond pricing

NU

effects. That is, if the presence of long-term owners is beneficial to bond pricing, does giving them more power lead to lower spreads? Conversely, does a more concentrated short-term

MA

ownership worsen bond pricing further? We find that for both short-term and long-term ownership, higher concentration worsens the situation: Whether short-term or long-term,

ED

larger shareholding by institutions with concentrated ownership always drives spreads higher. Overall, support for the private benefits hypothesis in the blockholding literature does not

PT

vary in institutional investment horizons (Barclay and Holderness, 1992). The lower spreads associated with long horizons are driven by more diffused institutional holdings.

CE

Our results are robust to a number of extended checks. We conduct (i) a two-stage model

AC

to capture investor preferences for bond-issuing firms, (ii) a propensity score matching to compare treatment firms of a certain type (i.e., the treatment effect is being dominated by short- or long-term investors) to those that are not of the same type, but closest in observable firm type determinants, (iii) a Heckman (1979) selection model to address the possibility that only creditors with certain characteristics choose to purchase bonds, and (iv) a three-stage estimation that utilizes full information. Our results remain qualitatively unchanged in all cases. We also employ alternative measures for (i) investor characteristics (i.e., dedicated, quasi-indexer, and transient investors from Bushee, 2001) and (ii) creditors’ perception of risk (e.g., unadjusted bond yields, credit ratings, and CDS spreads). Using these measures, we are able to reach intuitively similar conclusions as those in our main 5

ACCEPTED MANUSCRIPT presentation. Additionally, cutting samples differently (e.g., terciles) or using alternative model specifications (e.g., Fama and MacBeth, 1973, in addition to the two-way clustering

T

models of Petersen, 2009) does not change our findings.

RI P

Our study builds on Bhojraj and Sengupta (2003), who show that the presence of institutional investors is overall beneficial to bond pricing. By introducing heterogeneity in

SC

institutional investment horizons and finding economic and statistical differences among institutions in determining bond prices, we reinforce the importance of examining investor

NU

characteristics when pricing financial assets (Yan and Zhang, 2009). The focus on how investment horizons can drive institutional shareholder incentives is motivated by Hirschman

MA

(1970) and Holmstr¨om and Tirole (1993), where the impact of shareholders on firms can take on multiple forms (e.g., exit strategies vs. relationship investments).4 Our research

ED

also relates to the tradeoff between shareholder-manager conflicts (Jensen and Meckling, 1976) and shareholder-creditor conflicts (Myers and Majluf, 1984), where the resolution of

PT

the former can exacerbate the other and vice versa (Klock et al., 2005; Cremers et al., 2007). Further, we provide support for the private benefits argument in the blockholding literature,

CE

as our results indicate that concentrated ownership does not create shared benefits among

AC

debt and equity holders (Barclay and Holderness, 1992; McConnell and Servaes, 1990). More broadly, we contribute to the growing line of research regarding the impact of institutional investors on firm value (Agrawal and Mandelker, 1990) and policies (Jarrell and Poulsen, 1987; Gaspar et al., 2012). While prior research is rich in understanding how stock market participants learn from different types of creditors (e.g., Billett et al., 1995), the opposite receives relatively little attention. Beginning in the 1980s, the development of institutional ownership has created an 4

We relate short-term and long-term institutions to Hirschman’s (1970) “exit” and “voice,” respectively. That is, short-term (long-term) institutions are more likely passive (active) in their attitude toward management actions (Tirole, 1996, p. 375).

6

ACCEPTED MANUSCRIPT inherent information production mechanism in the stock market (Gillan and Starks, 2000; Monks and Minow, 2011). As the growth of institutional ownership drives management

T

decisions more toward shareholder interests, its effect on corporate debt pricing becomes

RI P

increasingly important.

For the remainder of this paper, we develop our hypotheses in Section 2, describe our

SC

data and sample in Section 3, perform our main empirical analyses in Section 4, present

NU

robustness checks of our findings in Section 5, and conclude in Section 6.

2. Development of hypotheses

MA

Figure 1 depicts the theoretical motivation of this paper. Jensen and Meckling (1976) characterize a firm as a “nexus of contracts” that binds together various parties. For sim-

ED

plicity, consider a manager, a creditor, and a shareholder in a one-year model under the Merton’s (1974) framework. At the end of the year, the payoffs of all parties are determined

PT

by the value of the firm (i.e., the underlying asset). If the firm does not use debt, then the shareholder’s payoff is equal to the firm value (linear blue line with a slope of 1). If

CE

the firm uses debt, then the shareholder of the levered firm (red kinked payoff of long call,

AC

where the strike is equal to the par of the bond) is subordinate to the creditor (green kinked payoff of short put) and acts as the residual claimant of firm value. As described, the creditor and the shareholder have very different payoff structures. In the levered case, suppose the shareholder has the power to impact management decisions so that the shareholder and the manager are aligned in incentives. Observing this, how would the creditor’s wealth be affected based on the observable characteristics of the shareholder? [Insert Figure 1 about here.] The literature provides evidence that institutional investors, in general, are beneficial to bondholders (Bhojraj and Sengupta, 2003). However, our understanding of how shareholder 7

ACCEPTED MANUSCRIPT characteristics affect this bond pricing effect is limited. Our question is the following: Are all of them beneficial to bondholders?5

T

With this foundation, we study how equity investment horizons of institutions affect

RI P

the corporate bond pricing of investee firms. The investment horizons of institutions signal (i) their investment objectives and (ii) how they impact management decisions. Investors

SC

that are short-term oriented likely care more about near-term profits and temporary price reactions than about long-term value Bushee (2001); Gaspar et al. (2012). Conversely, for

NU

investors who focus on long-term value, near-term profitability and price movements are not as important (Derrien et al., 2013). As long-term investors encourage management to

MA

shift focus toward the long-term value of the firm, they are more likely to produce benefits that can be shared by creditors (e.g., better reporting quality, less shirking of managers,

ED

etc. Sengupta, 1998; Healy et al., 1999; Jensen and Meckling, 1976). When pricing debt investments, bondholders take this dimension of shareholder characteristic into consideration.

PT

Therefore, we state the following hypothesis regarding the estimation of yield spreads based

CE

on institutional investment horizons: Hypothesis I: Larger short-term (long-term) institutional ownership is associated with

AC

lower (higher) bond yield spreads. Next, we verify our inference regarding the channels through which this effect takes place. We do so in two ways. First, if larger short-term ownership creates shareholder-creditor conflicts, then the adverse effect that they have on bond pricing should be driven by firms where financial distress risk is high. However, if larger long-term ownership provides benefits to all stakeholders, we do not expect to find such a pattern. Second, the pressure that short-term investors put on managers to temporarily boost performance is more likely through the threat 5

Although we are particularly interested in institutional investment horizons, other frequently examined aspects include investor identity (Dai, 2007), manager type (Hotchkiss and Strickland, 2003), and stakeholdings (Ali et al., 2008).

8

ACCEPTED MANUSCRIPT to sell as opposed to relationship investing (An and Zhang, 2013; Chidambaran and John, 1998). All else equal, more frequent trading leads to higher volatility, and exit strategies can

T

more easily be implemented when security prices are volatile. Therefore, we should expect

RI P

to see the adverse bond pricing effect by short-term institutions being driven by cases where equity volatility is high. The impact of long-term institutions on management decisions that

SC

aim for higher firm value – once the ownership is in place – should not vary in stock volatility. Our second and third hypotheses correspond to these arguments:

NU

Hypothesis II: The effect of short-term institutions on bond pricing is more pronounced

not exhibit such a pattern.

MA

in firms where financial distress risk is high; the effect of long-term institutions does

ED

Hypothesis III: The effect of short-term institutions on bond pricing is more pronounced in firms where stock volatility is high; the effect of long-term institutions does not exhibit

PT

such a pattern.

CE

The above hypotheses describe how bondholders, given the investment horizons of institutions, should react in different risk environments. They indicate that bondholders would

AC

be hurt most (i) when there are more short-term institutions and (ii) when risk is high. Clearly, the quality of information matters. We therefore explore how the market for corporate control interacts with our findings. Empirically, stronger shareholder rights (less antitakeover provisions) negatively affect public bond yields due to agency costs (Klock et al., 2005; Cremers et al., 2007). However, this external governance also presents a source of information that can be useful for bondholders. As opposed to the impact from “shareholders that bear the ability to perform a takeover” captured by measures constructed using antitakeover provisions, our investment horizon proxies describe the characteristics of “a subset of existing investors.” Whether the two forces interact, however, is an empirical question. For instance, a weaker governance from the market for corporate control may provide more room 9

ACCEPTED MANUSCRIPT for institutions to impact bond pricing but may also indicate an entrenched management that limits the importance of institutions. We state the following null hypothesis:

T

Hypothesis IV: The effect of institutions on bond pricing is stronger when external

RI P

governance from the market for corporate control is weaker.

SC

Finally, we test whether more powerful institutions affect bond pricing monotonically in their investment horizons. This relates our main finding to the blockholding literature

NU

and is an empirical question (Barclay and Holderness, 1992; McConnell and Servaes, 1990). If long-term ownership is beneficial to bond pricing, does giving those institutions more

MA

power help them to further lower spreads? Or does concentration give rise to greater private benefits that prevent them from doing so (e.g., greater benefits from moral hazard)? We

ED

state the following null hypotheses:

PT

Hypothesis Va: The adverse effect of short-term institutions on bond pricing is stronger when ownership is more concentrated.

CE

Hypothesis Vb: The beneficial effect of long-term institutions on bond pricing is stronger

3. Data

AC

when ownership is more concentrated.

3.1. Data sources and sample We obtain bond issue-level information from Bloomberg and TRACE (Trade Reporting and Compliance Engine). These data contain over-the-counter market activity information on secondary market transactions and quotes on publicly traded debt securities. We collect data on all corporate bond issues that are available from January 1995 to December 2012. We retain only nonconvertible, fixed-coupon bonds that are denominated in U.S. Dollars and

10

ACCEPTED MANUSCRIPT issued by U.S. firms. We require that observations have non-negative prices and maturity dates that are later than the quote/transaction dates.

T

Ownership data are from the Thomson Reuters (TFN; formerly CDA/Spectrum) Institu-

RI P

tional Holdings File, which is extracted from 13F filings by institutional investment managers to the U.S. Securities and Exchange Commission (SEC). This database contains information

SC

on common stock holdings and transactions of managers with $100 million or more in assets under management. For firms in the CRSP-Compustat intersection (CCM) that are not

NU

covered by the 13F, we set their institutional holdings to 0% because it is likely that their equity holdings simply do not meet the SEC filing requirements (Grinstein and Michaely,

MA

2005).

Financial statement items are obtained from the Compustat Quarterly Fundamentals File

ED

and are as of the fiscal-quarter-end date prior to a given TFN reporting date. Capitalization, equity returns, shares outstanding, and trade volume data are from the CRSP (Center for

PT

Research in Security Prices) Monthly Stocks File. We exclude financial (SIC codes from 6000 to 6999) and utility (SIC codes from 4900 to 4999) firms, since the fundamentals of these

CE

firms can be subject to regulatory supervision, rather than due to the economic reasons that

AC

we study (see, for instance, Fama and French, 1992, p. 429). Our final sample consists of 79,997 bond-quarter observations covering 5,424 issues of bonds by 1,040 unique firms. We winsorize our variables at the 1st and 99th percentiles to avoid the potential effect of outliers. To address inflationary concerns, we adjust all monetary variables to December 2010 dollars using consumer price index (CPI) data from the U.S. Bureau of Labor Statistics (BLS). Stock returns are also adjusted by inflation (i.e., percentage changes in the CPI).

11

ACCEPTED MANUSCRIPT 3.2. Variables 3.2.1. Bond pricing

T

Our primary dependent variable for bond pricing is corporate bond yield spreads (SPRD).

RI P

For each bond-quarter observation, we first infer the annualized raw yield (YLD) on the bond based on its price and remaining cash flows. Then, to capture the risk structure of the bond,

SC

we calculate its spread as the difference between its inferred raw yield and the yield on the Treasury security with a term to maturity closest to the remaining life of the bond.

