Short-termism Spillovers from the Financial Industry*

Andrew Bird† Aytekin Ertan‡

Stephen A. Karolyi†

Thomas G. Ruchti†

November 2016

* We thank Alex Edmans and seminar participants at London Business School for valuable comments and suggestions. † Bird, Karolyi, and Ruchti are at Tepper School of Business, Carnegie Mellon University (apmb@andrew.cmu.edu; skarolyi@andrew.cmu.edu; ruchti@andrew.cmu.edu). ‡ Ertan is at London Business School (aertan@london.edu).

Abstract To meet short-term benchmarks, lenders may alter their monitoring behavior, providing a channel for short-termism incentives to spill over into the corporate sector. We find that lenders with short-termism incentives enforce material covenant violations at abnormally high rates. Further, they are more likely to target high-quality borrowers with which they have a prior relationship and that are less financially constrained. Affected borrowers are more likely to switch lenders, receive worse loan terms on future loans, and reduce investment. Market reactions to announcements of material covenant violations when lenders have short-term incentives are 0.88% lower, suggesting that short-termism spillovers are value-decreasing. JEL Classifications: G21, G28, G32, M41 Keywords: lending, short-termism, covenants, monitoring, spillovers, real effects

1

1 Introduction

The transmission of corporate shocks is an important friction affecting intrafirm and

interfirm relationships. Capital and labor shocks are distributed through firms’ internal markets

(Silva 2013; Giroud and Mueller 2015, 2016); stockholders transmit shocks to corporations through

financing and governance channels;1 customers and suppliers transmit shocks through production

networks;2 perhaps most importantly, financial intermediaries transmit shocks. Financial markets

regulation and bank supervision spills over to the corporate sector through financial intermediaries

(Hirtle, Kovner, and Plosser 2016), and lenders transmit financial shocks to borrowers (Chava

and Purnanandam 2011; Chodorow-Reich 2014; Bord, Ivashina, and Taliaferro 2014; Carvalho,

Ferreira, and Matos 2015). In this paper, we investigate the transmission of short-term capital

market incentives from lenders to borrowers in the private loan market.

The importance of myopia in corporate decision-making raises questions about how these

corporate decisions spill over across markets more broadly. Because the financial sector interacts

with other sectors (through lending, equity issuance, acquisitions, etc.), financial sector shocks

and incentives could have far-reaching effects. In principle, because banks lend to private as well

as public firms, financial sector short-termism is a distinct channel through which capital markets

pressures that are typically associated with public markets could affect private firms. Similarly,

although managerial myopia is typically associated with poorly governed firms, even firms with

the best governance or highest institutional ownership may be affected by short-termism spillovers

from the financial sector.

We provide new evidence that short-term capital market incentives are transmitted

through financing. Collectively, we produce four novel findings. First, we find that lenders facing

1 See Cooper and Kaplanis (1994), Coval and Moskowitz (1999), Van Nieuwerburgh and Veldkamp (2009), Pedersen, Mitchell, and Pulvino (2007), Boyson, Stahel, and Stulz (2010), Shek, Shim, and Shin (2015), Belo, Lin, and Yang (2014), Dew-Becker and Giglio (2016). 2 See Wu (2016), Murfin and Njoroge (2015), Breza and Liberman (2015), Acemoglu, Carvalho, Ozdaglar, and Tahbaz-Salehi (2012), Gabaix (2011), Kelly, Lustig, and Van Nieuwerburgh (2013).

2

incentives to meet quarterly earnings benchmarks are more likely to extract material benefits from

borrowers. Second, lenders with short-termism incentives push relatively high-quality borrowers

into material covenant violations because these are precisely the borrowers from whom rents can

be extracted. Because unhealthy borrowers are already selected for material covenant violations

by lenders both with and without short-termism incentives, only relatively healthy borrowers are

left to be targeted by incremental attention. Third, affected borrowers are more likely to reduce

capital investment and research and development (R&D) expenditures. Given the selection of

higher quality borrowers, it is particularly likely that these real investment effects on borrowers

are value-destroying. Finally, we find that the market reaction to announcements of material

covenant violations is 88 basis points lower among borrowers whose lenders face short-termism

incentives, which suggests that the incremental attention from lenders with short-termism

incentives does not improve shareholder value.

We use the syndicated loan market as a setting in which myopic incentives are clear and

the consequences of lender short-termism for borrowers are measurable. To measure short-termism

incentives, we build on prior work on managerial myopia that focuses on the incentive to meet

short-term earnings benchmarks (Graham, Harvey, and Rajgopal 2005; Cheng and Warfield 2005;

Bergstresser and Philippon 2006; Peng and Roell 2008). As argued by this literature, short-term

earnings benchmarks provide a natural context in which to study managerial myopia because

firms close to the threshold stand to gain an immediate capital market benefit from increasing

reported earnings (Bhojraj, Hribar, Picconi, and McInnis 2009; Bird, Karolyi, and Ruchti 2016).

Income-seeking lenders have a clear incentive to push borrowers that have breached a

financial covenant threshold—defined as having negative covenant slack—into a material

covenant violation, resulting in the borrower being in technical default.3 The lender may then

pursue a formal waiver, which typically involves a waiver fee, or a loan amendment, which may

3 Throughout the paper, we use material covenant violation and technical default interchangeably. While potentially distinct events, they are observed simultaneously in our data.

3

involve an amendment fee and changes in loan terms, including spreads and amounts. In fact, the

lender may even accelerate the loan, requiring the borrower to immediately post principal

repayment. In many cases, loan acceleration is accompanied by a refinancing, but may also involve

payment default and, in rare cases, bankruptcy. Financial covenant thresholds and the underlying

financial ratios and amounts are observable, so we can calculate the distance to covenant

thresholds and identify negative slack. We also observe loan renegotiations and, from borrowers’

financial statements, material covenant violations.4 Of these outcomes, material covenant

violations and subsequent loan renegotiations are most relevant to lenders with short-termism

incentives because waiver and amendment fees are immediately recognizable as income and

incremental changes in loan spreads and amounts also boost short-term earnings.5

Consistent with the existence of short-term benefits of detecting and enforcing material

covenant violations, we provide novel evidence that lenders’ short-termism incentives are

transmitted to borrowers through changes in both the frequency of technical default and the types

of borrowers that are pushed into technical default. We find that, for borrowers with negative

slack, lenders with short-termism incentives enforce material covenant violations at a 40 percent

higher rate. Among borrowers with material covenant violations, the severity of their breach, as

measured by the distance to financial covenant thresholds, is 51 percent lower for borrowers with

myopic lenders. The decision to enforce a material covenant violation is discrete, and, despite the

short-term benefits, enforcing incrementally more material covenant violations may be costly for

the lender in the long run. Because myopic lenders select borrowers that they would not otherwise

select into a material covenant violation, enforcing such violations may increase the likelihood

that affected borrowers find a new lender for subsequent loans (Kang and Stulz 2000; Bharath,

Dahiya, Saunders, and Srinivasan 2011; Gopalan, Udell, and Yerramilli 2011). Thus, enforcing

4 We use a combination of Michael Roberts’ and Amir Sufi’s covenant violation data (Roberts and Sufi 2009 and Nini, Smith, and Sufi 2012). 5 As per SFAS 91 (1986), in sole lender deals, loan origination fees integral to the issuance as well as interest payments are deferred for recognition. In syndicated loans, lead arrangers are entitled to recognize fees related to the fraction of the loan that they sell if they charge the same spread as all syndicate participants. In renegotiations, changes in interest rates are generally amortized over the remaining life of the loan. See Ertan (2016) for a more detailed treatment of the relevant accounting issues.

4

covenant violations may reduce future loan income, particularly the incremental loan income

generated from fees. We provide evidence that this lender behavior is costly and consistent with

the incentive to temporarily increase reported earnings—loans renegotiated with lenders with

short-termism incentives have 7.6 percent higher spreads and 0.9 percent larger amounts relative

to loans renegotiated with other lenders, but affected borrowers are 1.4 percent more likely to

switch lenders for their next loan.

How does the behavior of myopic lenders spill over to the corporate sector? Lending

relationships may solve asymmetric information or moral hazard problems, but, in doing so, they

may increase the cost of switching lenders (Chava and Purnanandam 2011; Bord, Ivashina, and

Taliaferro 2014; Carvalho, Ferreira, and Matos 2015). Relationship lenders know more about

borrowers than outsiders, so their discretion on whether to enforce material covenant violations

can be beneficial because it promotes the use of this private information in a valuable way.

However, in times of stress, this discretion could promote opportunistic behavior by lenders. In

addition to the direct costs associated with material covenant violations, our evidence suggests

that borrowers switch lenders at a higher rate after observing such opportunistic behavior. The

incremental cost of funding renegotiated loans, combined with waiver fees and unobservable

switching costs, may lead borrowers to alter their investment behavior. We provide evidence that

lender short-termism has real effects on the investment policy of their borrowers. Relative to other

borrowers that breach their covenant thresholds, affected borrowers are 2.4 percent more likely to

cut R&D spending and 4.9 percent more likely to cut capital investment. Additionally, borrower’s

announcements of material covenant violations are met with 88 basis points lower market

reactions when their lenders face short-termism incentives. Together, these findings suggest that

the incremental attention from lenders facing short-termism incentives is value-decreasing for

borrowers’ shareholders.

