Does Borrowing from Banks Cost More than Borrowing from the Market?

• Author: MICHAEL SCHWERT
• Date: 01 April 2020
• DOI: http://doi.org/10.1111/jofi.12849
• Published date: 01 April 2020

THE JOURNAL OF FINANCE •VOL. LXXV, NO. 2 •APRIL 2020

Does Borrowing from Banks Cost More than

Borrowing from the Market?

MICHAEL SCHWERT∗

ABSTRACT

This paper investigates the pricing of bank loans relative to capital market debt. The

analysis uses a novel sample of loans matched with bond spreads from the same firm

on the same date. After accounting for seniority,lenders earn a large premium relative

to the bond-implied credit spread. In a sample of secured term loans to noninvestment-

grade firms, the average premium is 140 to 170 bps or about half of the all-in-drawn

spread. This is the first direct evidence of firms’ willingness to pay for bank credit and

raises questions about the nature of competition in the loan market.

THE TWO PRIMARY SOURCES OF debt for firms are private bank loans and public

bonds issued in the market. The academic literature offers a number of theories

on the interaction between the private and public debt markets (e.g., Diamond

(1991), Rajan (1992), Chemmanur and Fulghieri (1994)) and empirical evidence

on cross-sectional and time series variation in the quantities of loans and bonds

issued (e.g., Faulkender and Petersen (2006), Rauh and Sufi (2010), Becker and

Ivashina (2014)). We know less, however, about the pricing of bank loans and

the relative pricing of private and public debt. This paper fills that gap by

offering new evidence on the relative costs of loan and bond debt.

The central finding of this paper is that banks earn a substantial interest rate

premium relative to the price of credit risk implied by the bond market after

accounting for seniority. I arrive at this finding by constructing a novel data

set consisting of new loan facilities and secondary bond market quotes from

the same firm on the same date. These data allow for a clean comparison of

∗Michael Schwert is at the Wharton School, University of Pennsylvania. I thank Stefan Nagel

(the Editor); an anonymous associate editor; two anonymous referees; Felipe Aldunate; Bo Becker;

Darrell Duffie; Erik Gilje; Will Gornall; Daniel Green; Kewei Hou; Pete Johnson; Ryan Lewis; Bill

Maxwell; Mark Mitchell; Erwan Morellec; Greg Nini; Michael Roberts; Farzad Saidi; Bill Schwert;

Ilya Strebulaev; Ren´

e Stulz; and seminar participants at Arizona State, BI Oslo, Boston College,

Chicago Booth, Emory, EPFL and HEC Lausanne, the Federal Reserve Board, Ohio State, Okla-

homa, Penn State, Stockholm School of Economics, Wharton, the 2017 Colorado Finance Summit,

the 2018 FinanceUC Conference, the 2018 Columbia Junior Workshop in New Empirical Finance,

the 2018 Texas Forum on Corporate Finance, and the 2019 Utah Winter Finance Conference for

helpful suggestions. I am also grateful for discussions with employees of Greif, Huntington Bank,

and Scotts Miracle-Gro. Chuck Fang and Rick Ogden provided excellent research assistance. I

have read The Journal of Finance’s disclosure policy and have no conflicts of interest to disclose.

I have a consulting relationship with AQR Capital Management that provides access to the bond

quote data used in this paper.

DOI: 10.1111/jofi.12849

C2019 the American Finance Association

905

906 The Journal of Finance R

Tabl e I

Empirical Evidence on Loan and Bond Recovery Rates

This table reports summary statistics on defaulted debt recovery rates from Moody’s Ultimate

Recovery Database. The sample includes cases involving public firms rated by Moody’s that filed

for bankruptcy between 1997 and 2017, excluding debtor-in-possession loans and firms in the

financial and utilities industries. The first set of rows includes all defaults in which the firm had

both loans and bonds outstanding, while the second set restricts the sample to defaults with first-

lien term loans and senior unsecured bonds. Observations are aggregated by default event. The

reported recovery rates are value-weighted averages of instrument-level recoveries, with weights

based on amounts outstanding at the time of default. Instrument-level recoveries are based on

Moody’s suggested method (settlement value or trading price) and discounted from emergence to

the default date by the instrument’s interest rate. Recovery of 100% means the claim was paid

principal and accrued interest. Firm-level recovery is the family recovery rate reported by Moody’s.