NU

Specifically, the yield spread for corporate bond issue i at time t is

MA

SPRDi,t = YLDi,t − rtf,T −t ,

(1)

ED

where T denotes the maturity date of the corporate bond. rtf,T −t is the Treasury yield at time t with a term to maturity closest to T − t, the remaining life of the corporate bond.

PT

3.2.2. Investment Horizons

CE

Our main explanatory variables are equity ownerships of firms by short-term and longterm institutional investors. For simplicity and illustration purposes, we use the term “SIO”

AC

to abbreviate phrases “short-term institutional owners” and “short-term institutional ownership” throughout this paper. Likewise, “LIO” denotes those that are “long-term.” We measure investment horizons of institutional investors for each calendar quarter based on their average quarterly portfolio turnover over the past year using investor churn rates (Gaspar et al., 2005; Yan and Zhang, 2009). The churn rate of investor k for quarter t is defined as CRk,t

  sell min Churnbuy , Churn k,t k,t ≡ 1P , (N P + N P ) j,k,t j,t j,k,t−1 j,t−1 j∈J 2

(2)

X

(3)

where Churnbuy k,t =

|Nj,k,tPj,t − Nj,k,t−1Pj,t−1 − Nj,k,t−1∆Pj,t |

j∈J;Nj,k,t >Nj,k,t−1

12

ACCEPTED MANUSCRIPT and Churnsell k,t =

X

|Nj,k,tPj,t − Nj,k,t−1Pj,t−1 − Nj,k,t−1∆Pj,t |

(4)

j∈J;Nj,k,t ≤Nj,k,t−1

T

measure aggregate purchases and sales, respectively. Nj,k,t ≥ 0 is investor k’s shareholding

RI P

of firm j ∈ J for quarter t, where J is the set of all firms in our sample. Pj,t denotes the price per share of firm j at the end of quarter t. The quarterly churn rate CRk,t for investor

SC

k at time t is defined as the minimum of purchase- and sale-generated changes in number of shares valued using end-of-quarter prices at time t and scaled by average portfolio size

NU

during the past quarter from t−1 to t; this captures the portfolio turnover of an institutional

MA

investor during that past quarter.

The investment horizon for investor k at quarter t is determined using k’s average churn

ED

rate over the past four quarters, i.e.,

(5)

t =0

PT

avgCRk,t

3

1X CRk,t−t′ . = 4 ′

CE

Intuitively, a higher average churn rate implies a shorter investment horizon. For each quarter, we classify an institution as short-term (long-term) if it has an average churn rate

AC

that is above (below) the sample median for that quarter. A firm j would have a portion of its equity being held by institutional investors during each quarter. This portion, which ranges from 0% and 100%, is referred to as total institutional ownership (TIO). Using the categorization of investment horizons as short-term and long-term, we decompose firm-level TIO into SIO and LIO. Therefore,

TIOj,t = SIOj,t + LIOj,t

for firm j at time t.

13

(6)

ACCEPTED MANUSCRIPT 3.2.3. Control Variables Based on earlier studies, we control for various bond level characteristics (term left to

T

maturity, current clean price, return over the past quarter, and issue size), firm-level char-

RI P

acteristics (tangibility, z-score6 , debt-to-equity ratio, profitability, firm size, market-to-book ratio, stock beta, and stock return over the past quarter), and macroeconomic variables

SC

(credit spread and term spread) that have been shown to affect corporate debt pricing.7

NU

Detailed definitions of all variables used in this study can be found in the Appendix. 4. Analyses

MA

4.1. Descriptive statistics

Table 1 presents the descriptive statistics for our sample. The mean of corporate bond

ED

spreads (SPRD) is 2.43% with a standard deviation of 2.37%. On average, credit ratings are 8.02, approximately corresponding to BBB+/Baa1. On the scale from 1 to 22, the

PT

distribution of credit ratings in our sample is skewed toward smaller values, indicating that bonds in our sample are issued by firms of higher ratings. This is consistent with the

CE

empirical evidence that trustworthy firms are able to issue more bonds. The bond issues have

dollars.

AC

an average market size of $465.49 thousand and a standard deviation of $504.19 thousand

[Insert Table 1 about here.]

6

The original Altman’s (1968) z-score consists of five components, which includes the ratio of market value of equity to book value of long-term debt. Since we also control for a similar term, market-to-book, in our multivariate models as a separate variable, we follow Graham et al. (2008) and employ a modified z-score that does not include the term. Compared to the original z, while a higher modified z-score still indicates better financial health and thus lower default risk, the usual 1.81 and 2.99 cutoffs do not apply when using this measure. 7 There is debate as to whether duration or maturity should be used as a control variable in estimating yields. While the former more effectively captures the remaining life of debt securities, the calculation of duration accounts for the time value of money and requires yield, which is (partially) our dependant variable. For this reason, we use maturity. This is consistent with prior studies on debt pricing such as Bhojraj and Sengupta (2003) and Graham et al. (2008), among many others. Using duration instead does not change our results qualitatively (results available upon request).

14

ACCEPTED MANUSCRIPT At the bond level, the mean (median) short-term and long-term institutional ownership are 26% (24%) and 45% (46%), respectively. These make up 71% of the mean total institu-

T

tional ownership. Comparing this to the figure of 55% in the work by Bhojraj and Sengupta

RI P

(2003) during their earlier sample period from 1991 to 1996, it appears that institutions are gradually holding more shares of firms.

SC

Table 2 shows the correlation matrix of variables used in this study. Focusing on the key variables described in Section 3.2, the correlation coefficient between corporate bond spreads

NU

and SIO (LIO) is significantly positive at 0.09 (significantly negative at −0.13). These are

MA

consistent with what we expect.

[Insert Table 2 about here.]

ED

Figure 2 further presents preliminary evidence for our arguments using two sets of bond portfolios. For every quarter of our sample, we sort bond issues in quintiles based on SIO

PT

and LIO levels and estimate the average corporate spread for every portfolio. We graph the mean spreads for SIO (LIO) portfolios in the left (right) figure. Overall, as we move from

CE

the lowest to the highest SIO quintile, spreads gradually increase. When we do the same for

AC

the LIO portfolios, we find opposite patterns. [Insert Figure 2 about here.]

We now move to multivariate analysis. 4.2. Investment Horizons and Corporate Bond Spreads To test Hypothesis I, we use a baseline model that estimates lead yield spreads of corporate bonds as a function of institutional ownership and controls: ′

SPRDt+1 = β0 + βS SIOt + βL LIOt + Xt B + ǫt ,

15

(7)

ACCEPTED MANUSCRIPT where β and B are the estimated coefficients and ǫ is the vector of errors. X is the matrix of control variables. We report the results in Panel A of Table 3. All models are estimated

T

using Petersen’s (2009) two-way clustering methodology that simultaneously controls for

RI P

cross-sectional and time-series dependencies.

SC

[Insert Table 3 about here.]

In Model 1, the key explanatory variable is total institutional ownership (TIO). Con-

NU

sistent with Bhojraj and Sengupta (2003), we document that institutional ownership lowers corporate yield spreads on average. The estimated coefficient on TIO is −0.49, and it is

MA

statistically significant at the 10% level. Thus, institutional ownership as a whole benefits bondholders.8

ED

In Models 2 and 3, we add SIO and LIO, respectively, to the bond pricing estimation as additional explanatory variables. These show the incremental effects of SIO and LIO on

PT

bond pricing, conditional on the two having equal effects within the aggregated TIO. That is, the total SIO effect in Model 2 is the sum of the estimated coefficients for TIO and SIO,

CE

and the total LIO effect in Model 3 is the sum of the estimated coefficients for TIO and LIO. We see that short-term (long-term) institutions significantly increase (decrease) yield

AC

spreads and that both play dominating roles over aggregate ownership TIO as bond pricing determinants. The estimated coefficient for SIO (LIO) is 2.3808 (−2.3260). Both coefficients are statistically significant at the 0.01 level. compared to the −1.7190 (0.6825) estimated for TIO, which is significantly smaller in magnitude.9 8

The TIO coefficient in Model 1 provides only marginal support for results reported in Bhojraj and Sengupta (2003). This is, however, specific to our sample period. In untabulated subsample analyses, we find that the 10% significance level for our replication is the result of the stronger SIO effects in the post-2006 period, during and following the Great Recession. When we further limit our sample period to before 1997 as in Bhojraj and Sengupta (2003), we are able to find a positive TIO coefficient of −1.7463 that is statistically significant at the 1% level, with our baseline results for SIO and LIO continuing to hold. 9 Observing from Table 2 that our key explanatory variables are highly correlated with other variables (e.g., 0.68 between SIO and TIO and 0.69 between LIO and TIO), we closely monitor their variance inflation

16

ACCEPTED MANUSCRIPT In Model 4, we use SIO and LIO together in place of TIO. We continue to see that bonds issued by firms of larger SIO (LIO) are associated with higher (lower) lead yield

T

spreads. Notably, while both are statistically significant, the magnitude of LIO more than

RI P

doubles that of SIO, contributing to the overall lower spreads associated with TIO that are consistent with prior studies. The estimated coefficients for SIO and LIO are 0.7303

SC

and −1.6609, respectively. Putting this into perspective, a 100% increase in SIO (LIO) is associated with an increase (decrease) of 73 (166) basis points in corporate bond yield

NU

spreads. In other words, a one-standard-deviation increase in SIO (LIO) leads to a onequarter-ahead corporate yield spread increase (decrease) of 7.3 (16.6) basis points.

MA

We also provide some evidence that the effects of investment horizons on bond pricing are not merely short lived. We replace the one-quarter-ahead corporate yield spreads with

ED

those from two quarters ahead, one year ahead, two years ahead, and three years ahead. According to untabulated results we continue to find that SIO (LIO) remains significantly

PT

positive (negative) in explaining lead yield spreads. All control variables are included in these regressions.

CE

Taken together, our results suggest that corporate bond spreads decrease as institutional

AC

equity investment horizons become longer. LIO is the main driving force behind the negative relation between institutional ownership and yield spreads found in Bhojraj and Sengupta (2003). This finding is consistent with our argument that short-term institutions more likely create shareholder-creditor conflicts through the pursuit of near-term earnings goals. Longterm institutions aim for value and, because of this process, are more likely to avoid issues that raise the cost of debt financing (e.g., relationship investing). As the former approach factors (VIF) throughout this study to ensure that our results do not suffer from collinearity issues. All VIFs calculated are low in magnitude compared to the conventional rule-of-thumb threshold of 10. For instance, in Table 3, the VIFs for TIO and SIO are 2.24 and 2.39 in Model 2, and the VIFs for TIO and LIO are 2.23 and 2.23, respectively, in Model 3. Although control variables are less of a concern in this regard (Wooldridge, 2012), we do the same for them nonetheless.

17

ACCEPTED MANUSCRIPT pressures managers to achieve near-term goals even at the cost of long-term value and the wealth of other stakeholders, the latter approach more likely works with them for the long-

T

term prospects of the entire firm.

RI P

4.3. Validation of inferences

Following the above base results, we conduct two sets of subsample analyses to validate

SC

our inferences. We argue that when institutions are short-term oriented, they more likely

NU

exacerbate the agency costs of debt. In contrast, when institutions invest for the long-term, they more likely focus on the quality of cash flows and cost of capital. Although the two

MA

types of investors eventually exert an effect on bond prices, the channels, as we argue, are not identical. Through the following subsample analyses, we provide a better understanding of the bond pricing effect of institutions.

ED

In the first set, we observe the financial distress risk of firms to ensure that the SIO effect

PT

is indeed the results of the exacerbating agency costs of debt. In the second set, we examine firm-level equity volatility to verify that the channels through which institutional investors

CE

affect bond pricing are as we describe.