There is an extensive range of literature studying the role of myopia in finance. Asker,

Farre-Mensa, and Ljungqvist (2015) study the consequences of short-termism for investment

5

decisions by investigating the difference between public and private firms. They find that private

firms invest more than public firms and are more responsive to investment opportunities. Edmans,

Fang, and Lewellen (2016) develop a measure of managerial short-termism based on the impending

vesting of CEO equity and tie this to reductions in investment. Salitskiy (2015) uses a similar

measure to show that CEO compensation duration is positively associated with firm risk. Ladika

and Sautner (2016) use the implementation of FAS 123R, which required option compensation to

be expensed, to show that executives with more short-term incentives spend less on long-term

investment. Contrastingly, Laux (2012) shows that short-term incentives may provide benefits

through earlier feedback regarding CEO talent. Relative to this literature, we find that one of the

key determinants of myopic behavior, the incentive to meet short-term earnings benchmarks, is

transmitted from the financial sector to the corporate sector through financing relationships.

Earnings management is an important and well-documented manifestation of managerial

myopia and can occur through manipulating accruals (e.g., Burgstahler and Dichev 1997) or real

activities manipulation (e.g., Roychowdhury 2006). We look specifically at this behavior for banks

(Scholes, Wilson, and Wolfson 1990; Beatty, Ke, and Petroni 2002). Banks have been shown to

be particularly likely to engage in end-of-quarter transactions management in an effort to meet

their quarterly earnings benchmarks. For example, Dechow and Shakespeare (2009) study the

increased use of loan securitization toward fiscal quarter ends, while Ertan (2016) shows that

banks, in order to report higher income, initiate more (and conditionally cheaper) loans toward

the end of quarters in which this would help banks meet earnings benchmarks.

Our work is also closely related to the literature on lender control and loan contracting.

This literature finds that lenders exert either implicit or explicit control over borrower actions

while borrowers are in violation of covenants, including on governance and executive

compensation, investment, employment, innovation, and capital structure (Chava and Roberts

2008; Roberts and Sufi 2009; Nini, Smith, and Sufi 2009, 2012; Falato and Liang 2013; Gu, Mao,

and Tian 2016; Chakraborty, Chava, and Ganduri 2016). Technical defaults also have direct and

6

negative financing consequences for borrowers, such as covenant waiver fees, spread increases,

renegotiation, refinancing, and potentially payment default (Roberts 2014, Freudenberg,

Imbierowicz, Saunders, and Steffen 2015). Some of these negative consequences are persistent;

renegotiation often involves increasingly strict covenants or collateral requirements, which make

future default more likely. Our results suggest that lenders’ incentives to meet short-term earnings

benchmarks provide within-lender variation in their detection and enforcement of covenant

violations.

Bank financial health affects lending terms, and therefore borrowers (Hubbard, Kutner,

and Palia 2002). In line with this evidence, shocks to lenders may spread to their respective

borrowers. For example, Berger, Saunders, Scalise, and Udell (1998) find that bank mergers lead

to a reduction in small business lending by the newly-merged lender, though this is largely offset

by the actions of other lenders. In particular, reduction in lending is mitigated by increased lending

by banks not affected by the shock (Bord, Ivashina, and Taliaferro 2015). Gan (2007) shows that

borrowers whose lenders were exposed to the land market collapse in Japan experienced declines

in market values and reduced investment. Further, Peek and Rosengren (1997) study the

transmission of this shock to the lending activities of US branches of Japanese banks, finding

similar negative outcomes. However, Ongena, Smith, and Michelsen (2003) study a crisis in the

Norwegian banking system and find relatively small effects on borrowers. Chodorow-Reich (2014)

finds that negative shocks to lenders lead to negative outcomes for borrowers, particularly a

reduction in employment. This transmission is worse for bank-dependent borrowers (Chava and

Purnanandam 2011), and this contagion is not mitigated by public debt markets (Carvalho,

Ferrera, and Matos 2015). Murfin (2012) shows that, following negative shocks to their loan

portfolios, banks impose stricter covenants on borrowers. However, Loutskina and Strahan (2009)

find that securitization has reduced the effect of bank characteristics on borrowers. Our results

provide new evidence that lenders not only transmit financial shocks to borrowers, but also their

own short-term capital markets incentives. These results suggest that, through financing

7

relationships, the pressures of public markets can be transmitted across governance structures and

to private firms.

2 Data and Sample Selection

We construct our main estimation sample using three sources: the Center for Research in

Security Prices (CRSP), Standard & Poor’s Compustat, and Thomson Reuters’ DealScan. We

obtain market data from CRSP, quarterly firm financials from Compustat, and loan details from

DealScan. In addition to these sources, we rely on: i) Michael Roberts’ link table to match

DealScan borrowers to Compustat/CRSP, ii) Aytekin Ertan’s link table to match DealScan lead

lenders to Compustat/CRSP, and iii) Amir Sufi’s and Michael Roberts’ data to identify material

covenant violations. Since the data on material covenant violations span 1996 to 2008, we

concentrate on this period in our main tests. We also exclude borrowers from the financial and

utilities sectors from our analysis.6 These considerations and matching procedures yield a total

sample of 8,523 distinct loan packages and 3,466 borrowers.

Since we want to understand lender behavior on a time-varying basis (i.e., rather than on

a loan-initiation basis), we trace loans over time. Consistent with this objective, we construct our

main sample at the package-quarter level. We opt for loan packages rather than tranches because

covenants are defined at the package level, and for loan packages rather than the borrowing entity

since the same borrower may have multiple loans outstanding from different lenders in a given

quarter. With these objectives in mind, we construct the sample as follows. We match each loan

package to the underlying Compustat borrower, in order to compute the borrower’s time-varying

covenant slack and financial characteristics. We also match loans to the Compustat lead lender,

to assign time-varying lender performance to each package-quarter. We convert packages to

package-quarters using the stated start and end dates, which, after other data requirements, yields

6 Two-digit SIC codes between 60 and 69, and 44 and 50, respectively.

8

a total of 90,125 observations.7 Appendix A includes variable definitions, and Panel A of Table 1

reports both fixed and time-varying characteristics of corporate borrowers and loans.

We proxy for lender short-termism using MBLender, an indicator variable that equals one

for lender-quarters in which reported earnings-per-share (EPS) either meets or beats the

Institutional Brokers' Estimate System (IBES) consensus by less than two cents. In these

calculations, we use unadjusted and unscaled one-quarter-ahead forecasts, following the concerns

raised by Diether, Malloy, and Scherbina (2002) and Cheong and Thomas (2011), respectively.

From a broad perspective, just-beating the analyst forecast consensus is indicator of corporate

short-termism. The idea is that in various bins comprising the earnings surprise distribution, the

bin on the right of zero, on average, includes more firms that follow myopic actions than other

bins do (Healy 1985; Roychowdhury 2006). This behavior manifests itself in operating, financing,

and investment decisions; therefore, it is sensible to see banks changing how they handle their

borrowers in the event of technical default. From a specific standpoint, banks in need of enhanced

accounting income performance are likely to try to acquire additional rents (e.g., waiver and

amendment fees) from the borrower to enjoy a boost in contemporaneous earnings.

Turning to the borrower side of our analysis, our efforts to construct a time-varying

measure of covenant slack yield a total of 90,125 firm-quarters. For each of these observations, we

calculate the slack of firm i’s jth covenant in quarter t as:

Slackijtmin = (ui,j,t – ui,j,t) / σijt (1)

for minimum covenants, such as minimum interest coverage ratio, and as:

Slackijtmax = (ui,j,t – ūi,j,t) / σijt (2)

7 We define package maturity as the stated maturity date of the largest tranche.

9

for maximum covenants, such as maximum debt-to-EBITDA ratio. In these equations, u

represents the underlying financial ratio or amount, u [ū] the minimum [maximum] covenant

threshold, and σ the past eight-quarter volatility of the underlying ratio or amount.8 Among these

standardized values, we code the minimum as the firm-quarter Slack. We also code an indicator

variable, NegativeSlack, denoting firm-quarters with Slack less than zero. One of the dependent

variables of interest in our analysis of lender short-termism is Violation, an indicator for covenant

violations identified in the data sets provided by Michael Roberts’ and Amir Sufi. This metric

equals one for package-quarters with material covenant violations, and zero otherwise.9

3 Short-termism Spillovers

The implications of short-termism arising from the importance of quarterly earnings

benchmarks have been debated for many years. On one hand, capital markets may discipline

corporate decision-makers, which could produce positive effects, including constraining manager-

specific investments (Shleifer and Vishny 1989). However, they may also induce decision-makers

to focus on short-term cash flows at the expense of long-term investments (Stein 1988, 1989). In

the corporate sector, much of the empirical evidence is consistent with the latter, more negative,

aspects of these capital market incentives (Graham, Harvey, and Rajgopal 2005; Cheng and

Warfield 2005; Benmelech, Kandel, and Veronesi 2010; Gopalan, Milbourn, Song, and Thakor

2014), but less is known about this effect in the financial sector (Ertan 2016). Like other public

corporations, public lenders are subject to capital markets pressure to meet quarterly earnings

benchmarks (Beatty, Ke, and Petroni 2002). Similarly, these short-term capital market incentives

may discipline lenders or produce value-reducing short-termism. We conjecture that this pressure

8 Covenant threshold calculations appear in Appendix B. These are broadly in line with Demerjian and Owens (2016). In untabulated tests, we focus only on covenants for which the calculation of negative slack is much less likely to be associated with significant measurement error—net worth and current ratio figures (e.g., Chava and Roberts 2008, Roberts and Sufi 2009a) and find similar results throughout. 9 A limitation of our analysis is that we observe Violation at the firm-quarter level, which may create measurement error in our loan package-quarter level regressions. We argue that this measurement error is likely to produce attenuation bias because it means loan package-quarters that are not in violation may be assigned Violation equal to one if the borrower is in violation of a financial covenant in a separate loan package in that quarter. Hence, loan package-quarters that are not in violation may be assigned to Violation despite not being subject to lender influence.