Mean SD p10 p50 p90 Defaults

All defaults with loans and bonds outstanding

Firm-level recovery 0.51 0.26 0.16 0.48 0.90 316

All loan types 0.84 0.26 0.38 1.00 1.00 316

Line of credit 0.87 0.24 0.44 1.00 1.00 289

Term loan 0.76 0.30 0.29 1.00 1.00 202

All bond types 0.32 0.30 0.01 0.22 0.80 316

Senior secured 0.51 0.34 0.05 0.49 1.00 88

Senior unsecured 0.35 0.33 0.01 0.27 1.00 163

Defaults with first-lien term loans and senior unsecured bonds

Firm-level recovery 0.54 0.26 0.18 0.56 0.90 95

First-lien term loan 0.83 0.28 0.32 1.00 1.00 95

Senior unsecured bond 0.33 0.33 0.003 0.23 0.93 95

pricing in the loan and bond markets that is unaffected by firm-time observable

factors.1I account for the firm’s priority structure of debt using both reduced-

form and structural models of credit risk.

From a credit risk standpoint, the key difference between loans and bonds is

that loans are senior in bankruptcy.2Defaultis the only state in which creditors

are not paid in full, so expected payoffs in default are a crucial determinant of

the cost of credit. Table Ipresents evidence from Moody’s Ultimate Recovery

Database on bankruptcies of firms with both loans and bonds outstanding at

the time of default from 1997 to 2017. The average recovery rate for loans

is 84%, with lenders recovering principal and accrued interest in most cases,

whereas senior unsecured bonds recover 35% on average. This difference in

exposure to default losses implies that loan credit spreads should be smaller

than bond credit spreads.

Duffie and Singleton (1999) develop a reduced-form default intensity model

that serves as a useful benchmark. In their model, the credit spread on a risky

1The sample consists of large firms with access to public debt markets, so it does not represent

the population of corporate borrowers. I discuss external validity in Section I.

2Besides seniority, there exist many other differences in the cash flows and economic features

of loans and bonds that I discuss in this paper.

Pricing of Bank Loans Relative to Capital Market Debt 907

zero-coupon bond equals the “risk-neutral mean-loss rate,” or the probability of

default times the expected loss given default. The probability of default is the

same for all debt instruments issued by the same firm, assuming that cross-

default provisions are in place, so the relative spreads on bonds and loans

depend only on their expected losses given default. Table Ishows that the

unconditional expected loss given default for senior unsecured bonds is four

times higher than the expected loss given default for loans. The Duffie and

Singleton (1999) model therefore predicts that bond spreads should be about

four times as large as loan spreads.3

Figure 1visually summarizes the relative pricing of corporate bonds and

loans, uncovering facts not previously reported in the literature. The top panel

plots bond and loan spreads relative to the London Interbank Offered Rate

(LIBOR) swap curve as nonparametric functions of distance-to-default

(Bharath and Shumway (2008)), and the bottom panel plots the ratio of the

spreads. The sample comprises new loans with secondary market quotes for

senior unsecured bonds from the same firm on the loan’s start date.4Maturi-

ties under three years and loan-bond pairs with a maturity difference greater

than two years are excluded to mitigate the effect of maturity structure on the

relative spreads.

At first glance, the plot in Panel A appears intuitive. When the firm is close to

default, bond spreads are significantly higher than loan spreads but fall short of

the bond-loan spread ratio of four-to-one predicted by the Duffie and Singleton

(1999) model when the distance-to-default is zero. However, the loan and bond

spreads are similar when the firm is far from default, which is puzzling in light

of the bank’s senior position in the priority structure. Panel B of Figure 1shows

that the ratio of bond spreads to loan spreads is significantly lower than four in

all rating categories. Strikingly,the loan and bond spreads of investment-grade

firms are statistically indistinguishable.5This suggests that bank lenders earn

a substantial premium relative to the market pricing of their risk exposure.

To quantify the magnitude of this premium, I apply a structural model of

credit risk. This analysis focuses on a subsample of the data that contain term

loans secured by a first lien and senior unsecured bonds, so that seniority is

3The expected loss given default in Duffie and Singleton (1999) is under the risk-neutral mea-

sure, whereas Table Iprovides an estimate of the physical expectation. The current discussion is

for illustration and may be affected by systematic risk in recoveries. Section III.Bdiscusses the

risk premium on recoveries necessary to explain loan pricing.

4I address the difference between primary and secondary market pricing in Section II.A.2.The

comparison of fixed-rate bond spreads and floating-rate loan spreads is innocuous, as Duffie and

Liu (2001) show that the fixed-floating credit spread basis is on the order of one basis point.

5While loans to noninvestment-grade firms are generally secured and therefore senior to un-

secured bonds, loans to investment-grade firms are typically unsecured and on equal footing with

bondholders. However, it is unusual for an investment-grade firm to default before being down-

graded. The Internet Appendix, available in the online version of this article on The Journal of

Finance website, shows that loans were unsecured in only 7% of the 316 default events in Moody’s

Ultimate Recovery Database from 1997 to 2017 that involved both loans and bonds. In this sample

of defaults, there is one instance in which bonds received a materially higher recovery than loans,

and there are no instances in which unsecured bonds recovered more than secured loans.