AC

4.3.1. Financial distress risk

An important characteristic of the agency costs of debt is that these problems are more severe when financial distress risk is high. If short-term institutions indeed exacerbate these problems, their adverse effect on bond yield spreads should be significantly larger for cases where default risk is high. The effect from long-term institutions should not exhibit this pattern. To test Hypothesis II, we categorize firms in our sample as high and low financial distress risk and test whether the two groups have equal SIO and LIO effects on future corporate yield spreads. The results are reported in Table 4. We capture financial distress risk using five separate measures to ensure robustness: Model 1 uses the naive probability of default 18

ACCEPTED MANUSCRIPT from Bharath and Shumway (2008); Model 2 uses a default probability estimated from the hazard model of Shumway (2001) and Chava and Jarrow (2004); Model 3 uses the distance-

T

to-default measure based on Merton (1974) and Vassalou and Xing (2004); Model 4 uses

RI P

investment grading (Altman, 1989); and Model 5 uses leverage (Frank and Goyal, 2009). In Models 1–2, 3, and 5, a firm with a higher probability of default, a shorter distance to

SC

default, or a higher leverage ratio than the sample median is categorized as high default risk, respectively. Because we are in particular considering the issue of financial distress risk

NU

in these regression models, we exclude the z-score as a control variable. All other previous

MA

controls are included.10

[Insert Table 4 about here.]

ED

For all models, SIO drives bond yield spreads up only in the subsample of firms with high financial distress risk. The SIO coefficients for bond issues by firms with higher probabilities

PT

of default (Models 1 and 2), with closer distance-to-default (Model 3), with non-investment grading (Model 4), and with higher leverage (Model 5) are 1.0400 (naive) and 0.7788 (haz-

CE

ard), 0.7927, 1.0379, and 0.9945, respectively. All coefficients are statistically significant at the 5% level. For firms with low risk, SIO is mostly insignificant at conventional levels.11

AC

These results are consistent with our argument that a larger SIO is associated with higher agency costs of debt. The results in Table 4 also show that LIO maintains its significantly negative bond pricing effect across all subsamples. That is, the existence of a beneficial bond pricing effect exerted by long-term institutions is a more systematic phenomenon. For firms with lower risk, however, the magnitude of this benefit seems larger. In Models 1 and 2, 10

Including the z-score does not change our results qualitatively. We thank Emmanuel Alanis for help with estimations of the distance-to-default measure. 11 This insignificance holds for four out of five of the specifications. The only exception is Model 4 (investment grading), where the estimated coefficient for SIO becomes negative. While this may indicate that SIO can also be beneficial to corporate bond pricing under certain situations, the effect is not robust throughout the specifications for our research purpose. Further, the magnitude of this SIO effect is much smaller than the LIO effect.

19

ACCEPTED MANUSCRIPT which use median probabilities of default when determining subsamples, for instance, the estimated coefficients for LIO are −1.9801 and −1.7318 for the high-risk group and −2.1930

T

and −2.5001 for the low-risk group, respectively.

RI P

4.3.2. Stock volatility

The trading activities of short-term institutions can be linked to equity volatility in two

SC

ways. First, they trade more frequently and thus introduce higher volatility. Second, by seeking quick profits, they more likely incentivize managers through their trading behavior

NU

when observing or expecting changes in stock prices, especially when there are larger price

MA

fluctuations. We therefore expect to see the bond pricing effect of SIO to be mainly driven by stocks that are more volatile. In contrast, if long-term institutions promote the pursuit of value, their bond pricing effect should not vary as much in temporary fluctuations of security

ED

prices.

To test Hypothesis III, we first determine if firms are of low or high equity volatility using

PT

two measures. The first measure is stock volatility, calculated as the standard deviation of

CE

historical stock returns over the last 180 days (Campbell and Taksler, 2003). The second measure is idiosyncratic volatility, calculated using residuals from Fama-French estimations

AC

over the past quarter (Ang et al., 2006). We sort firms on these measures and define those with a calculated measure that is above (below) the sample median as high (low) equity volatility.

We report the results in Models 1 (stock volatility) and 2 (idiosyncratic volatility) of Table 5, respectively. In both models, we find that the effect of short-term investors on bond pricing is driven by firms with higher equity volatility. The effect of long-term investors does not vary statistically across the subsamples. [Insert Table 5 about here.] In Model 1, the estimated coefficients for SIO and LIO are 1.0847 (significant at the 20

ACCEPTED MANUSCRIPT 1% level) and −1.4983 (significant at the 1% level), respectively, for the observations with relatively high stock volatility. Compared to these numbers, the estimated coefficients for

T

SIO and LIO are −0.6466 (significant at the 10% level) and −1.7062 (significant at the

RI P

1% level), respectively, for the observations with low stock volatility. The difference in SIO (LIO) between the two groups is statistically significant (insignificant). We find qualitatively

SC

identical results when using idiosyncratic volatility in Model 2.

NU

4.4. Extensions: two empirical questions 4.4.1. External governance

MA

Our findings thus far show that bondholders are hurt most when short-term institutional ownership is large and when risk is high. Risk can stem from the financial distress of borrowers and/or the volatility of equity returns. In high-risk states, information is especially

ED

important. We therefore explore how the market for corporate control, which simultaneously

PT

plays the roles of both (i) an alternative source of agency costs of debt and (ii) an information production mechanism, interacts with the bond pricing effect of different institutional

CE

investors. Whether either of the two roles dominates is an empirical question, and different types of investors might see different effects. In the case of short-term investors, for instance,

AC

weaker external governance can indicate less information provided by the market for corporate control (thus driving spreads further up, as implied by Hypothesis IV ) but may also create less shareholder-creditor conflicts (thus driving spreads down). We employ both the Governance Index (G-Index) of Gompers et al. (2003) and the Entrenchment Index (E-Index) of Bebchuk et al. (2009) to measure the strength of external governance provided by the market for corporate control. Both are calculated as the count of selected anti-takeover provisions, with the latter a subset of the former. Intuitively, firms with less anti-takeover provisions in place are more vulnerable to involuntary management replacements and are thus viewed as more strongly governed externally. We define a firm

21

ACCEPTED MANUSCRIPT as weakly (strongly) governed if it has an index value that is above or equal (below) to the annual sample median. Because these indices capture the power of all shareholders that bear

T

the ability to perform a takeover, using them to categorize firms allows us to examine how

RI P

the strength of external governance (i.e., the market for corporate control) interacts with the characteristics of a subset of existing investors (i.e., institutions that currently hold equity).

SC

The results are reported in Models 1 (G-Index) and 2 (E-Index) of Table 6. In both models, we find that when external governance is weak, the effects of both SIO and LIO

NU

on bond pricing are larger in magnitudes, i.e., the harm from short-term institutions and the benefits from long-term institutions are both significantly more pronounced. For bonds

MA

issued by firms with strong external governance, the SIO effect is statistically insignificant and the LIO effect is less than half of that found in the strong governance group. For

ED

instance, in Model 1, the estimated coefficients for SIO and LIO are 1.3126 (significant at the 1% level) and −1.9614 (significant at the 1% level), respectively, for bonds issued by

PT

weakly governed firms. In contrast, the estimated coefficients for SIO and LIO are −0.5956 (the p-value is 15.6%) and −0.7274 (significant at the 5% level), respectively, when external

AC

CE

governance is strong. We find qualitatively identical results in Model 2. [Insert Table 6 about here.]

When external governance from the market of corporate control is weak, existing institutional ownership is more powerful because it is less likely that outsiders can share or replace their power. Further, when the possibility of the takeover as a monitoring mechanism is not as effective, creditors rely more on their observation of existing institutional ownership in the pricing of debt. As a result, under weak external governance, it is more likely that short-term investors worsen agency conflicts of debt and there is more room for long-term investors to make improvements.

22

ACCEPTED MANUSCRIPT 4.4.2. Ownership concentration Because the results indicate that short-term (long-term) institutions are detrimental (ben-

T

eficial) to bond pricing, we further ask whether this effect is monotonic in concentration.

RI P

That is, with more power, do institutions retain their bond pricing effect in the same direction and make it stronger? Given our earlier findings, we are especially interested in

SC

the case of long-term institutions (i.e., Hypothesis Vb).12 When a long-term institution holds more shares of a firm, it has more voting power to push for improvements, but the

NU

increased holdings may also introduce entrenchment and incentives for securing private ben-

benefits is an empirical question.

MA

efits. Therefore, whether more concentrated long-term institutional investors create more

Essentially, in Hypotheses Va and Vb, we test whether the short-term and long-term

ED

institutions in our data support the private or shared benefits hypothesis in the blockholding literature.13 We define an institution as concentrated at the firm level if it holds at least 5%

PT

of the shares outstanding of that firm (i.e., blockholding in Holderness, 2009). Empirically, this threshold also appears to draw the line for outside directors in becoming entrenched

CE

(Morck et al., 1988). For robustness, we also use 3% and 1% as cutoffs. We report the

AC

results for 5%, 3%, and 1% in Models 3, 4, and 5 of Table 6, respectively. For both short-term and long-term ownership, we find that higher concentration worsens the situation: Larger holdings by more concentrated owners always lead to larger bond yield spreads. The lower corporate bond spreads associated with larger LIO is limited to diffused ownership (i.e., ownership below the cutoff). For instance, the estimated coefficients for 12

From our interpretations, short-term institutions do not provide “shared benefits” in the first place. Therefore, the outcome of testing Hypothesis Va is more easily expected. 13 The “shared benefits hypothesis” states that concentrated ownership produces efficient monitoring by shareholders, and that this benefit spills over to other claimants of firm value (e.g., small shareholders and creditors). The “private benefits hypothesis” suggests that concentrated shareholders use their power to extract private benefits that do not extend to other parties. Holderness (2003) and Edmans (2014) provide thorough surveys on studies related to blockholding and governance.

23

ACCEPTED MANUSCRIPT LIO in Model 3 (where the cutoff is 5%) is negatively significant at −4.0027 (significant at the 1% level) for the diffused component and positively insignificant at 0.4838 (the p-value

T

of 14.7%) for the concentrated component. This pattern remains and gradually becomes

RI P

stronger as we lower the threshold from 5% in Model 3 to 3% and 1% in Models 4 and 5, respectively. The higher spreads associated with SIO are driven by concentrated ownerships

SC

(i.e., ownerships equal to or above the cutoff). The estimated coefficients for SIO in Model 3 are 0.5548 (the p-value is 12.5%) and 1.9248 (significant at the 1% level) for the diffused

NU

and concentrated components, respectively. As we incrementally set the threshold to lower

MA

values in Models 4 and 5, both coefficients become lower.14

5. Robustness checks

ED

We run a battery of robustness checks to ensure the strength of our results. Broadly, we first address potential endogeneity and selection concerns. We then make sure that the

PT

results are not specific to our choices of key variables, empirical methods, and sample periods.

CE

5.1. Endogeneity

Our results suggest that heterogeneity in institutional ownership plays an important

AC

role in bond pricing. In particular, firms with larger short-term (long-term) institutional ownership are associated with higher (lower) future bond yield spreads. While the results support our hypotheses, they may simply be an outcome of self-selection by institutions. There are two possibilities. First, different types of institutions can have different preferences for the credit risk of firms. Long-term institutions, for example, may prefer to invest in large and established firms that have low default risk. Further, unsystematic shocks may affect both bond spreads 14

Eventually, as shown in Model 5, the estimated coefficient for the diffused component of SIO becomes negative. This may indicate that SIO can be beneficial to bond pricing but only when the holdings of individual institutions are very diffused.

24

ACCEPTED MANUSCRIPT and institutional ownership at the same time (i.e., spurious correlation). Both can create non-zero correlations between the ownership variables and error terms in our models and

T

introduce endogeneity. We employ a two-stage least squares estimation (2SLS) to address

RI P

this concern.