10

induces a change in behavior, potentially either positive or negative, with respect to their loan

portfolios. However, we first statistically test whether lenders appear to meet earnings benchmarks

systematically.

Within our sample of public lenders, diagnostic tests suggest statistically significant levels

of EPS manipulation. Panel A of Figure 1 presents a histogram of lenders EPS relative to equity

analysts’ consensus EPS forecasts, or “earnings surprise”. The figure shows a large discontinuity

in the frequency of lender-quarters just above zero earnings surprise, consistent with EPS

manipulation. Panel B of Figure 1 presents a McCrary test of this discontinuity, and rejects the

null hypothesis of no discontinuity at the one percent significance level. In economic terms, the

McCrary test discontinuity estimates suggest that an abnormal mass of 4.38 percent of lender-

quarters exists just above the zero earnings surprise cutoff. Relative to the 29.71 percent of lender-

quarters within two cents of the zero earnings surprise cutoff, this abnormal mass represents an

economically meaningful set of lender-quarter observations. Overall, these tests provide evidence

in keeping with changes in lender behavior due to short-term pressure from capital markets.

Additionally, this finding complements evidence in Ertan (2016) that lenders facing similar short-

term incentives trade off spread-based earnings for fee-based income in new loans.

As the delegated monitor, the lead arranger of a loan syndicate is obligated to engage

borrowers frequently on behalf of the syndicate participants. These interactions fall under the

broad scope of monitoring, but include coordinating and observing payments and compliance with

contractual terms. Among this set of interactions, one of the most common involves an assessment

of the borrower’s conformity to financial covenants.

Lead arrangers may observe borrowers to be in violation of covenant thresholds (e.g.,

maximum Debt/EBITDA), and may then relay this information to syndicate participants.

Although the loan syndicate may vote to accelerate the loan, which would force the borrower to

refinance, renegotiate, or incur a payment default, in practice, it would typically waive the

covenant violation (Gopalakrishnan and Parkash 1995; Dichev and Skinner 2002). In this process,

11

the lead arranger has discretion over the intensity of their detection technology and, to some

extent, over the ex-post demands on the borrower given their individual syndicate voting

procedure (Lee and Mullineaux 2004).10

Importantly, in exchange for the covenant violation waiver, borrowers typically pay a fee,

which, for the lenders, may be immediately recognizable as income. They may alternatively agree

to change the interest spread for the duration of the loan (Beneish and Press 1993, 1995; DeAngelo,

DeAngelo, and Wruck 2002; Sufi 2009). Lenders, including the lead arranger, stand to increase

contemporaneous income in the form of fees, whether amortized over the remaining life of the

loan or not, and increased spreads.11

Finally, the syndicated loan market is dominated by a small subset of publicly-listed

lenders. Members of this group typically serve as the lead arranger, particularly for larger loans,

which enables them to conduct primary negotiations with the borrower, collect incremental fee

income, and monitor the borrower. As a result, pressure to beat short-term earnings benchmarks

is a plausible mechanism through which lenders’ capital market incentives can affect borrowers in

the syndicated loan market.

Given the unique discretion over the detection and outcomes of covenant violations

attributed to lead arrangers as well as the frequency of publicly-listed lenders acting as lead

arrangers, we investigate the effects of the capital markets incentives of short-term earnings

benchmarks on technical default frequency and selection. Because we can observe the distance to

covenant thresholds for publicly-listed borrowers, we can construct a measure of covenant

10 For covenant waivers, simple majority or super majority voting rules apply. As lead arrangers typically retain between one-quarter and one-half of the loan amount, they have a proportionally high voting stake in technical default proceedings. 11 As per SFAS 91 (1986), lead arrangers are entitled to recognize fees related to the fraction of the loan that they sell if they charge the same spread as all syndicate participants. Changes in interest rates are generally amortized over the remaining life of the loan. See Ertan (2016) for a more detailed treatment of the relevant accounting issues.

12

violation selection intensity and characterize the selection of firms into technical default at the

lender-quarter level of observation.

We begin our analysis with two sets of motivational nonparametric evidence that lenders’

short-term incentives affect borrowers in their loan portfolio. First, we document that lenders that

just beat their short-term earnings benchmarks are significantly more likely to push lenders with

negative slack (those who are in breach of a covenant) into technical default than lenders that

just miss, significantly miss, or significantly beat their quarterly earnings benchmarks. Figure 2

presents evidence that, conditional on the distance to covenant thresholds, borrowers are

significantly more likely to be pushed into technical default by lenders with short-term incentives

compared to other lenders. Second, we document that the lenders with short-term incentives do

not randomly select violating borrowers; instead, they select and push higher-quality borrowers

into technical default. Figure 3 presents evidence that the average distance to covenant thresholds

among borrowers in technical default is significantly higher for borrowers whose lender faces short-

term incentives than for other borrowers. Because the distance to covenant thresholds is both the

revealed preferred measure of borrower quality and salient in this market, this evidence suggests

that (i) on average, lenders are more likely to push low-quality borrowers into technical default,

and (ii) short-term incentives constrain lenders and lead them to push more, higher-quality

borrowers into technical default than they otherwise would.

The two pieces of motivating evidence from Figures 2 and 3 suggest that lenders alter

their treatment of borrowers in their loan portfolio when they face short-term incentives. However,

it is also important to consider the unobservable characteristics of lenders or borrowers that might

otherwise explain these observations. For a more formal treatment of such issues, we turn to a

saturated fixed-effects regression analysis.

13

3.1 Technical default frequency

A lender has a discontinuous increase in control at the covenant’s thresholds because when

the borrower’s underlying financial ratio exceeds the maximum or falls below the minimum

threshold, the borrower is in violation of the covenant and the lender has the right to accelerate

the loan. Figure 2 shows that the probability of experiencing a material covenant violation

increases significantly at the covenant threshold (i.e., when covenant slack is zero). In our

regression analysis, to focus on the region in which the lender has the discrete choice (i.e., to be

precise, the loan syndicate votes) to accelerate the loan and extract some benefit from the

borrower, we rely on NegativeSlackit, an indicator variable that equals one if borrower i has a

negative slack of a covenant in quarter t and zero otherwise.

Similarly, in keeping with the literature on short-termism and earnings manipulation, we

focus on lender-quarter observations in which the lender has an earnings surprise in the [0,2) cent

interval. As defined in the previous section, MBLenderjt is an indicator variable that equals one if

lender j has an earnings surprise of zero or one cent in quarter t and zero otherwise. To avoid

inference problems associated with behavioral differences among lenders at different places within

the earnings surprise distribution, we semi-parametrically control for lender earnings surprise

using five additional indicators. They identify lender-quarter observations in which the lender has

an earnings surprise between negative five and negative two cents, between negative two cents

and zero cents, between positive two and positive five cents, and greater than five cents,

respectively (misses by more than five cents is the omitted category).

Although the graphical evidence in Figure 2 is consistent with our conjecture that lenders

alter their behavior by enforcing material covenant violations (Violate) on a larger fraction of

borrowers with NegativeSlack, one may be concerned that unobservable characteristics of

borrowers or lenders, or even unobservable trends in the syndicated loan market, may drive this

finding. To that end, we incrementally saturate our baseline regression model with fixed effects

14

to isolate variation that cannot be explained by these unobservable characteristics. Our baseline

regression model is as follows:

Violationit = a + b1MBLenderjt + b2NegativeSlackit + b3MBLenderjt×NegativeSlackit + eijt (3)

in which Violationit is an indicator that equals one if borrower i discloses a material covenant

violation in quarter t and zero otherwise. We present estimates of this model in Table 2.

The estimates in Table 2 correspond to an increasingly restrictive set of fixed effects. These

fixed effects are meant to isolate and eliminate various confounding explanations for the baseline

result. In column (1), we present our baseline estimates, which include no fixed effects. This shows

that a borrower with negative slack is 5.6 percentage points more likely to be pushed into violation

if its lender is in a just-meet quarter, relative to borrowers whose lenders are not in a just-meet

quarter. This effect is statistically and economically significant, corresponding to a 25.1 percent

marginal effect given the average violation probability for negative slack borrowers of 22.3 percent.

Columns (2) through (4) add successively more fixed effects to the specification, starting

with borrower, lender and quarter fixed effects in column (2). These fixed effects remove fixed

differences across borrowers or lenders. This is important given that, for example, some lenders

may persistently be close to their earnings benchmarks, perhaps because some firms are more

easily understood by analysts than others, while also exhibiting consistent treatment of borrowers

with negative slack. Further, the inclusion of quarter fixed effects alleviates the concern that just-

meet quarters are more likely during certain times, such as recessions, which are associated with

greater incidence of covenant violation. Column (3) adds fixed effects for lender-borrower pairs as

well as industry-quarters. The former address the possibility of endogenous matching between

lenders and borrowers, so that the identification of the effect of a just-meet lender on the

probability of violation comes from variation within lending relationships, rather than across them.

The latter ensure that differences in lender portfolios, which may change over time, do not interact

with the business cycle (or other time series variation) to produce spurious results. Finally, in

15

column (4) we also include loan, borrower and lender characteristics to try to pick up remaining,

time-varying, differences across loans, which are not accounted for by the fixed effects. Across the

columns in this table, we maintain the consistent result that borrowers with negative slack are

significantly more likely to have a material covenant violation enforced if their lender is in a just-

meet quarter.

