Do Creditor Rights Increase Employment Risk? Evidence from Loan Covenants

• Published date: 01 December 2016
• DOI: http://doi.org/10.1111/jofi.12435
• Author: ANTONIO FALATO,NELLIE LIANG
• Date: 01 December 2016

THE JOURNAL OF FINANCE •VOL. LXXI, NO. 6 •DECEMBER 2016

Do Creditor Rights Increase Employment Risk?

Evidence from Loan Covenants

ANTONIO FALATO and NELLIE LIANG∗

ABSTRACT

Using a regression discontinuity design, we provide evidence that there are sharp and

substantial employment cuts following loan covenant violations, when creditors gain

rights to accelerate, restructure, or terminate a loan. The cuts are larger at firms with

higher financing frictions and with weaker employee bargaining power,and during in-

dustry and macroeconomic downturns, when employees have fewer job opportunities.

Union elections that create new labor bargaining units lead to higher loan spreads,

consistent with creditors requiring compensation when employees gain bargaining

power. Overall, binding financial contracts have a large impact on employees and are

an amplification mechanism of economic downturns.

AFUNDAMENTAL QUESTION IN BOTH finance and macroeconomics is whether fi-

nancing frictions and, more broadly,firm financial conditions, have real effects.

Existing empirical research on this question focuses primarily on corporate in-

vestment (e.g., Whited (1992), Rauh (2006), and Chava and Roberts (2008);

Benmelech, Bergman, and Seru (2011) is a recent exception that instead fo-

cuses on employment). However, a long tradition of theoretical research in

both macro (e.g., Bernanke and Gertler (1989)) and corporate finance (e.g.,

Berk, Stanton, and Zechner (2010)) as well as the exceptional job losses in

the aftermath of the financial crisis and in the Great Recession of 2008 and

2009 highlight the potential importance of financing effects on employment,

∗Antonio Falato and Nellie Liang are with the Federal Reserve Board. Viewsexpressed are those

of the authors and do not represent the views of the Board or its staff. Special thanks to Mark

Carey and Greg Nini for their help with Dealscan and for kindly sharing their Compustat-Dealscan

key. We thank the Editor and two anonymous referees, Effi Benmelech (discussant), Sudheer

Chava (discussant), Steve Davis (discussant), Edward Morrison (discussant), Greg Nini (discus-

sant), Michael Roberts (discussant), Amit Seru (discussant), Hayong Yun(discussant), Bill Bassett,

Jonathan Berk, Gabriel Chodorow-Reich, Kai Li, Atif Mian, Marco Pagano, Martin Schmalz, Steve

Sharpe, Jeremy Stein, Amir Sufi, Till von Wachter, and Toni Whited, seminar participants at

the Federal Reserve Board, and conference participants at the annual meetings of the American

Economic Association, American Finance Association, the Conference on Empirical Legal Studies,

the UBC Winter Finance Conference, the Financial Intermediation Research Society, the NBER

Capital Markets and the Economy, the NBER Monetary Economics, and the CSEF-EIEF-SITE

conference on labor and finance for helpful comments and discussions. Brandon Nedwek, Nicholas

Ryan, Richard Verlander, Xavy San Gabriel, and especially Suzanne Chang provided excellent

research assistance. All remaining errors are ours. The first version of this paper was released as

a working paper in December 2012. The authors do not have any potential conflicts of interest to

disclose, as identified in the JF Disclosure Policy.

DOI: 10.1111/jofi.12435

2545

2546 The Journal of Finance R

which raises two important empirical questions: Are corporate financing and

labor policies related, and how? The goal of this paper is to make progress on

these questions by examining the response of corporate labor policies to loan

covenant violations.

There are several reasons why loan covenant violations provide a unique

opportunity for studying the employment impact of financing. First, violations

give creditors the right to demand immediate repayment and withhold further

credit. Thus, they provide a specific channel through which financing can im-

pact employment, whereby firms may lay off employees in order to improve

net cash flows and appease creditors who are concerned about the value of

their claims. Second, violations are frequent and rarely lead to outright de-

fault, suggesting that they are well suited to study employment effects outside

of bankruptcy. Third, because of their discrete nature, violations enable us to

employ a Regression Discontinuity (RD) design analogous to that in Chava

and Roberts (2008), which helps to overcome the identification challenge of dis-

tinguishing financing effects from changes in firm fundamentals and outlook.