We use two instrumental variables (IVs) for institutional ownership. The first is the

SC

industry median of institutional ownership based on the Fama-French 48-industries classification (FF48). Michaely and Vincent (2012) use this same IV to address endogeneity issues

NU

associated with total institutional ownership. Choi and Sias (2009) also document that institutions tend to herd within industries. Importantly, the median ownership level of an

MA

industry is not likely related to the yield spreads of a particular firm or bond issue within that industry and thus qualifies as a good IV. Similar to Grinstein and Michaely (2005), our

ED

second instrument is lagged ownership. As in our main analyses, we follow Petersen (2009) and Gow et al. (2010) by applying two-way clustered standard errors that simultaneously

PT

correct for time-series and cross-sectional dependencies.

CE

In the first stage, we estimate SIO and LIO as follows: ′

′

′

′

AC

SIOt = α0S + Zt ΛS + Xt ΘS + uSt

LIOt = α0L + Zt ΛL + Xt ΘL + uLt ,

(8)

where Z and X are the matrices of the IVs and other control variables, respectively. Λ and Θ denote the vectors of estimated coefficients, and u presents the vectors of first-stage errors. We then use these predicted ownership values in place of the actual ownership variables for our second stage estimation of the effect of institutional ownership on bond yield spreads. Specifically, d t + βL LIO d t + Xt B ′ + ǫt+1 , SPRDt+1 = β0 + βS SIO

(9)

d and LIO d are the predicted values of instiwhere SPRDt+1 denotes bond yield spread. SIO 25

ACCEPTED MANUSCRIPT tutional ownership from Equation 8. X, as in the previous stage, is the matrix of control

RI P

[Insert Table 7 about here.]

T

variables for bond pricing.

Model 1 in Table 7 reports the results from this estimation. We confirm our earlier

SC

results under the two-stage setting. The estimated coefficients for the predicted ownership d and LIO d are 1.0690 (significant at the 5% level) and −1.7906 (significant at variables SIO

NU

the 1% level), respectively. These patterns are similar to those reported in Table 3. In untabulated results, we extend our multistage estimation using a three-stage least squares

MA

model (3SLS) to address potential concerns regarding the contemporaneous correlation of error terms across equations. By simultaneously estimating ownerships and spreads, we

ED

continue to obtain qualitatively similar results.15 We use a matching estimator based on Rosenbaum and Rubin (1983) to further address

PT

the concerns caused by investor preferences. We define the treatment effect as whether a firm is held by significantly more short-term or long-term institutions compared to other firms.

CE

Specifically, if a firm has a SIO (LIO) level that is in the top quartile of our sample, we

AC

categorize it as a SIO (LIO) firm (i.e., a treatment firm). Based on a probit estimation using firm-level characteristics, we find a match for each treatment firm from the non-treatment pool that has the closest propensity to the treatment (i.e., a control firm). We compare the corporate bond spreads of the treatments and controls. The results are reported in Model 3 of Table 7. In both the SIO-treated and LIO-treated cases, we observe that our main results persist. The difference in corporate bond spreads between the SIO-treated and matched firms is 0.4403% (significant at the 1% level); the 15

We estimate the system of equations for every cross section in our sample and report the average of all cross-sectional coefficients using Newey and West (1987) adjusted errors with three lags. We continue to find opposite effects on bond pricing from SIO and LIO. The estimated coefficients on SIO and LIO are 1.0460 and −2.8700, respectively, and both remain statistically and economically significant.

26

ACCEPTED MANUSCRIPT same difference between the LIO-treated and matched firms is −0.3036% (significant at the 1% level). We also employed alternative cut-offs (70% and 80%) and obtain quantitatively

T

similar results.

RI P

Finally, bond issuance is to some extent a choice of the firm.16 That is, firms may self-select to issue bonds and consequently become an observation in our sample. The non-

SC

issuing firms that are unobservable in our sample may provide different implications on the relation between their shareholder base and yield spread on bonds – should they choose to

NU

issue. Thus, our sample is potentially non-random. We address this selection problem using a Heckman (1979) two-stage procedure, where the inverse Mills ratio is obtained from the

MA

first-stage estimation of the likelihood of bond issuance. As shown in Model 2 of Table 7, our results continue to hold.

ED

5.2. Alternatives for key variables

To ensure that our results are not driven by the choices of the particular empirical mea-

PT

sures that we use in the main analysis, we re-estimate our models using various alternatives

CE

for the key variables (i.e., bond spreads and investment horizons). 5.2.1. Creditors’ perception of risk

AC

In addition to corporate bond yield spreads, we further use raw yields, credit ratings, and spreads on credit default swaps (CDS) to capture creditors’ perception of risk. The raw yields are dynamically inferred from bond prices by quarter, as described in Section 3.2. Results from using raw yields are qualitatively identical to those from using yield spreads. These results are untabulated but are available from the authors upon request. Credit ratings data are from Compustat and Bloomberg. We use two measures for ratings. For the first measure, we follow Avramov et al. (2007) by converting letter ratings 16

Empirically, larger and more established firms are more capable of issuing marketable debt securities due to their less severe lemons problem. Given our CCM (CRSP/Compustat Merged) universe, we believe that many of the firms that are not in the Bloomberg and Trace databases do have such choice.

27

ACCEPTED MANUSCRIPT (AAA/Aaa, AA+/Aa1, . . . , D/C) directly to numerical values (1, 2, . . . , 22). The second measure is a coarser version of numerics with a scale from 1 to 6 as used in Bhojraj and Sengupta

T

(2003), where 1 represents ratings of Aaa/AAA (prime), 2 represents ratings of Aa/AA (high

RI P

grade), 3 represents ratings of A/A (upper medium grade), 4 represents ratings of Baa/BBB (lower medium grade), 5 represents ratings of Ba/BB (non-investment grade speculative),

SC

and 6 represents ratings of B/B (highly speculative).17

We test the relation between these bond rating categories and institutional investment

NU

horizons using an ordered logit with two-way clustering standard errors of the following form: ′

(10)

MA

Ratingt+1 = β0 + βS SIOt + βL LIOt + Xt B + ǫt ,

where β and B are the estimated coefficients and ǫ is the vector of errors. As before, X is

ED

the matrix of control variables that include bond, firm, and macroeconomic level controls.

PT

All variable definitions can be found in the Appendix.

CE

[Insert Table 8 about here.]

The results are reported in Models 1 and 2 of Table 8 for the 1–22 and 1–6 measures,

AC

respectively. Consistent with previously documented results on yield spreads, we show that as long-term institutional ownership increases, the credit rating of the firm improves as well. In particular, Model 1 shows that the SIO and LIO coefficients are 3.2142 and −1.2873, respectively. Both coefficients are statistically significant at the 1% level. The second credit rating measure in Model 2 produces similar results. When restricting our sample to firms that have available data on their use of CDS, we lose more than half of our sample. We nonetheless conduct the analysis and report our results, 17

Bhojraj and Sengupta (2003) use a decreasing scale, where 1 represents the highest risk. However, we use an increasing scale that provides consistency in the interpretation of estimated coefficients across all measure of risk that we employ (e.g., higher spreads indicate higher risk).

28

ACCEPTED MANUSCRIPT because using CDS allows us to capture a different dimension of risk. Blanco et al. (2005) argue that swap prices lead the credit spread market and that an important reason for price

T

discovery in the CDS market is largely due to its high volume of informed trading. Given

RI P

that institutional investors are more likely to be informed traders, they should be able to more easily adjust their equity positions based on information flows from the CDS market.

SC

We test the relation between CDS spreads and institutional investment horizons using

NU

the following specification:

′

(11)

MA

CDSt+1 = β0 + βS SIOt + βL LIOt + Xt B + ǫt ,

where β, B, X, and ǫ are as previously defined. CDSt+1 is the lead annualized CDS spread. We report the results in Model 3 of Table 8. Overall, we show that the insurance premium

ED

for firms held by mainly short-term institutional investors is higher. In particular, a 100%

PT

increase in SIO leads to an increase of 66 basis points in the premium. We also see that LIO does not appear to be significantly related to the future CDS prices. This is, however,

CE

subject to the sub-sample that has CDS information available. Overall, we continue to see from CDS spreads that creditors’ perception of risk is related to firm-level institutional

AC

investment horizons.

5.2.2. Investor heterogeneity We incorporate two additional measures that capture investor heterogeneity. The first is an alternative measure for investment horizons used in Gaspar et al. (2005) and Gaspar et al. (2012). Instead of using the minimum of aggregate buy and sell in calculating churn rates, this measure uses the sum of aggregate buys and sells. Specifically, the quarterly churn rates are calculated as CRGMM ≡ k,t

|Nj,k,tPj,t − Nj,k,t−1Pj,t−1 − Nj,k,t−1∆Pj,t | P , 1 j∈J (Nj,k,t Pj,t + Nj,k,t−1 Pj,t−1 ) 2 29

(12)

ACCEPTED MANUSCRIPT instead of according to the functional form specified in Equation 2. The second alternative measure of institutional heterogeneity is the categorization of insti-

T

tutions in Bushee (2001) and Bushee and Noe (2000). These studies use a k-means clustering

RI P

methodology to categorize investors as dedicated (high concentration and low turnover in equity ownership, with little trading sensitivity to earnings), quasi-indexers (highly diversi-

SC

fied ownership and low turnover, long-term buy-and-hold strategies), and transient (highly diversified portfolios with high turnover, momentum strategies following earnings news).

NU

We re-estimate our models using these alternative measures. Model 4 in Table 8 presents the results when using Gaspar et al.’s (2005) measure. The estimated coefficients for SIO

MA

and LIO are 1.5249 and −1.5627, respectively. Intuitively, a 100% increase in SIO (LIO) should increase (decrease) future yield spreads by 2.67% (2.16%). Both coefficients are

ED

statistically significant at the 1% level. Similarly, Model 5 in Table 8 shows that transient institutions increase future yield spreads and quasi-indexer institutions decrease them. The

PT

estimated coefficient for transient (quasi-indexer) ownership is 2.6707 (−2.1591). Dedicated

next quarter.

CE

institutional ownership does not appear to have an effect on the firms’ yield spread for the

AC

These results are consistent with those reported earlier in two ways. First, from the perspective of investment horizons, in general, short-term (long-term) ownership increases (decreases) bond yield spreads. Second, the concentration of ownership leads to an adverse effect on bond pricing. Altogether, our findings are not specific to our choices of empirical proxies. 5.3. Other robustness checks Throughout this study, we categorize institutions into two groups: short-term and longterm. While this categorization eliminates the “ease” in finding results if they are monotonic, there is no way to identify whether there is indeed monotonicity. In other words, cutting the

30

ACCEPTED MANUSCRIPT sample at the median does not allow us to capture any non-linearity. We therefore repeat our main analysis using a tercile split that divides total institutions into three groups: SIO, MIO,

T

and LIO. MIO denotes institutions in the middle tercile of investment horizons. Model 1 in

RI P

Table 9 presents the results from this tercile cut. SIO (LIO) continues to increase (decrease) yield spreads. The estimated coefficient for MIO is not significantly different from that for

SC

LIO.

NU

[Insert Table 9 about here.]

We also address the concern that our results can be sensitive to the empirical methods

MA

that we use. We re-estimate our results using the regression of Fama and MacBeth (1973) and Newey and West (1987) adjusted errors with three lags. Model 2 in Table 9 shows

ED

the results. Once again, our findings are qualitatively similar across different estimation methods. The estimated coefficients for SIO and LIO are 0.7977 and −2.0264, respectively.

PT

This indicates that, on average, a 100% increase in SIO (LIO) leads to a 0.8% increase (2.0% decrease) in bond spreads.

CE

Further, empirical evidence suggests that the relative importance of different types of in-

AC

stitutions changes over time. Chichernea et al. (2015), for instance, find that LIO dominates SIO after 2001, thus suggesting a structural break in our sample period. It is possible that our results hold for only certain sub-periods and are not a general phenomenon. We address this issue by presenting separate results for the time periods before and after 2001 – when the structural break in institutional ownership was observed in prior studies – in Models 3 and 4 of Table 9, respectively. Overall, the bond pricing effect by investment horizon remains qualitatively similar over time. It appears, however, that both the SIO and LIO effects become more positive through time. The effect of SIO (LIO) before and after 2001 is 0.5946 and 0.9183 (−3.0043 and −2.0469), respectively.