3.2 Technical default selection

Statistically, borrowers are no more likely to have negative slack in quarters in which their

lender faces short-term incentives than during quarters in which their lender does not.12 Therefore,

if lenders detect and enforce covenant violations for borrowers that violate covenants most

egregiously, then this limited supply of borrowers with negative slack implies that lenders must

select borrowers of higher quality to push into technical default. To quantify this selection, we

focus first on a salient and observable measure of quality, Slack, which is the standardized distance

to covenant threshold as defined in Section 2. Because lenders detect and contractually enforce

covenant violations using Slack, it is reasonable to assume that differences in the distribution of

Slack conditional on Violation for lenders currently facing or not facing short-termism incentives

should reflect changes in lender behavior as a result of short-termism. For example, if, on average,

lenders push borrowers with less negative Slack into technical default when they face short-

termism incentives, then we would infer that they select higher-quality borrowers into technical

default when they face short-termism incentives. Our baseline regression model is as follows:

Slackit = a + b1MBLenderjt + b2Violationit + b3MBLenderjt×Violationit + eijt (4)

We present estimates of the above model in Table 3. As in the previous section, the

estimates in Table 3 correspond to an increasingly restrictive set of fixed effects. These fixed

effects are meant to isolate and eliminate various confounding explanations for the baseline results.

12 In untabulated results, we find that the within-pair correlation between MBLender and NegativeSlack is -0.0000502 (p = 0.990), which means that within a lending relationship, borrowers are no more likely to experience NegativeSlack in MBLender quarters than non-MBLender quarters.

16

Column (1) presents the baseline estimates, which include no fixed effects. The positive and

significant coefficient on the interaction of Violation with MBLender shows that short-termism

incentives on the part of the lender lead them to push healthier borrowers, that is, those with less

NegativeSlack, into violation. The coefficient in column (1), which has the most restrictive set of

fixed effects, suggests that violating borrowers that have lenders with short-termism incentives

have, on average, covenant slack that is 1.06 standard deviations higher than other violating

borrowers. Relative to the mean slack of borrowers in violation of -2.06, this difference represents

a 51 percent increase in borrower quality.

As more fixed effects are included in the regressions, moving from column (2) through

column (4), we find consistent results. This suggests that neither unobservable differences across

borrowers, lenders, and time, nor those within borrower-lender pairs, nor any differences in

borrower industries over time can explain the main result in Table 3 that short-termism incentives

cause lenders to push relatively financially healthy borrowers into default. If violation has negative

consequences for borrowers and for the borrower-lender relationship, which we investigate in the

next section, this result suggests that such consequences are particularly harmful. This is because

they occur in cases where the borrower was less likely to otherwise experience financial difficulty

and where the lending relationship was more likely to continue.

To complement these results on lenders’ technical default selection on Slack, we investigate

alternative measures of borrower quality, which are intended to capture borrowers’ willingness

and ability to pay covenant waiver fees or change loan terms in response to a technical default.

These regressions take the following form:

Yit = a + b1MBLenderjt + b2Violationit + b3MBLenderjt×Violationit + eijt (5)

where Y represents a variety of relevant borrower and loan characteristics. Panel A of Table 4

presents estimates that correspond to borrower characteristics. In Panel A, Y alternatively

represents lnSize, the natural log of total assets, Whited-Wu, the Whited-Wu index of financial

17

constraints, PriorRelationship, an indicator that equals one if the borrower-lender pair have

transacted in the past and zero otherwise, and M/B, the ratio of market capitalization to book

equity. Panel B of Table 4 presents estimates that correspond to loan characteristics. In Panel B,

Y alternatively represents lnSyndicationFees, the natural log of total syndication fees in basis

points, Loan/Assets, the ratio of loan amount to total assets, and lnSpread, the natural log of the

all-in-drawn spread in basis points. All of the estimates presented in Table 4 correspond to

regression models with Lender and Industry×Quarter fixed effects, as it investigates cross-sectional

variation in borrower and loan characteristics, making borrower fixed effects unnecessary.

The results in Table 4 generally correspond with those in Table 3 in showing that ‘better’

borrowers are targeted for violations when lenders face incentives for short-termism. In particular,

larger borrowers, those with less severe financial constraints, and those with higher growth

prospects—as measured by market-to-book ratios—are incrementally more likely to be pushed

into violation in these cases. Perhaps most surprisingly, column (3) shows that this is also true

for borrowers with a prior relationship with their lender. Panel B of Table 5 yields a similar set

of findings, such as that loans that yielded greater syndication fees, that reflect a greater fraction

of the borrower’s assets, and that had smaller spreads, are more likely to be affected by the

lender’s incentives. These effects may also be related to a greater possibility for extracting rents

from some loans than others.

These results reflect a tradeoff between the short-term incentives created by the market

rewards to meeting earnings benchmarks and the long-term maintenance of profitable lending

relationships. The signs on the Violation coefficient in these regressions show that, in the absence

of short-term incentives, lenders are much more likely to exhibit forbearance for these better

quality borrowers. Short-termism appears to induce lenders to compromise on these

considerations.

18

4 Consequences of Short-termism Spillovers

The low baseline conversion rate from NegativeSlack to Violation indicates that lenders

exercise discretion in detecting and enforcing covenant violations, which suggests that they may

anticipate consequences from such behavior. In particular, the relationship lending literature

suggests that one reason lead arrangers value their position is the incremental fees that accrue to

them in their role as designated monitor for the loan syndicate. A natural and salient consequence

from the perspective of the lender is, therefore, the borrower’s decision to choose a new lender for

subsequent borrowing, which would reflect lost fees.

The mechanism through which lender short-termism is transmitted to borrowers involves

increased lender attention and a subsequent increase in the enforcement of material covenant

violations. From the borrower’s perspective, therefore, these are the natural consequences to

investigate. On one hand, the incremental attention from lenders facing short-termism incentives

may benefit borrowers because lenders may exercise value-increasing discipline over corporate

operations or financing. Indeed, the risk-shifting effects of lender control associated with borrower

technical default include governance and executive compensation, investment, employment,

innovation, and capital structure (Chava and Roberts 2008; Roberts and Sufi 2009; Nini, Smith,

and Sufi 2009, 2012). On the other hand, technical default has direct and negative financing

consequences, including covenant waiver fees, spread increases, renegotiation, refinancing, and

potentially payment default (Roberts 2014; Freudenberg, Imbierowicz, Saunders, and Steffen

2015). Some of these negative consequences are persistent; renegotiation often involves

increasingly strict covenants or collateral requirements, which make future default more likely.

To disentangle the effects of lender attention and default itself, we examine differences in

contemporaneous renegotiation, investment, and research and development expenditures as well

as longer-run default rates between borrowers whose lenders face short-termism incentives and

those who do not. Lender attention makes no clear prediction about these short-term

consequences; lender attention may improve project selection by reducing overinvestment in bad

19

projects, but it also may reduce investment overall to reduce the risk of loss at the expense of

firm value. Although both lender attention and default predict differences in the short-run

consequences we study, the literature on lender monitoring in technical default suggests that

lender attention is associated with lower future default rates whereas direct default costs are

associated with higher future default rates. In this way, inferences we make about the net benefits

of lender intervention due to short-termism rely upon differences in future default rates.

4.1 Lender choice and renegotiation

Because lead arrangers may exercise discretion over how intensely they seek out covenant

violations and their communication of such violations to the rest of the loan syndicate, borrowers

may update their expectations regarding future treatment based on lead arranger behavior when

the borrower is in violation of covenant thresholds. That is, borrowers may prefer to switch lenders

altogether if they believe they will receive harsher treatment from their current lender during

instances of financial distress. However, despite facing this potential consequence, lenders may be

willing to exercise this discretion to achieve short-term objectives. Typical renegotiation outcomes

like waiver fees and spread increases would improve contemporaneous (and future) loan income.

Another common outcome of renegotiations is a reduction in loan amount. However,

because the dollar contribution to income from waiver fees and spread increases is proportional to

the loan amount, lenders facing short-termism incentives should, in theory, prefer to avoid

decreasing loan amounts in renegotiation. Therefore, a differentiating feature of lender behavior

when facing short-termism incentives is that they not only impose technical default more

frequently, but that they also prefer to increase waiver fees and spreads as opposed to the loan

amounts. Our tests on changes in loan amount distinguish strategic short-termism from simply

harsher detection or enforcement of borrowers in violation of covenants.

We examine renegotiation outcomes and the propensity of borrowers to switch lenders in

subsequent borrowing using the following regression models:

20

LenderSwitchit = a + b1MBLenderjt + b2NegativeSlackit + b3MBLenderjt×NegativeSlackit + eijt (6)

ΔlnSpreadit = a + b1MBLenderjt + b2NegativeSlackit + b3MBLenderjt×NegativeSlackit + eijt (7)

ΔlnAmountit = a + b1MBLenderjt + b2NegativeSlackit + b3MBLenderjt×NegativeSlackit + eijt (8)

where LenderSwitch is an indicator that equals one if borrower i chooses a different lead arranger

in its next loan, as long as that loan is initiated within five years of quarter t, and zero otherwise,

and ΔlnSpread (ΔlnAmount) is the difference in the natural logs of the spread (amount) on

borrower i’s current loan and the spread on borrower i’s renegotiated loan spread (amount), where

renegotiated loans are identified using a combination of the renegotiation flag in Dealscan, the

methodology in Roberts (2014), and an additional filter that they are announced within one year

of quarter t. We present estimates of the LenderSwitch model in Table 5 and the renegotiation

outcomes models in Table 6. Estimates with ΔlnSpread and ΔlnAmount are presented in Panel A

and Panel B of Table 6, respectively. Again, the estimates in these tables correspond to an

increasingly restrictive set of fixed effects.