Finally, violations allow us to examine time-series variation in the effect of

financing on employment, which has important implications for the academic

and policy debate on the influence of financing on macroeconomic stability.

Using an RD, we document robust evidence of sizable job cuts following loan

covenant violations.1Our baseline estimates indicate that covenant violations

lead to a sharp drop of about 13 log points or about 10% of the workforce in

a given year, an economically significant effect that corresponds to a quartile

of the within-firm distribution of log employment. Violations also lead to an

increase in the likelihood of a layoff of about three percentage points, which

is economically significant relative to the 9% unconditional probability of a

layoff and suggests that employment reductions are not due to just voluntary

departures. These results are robust to restricting the sample to those observa-

tions that are “close” to the covenant threshold, where violations can plausibly

be considered a “quasi-random” treatment, following an approach similar to

Chava and Roberts (2008). They also hold up to a variety of falsification and

sensitivity tests, which include using alternative measures of employment and

covenants. Taken together, our evidence suggests that financial distress entails

large costs for employees.

The employment impact of violations varies predictably with various prox-

ies for a firm’s financial condition and for the relative bargaining strength of

creditors versus the firm and its employees. Employment cuts subsequent to

violations are as deep as 17 log points for highly levered firms and for firms

with no credit rating, which have fewer alternative sources of external financ-

ing, as well as for firms in which labor has less bargaining power, based on

1Our sample includes 9,190 firm-year observations for 2,153 unique U.S. firms that have infor-

mation on loan covenants in Dealscan and employment and balance sheet information in Compu-

stat between 1994 and 2010. We complement our employment data with a large sample of layoff

announcements for 7,649 firm-year observations in Compustat, which we assemble by combining

the results of keyword searches from major news sources and information on layoff announcements

from Capital IQ.

Do Creditor Rights Increase Employment Risk? 2547

industry-level measures of union membership and coverage. Thus, the acceler-

ation and termination rights granted upon violation have more bite on employ-

ment when creditors are relatively stronger and labor is weaker. Employment

cuts also are deeper for firms in industries with relatively high domestic con-

centration and low import penetration, suggesting that violations may help

mitigate “quiet-life”-type agency distortions that make managers reluctant to

fire employees (see Bertrand and Mullainathan (2003) and Atanassov and Kim

(2009)). Investment cuts also vary predictably with labor bargaining power and

are bigger when labor is stronger, which highlights an interesting implication

of our findings, namely, the substitutability between labor and capital.

The impact of violations on employment also varies predictably in the time

series, with employment cuts following violations being outsized in bad times

and in industry downturns, when employees have fewer alternative job oppor-

tunities.2For example, violations lead to employee cuts that are more than

twice as large in NBER recession periods as in nonrecession periods. The com-

bination of a higher likelihood of violations and bigger job cuts when violations

occur leads to an estimated expected impact on employment that is about three

times larger in recession than in nonrecession times. Since employment cuts are

concentrated exactly in those times when creditors have the most bargaining

power and labor has the least, the time-series evidence corroborates our inter-

pretation that the impact of violations reflects the relative strength of creditor

and labor rights. The time-series results suggest that financial covenants are

an important amplification mechanism of economic downturns since violations

are both more likely to occur and, if they occur, have an even larger adverse

impact on employment in bad times.

In our final set of tests, we examine whether loan prices impound labor rights,

that is, whether firms face a higher cost of capital when labor is stronger. Our

identification is an RD that exploits the requirement of U.S. labor laws that,

in order to create a new labor bargaining unit, an election must be held in

which workers vote by majority rule for or against union representation.3We

find that union wins are reliably associated with higher spreads on loans origi-

nated within two years of the election. The effect of unionization is particularly

large for low-rated and unrated borrowers, for which the allocation of bargain-

ing rights should be expected to matter the most because these borrowers have

higher risk of financial distress. This result is robust to a matched sample

methodology and continues to hold when we consider “close” elections to ad-

dress potential concerns about the anticipation of election outcomes and omit-

ted variables. The evidence together suggests that there is a trade-off between

credit pricing and the allocation of control rights, as creditors demand ex-ante

compensation when the ex-post bargaining power of employees is higher.

2We thank Michael Roberts for suggesting these tests of time-series variation.

3Wematch administrative data from the National Labor Relations Board (NLRB) on all elections

that took place in the United States between 1985 and 2010 with pricing information from Dealscan

and accounting data from Compustat, which results in a sample of 3,914 loans for 1,756 unique

election results.