31

ACCEPTED MANUSCRIPT Finally, to test whether shorter- and longer-term bonds have different sensitivities toward ownership types, we split our sample of bond issues by remaining maturity for each

T

quarter and repeat our main analysis for the two subsamples. Bonds with remaining term

RI P

to maturity below (above) the sample median are categorized as relatively short-term (longterm). Models 5 and 6 in Table 9 present the results for short-term and long-term bonds,

SC

respectively. Based on the results, short-term bonds seem to be more sensitive to the effects of institutional ownership. Importantly, our earlier finding that SIO (LIO) increases

NU

(decreases) future bond yield spreads remains in both groups.

MA

6. Conclusions

We examine how heterogeneity in institutional investment horizons affects bond pricing.

ED

We find that firms with larger short-term (long-term) institutional ownership are associated with higher (lower) future bond yield spreads. The effect of short-term ownership primarily

PT

takes place in firms with higher financial distress risk and higher stock volatility; the effect of long-term ownership, in contrast, appears to be more systematic across subsamples. The

CE

bond pricing effect by institution type (short-term or long-term) is stronger when external

AC

governance is weak. Further, concentrated ownerships always lead to a costlier debt in our sample, even for long-term investors. Our results support the argument that different types of institutions affect bond pricing differently. Further, the dynamics that different institutions play in this pricing relation are through different channels. An institution’s investment horizon signals its investment objectives and how it may impact management decisions. By incentivizing management, institutions influence bondholders’ perception of the risk associated with their investments, leading them to price accordingly.

32

ACCEPTED MANUSCRIPT References Agrawal, Anup, and Gershon N. Mandelker, 1990, Large shareholders and the monitoring

T

of managers: The case of antitakeover charter amendments, Journal of Financial and

RI P

Quantitative Analysis 25, 143–161.

SC

Ali, Ashiq, Sandy Klasa, and Oliver Zhen Li, 2008, Institutional stakeholdings and betterinformed traders at earnings announcements, Journal of Accounting and Economics 46,

NU

47–61.

Altman, Edward I., 1968, Financial ratios, discriminant analysis and the prediction of cor-

MA

porate bankruptcy, Journal of Finance 23, 589–609. Altman, Edward I., 1989, Measuring corporate bond mortality and performance, Journal of

ED

Finance 44, 909–922.

PT

An, Heng, and Ting Zhang, 2013, Stock price synchronicity, crash risk, and institutional

CE

investors, Journal of Corporate Finance 21, 1–15. Ang, Andrew, Robert J. Hodrick, Yuhang Xing, and Xiaoyan Zhang, 2006, The cross-section

AC

of volatility and expected returns, Journal of Finance 61, 259–299. Avramov, Doron, Tarun Chordia, Gergana Jostova, and Alexander Philipov, 2007, Momentum and credit rating, Journal of Finance 62, 2503–2520. Barclay, Michael J., and Clifford G. Holderness, 1992, The law and large-blocktrades, Journal of Law and Economics 35, 265–294. Bebchuk, Lucian Arye, Alma Cohen, and Allen Ferrell, 2009, What matters in corporate governance?, Review of Financial Studies 22, 783–827.

33

ACCEPTED MANUSCRIPT Bharath, Sreedhar T., and Tyler Shumway, 2008, Forecasting default with the merton distance to default model, Review of Financial Studies 21, 1339–1369.

T

Bhojraj, Sanjeev, and Partha Sengupta, 2003, Effect of corporate governance on bond ratings

RI P

and yields: The role of institutional investors and outside directors, Journal of Business 76, 455–475.

SC

Billett, Matthew T., Mark J. Flannery, and Jon A. Garfinkel, 1995, The effect of lender

NU

identity on a borrowing firm’s equity return, Journal of Finance 50, 699–718. Blanco, Roberto, Simon Brennan, and Ian W. Marsh, 2005, An empirical analysis of the

MA

dynamic relation between investment-grade bonds and credit default swaps, Journal of Finance 60, 2255–2281.

ED

Bushee, Brian, 2001, Do institutional investors prefers near-term earnings over long-run

PT

value?, Contemporary Accounting Research 18, 207–246. Bushee, Brian, and Chris Noe, 2000, Corporate disclosure practices, institutional investors,

CE

and stock return volatility, Journal of Accounting Research 38, 171–202.

AC

Bushee, Brian J., 1998, The influence of institutional investors on myopic R&D investment behavior, Accounting Review 73, 305–333. Campbell, John Y., and Glen B. Taksler, 2003, Equity volatility and corporate bond yields, Journal of Finance 58, 2321–2350. Chava, Sudheer, and Robert A. Jarrow, 2004, Bankruptcy prediction with industry effects, Review of Finance 8, 537–569. Chichernea, Doina C., Alex Petkevich, and Blerina Bela Zykaj, 2015, Idiosyncratic volatility, institutional ownership, and investment horizon, European Financial Management 21, 613–645. 34

ACCEPTED MANUSCRIPT Chidambaran, N. K., and Kose John, 1998, Relationship investing: Large shareholder monitoring with managerial cooperation, New York University working paper.

T

Choi, Nicole, and Richard Sias, 2009, Institutional industry herding, Journal of Financial

RI P

Economics 94, 469–491.

SC

Cremers, K. J. Martijn, Vinay B. Nair, and Chenyang Wei, 2007, Governance mechanisms and bond prices, Review of Financial Studies 20, 1359–1388.

NU

Cremers, Martijn, and Ankur Pareek, 2015, Short-term trading and stock return anomalies:

MA

Momentum, reversal, and share issuance, Review of Finance 19, 1649–1701. 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,

ED

538–563.

PT

Derrien, Fran¸cois, Ambrus Kecsk´es, and David Thesmar, 2013, Investor horizons and corpo-

CE

rate policies, Journal of Financial and Quantitative Analysis 48, 1755–1780. Edmans, Alex, 2014, Blockholders and corporate governance, Annual Review of Financial

AC

Economics 6, 23–50.

Fama, Eugene F., and Kenneth R. French, 1992, The cross-section of expected stock returns, Journal of Finance 47, 427–465. Fama, Eugene F., and James D. MacBeth, 1973, Risk, return, and equilibrium: Empirical tests, Journal of Political Economy 81, 607–636. Frank, Murray Z., and Vidhan K. Goyal, 2009, Capital structure decisions: Which factors are reliably important?, Financial Management 38, 1–37.

35

ACCEPTED MANUSCRIPT Gaspar, Jos´e-Miguel, Massimo Massa, and Pedro Matos, 2005, Shareholder investment horizons and the market for corporate control, Journal of Financial Economics 76, 135–165.

T

Gaspar, Jos´e-Miguel, Massimo Massa, Pedro Matos, Rajdeep Patgiri, and Zahid Rehman,

RI P

2012, Payout policy choices and shareholder investment horizons, Review of Finance 17, 261–320.

SC

Gillan, Stuart L., and Laura T. Starks, 2000, Corporate governance proposals and share-

NU

holder activism: The role of institutional investors, Journal of Financial Economics 57, 275–305.

MA

Gompers, Paul A., Joy Ishii, and Andrew Metrick, 2003, Corporate governance and equity prices, Quarterly Journal of Economics 118, 107–155.

ED

Gow, Ian D., Gaizka Ormazabal, and Daniel J. Taylor, 2010, Correcting for cross-sectional

PT

and time-series dependence in accounting research, Accounting Review 85, 483–512. Graham, John R., Si Li, and Jiaping Qiu, 2008, Corporate misreporting and bank loan

CE

contracting, Journal of Financial Economics 89, 44–61.

AC

Grinstein, Yaniv, and Roni Michaely, 2005, Institutional holdings and payout policy, Journal of Finance 60, 1389–1426. Grossman, Sanford J., and Oliver D. Hart, 1980, Takeover bids, the free-rider problem, and the theory of the corporation, Bell Journal of Economics 11, 42–64. Hackethal, Andreas, and Reinhard H. Schmidt, 2004, Financing patterns: Measurement concepts and empirical results, Goethe University Frankfurt working paper. Healy, Paul M., Amy P. Hutton, and Krishna G. Palepu, 1999, Stock performance and intermediation changes surrounding sustained increases in disclosure, Contemporary Accounting Research 16, 485–520. 36

ACCEPTED MANUSCRIPT Heckman, James, 1979, Sample selection bias as a specification error, Econometrica 47, 153–61.

RI P

T

Hirschman, Albert O., 1970, Exit, Voice, and Loyalty. (Harvard University Press). Holderness, Clifford G., 2003, A survey of blockholders and corporate control, Economic

SC

Policy Review 9, 51–63.

Holderness, Clifford G., 2009, The myth of diffuse ownership in the united states, Review of

NU

Financial Studies 22, 1377–1408.

MA

Holmstr¨om, Bengt, and Jean Tirole, 1993, Market liquidity and performance monitoring, Journal of Political Economy 101, 678–709.

ED

Hotchkiss, Edith S., and Deon Strickland, 2003, Does shareholder composition matter? Evidence from the market reaction to corporate earnings announcements, Journal of Finance

PT

58, 1469–1498.

CE

Jarrell, Gregg A., and Annette B. Poulsen, 1987, Shark repellents and stock prices : The

168.

AC

effects of antitakeover amendments since 1980, Journal of Financial Economics 19, 127–

Jensen, Michael C., 1986, Agency costs of free cash flow, corporate finance, and takeovers, American Economic Review 76, 323–329. Jensen, Michael C., and William H. Meckling, 1976, Theory of the firm: Managerial behavior, agency costs and ownership structure, Journal of Financial Economics 3, 305–360. Klock, Mark S., Sattar A. Mansi, and William F. Maxwell, 2005, Does corporate governance matter to bondholders?, Journal of Financial and Quantitative Analysis 40, 693–719.

37

ACCEPTED MANUSCRIPT McConnell, John J., and Henri Servaes, 1990, Additional evidence on equity ownership and corporate value, Journal of Financial Economics 27, 596–612.

T

Merton, Robert C., 1974, On the pricing of corporate debt: The risk structure of interest

RI P

rates, Journal of Finance 29, 449–470.

SC

Michaely, Roni, and Christopher Vincent, 2012, Do institutional investors influence capital structure decisions?, Cornell University working paper.

NU

Monks, Robert A. G., and Nell Minow, 2011, Corporate Governance. (Blackwell Publishers).

MA

Morck, Randall, Andrei Shleifer, and Robert W. Vishny, 1988, Management ownership and market valuation: An empirical analysis, Journal of Financial Economics 20, 293–315.

ED

Murphy, Kevin J., 1985, Corporate performance and managerial remuneration: An empirical analysis, Journal of Accounting and Economics 7, 11–42.

CE

nomics 5, 147–175.

PT

Myers, Stewart C., 1977, Determinants of corporate borrowing, Journal of Financial Eco-

Myers, Stewart C., and Nicholas S. Majluf, 1984, Corporate financing and investment de-

AC

cisions when firms have information that investors do not have, Journal of Financial Economics 13, 187–221. Newey, Whitney K., and Kenneth D. West, 1987, A simple positive definite, heteroscedasticity and autocorrelation consistent covariance matrix, Econometrica 55, 703–705. Petersen, Mitchell A., 2009, Estimating standard errors in finance panel data sets: Comparing approaches, Review of Financial Studies 22, 435–480. Rosenbaum, Paul R, and Donald B Rubin, 1983, The central role of the propensity score in observational studies for causal effects, Biometrika 70, 41–55. 38

ACCEPTED MANUSCRIPT Sengupta, Partha, 1998, Corporate disclosure quality and the cost of debt, Accounting Review 73, 459–474.

T

Shleifer, Andrei, and Robert W. Vishny, 1986, Large shareholders and corporate control,

RI P

Journal of Political Economy 94, 461–488.

SC

Shumway, Tyler, 2001, Forecasting bankruptcy more accurately: A simple hazard model, Journal of Business 74, 101–124.