In Table 5, column (1) presents the baseline estimates, which include no fixed effects, of

the propensity of borrowers in violation of a financial covenant to switch lenders conditional on

whether their lender faces short-termism incentives or not. The coefficient on the interaction of

NegativeSlack with MBLender shows that lender short-term incentives increase the likelihood that

the borrower will switch lenders for its next loan by 1.4 percentage points. Given the importance

of lending relationships (Petersen and Rajan 1994; Berger and Udell 1995; Bharath, Dahiya,

Saunders, and Srinivasan 2007; Gopalan, Udell, and Yerramilli 2011) to both lenders and

borrowers, combined with the fact that the affected lending relationships are on average of better

quality (see Table 4), this increase in switching has negative long-term effects for both parties to

the loan.

This consequence of lender short-termism is paralleled in column (1) of Panels A and B of

Table 6, which show that just-meet quarters are associated with increases in both spreads and

21

loan amounts for borrowers with negative slack. This is just what we would expect if the lender’s

primary goal in renegotiating debt contracts following default is to extract fees, since both higher

spreads and increased loan amounts contribute to their income. Note that the actual cash-flow

effect for borrowers is uncertain, since they get to borrow additional funds but at the cost of

higher interest expense (in addition to the payment of waiver/amendment fees). The higher spread

paid by borrowers affected is likely a key component in their subsequent decision to switch lenders.

For each of Tables 5 and 6, we also include the suite of fixed effects as in the earlier tables,

and find consistent results, alleviating concern about omitted variables bias as an alternative

explanation of our results. Generally, this suggests that unobservable differences in lender loan

portfolios (and the behavior of borrowers therein) are not strongly correlated with the incidence

of short-termism incentives, as measured by just-meet quarters.

4.2 Future default rates

To disentangle the effects of lender attention and default itself, we examine differences in

contemporaneous renegotiation, investment, and longer-run default rates between borrowers

whose lenders face short-termism incentives and those who do not. Lender attention makes no

clear prediction about these short-term consequences; lender attention may improve project

selection by reducing overinvestment in bad projects, but it may also reduce investment overall

to reduce the risk of loss at the expense of firm value. Although both lender attention and default

can predict differences in the short-run consequences we study, the literature on lender monitoring

in technical default suggests that lender attention is associated with lower future default rates,

whereas direct default costs are associated with higher future default rates. In this way, inferences

we make about the net benefits or costs of lender intervention due to short-termism rely upon

differences in future default rates.

Three features of the behavior of lenders facing short-termism incentives make evaluating

the net benefits of said behavior difficult. First, incremental attention from lenders with short-

22

termism incentives may improve or worsen borrower project selection. Second, lenders with short-

termism incentives push borrowers into technical default at higher rates, which also means that

the set of borrowers in technical default is of higher quality. Third, the negative externalities of

default are significant and may be an early warning sign of future distress.

To disentangle the effects of incremental attention from lenders with short-termism

incentives, we make two sets of comparisons. First, we compare the future bankruptcy rates of

borrowers in violation of a financial covenant between a quarter in which their lender has short-

termism incentives and a quarter in which their lender does not. This captures the effects of

selection on borrower quality, incremental attention, and the negative externalities of default.

Second, among borrowers in violation of a financial covenant, we compare the future bankruptcy

rates of borrowers that are pushed into technical default by lenders with and without short-

termism incentives. This specification highlights the selection and externalities effects, so any

differences in estimates between these specifications should be informative particularly about the

direct effects of lender attention on borrowers. If, for example, we observe a reduction in

bankruptcy propensity among affected borrowers, we would infer that the selection on borrower

quality and incremental lender attention channels dominate the negative externalities of default.

Further, because the first set of estimates incorporate the incremental attention channel and the

second set do not, differences in coefficients yields inference about the direction and magnitude of

the incremental attention effect. If the first set of estimates are similar to the second set, or if the

second set exceeds the first set in magnitude, we would conclude that the incremental attention

of lenders with short-termism incentives does not have net benefits.

Table 7 presents results of the following two regression models:

Bankruptcyit+1→t+5 = a + b1MBLenderjt + b2NegativeSlackit + b3MBLenderjt×NegativeSlackit + eijt

(9)

Bankruptcyit+1→t+5 = a + b1MBLenderjt + b2Violationit + b3MBLenderjt×Violationit + eijt (10)

23

where Bankruptcyt+1→t+5 is an indicator that equals one if borrower i experiences at least one

bankruptcy within five years of quarter t and zero otherwise. We choose to measure future defaults

using bankruptcies both because this picks up borrowers in serious distress and because this

measure is less susceptible to lender discretion than are other measures of default. Panel A presents

estimates of the effects of NegativeSlack and Panel B present estimates of the effects of Violation,

conditional on NegativeSlack.

As before, in each panel of Table 7, column (1) presents the baseline estimates, which

include no fixed effects, column (2) presents estimates with Borrower, Lender, and Quarter fixed

effects, column (3) presents estimates with Lender×Borrower and Industry×Quarter fixed effects,

and column (4) presents estimates with with Lender×Borrower and Industry×Quarter fixed

effects and additional control variables based on lender, borrower, and loan characteristics.

The results in Panel A suggest that, relative to other borrowers in technical default, those

with myopic lenders are more likely to declare bankruptcy in the future. This inference holds

within borrower, within borrower-lender pairs, and controlling for industry trends in

NegativeSlack, default, and lender short-termism. Similarly, the results in Panel B suggest that,

relative to other borrowers with material covenant violations, those with myopic lenders are also

less likely to experience bankruptcy in the future. Again, this inference holds with each layer of

fixed effects, suggesting that the relationship cannot be explained by characteristics that might

determine matching between borrowers and lenders or industry trends in performance.

These results could be explained either by the higher ex ante quality of borrowers pushed

into default by myopic lenders, or the direct effect of incremental monitoring. However, it appears

unlikely that this behavior is beneficial for the lender in the long run. One possibility would have

been that, in the absence of strong capital markets incentives, lenders underinvest in monitoring

their borrowers. If that were the case, we would expect to see a higher incidence of future distress

among selected borrowers, or at least a rate similar to those which would be selected regardless.

Since this is not the case, it appears that lenders are hurting their relationships with potentially

24

profitable future borrowers, rather than pruning their portfolios of poor quality borrowers. We

discuss whether the possibility that this behavior is beneficial to borrowers in Section 4.4.

4.3 Capital expenditures and R&D investment

The change in control rights associated with technical default allows lenders to influence

corporate decisions. In line with their incentive to reduce risk-taking, lenders influence borrowers

to reduce capital investment and R&D (Chava and Roberts 2008; Nini, Smith, and Sufi 2009; Gu,

Mao, and Tian 2016). However, what remains unclear is the role of lenders’ short-termism

incentives. To minimize the cost of enforcing material covenant violations at a higher rate, lenders

should hypothetically prefer to intervene less in their borrowers’ optimal investment policies. For

example, because lenders with short-termism incentives select higher-quality borrowers into

technical default, these borrowers are likely to have better investment opportunities, so

intervening in these investments would be inefficient. However, lenders with short-termism

incentives may continue to exercise their control rights if borrowers’ investment policies are, from

their perspective, excessively risky.

To investigate the incremental propensity of borrowers in violation of a financial covenant

to cut investment, due to the short-termism incentives of their lenders, we estimate the following

two regression models:

CutR&Dit = a + b1MBLenderjt + b2NegativeSlackit + b3MBLenderjt×NegativeSlackit + eijt (11)

CutCAPEXit = a + b1MBLenderjt + b2NegativeSlackit + b3MBLenderjt×NegativeSlackit + eijt

(12)

where CutR&D is an indicator that equals one if borrower i’s seasonally-adjusted R&D expenses

are lower in quarter t+1 than in quarter t and zero otherwise, and CutCAPEX is indicator that

equals one if borrower i’s seasonally-adjusted capital expenditures are lower in quarter t+1 than

in quarter t and zero otherwise. Panel A of Table 8 presents estimates of the lender short-termism

25

effect on CutR&D, and Panel B of Table 8 presents estimates of the lender short-termism effect

on CutCAPEX.

The estimates in column (1) of Table 8 suggest that, in the cross-section of loan-quarters,

those in which borrowers are in violation of a financial covenant are associated with a 7.0

percentage point increase in the propensity of cutting R&D expenses and a 1.0 percentage point

increase in the propensity of cutting capital expenditures. This effect is statistically and

economically larger when lenders have short-termism incentives. For the set of borrowers in

violation of a financial covenant, having a lender with short-termism incentives is associated with

a 2.4 percentage point increase in the propensity of cutting R&D expenses and 4.9 percentage

point increase in the propensity of cutting capital expenditures. These effects suggest that lender

incentives, specifically those that make lenders act myopically, play a large role in their detection

and enforcement of covenant violations.

Again, the other columns in Table 8 incrementally incorporate more restrictive sets of

fixed effects to eliminate alternative explanations based on unobservable borrower characteristics

(Borrower fixed effects), lender characteristics (Lender fixed effects), time trends in covenant

violation propensity and investment opportunities (Quarter and Industry×Quarter fixed effects),

and characteristics of the borrower-lender pair (Lender×Borrower fixed effects), including

characteristics that lead to endogenous matching in the loan market (Schwert 2016).

These results have significant implications for the net benefits of lender short-termism for

borrowers. In particular, myopic lenders enforce technical default more strictly and for a higher

quality pool of borrowers, so a counterfactually large and high quality group of borrowers faces

the direct costs of material covenant violations and the indirect costs of switching lenders.