NU

Tirole, Jean, 1996, A theory of collective reputations (with applications to the persistence

MA

of corruption and to firm quality), Review of Economic Studies 63, 1–22. Vassalou, Maria, and Yuhang Xing, 2004, Default risk in equity returns, Journal of Finance

ED

59, 831–868.

von Thadden, Ernst-Ludwig, 1995, Long term contracts, short term investment and moni-

PT

toring, Review of Economic Studies 62, 557–575.

Learning).

CE

Wooldridge, Jeffrey M., 2012, Introductory Econometrics: A Modern Approach. (Cenage

AC

Yan, Xuemin (Sterling), and Zhe Zhang, 2009, Institutional investors and equity returns: Are short-term institutions better informed?, Review of Financial Studies 22, 893–924.

39

RI P

T

ACCEPTED MANUSCRIPT

Figure 1Creditor and the Shareholder-Manager Link

NU

SC

This figure depicts the motivation of this paper using Merton’s (1974) foundation of corporate debt pricing. Without borrowing, the payoff for an unlevered owner (blue) is equal to firm value. If the firm borrows, then the firm value is split between the creditor (green) and the levered owner (red). The creditor’s payoff structure is essentially a put option written on firm assets and is affected by the connection between the levered owner and the manager (black dashed).

Owner of Unlevered Firm

ED

MA

Payoff

PT

Owner of Levered Firm

CE

Manager...?

AC

Creditor

Firm Value

Par

40

RI P

T

ACCEPTED MANUSCRIPT

SC

Figure 2Credit risk and institutional ownership

Low SIO

2

2.8

YLDSPRD 2

PT

2.8 2.4

2.4

3

4

0

1

CE AC

0

1

YLDSPRD 2

2.5

2.6

3

2.9 2.8

4.2

4

MA 3.9

ED

3

4

NU

This figure documents average corporate yield spreads of portfolios sorted on short-term (SIO) and long-term (LIO) institutional ownership quintiles in the left and right panels, respectively, over the sample period of 1995–2012.

3

4

High SIO

Low LIO

YLDSPRD

2

YLDSPRD

41

High LIO

SC

RI P

T

ACCEPTED MANUSCRIPT

Table 1 Summary statistics.

AC

10th Pct 0.55 0.55 0.15 0.33 1.54 5.00 0.23 94.58 -2.11 11.67 0.11 0.17 0.12 0.02 5.89 0.61 0.20 -17.33 0.77 -0.05

ED

MA

Std Dev 2.37 0.15 0.10 0.10 11.46 2.72 1.75 11.09 5.47 504.19 0.23 0.56 0.77 0.02 40.79 0.62 0.57 19.72 0.59 1.15

PT

Mean 2.43 0.71 0.26 0.45 10.09 8.02 1.19 104.60 1.91 465.49 0.38 0.77 0.50 0.04 38.17 1.21 0.78 2.78 1.22 1.91

CE

SPRD TIO SIO LIO Maturity Credit Rating (1–22) CDS Spread Bond Price Bond Return Size of Issue Tangibility z-score D/E Profitability Firm Size M/B Beta Stock Return Credit Spread Term Spread

NU

This table shows the descriptive statistics of variables used in the study. The definition and construction details for each variable can be found in the Appendix, Section Appendix A.

42

25th Pct 0.95 0.64 0.18 0.40 3.21 6.00 0.36 99.45 0.31 149.43 0.18 0.50 0.19 0.03 11.40 0.78 0.42 -6.85 0.90 1.20

50th Pct 1.74 0.71 0.24 0.46 6.46 8.00 0.61 103.68 1.59 318.00 0.31 0.78 0.31 0.04 27.76 1.03 0.69 2.52 1.08 2.01

75th Pct 3.23 0.80 0.32 0.52 13.80 10.00 1.21 109.80 3.10 590.20 0.54 1.07 0.56 0.05 47.81 1.46 0.97 11.76 1.32 2.96

90th Pct 5.36 0.88 0.41 0.57 24.36 11.00 2.76 117.39 6.31 1052.97 0.72 1.44 0.91 0.06 69.04 2.05 1.49 21.43 1.66 3.21

PT

ED

SPRD TIO SIO LIO Maturity Rating CDS Spread Bond Price Bond Return Size of Issue Tangibility z-score D/E Profitability Firm Size M/B Beta Stock Return

SPRD 1.00 -0.03 0.09 -0.13 -0.02 0.54 0.66 -0.54 0.01 -0.13 0.06 -0.34 0.49 -0.24 -0.17 -0.23 0.32 -0.01 1.00 0.68 0.69 -0.02 0.15 0.09 0.05 -0.02 -0.03 0.01 0.08 -0.11 -0.05 -0.23 -0.12 0.12 0.00

TIO

1.00 -0.04 -0.08 0.39 0.17 -0.07 0.00 -0.13 0.06 -0.08 0.04 -0.10 -0.33 -0.12 0.20 0.02

SIO

1.00 0.05 -0.18 -0.03 0.13 -0.03 0.08 -0.05 0.19 -0.18 0.02 0.01 -0.04 -0.03 -0.02

LIO

1.00 -0.15 -0.06 0.11 0.02 0.07 0.02 0.07 -0.07 0.05 0.14 0.00 -0.06 -0.00

Maturity

1.00 0.64 -0.25 0.07 -0.11 0.11 -0.54 0.49 -0.36 -0.40 -0.42 0.51 0.06

Rating

1.00 -0.38 0.03 -0.01 0.01 -0.35 0.71 -0.31 -0.13 -0.29 0.44 -0.04

CDS Spread

1.00 0.03 0.22 -0.04 0.17 -0.31 0.13 0.12 0.06 -0.12 0.04

Bond Price

1.00 0.02 0.01 -0.04 0.02 -0.01 -0.01 -0.03 0.07 0.27

Bond Return

43 1.00 -0.06 -0.09 -0.06 0.01 0.34 -0.08 -0.06 -0.01

Size of Issue

1.00 -0.15 0.10 0.13 -0.04 -0.10 -0.03 0.02

1.00 -0.41 0.37 -0.01 0.35 -0.33 -0.02

1.00 -0.27 -0.11 -0.27 0.29 -0.02

D/E

T

1.00 -0.10 -0.14 -0.02

Firm Size

RI P 1.00 0.02 0.54 -0.26 0.04

Profitability

SC

z-score

NU

Tangibility

MA

1.00 -0.25 0.03

M/B

1.00 0.08

Beta

1.00

Stock Return

This table presents the correlation matrix of variables used in this study. The definition and construction details for each variable can be found in the Appendix, Section Appendix A.

Table 2 Correlations.

CE

AC

ACCEPTED MANUSCRIPT

ACCEPTED MANUSCRIPT

Table 3 Corporate bond yield spreads and institutional ownership.

RI P

T

This table presents results from cross-sectional regressions of future corporate bond yield spreads on different combinations of institutional investment horizons as the key explanatory variable(s). The model specifications are variations of the following form: ′

SPRDt+1 = β0 + βS SIOt + βL LIOt + Xt B + ǫt ,

Model 1 -0.4925 * [0.058]

SIO LIO

0.0211 ** [0.038] -0.1144 *** [0.000] -0.0110 [0.591] -0.0003 *** [0.002] -0.0248 [0.850] -0.6725 *** [0.000] 0.3804 *** [0.000] 3.2292 [0.116] -0.0093 *** [0.000] -0.2599 *** [0.000] 0.4994 *** [0.000] -0.0058 *** [0.001] 0.5447 ** [0.037] 0.4532 *** [0.000] 13.9129 *** [0.000] 0.4574 79997

ED

Maturity Bond Price

PT

Bond Return Size of Issue

CE

Tangibility z-score D/E

AC

Profitability Firm Size M/B Beta Stock Return Credit Spread Term Spread Intercept R2 N

Model 2 Model 3 -1.7190 *** 0.6825 ** [0.000] [0.026] 2.3808 *** [0.000] -2.3260 *** [0.000] 0.0230 ** 0.0230 ** [0.024] [0.025] -0.1138 *** -0.1139 *** [0.000] [0.000] -0.0116 -0.0116 [0.568] [0.568] -0.0002 *** -0.0002 *** [0.007] [0.006] -0.0639 -0.0682 [0.626] [0.603] -0.6195 *** -0.6182 *** [0.000] [0.000] 0.3739 *** 0.3756 *** [0.000] [0.000] 3.3193 3.2979 [0.103] [0.105] -0.0082 *** -0.0082 *** [0.000] [0.000] -0.2493 *** -0.2479 *** [0.000] [0.000] 0.4570 *** 0.4626 *** [0.000] [0.000] -0.0060 *** -0.0060 *** [0.001] [0.001] 0.5880 ** 0.5880 ** [0.019] [0.020] 0.4880 *** 0.4846 *** [0.000] [0.000] 13.8624 *** 13.8326 *** [0.000] [0.000] 0.4622 0.4618 79997 79997

MA

TIO

NU

SC

where SPRDt+1 is the lead corporate bond yield spread. TIO, SIO, and LIO represents the total, short-term, and long-term institutional ownerships of issuing firms, respectively. All models are estimated using Petersen’s (2009) two-way clustering methodology that simultaneously controls for cross-sectional and time-series dependencies. The sample period is from 1995 to 2012. Variable definitions can be found in the Appendix.

44

Model 4

0.7303 ** [0.019] -1.6609 *** [0.000] 0.0230 ** [0.024] -0.1139 *** [0.000] -0.0116 [0.569] -0.0002 *** [0.006] -0.0698 [0.595] -0.6184 *** [0.000] 0.3756 *** [0.000] 3.3018 [0.105] -0.0082 *** [0.000] -0.2469 *** [0.000] 0.4590 *** [0.000] -0.0060 *** [0.001] 0.5879 ** [0.019] 0.4863 *** [0.000] 13.8242 *** [0.000] 0.4620 79997

ACCEPTED MANUSCRIPT Table 4 Financial distress risk.

SC

RI P

T

This table presents results from cross-sectional regressions of future corporate bond yield spreads on institutional investment horizons. In each model, issuing firms are categorized into those that have low or high financial distress risk. Model 1 uses Bharath and Shumway’s (2008) naive probability of default; Model 2 uses a default probability estimated from the hazard model of Shumway (2001) and Chava and Jarrow (2004); Model 3 uses the distance-to-default measure based on Merton (1974) and Vassalou and Xing (2004); Model 4 uses investment grading (Altman, 1989); and Model 5 uses leverage (Frank and Goyal, 2009). The model specification takes the following form:  FDR  X FDR ′ SPRDt+1 = β0 + dFDR βSd SIOt + βLd LIOt + Xt B + ǫt , dFDR ∈{dFDR ,dFDR } hi lo

MA

NU

where SPRDt+1 is the lead corporate bond yield spread. SIO and LIO represents the short-term FDR and long-term institutional ownerships of issuing firms, respectively. dFDR (dlo ) is a dummy hi variable that indicates whether individual firms are of relatively high (low) financial distress risk. All models are estimated using Petersen’s (2009) two-way clustering methodology. The sample period is from 1995 to 2012. Variable definitions can be found in the Appendix.