Especially because these higher quality borrowers are those that would not face default costs if it

were not for the short-termism incentives of their lenders and because higher quality borrowers

are likely to have higher quality projects, these incremental effects on investment and R&D

suggest that lender short-termism has negative net benefits. Thus, although short-termism

26

incentives encourage lenders to increase attention to their borrowers, they reduce economic growth

through enforcement of technical defaults.

4.4 Material covenant violation announcement returns

The incremental attention from lenders with short-termism incentives impacts the

financing and real investment of their borrowers. Although we have argued that these effects are

likely to be value-decreasing in previous sections, they could, in principle, be value-increasing.

Incremental lender attention may improve borrower performance, which could benefit

shareholders as well as debtholders. Even if incremental lender attention curtails investment and

R&D expenditures for affected borrowers, this may be optimal. For example, affected borrowers

may have unobservably poor investment opportunities that are uncorrelated with the observable

characteristics we study in earlier sections. To address the optimality of incremental lender

attention, we focus on the market reaction to announcements of material covenant violations.

If the market reaction to these announcements by borrowers with myopic lenders is more

positive than that of borrowers without myopic lenders, then the incremental attention due to

short-termism incentives is value-increasing for shareholders. However, if the market reactions are

insignificantly different or more negative for borrowers with myopic lenders, then we would infer

that the net benefits of incremental attention from myopic lenders are value-decreasing. To

investigate differences in the market reaction for borrowers with lenders with and without short-

termism incentives, we estimate the following regression model:

CMARit = a + b1MBLenderjt + b2Violationit + b3MBLenderjt×Violationit + eijt (13)

where CMAR is the sum of the cumulative market-adjusted announcement returns over the three

day periods around the announcement of covenant breach (negative slack) and the announcement

of the material covenant violation itself. We include both dates to account for the fact that the

market may anticipate the future violation upon observing the breach of the covenant. Covenant

breaches are publicly observable on the borrower’s quarterly earnings announcement date and

27

material covenant violations are observable in public SEC filings, including 8-K, 10-Q, and 10-K

forms. Table 9 presents estimates of the market reaction to material covenant violation

announcements conditional on lender short-termism.

The estimates in Table 9 suggest that, in the cross-section of loan-quarters, those in which

borrowers have a material covenant violation are associated with a 73 basis point decrease in firm

value. In all specifications, we find that material covenant violations are met with more negative

market reactions for borrowers whose lenders face short-termism incentives. Our most restrictive

and preferred specification suggests that the market reaction to technical default is 88 basis points

more negative for borrowers whose lenders face short-termism incentives. This finding is

statistically significant at the 5% level. Together, these findings suggest that, despite the decline

in future bankruptcy discussed in Section 4.2, lender myopia is, on net, value-decreasing for

borrowers.

5 Conclusion

In this paper, we provide novel evidence that short-term capital market incentives are

transmitted through financing relationships. Specifically, we find that lenders facing pressure to

meet short-term earnings benchmarks are more likely to extract material benefits from borrowers,

which, in turn, leads borrowers to incur not only the direct costs of waiver fees and loan

renegotiations, but also the indirect costs of switching lenders for subsequent loans. Consistent

with this behavior, we find that affected borrowers reduce capital investment and research and

development expenditures. Because lenders with short-term incentives enforce material covenant

violations for higher quality borrowers and at higher rates, and because investors react more

negatively to default announcements for affected borrowers, we infer that these real investment

effects are value-destroying overall.

28

Appendix A: Variable Definitions Variable Name Definition (database codes) Source

Violation One for material violation Michael Roberts’ and Amir

Sufi's websites Negative slack One if min(slack)<0 Compustat & DealScan MBlender Surprise 0 or 1 cent IBES

Market cap Log(Prccq × Csho) Compustat Market to book Log(Market cap / Ceqq) Compustat Lev (Dlcq + Dlttq) / (Atq) Compustat Refinancing Dlcq/(Dlcq + Dlttq) Compustat WW Whited and Wu Index Compustat Roa Ibq / Atq Compustat Loan amount Natural log of package amount DealScan Loan spread Natural log of weighted average spread DealScan Loan maturity Natural log of weighted average maturity DealScan Collateral Indicator equals one for collateral DealScan Number of tranches number of facilities in package DealScan Bank assets log(bhck2170) of the bank Call reports & FR Y-9C filings Bank loans log(bhck2122) of the bank Call reports & FR Y-9C filings Capital ratio bhck3210/bhck2170 Call reports & FR Y-9C filings

Lender Switch Next lead lenders are not in current loan DealScan ΔlnSpread percentage change in loan spreads between

current and next loan DealScan

ΔlnAmount percentage change in loan amount between

current and next loan DealScan

Default One for borrowers that default within the next

five years Compustat

Cut R&D One if borrower's seasonally-adjusted Xrdq

lower in quarter t+1 than in quarter t Compustat

Cut CAPEX One if borrower's seasonally-adjusted Capxy

lower in quarter t+1 than in quarter t Compustat

29

Appendix B: Covenant Calculations Covenant Name Calculation (Compustat codes)

Debt-to-EBITDA (Dlcq + Dlttq) / Rolling EBITDA Debt-to-Equity (Dlcq + Dlttq) / Seqq Debt-to-Tangible NW (Dlcq + Dlttq) / (Atq – Intanq – Ltq) Leverage (Dlcq + Dlttq) / Atq Current ratio Actq/Lctq Quick ratio (Rectq + Cheq) / Lctq Cash interest coverage Rolling EBITDA/Rolling interest paid Interest coverage Rolling EBITDA/Rolling interest expense Debt service coverage Rolling EBITDA/(Rolling interest expense and principal payment)

Fixed charge coverage Rolling EBITDA/(Rolling interest expense, principal payment, and rent payment)

Net worth Atq – Ltq Tangible net worth Atq – Intanq – Ltq EBITDA Rolling EBITDA

Rolling EBITDA, interest expense, interest paid, principal paid are the sum of the firm’s past four quarters.

30

References

Acemoglu, D., Carvalho, V. M., Ozdaglar, A., & Tahbaz-Salehi, A. (2012). The network origins of aggregate fluctuations. Econometrica, 80(5), 1977-2016.

Asker, J., Farre-Mensa, J., & Ljungqvist, A. (2015). Corporate Investment and Stock Market Listing: A Puzzle? Review of Financial Studies, 28(2), 342-390.

Beatty, A. L., Ke, B., & Petroni, K. R. (2002). Earnings management to avoid earnings declines across publicly and privately held banks. The Accounting Review, 77(3), 547-570.

Belo, F., Lin, X., & Yang, F. (2014). External equity financing shocks, financial flows, and asset prices. Working Paper.

Beneish, M. D., & Press, E. (1993). Costs of technical violation of accounting-based debt covenants. The Accounting Review, 233-257.

Beneish, M. D., & Press, E. (1995). The resolution of technical default. The Accounting Review, 337-353.

Berger, A. N., & Udell, G. F. (1995). Relationship lending and lines of credit in small firm finance. Journal of Business, 351-381.

Berger, A. N., Saunders, A., Scalise, J. M., & Udell, G. F. (1998). The effects of bank mergers and acquisitions on small business lending. Journal of Financial Economics, 50(2), 187-229.

Bharath, S., Dahiya, S., Saunders, A., & Srinivasan, A. (2007). So what do I get? The bank's view of lending relationships. Journal of Financial Economics, 85(2), 368-419.

Bhojraj, S., Hribar, P., Picconi, M., & McInnis, J. (2009). Making sense of cents: An examination of firms that marginally miss or beat analyst forecasts. The Journal of Finance, 64(5), 2361-2388.

Bird, A., Karolyi, S. A., & Ruchti, T. G. (2016). Understanding the "Numbers Game". Working Paper.

Bord, V., Ivashina, V., & Taliaferro, R. (2015). Large banks and the transmission of financial shocks. Working Paper.

31

Boyson, N. M., Stahel, C. W., & Stulz, R. M. (2010). Hedge fund contagion and liquidity shocks. The Journal of Finance, 65(5), 1789-1816.

Breza, E., & Liberman, A. (2016). Financial contracting and organizational form: Evidence from the regulation of trade credit. Journal of Finance (forthcoming).

Burgstahler, D., & Dichev, I. (1997). Earnings management to avoid earnings decreases and losses. Journal of Accounting and Economics, 24(1), 99-126.

Carvalho, D., Ferreira, M. A., & Matos, P. (2015). Lending relationships and the effect of bank distress: evidence from the 2007--2009 financial crisis. Journal of Financial and Quantitative Analysis, 50(06), 1165-1197.

Chakraborty, I., Chava, S., & Ganduri, R. (2016). Credit default swaps and moral hazard in bank lending. Working Paper.

Chava, S., & Purnanandam, A. (2011). The effect of banking crisis on bank-dependent borrowers. Journal of Financial Economics, 99(1), 116-135.

Chava, S., & Roberts, M. R. (2008). How does financing impact investment? The role of debt covenants. The Journal of Finance, 63(5), 2085-2121.

Cheong, F. S., & Thomas, J. (2011). Why do EPS forecast error and dispersion not vary with scale? Implications for analyst and managerial behavior. Journal of Accounting Research, 49(2), 359-401.

Chodorow-Reich, G. (2014). The employment effects of credit market disruptions: Firm-level evidence from the 2008--9 financial crisis. The Quarterly Journal of Economics, 129(1), 1-59.

Cooper, I., & Kaplanis, E. (1994). Home bias in equity portfolios, inflation hedging, and international capital market equilibrium. Review of Financial Studies, 7(1), 45-60.

Coval, J. D., & Moskowitz, T. J. (1999). Home bias at home: Local equity preference in domestic portfolios. The Journal of Finance, 54(6), 2045-2073.