AC

CE

PT

ED

Model 1 Model 2 Naive Shumway SIO (Hi FDR) 1.0400 *** 0.7788 ** [0.002] [0.023] LIO (Hi FDR) -1.9801 *** -1.7318 *** [0.000] [0.000] SIO (Low FDR) -0.0779 0.5028 [0.833] [0.113] LIO (Low FDR) -2.1930 *** -2.5001 *** [0.000] [0.000] Maturity 0.0222 ** 0.0221 ** [0.032] [0.031] Bond Price -0.1152 *** -0.1152 *** [0.000] [0.000] Bond Return -0.0109 -0.0099 [0.604] [0.635] Size of Issue -0.0001 -0.0001 [0.224] [0.227] Tangibility 0.1083 0.2035 [0.393] [0.107] D/E 0.4179 *** 0.4061 *** [0.000] [0.000] Profitability -0.1430 -0.7181 [0.941] [0.710] Firm Size -0.0066 *** -0.0071 *** [0.000] [0.000] M/B -0.2350 *** -0.2908 *** [0.000] [0.000] Beta 0.4975 *** 0.5657 *** [0.000] [0.000] Stock Return -0.0055 *** -0.0052 *** [0.003] [0.006] Credit Spread 0.5585 ** 0.5711 ** [0.031] [0.029] Term Spread 0.4922 *** 0.4770 *** [0.000] [0.000] Intercept 13.6788 *** 13.7026 *** [0.000] [0.000] R2 0.4549 0.4555 N 79997 79997

45

Model 3 DD 0.7927 ** [0.031] -1.7359 *** [0.000] 0.3029 [0.342] -2.4632 *** [0.000] 0.0221 ** [0.032] -0.1145 *** [0.000] -0.0106 [0.612] -0.0001 [0.166] 0.0954 [0.444] 0.4189 *** [0.000] -0.0799 [0.967] -0.0066 *** [0.000] -0.2266 *** [0.000] 0.4968 *** [0.000] -0.0054 *** [0.003] 0.5829 ** [0.026] 0.4822 *** [0.000] 13.6100 *** [0.000] 0.4553 79997

Model 4 Model 5 IG Leverage 0.8550 ** 0.9945 *** [0.018] [0.004] -0.1722 -1.9205 *** [0.656] [0.000] -1.2825 *** 0.1566 [0.000] [0.643] -2.5954 *** -2.2951 *** [0.000] [0.000] 0.0301 *** 0.0225 ** [0.002] [0.028] -0.1011 *** -0.1150 *** [0.000] [0.000] -0.0201 -0.0107 [0.303] [0.609] -0.0001 * -0.0001 [0.086] [0.179] 0.0166 0.0721 [0.884] [0.569] 0.3568 *** 0.4005 *** [0.000] [0.000] -1.5481 -0.9278 [0.382] [0.633] -0.0040 *** -0.0065 *** [0.000] [0.000] -0.1883 *** -0.3344 *** [0.000] [0.000] 0.3266 *** 0.5564 *** [0.000] [0.000] -0.0059 *** -0.0058 *** [0.001] [0.001] 0.6633 *** 0.5650 ** [0.008] [0.028] 0.4666 *** 0.4774 *** [0.000] [0.000] 12.4549 *** 13.8067 *** [0.000] [0.000] 0.4948 0.4553 78743 79997

ACCEPTED MANUSCRIPT Table 5 Equity volatility.

SC

RI P

T

This table presents results from cross-sectional regressions of future corporate bond yield spreads on short-term and long-term institutional investment horizons. The models break down issuing firms into those that have high and low equity volatility. Model 1 uses stock volatility, calculated as the standard deviation of historical stock returns over the last 180 days (Campbell and Taksler, 2003). Model 2 uses idiosyncratic volatility, calculated using residuals from Fama-French estimations over the past quarter (Ang et al., 2006). The model specification takes the following form:  EqVol  X EqVol ′ dEqVol βSd SIOt + βLd LIOt + Xt B + ǫt , SPRDt+1 = β0 + ,dEqVol dEqVol ∈{dEqVol } hi lo

MA

NU

where SPRDt+1 is the lead corporate bond yield spread. SIO and LIO represent the short-term EqVol and long-term institutional ownerships of issuing firms, respectively. dEqVol (dlo ) is a dummy hi variable that indicates whether individual firms are of relatively high (low) equity volatility. Both models are estimated using Petersen’s (2009) two-way clustering methodology that simultaneously controls for cross-sectional and time-series dependencies. The sample period is from 1995 to 2012. Variable definitions can be found in the Appendix.

AC

CE

PT

ED

Model 1 Model 2 Stock Idiosyncratic Volatility Volatility SIO (High Volatility) 1.0847 *** 1.0016 *** [0.002] [0.003] LIO (High Volatility) -1.4983 *** -1.4900 *** [0.000] [0.000] SIO (Low Volatility) -0.6466 * -0.6136 * [0.075] [0.088] LIO (Low Volatility) -1.7062 *** -1.6611 *** [0.000] [0.000] Maturity 0.0241 ** 0.0236 ** [0.019] [0.021] Bond Price -0.1136 *** -0.1131 *** [0.000] [0.000] Bond Return -0.0131 -0.0131 [0.513] [0.514] Size of Issue -0.0002 *** -0.0002 *** [0.006] [0.004] Tangibility -0.1240 -0.1302 [0.337] [0.310] z-score -0.5792 *** -0.5795 *** [0.000] [0.000] D/E 0.3650 *** 0.3643 *** [0.000] [0.000] Profitability 3.3757 * 3.2882 [0.099] [0.103] Firm Size -0.0075 *** -0.0074 *** [0.000] [0.000] M/B -0.2284 *** -0.2348 *** [0.000] [0.000] Beta 0.3113 *** 0.3525 *** [0.000] [0.000] Stock Return -0.0060 *** -0.0058 *** [0.000] [0.001] Credit Spread 0.5174 ** 0.5437 ** [0.037] [0.031] Term Spread 0.5092 *** 0.4962 *** [0.000] [0.000] Intercept 14.0009 *** 13.9262 *** [0.000] [0.000] 46 R2 0.4692 0.4683 N 79997 79997

ACCEPTED MANUSCRIPT Table 6 External governance and ownership concentration.

SIO (Weak Ext Gov) LIO (Weak Ext Gov)

MA

SIO (Strong Ext Gov) LIO (Strong Ext Gov)

ED

SIO (Diffused; < x) LIO (Diffused; < x)

SIO (Concentrated; ≥ x)

CE

Bond Price

PT

LIO (Concentrated; ≥ x) Maturity

Bond Return Size of Issue

AC

Tangibility z-score D/E

Profitability Firm Size M/B Beta Stock Return Credit Spread Term Spread Intercept R2 N

Model 2 Model 3 E-Index x = 5% 0.9969 *** [0.002] -1.8196 *** [0.000] -0.5485 [0.199] -0.9955 ** [0.012] 0.5548 [0.125] -4.0027 *** [0.000] 1.9248 *** [0.000] 0.4838 [0.147] 0.0225 ** 0.0252 ** [0.028] [0.013] -0.1142 *** -0.1115 *** [0.000] [0.000] -0.0113 -0.0134 [0.577] [0.503] -0.0002 *** -0.0002 *** [0.007] [0.009] -0.0427 0.0277 [0.745] [0.824] -0.6139 *** -0.4810 *** [0.000] [0.000] 0.3782 *** 0.3349 *** [0.000] [0.000] 3.2696 3.4554 [0.102] [0.104] -0.0081 *** -0.0049 *** [0.000] [0.000] -0.2469 *** -0.1843 *** [0.000] [0.000] 0.4559 *** 0.4489 *** [0.000] [0.000] -0.0061 *** -0.0060 *** [0.001] [0.001] 0.5814 ** 0.6520 *** [0.021] [0.009] 0.4868 *** 0.5180 *** [0.000] [0.000] 13.8604 *** 13.6631 *** [0.000] [0.000] 0.4631 0.4723 79997 79997

0.0228 ** [0.026] -0.1146 *** [0.000] -0.0115 [0.571] -0.0002 *** [0.006] -0.0153 [0.907] -0.6078 *** [0.000] 0.3843 *** [0.000] 3.2085 [0.113] -0.0084 *** [0.000] -0.2470 *** [0.000] 0.4614 *** [0.000] -0.0060 *** [0.001] 0.5819 ** [0.021] 0.4823 *** [0.000] 13.8537 *** [0.000] 0.4642 79997

Model 4 x = 3%

Model 5 x = 1%

NU

Model 1 G-Index 1.3126 *** [0.000] -1.9614 *** [0.000] -0.5956 [0.156] -0.7274 ** [0.042]

SC

RI P

T

This table presents results from cross-sectional regressions of future corporate bond yield spreads on short-term and long-term institutional investment horizons. To examine external governance, Models 1 and 2 break down ownership into those that are strongly and weakly governed in the market for corporate control, as indicated by Gompers et al.’s (2003) Governance and Bebchuk et al.’s (2009) Entrenchment Indices, respectively. Firms with higher index values are relatively weaker in external governance. To examine ownership concentration, Models 3, 4, and 5 split ownership into institutional holdings that are less than 5%, 3%, and 1%, respectively, and those that are at least as much. All models are estimated using Petersen’s (2009) two-way clustering methodology that simultaneously controls for cross-sectional and time-series dependencies. The sample period is from 1995 to 2012. Variable definitions can be found in the Appendix.

47

0.2756 [0.522] -6.3322 *** [0.000] 1.8097 *** [0.000] 0.1926 [0.589] 0.0248 ** [0.015] -0.1119 *** [0.000] -0.0134 [0.501] -0.0002 ** [0.019] 0.0090 [0.944] -0.4576 *** [0.000] 0.3262 *** [0.000] 3.2262 [0.131] -0.0051 *** [0.000] -0.1345 *** [0.005] 0.3706 *** [0.000] -0.0062 *** [0.000] 0.6011 ** [0.013] 0.4886 *** [0.000] 14.0452 *** [0.000] 0.4763 79997

-2.2997 ** [0.016] -10.3115 *** [0.000] 1.4326 *** [0.000] 0.0377 [0.911] 0.0265 *** [0.009] -0.1105 *** [0.000] -0.0137 [0.487] -0.0002 ** [0.034] 0.1382 [0.279] -0.3809 *** [0.000] 0.3152 *** [0.000] 4.1746 * [0.065] -0.0023 *** [0.006] -0.0927 * [0.063] 0.3838 *** [0.000] -0.0063 *** [0.001] 0.6867 *** [0.006] 0.5104 *** [0.000] 13.6697 *** [0.000] 0.4808 79997

ACCEPTED MANUSCRIPT Table 7 Robustness – endogeneity and sample selection.

RI P

T

Model 1 of the table addresses endogeneity concerns. It shows the results from a two-stage least squares (2SLS) estimation, where ownership variables SIO, and LIO are treated as endogenous. In the first stage, we estimate SIO and LIO. The predicted ownership values are then used in the second stage estimation for the effect of institutional ownership on bond yield spreads. Specifically, d t + Xt B ′ + ǫt+1 , d t + βL LIO SPRDt+1 = β0 + βS SIO

Model 1 2SLS

1.0690 ** [0.010] LIO -1.7906 *** [0.000] Maturity 0.0176 ** [0.013] Bond Price -0.1094 *** [0.000] Bond Return -0.0092 [0.447] Size of Issue -0.0002 ** [0.023] Tangibility -0.0963 [0.460] z-score -0.6014 *** [0.000] D/E 0.3601 *** [0.000] Profitability 3.2048 [0.108] Firm Size -0.0078 *** [0.000] M/B -0.2482 *** [0.000] Beta 0.4366 *** [0.000] Stock Return -0.0059 *** [0.001] Credit Spread 0.5936 ** [0.015] Term Spread 0.4897 *** [0.000] λ

Model 2 Heckman

Model 3 Propensity Score Matching Treated Control Difference 0.7826 ** 3.4961 3.0559 0.4403 *** [0.011] [0.000] -1.5977 *** 2.6865 2.9901 -0.3036 *** [0.000] [0.000] 0.0203 ** [0.043] -0.1152 *** [0.000] -0.0110 [0.584] -0.0002 *** [0.004] -0.0872 [0.497] -0.6308 *** [0.000] 0.3715 *** [0.000] 3.3945 * [0.090] -0.0080 *** [0.000] -0.2431 *** [0.000] 0.4548 *** [0.000] -0.0060 *** [0.001] 0.5815 ** [0.022] 0.4886 *** [0.000] -0.3674 [0.410] 13.9848 *** [0.000] 0.4617 80930

AC

CE

PT

ED

SIO

MA

NU

SC

d and LIO d are the predicted values of institutional where SPRDt+1 denotes bond yield spread. SIO ownership obtained from the previous stage. X is the matrix of control variables for bond pricing. The results are reported using two-way clustered errors that simultaneously correct for time-series and cross-sectional dependencies. Model 2 addresses the selection bias (Heckman, 1979) by including the inverse Mills ratio (λ) as a control variable. Model 3 presents the propensity matching methodology. The treated group is defined as top quartile of the SIO or LIO ownership. The control group is matched based on the propensity scores matching using the Rosenbaum and Rubin (1983) methodology. The sample period is from 1995 to 2012. Variable definitions can be found in the Appendix.