DeAngelo, H., DeAngelo, L., & Wruck, K. H. (2002). Asset liquidity, debt covenants, and managerial discretion in financial distress:: the collapse of LA Gear. Journal of Financial Economics, 64(1), 3-34.

32

Dechow, P. M., & Shakespeare, C. (2009). Do managers time securitization transactions to obtain accounting benefits? The Accounting Review, 84(1), 99-132.

Dew-Becker, I., & Giglio, S. (2016). Asset pricing in the frequency domain: theory and empirics. Review of Financial Studies, hhw027.

Dichev, I. D., & Skinner, D. J. (2002). Large--sample evidence on the debt covenant hypothesis. Journal of Accounting Research, 40(4), 1091-1123.

Diether, K. B., Malloy, C. J., & Scherbina, A. (2002). Differences of opinion and the cross section of stock returns. The Journal of Finance, 57(5), 2113-2141.

Edmans, A., Fang, V. W., & Lewellen, K. A. (2016). Equity vesting and managerial myopia. Working Paper.

Ertan, A. (2016). Real earnings management in the financial industry. Working Paper.

Falato, A., & Liang, N. (2016). Do Creditor Rights Increase Employment Risk? Evidence from Debt Covenants. Journal of Finance (forthcoming).

Freudenberg, F., Imbierowicz, B., Saunders, A., & Steffen, S. (2015). Covenant violations and dynamic loan contracting. Working Paper.

Gabaix, X. (2011). The granular origins of aggregate fluctuations. Econometrica, 79(3), 733-772.

Gan, J. (2007). The real effects of asset market bubbles: Loan-and firm-level evidence of a lending channel. Review of Financial Studies, 20(6), 1941-1973.

Giroud, X., & Mueller, H. M. (2015). Capital and labor reallocation within firms. The Journal of Finance, 70(4), 1767-1804.

Giroud, X., & Mueller, H. M. (2016). Firm Leverage, Consumer Demand, and Unemployment during the Great Recession. Quarterly Journal of Economics (forthcoming).

Gopalakrishnan, V., & Parkash, M. (1995). Borrower and lender perceptions of accounting information in corporate lending agreements. Accounting Horizons, 9(1), 13.

Gorton, G., & Metrick, A. (2012). Securitized banking and the run on repo. Journal of Financial Economics, 104(3), 425-451.

Gu, Y., Mao, C. X., & Tian, X. (2014). Bank interventions and firm innovation: Evidence from debt covenant violations. Working Paper.

33

Healy, P. M. (1985). The effect of bonus schemes on accounting decisions. Journal of Accounting and Economics, 7(1), 85-107.

Hirtle, B., Kovner, A., & Plosser, M. C. (2016). The impact of supervision on bank performance. FRB of NY Staff Report.

Hubbard, R. G., Kuttner, K. N., & Palia, D. N. (2002). Are there bank effects in borrowers' costs of funds? Evidence from a matched sample of borrowers and banks. The Journal of Business, 75(4), 559-581.

Ivashina, V., & Scharfstein, D. (2010). Bank lending during the financial crisis of 2008. Journal of Financial Economics, 97(3), 319-338.

Ivashina, V., & Scharfstein, D. (2010). Loan Syndication and Credit Cycles. American Economic Review, 100(2), 57-61.

Kang, J.-K., & Stulz, R. M. (2000). Do Banking Shocks Affect Borrowing Firm Performance? An Analysis of the Japanese Experience*. Journal of Business, 73(1), 1-23.

Kelly, B., Lustig, H., & Van Nieuwerburgh, S. (2013). Firm volatility in granular networks. Working Paper.

Ladika, T., & Sautner, Z. (2016). Managerial short-termism and investment: Evidence from accelerated option vesting. Working Paper.

Laux, V. (2012). Stock option vesting conditions, CEO turnover, and myopic investment. Journal of Financial Economics, 106(3), 513-526.

Lee, S. W., & Mullineaux, D. J. (2004). Monitoring, Financial Distress, and the Structure of Commercial Lending Syndicates. Financial Management, 33(3), 107-130.

Li, N., Vasvari, F. P., & Wittenberg-Moerman, R. (2016). Dynamic threshold values in earnings-based covenants. Journal of Accounting and Economics, 61(2--3), 605-629.

Loutskina, E., & Strahan, P. E. (2009). Securitization and the declining impact of bank finance on loan supply: Evidence from mortgage originations. The Journal of Finance, 64(2), 861-889.

Murfin, J. (2012). The supply-side determinants of loan contract strictness. The Journal of Finance, 67(5), 1565-1601.

34

Murfin, J., & Njoroge, K. (2015). The implicit costs of trade credit borrowing by large firms. Review of Financial Studies, 28(1), 112-145.

Ongena, S., Smith, D. C., & Michalsen, D. (2003). Firms and their distressed banks: lessons from the Norwegian banking crisis. Journal of Financial Economics, 67(1), 81-112.

Pedersen, L. H., Mitchell, M., & Pulvino, T. (2007). Slow Moving Capital. American Economic Review, 97(2), 215-220.

Peek, J., & Rosengren, E. S. (1997). The international transmission of financial shocks: The case of Japan. American Economic Review, 87(4).

Roberts, M. R. (2015). The role of dynamic renegotiation and asymmetric information in financial contracting. Journal of Financial Economics, 116(1), 61-81.

Roberts, M. R., & Sufi, A. (2009). Control rights and capital structure: An empirical investigation. Journal of Finance, 64(4), 1657-1695.

Roberts, M. R., & Sufi, A. (2009). Renegotiation of financial contracts: Evidence from private credit agreements. Journal of Financial Economics, 93(2), 159-184.

Roychowdhury, S. (2006). Earnings management through real activities manipulation. Journal of Accounting and Economics, 42(3), 335-370.

Salitskiy, I. (2015). Compensation Duration and Risk Taking. Working Paper.

Scholes, M. S., Wilson, G. P., & Wolfson, M. A. (1990). Tax planning, regulatory capital planning, and financial reporting strategy for commercial banks. Review of Financial Studies, 3(4), 625-650.

Schwert, M. (2016). Bank Capital and Lending Relationships. Working Paper.

Shek, J., Shim, I., & Shin, H. S. (2015). Investor redemptions and fund manager sales of emerging market bonds: how are they related? Working Paper.

Shleifer, A., & Vishny, R. W. (1997). The limits of arbitrage. Journal of Finance, 52(1), 35-55.

Silva, R. C. (2013). Internal Labor Markets, Wage Convergence and Investment. Working Paper.

Stein, J. C. (1988). Takeover threats and managerial myopia. The Journal of Political Economy, 61-80.

35

Stein, J. C. (1989). Efficient capital markets, inefficient firms: A model of myopic corporate behavior. The Quarterly Journal of Economics, 655-669.

Sufi, A. (2007). Information asymmetry and financing arrangements: Evidence from syndicated loans. Journal of Finance, 62(2), 629-668.

Sufi, A. (2009). Bank lines of credit in corporate finance: An empirical analysis. Review of Financial Studies, 22(3), 1057-1088.

Sufi, A. (2009). The real effects of debt certification: Evidence from the introduction of bank loan ratings. Review of Financial Studies, 22(4), 1659-1691.

Van Nieuwerburgh, S., & Veldkamp, L. (2009). Information immobility and the home bias puzzle. The Journal of Finance, 64(3), 1187-1215.

Wu, D. (2016). Shock Spillover and Financial Response in Supply Chain Networks: Evidence from Firm-Level Data. Working Paper.

36

Panel A. Lender EPS Surprise Distribution

Panel B. McCrary (2008) test of EPS manipulation

Figure 1. Lender short-termism Panel A of this figure presents a histogram of the earnings surprise distribution in our sample of lenders-quarter observations. Zero earnings surprise indicates that the lender’s realized earnings-per-share (EPS) is equal to the consensus analyst EPS forecast in that quarter. Panel B presents a wider view of this distribution, but includes lines that represent splines on each side of the zero EPS cutoff. The McCrary (2008) test rejects the null hypothesis of no discontinuity at the zero EPS cutoff at the 5% significance level.

37

Figure 2. Material covenant violation frequency This figure presents the probability of material covenant violation conditional on (i) the distribution of covenant slack (i.e., standardized distance to covenant threshold), and (ii) lender short-termism incentives, which are measured using the distribution of the lender’s quarterly earnings-per-share (EPS) surprise. At zero covenant slack and below, lenders gain the contractual right to enforce a covenant violation. Large Miss indicates that the lender’s realized earnings were less than the consensus analyst EPS forecast by more than five cents. Miss indicates that the lender’s realized earnings were less than the consensus analyst EPS forecast by between five and two cents. Just Miss indicates that the lender’s realized earnings were exceeded by the consensus analyst EPS forecast by less than or equal to two cents and greater than zero cents. Just Beat indicates that the lender’s realized earnings exceeded the consensus analyst EPS forecast by at least zero and less than two cents. Beat indicates that the lender’s realized earnings exceeded the consensus analyst EPS forecast by at least two cents and less than five cents. Large Beat indicates that the lender’s realized earnings exceeded the consensus analyst EPS forecast by at least five cents.

38

Figure 3. Material covenant violation selection This figure presents the mean and 95% confidence interval of the distribution of covenant slack (i.e., standardized distance to covenant threshold) among borrowers with a material covenant violation conditional on lender short-termism incentives, which are measured using the distribution of the lender’s quarterly earnings-per-share (EPS) surprise. Large Miss indicates that the lender’s realized earnings were less than the consensus analyst EPS forecast by more than five cents. Miss indicates that the lender’s realized earnings were less than the consensus analyst EPS forecast by between five and two cents. Just Miss indicates that the lender’s realized earnings were exceeded by the consensus analyst EPS forecast by less than or equal to two cents and greater than zero cents. Just Beat indicates that the lender’s realized earnings exceeded the consensus analyst EPS forecast by at least zero and less than two cents. Beat indicates that the lender’s realized earnings exceeded the consensus analyst EPS forecast by at least two cents and less than five cents. Large Beat indicates that the lender’s realized earnings exceeded the consensus analyst EPS forecast by at least five cents.