Constant R2 N

13.3423 *** [0.000] 0.4698 78074

48

ACCEPTED MANUSCRIPT Table 8 Robustness – alternative measures of key variables.

RI P

T

This table presents results from cross-sectional regressions of creditors’ perception of risk on institutional ownership type. Particularly, alternative measures of the key variables in this study are used to ensure robustness. The model specifications are variations of the following form: X ′ RISKCreditor = β0 + βH HIOt + Xt B + ǫt , t+1 H

MA

NU

SC

is the measure for creditors’ perception of risk and HIO are measures of instiwhere RISKCreditor t+1 tutional ownership categorized in different ways. Models 1, 2, and 3 use Avramov et al.’s (2007) 1–22 credit ratings, Bhojraj and Sengupta’s (2003) 1–6 credit ratings, and CDS spreads in place of corporate bond yield spreads. Models 4 and 5 use Gaspar et al.’s (2005) investment horizons and Bushee’s (2001) categorization of institutions (dedicated, transient, and quasi-indexers) as alternative measures for investor heterogeneity. Models 1 and 2 are ordered logit models. All models are estimated using Petersen’s (2009) two-way clustering methodology. The sample period is from 1995 to 2012. Variable definitions can be found in the Appendix.

AC

CE

PT

ED

Model 1 Model 2 Model 3 Model 4 Model 5 Ratings 1–22 Ratings 1–6 CDS Gaspar et al. Bushee SIO 3.2142 *** 3.5532 *** 0.6575 ** [0.000] [0.000] [0.023] LIO -1.2873 *** -1.5134 *** 0.2417 [0.000] [0.000] [0.298] SIO (Gaspar et al.) 1.5249 *** [0.000] LIO (Gaspar et al.) -1.5627 *** [0.000] Transient IO 2.6707 *** [0.000] Quasi-Indexer IO -2.1591 *** [0.000] Dedicated IO -0.1623 [0.739] Maturity -0.0142 *** -0.0151 *** -0.0012 0.0228 ** 0.0236 ** [0.000] [0.000] [0.604] [0.026] [0.022] Bond Price -0.0172 *** -0.0107 ** -0.0261 *** -0.1141 *** -0.1154 *** [0.000] [0.011] [0.000] [0.000] [0.000] Bond Return 0.0131 ** 0.0117 * 0.0039 -0.0123 -0.0107 [0.030] [0.061] [0.694] [0.548] [0.597] Size of Issue 0.0003 *** 0.0004 *** 0.0002 *** -0.0002 *** -0.0002 *** [0.000] [0.000] [0.000] [0.009] [0.007] Tangibility -0.2374 * -0.4192 *** 0.1591 -0.0388 -0.0528 [0.069] [0.004] [0.158] [0.765] [0.681] z-score -1.2233 *** -1.1708 *** -0.1262 ** -0.5961 *** -0.5666 *** [0.000] [0.000] [0.027] [0.000] [0.000] D/E 0.3069 *** 0.5712 *** 1.2302 *** 0.3769 *** 0.3694 *** [0.000] [0.000] [0.000] [0.000] [0.000] Profitability 1.5177 2.3697 -6.0714 *** 2.7789 2.7644 [0.385] [0.231] [0.003] [0.184] [0.166] Firm Size -0.0210 *** -0.0236 *** -0.0047 *** -0.0079 *** -0.0080 *** [0.000] [0.000] [0.000] [0.000] [0.000] M/B -1.0013 *** -0.9227 *** -0.0113 -0.2504 *** -0.2736 *** [0.000] [0.000] [0.804] [0.000] [0.000] Beta 0.8241 *** 0.9098 *** 0.5225 *** 0.4480 *** 0.3738 *** [0.000] [0.000] [0.000] [0.000] [0.000] Stock Return 0.0012 0.0021 -0.0048 ** -0.0062 *** -0.0066 *** [0.403] [0.259] [0.012] [0.001] [0.000] Credit Spread -0.2677 *** -0.2606 *** 0.4008 *** 0.5908 ** 0.5832 ** [0.000] [0.000] [0.000] [0.019] [0.022] Term Spread 0.0343 0.0200 0.1092 *** 0.4836 *** 0.4613 *** [0.317] [0.593] [0.000] [0.000] [0.000] 49 Intercept Omitted Omitted 2.2966 *** 13.8339 *** 14.2404 *** [0.000] [0.000] [0.000] R2 0.1808 0.2908 0.6308 0.4635 0.4666 N 78731 78731 38245 79997 79997

ACCEPTED MANUSCRIPT Table 9 Robustness – subsamples and alternative estimation methods.

RI P

T

This table presents results from cross-sectional regressions of future corporate bond yield spreads on short-term and long-term institutional investment horizons as the key explanatory variables of interest. The earlier baseline model is re-estimated using different cuts in investor categorization, methodology, and subsamples to ensure robustness. The model specification takes the following form: ′ SPRDt+1 = β0 + βS SIOt + βL LIOt + Xt B + ǫt ,

MA

NU

SC

where SPRDt+1 is the lead corporate bond yield spread. TIO, SIO, and LIO represents the total, short-term, and long-term institutional ownerships of issuing firms, respectively. Model 1 reports results using the same institutional investment horizon proxy of Yan and Zhang (2009) as in the baseline results, but cutting the sample investors by terciles. Model 2 reports estimations using the Fama and MacBeth’s (1973) method. Models 3 and 4 conduct subsample analyses given prior empirical findings of a structural break in the importance of institutions. Models 5 and 6 conduct subsample analyses for differences in the maturity of bond issues. All models are estimated using Petersen’s (2009) two-way clustering methodology. The sample period is from 1995 to 2012. Variable definitions can be found in the Appendix.

AC

CE

PT

ED

Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Terciles Fama-MacBeth Before 2001 After 2001 ST Debt LT Debt SIO (YZ Tercile) 3.2141 *** [0.000] MIO (YZ Tercile) -1.4629 *** [0.000] LIO (YZ Tercile) -1.4224 *** [0.004] SIO 0.7977 *** 0.5946 ** 0.9183 *** 1.4086 *** 0.3498 [0.001] [0.034] [0.009] [0.000] [0.291] LIO -2.0264 *** -3.0043 *** -2.0469 *** -1.6911 *** -1.1868 *** [0.000] [0.000] [0.000] [0.000] [0.003] Maturity 0.0185 ** 0.0033 -0.0126 * 0.0209 *** 0.1377 *** 0.0129 ** [0.010] [0.761] [0.074] [0.008] [0.001] [0.037] Bond Price -0.1096 *** -0.1008 *** -0.1732 *** -0.1005 *** -0.2064 *** -0.0717 *** [0.000] [0.000] [0.000] [0.000] [0.000] [0.000] Bond Return -0.0090 0.0192 ** -0.0237 ** -0.0123 -0.0305 ** -0.0145 [0.457] [0.018] [0.019] [0.438] [0.018] [0.230] Size of Issue -0.0001 * 0.0000 0.0003 -0.0003 *** -0.0002 -0.0001 [0.062] [0.799] [0.106] [0.000] [0.104] [0.337] Tangibility -0.0460 -0.1106 -0.3855 * 0.0292 -0.0547 -0.1332 [0.725] [0.283] [0.071] [0.831] [0.717] [0.339] z-score -0.5627 *** -0.5004 *** -0.7475 *** -0.6352 *** -0.6669 *** -0.5677 *** [0.000] [0.000] [0.000] [0.000] [0.000] [0.000] D/E 0.3581 *** 0.4582 *** 0.3365 *** 0.3489 *** 0.2468 *** 0.3594 *** [0.000] [0.000] [0.000] [0.000] [0.000] [0.000] Profitability 2.2227 -0.2213 0.8115 3.7327 * 3.4234 1.8551 [0.262] [0.917] [0.816] [0.095] [0.136] [0.395] Firm Size -0.0075 *** -0.0110 *** -0.0158 *** -0.0083 *** -0.0127 *** -0.0059 *** [0.000] [0.000] [0.000] [0.000] [0.000] [0.000] M/B -0.2534 *** -0.1196 *** -0.1107 * -0.2798 *** -0.3006 *** -0.1956 *** [0.000] [0.005] [0.080] [0.000] [0.000] [0.000] Beta 0.4175 *** 0.5990 *** 0.6049 *** 0.3823 *** 0.4588 *** 0.4806 *** [0.000] [0.000] [0.000] [0.000] [0.000] [0.000] Stock Return -0.0063 *** -0.0004 -0.0038 -0.0061 *** -0.0040 ** -0.0077 *** [0.000] [0.814] [0.101] [0.003] [0.045] [0.000] Credit Spread 0.6285 ** 0.9361 0.5977 ** 0.7065 *** 0.4484 * [0.015] [0.332] [0.016] [0.002] [0.088] Term Spread 0.4534 *** 0.2736 0.4528 *** 0.6758 *** 0.3247 *** [0.000] [0.302] [0.000] [0.000] [0.000] Intercept 13.4409 *** 13.3609 *** 19.9592 *** 12.7816 *** 22.6148 *** 9.7152 *** [0.000] [0.000] [0.000] [0.000] [0.000] [0.000] R2 0.4742 0.5324 0.5856 0.4612 0.5721 0.4485 50 13132 N 79997 79997 66865 38536 41461

ACCEPTED MANUSCRIPT Appendix A. Appendix Appendix A.1. Variable Definitions

RI P

appear in typewriter fonts represent Compustat item names. Panel A: Key Variables

TIO SIO/LIO

Annualized raw yield of a bond, less the Treasury yield with closest term to maturity Total institutional ownership Short/long-term institutional ownership

NU

Panel B: Issue Level Control Variables

SC

SPRD

Years left to maturity of the bond

Duration

Duration of the remaining life of the bond

Bond Price Size of Issue

MA

Maturity

Bond Return

T

This table below explains the construction of variables used in this study. Texts that

Realized bond return during the past quarter Bond price as percentage of par Issue size in $ millions

z-score

ED

Panel C: Firm-Level Control Variables

Modified Altman’s (1968) z-score (see Footnote 4) ((1.2*wcapq + 1.4*req + 3.3*piq + 0.999*saleq)/atq)

Tangibility Firm Size M/B

Operating income before depreciation to assets (oibdpq/atq) Net property, plant, and equipment to assets (ppentq/atq) Natural log of assets, CPI adjusted (ln(atq))

Market-to-book assets ((prccq*cshprq + dlcq + dlttq - txditcq)/atq) CAPM stock beta

AC

Beta

PT

Profitability

Market leverage ratio ((dlcq + dlttq)/(prccq*cshoq))

CE

D/E

Stock Return

Stock return of the issuing firm over the past quarter

Panel D: Macroeconomic Level Control Variables Credit Spread Term Spread

Difference in yields between Baa and Aaa corporate bonds Difference in yields between the 10-year and 1-year Treasury securities

51

ACCEPTED MANUSCRIPT

Highlights for “Corporate Bond Pricing and Ownership Heterogeneity”

RI P

Manuscript submitted to The Journal of Corporate Finance

T

Kershen Huang and Alex Petkevich

We examine how heterogeneity in equity ownership is related to yield spreads.

•

Firms with larger short-term (long-term) institutional ownership are associated with higher (lower) yield spreads.

•

The positive (negative) relation between yield spreads and short-term institutions is primarily driven by concentrated (diffused) holdings

•

The adverse effect of short-term ownership takes place in borrowing firms with higher financial distress risk, stronger management rights, and higher stock volatility.

•

Overall, our results suggest bond investors price risk associated with high short-term ownership.

AC

CE

PT

ED

MA

NU

SC

•