39

Table 1. Summary Statistics

This table presents summary statistics for the regression variables of interest. All variables are from DealScan, Compustat, and the Michael Roberts’ covenant violations dataset. Panel A presents summary statistics of the main variables of interest. Panel B presents univariate differences in these variables across subsamples with and without MBLender, which is an indicator that identifies observations with a lender that just-meet-or-beat its earnings benchmark.

Panel A. Summary statistics

Mean SD P25 Median P75 Violation 5.83%

NegativeSlack 26.11%

Slack 1.611 8.238 -0.08 0.55 1.88 LenderSwitch 33.33%

MBLender 0.24

Spread (bps) 159.03 108.64 75 140 225 Amount ($mm) 467 813 70 200 500 PriorRelationship 73.79%

CutCAPEX 41.14%

CutR&D 7.59%

Defaultt+1→t+5 7.31%

ΔlnSpreadit 0.021 0.289 0 0 0.035 ΔlnAmountit -0.009 0.459 -0.083 0 0 Whited-Wu -0.32 0.09 -0.40 -0.33 -0.26 M/B 2.52 1.81 1.26 2.03 3.22

40

Panel B. Univariate differences on short-termism incentives

MBLender=1 MBLender=0 Mean Mean Difference p-value Violation 8.45% 4.97% 3.48% <0.01 NegativeSlack 28.88% 25.18% 3.70% 0.114 Slack 1.45 1.67 -0.22 0.096 LenderSwitch 38.69% 31.59% 7.10% <0.01 Spread (bps) 174.32 154.05 20.27 <0.01 Amount ($mm) 286 527 241 <0.01 PriorRelationship 74.52% 73.55% 0.97 0.032 CutCAPEX 43.87% 41.14% 2.73% 0.218 CutR&D 11.42% 6.31% 5.11% 0.062 Defaultt+1→t+5 10.05% 6.42% 3.65% <0.01 ΔlnSpread(%)it 0.014 0.024 0.009 0.087 ΔlnAmount(%)it -0.013 -0.008 -0.004 0.473 Whited-Wu -0.296 -0.339 0.044 0.049 M/B 2.43 2.56 -0.13 0.004

41

Table 2. Lender Short-termism and Technical Default Frequency This table presents loan package-quarter level linear probability model estimates of Violation on NegativeSlack, MBLender, NegativeSlack × MBLender, and control variables. We semiparametrically control for lender earnings surprise using bin indicators for large miss, just miss, and large beat, and we interact each of these with NegativeSlack to ensure that the coefficient on NegativeSlack × MBLender is identified from differences in lender behavior to borrowers with NegativeSlack when lenders just beat earnings benchmarks and when they do not. Other controls, Xit, include loan characteristics (i.e., amount, spread, maturity, collateral indicator, term loan indicator, tranche B indicator, loan purprose indicators, number of tranches, number of covenants, and refinancing indicator), borrower characteristics (i.e., leverage, Whited-Wu index, return-on-assets, M/B, loss firm indicator), and lender characteristics (i.e., percentage nonperforming loans, loan loss reserves, provision for loan losses, tier 1 capital ratio, total assets, number of loans). We incrementally include borrower, lender, quarter, lender-by-borrower, and industry-by-quarter fixed effects to eventually isolate within-pair variation in lenders’ treatment of borrowers conditional on short-termism behavior. Heteroskedasticity-robust standard errors are clustered by lender, and presented in parentheses. ***, **, and * denote results significant at the 1%, 5%, and 10% levels.

Dependent variable: Violationijt

(1) (2) (3) (4)

NegativeSlackit 0.089*** 0.083*** 0.080*** 0.022** (0.014) (0.014) (0.013) (0.011) MBLenderjt 0.021*** -0.007* -0.009** -0.013*** (0.004) (0.004) (0.004) (0.004) NegativeSlackit × MBLenderjt 0.056*** 0.051*** 0.051*** 0.070*** (0.015) (0.014) (0.013) (0.013)

Lender surprise bins Yes Yes Yes Yes Xit No No No Yes Fixed Effects:

__Borrower No Yes No No __Lender No Yes No No __Quarter No Yes No No __Lender × Borrower No No Yes Yes __Industry × Quarter No No Yes Yes R2 0.0494 0.0792 0.1374 0.2016

Obs. 90,125 90,042 89,926 54,331

42

Table 3. Lender Short-termism and Technical Default Selection on Default Severity This table presents loan package-quarter level linear probability model estimates of Slack, winsorized at the first and ninety-ninth percentiles, on Violation, MBLender, Violation × MBLender, and control variables in the sample of loan package-quarters with NegativeSlack. We semiparametrically control for lender earnings surprise using bin indicators for large miss, just miss, and large beat, and we interact each of these with Violation to ensure that the coefficient on Violation × MBLender is identified from differences in lender behavior to borrowers with Violation when lenders just beat earnings benchmarks and when they do not. Other controls, Xit, include loan characteristics (i.e., amount, spread, maturity, collateral indicator, term loan indicator, tranche B indicator, loan purprose indicators, number of tranches, number of covenants, and refinancing indicator), borrower characteristics (i.e., leverage, Whited-Wu index, return-on-assets, M/B, loss firm indicator), and lender characteristics (i.e., percentage nonperforming loans, loan loss reserves, provision for loan losses, tier 1 capital ratio, total assets, number of loans). We incrementally include borrower, lender, quarter, lender-by-borrower, and industry-by-quarter fixed effects to eventually isolate within-pair variation in lenders’ treatment of borrowers conditional on short-termism behavior. Heteroskedasticity-robust standard errors are clustered by lender, and presented in parentheses. ***, **, and * denote results significant at the 1%, 5%, and 10% levels.

Dependent variable: Slackit

(1) (2) (3) (4)

Violationit -2.602*** -2.572*** -2.248*** -2.097*** (0.535) (0.553) (0.407) (0.464) MBLenderjt -0.106 -0.153 -0.087 -0.083 (0.141) (0.118) (0.123) (0.189) Violationit × MBLenderjt 1.059** 1.180** 0.966** 1.001** (0.536) (0.531) (0.430) (0.466)

Lender surprise bins Yes Yes Yes Yes Xit No No No Yes Fixed Effects:

__Borrower No Yes No No __Lender No Yes No No __Quarter No Yes No No __Lender × Borrower No No Yes Yes __Industry × Quarter No No Yes Yes R2 0.0134 0.0864 0.2480 0.2836 Obs. 23,515 23,486 23,091 13,653

43

Table 4. Lender Short-termism and Default Selection on Borrower and Loan Types

This table presents loan package-quarter level linear probability model estimates of borrower (Panel A) and loan (Panel B) characteristics on Violation, MBLender, Violation × MBLender, and control variables in the sample of loan package-quarters with NegativeSlack. Dependent variables include lnSize, the natural log of total assets, the Whited-Wu index, PriorRelationship, an indicator for borrower-lender pairs with previous transactions, and M/B, measured as market capitalization over book equity, SyndicationFees, Loan/Assets, and lnSpread. We semiparametrically control for lender earnings surprise using bin indicators for large miss, just miss, and large beat, and we interact each of these with Violation. Other controls, Xit, include loan characteristics (i.e., amount, spread, maturity, collateral indicator, term loan indicator, tranche B indicator, loan purprose indicators, number of tranches, number of covenants, and refinancing indicator), borrower characteristics (i.e., leverage, Whited-Wu index, return-on-assets, M/B, loss firm indicator), and lender characteristics (i.e., percentage nonperforming loans, loan loss reserves, provision for loan losses, tier 1 capital ratio, total assets, number of loans). We incrementally include borrower, lender, quarter, lender-by-borrower, and industry-by-quarter fixed effects to eventually isolate within-pair variation in lenders’ treatment of borrowers conditional on short-termism behavior. Heteroskedasticity-robust standard errors are clustered by lender, and presented in parentheses. ***, **, and * denote results significant at the 1%, 5%, and 10% levels.

Panel A. Technical Default Selection on Borrower Types

Dependent variable: lnSizeijt Whited-Wuit Prior

Relationshipit M/Bit

(1) (2) (3) (4)

Violationit -1.192*** 0.046*** -0.010** -0.248*** (0.119) (0.012) (0.005) (0.052) MBLenderjt -0.106** 0.005 -0.002 -0.029 (0.049) (0.006) (0.002) (0.030)

Violationit × MBLenderjt 0.309*** -0.016** 0.012** 0.057** (0.110) (0.007) (0.006) (0.024)

Lender surprise bins Yes Yes Yes Yes

Xit Yes Yes Yes Yes Fixed Effects:

__Lender Yes Yes Yes Yes

__Industry × Quarter Yes Yes Yes Yes

R2 0.5242 0.5572 0.5791 0.4709 Obs. 22,928 21,078 23,093 22,910

44

Panel B. Technical Default Selection on Loan Types

Dependent variable: lnSyndicationFeesit Loan/Assetsijt lnSpreadijt

(1) (2) (3)

Violationit -0.042* -0.267* 0.187*** (0.022) (0.163) (0.040) MBLenderjt 0.006 0.080 -0.008 (0.026) (0.171) (0.020)

Violationit × MBLenderjt 0.039* 0.055** -0.032*

(0.020) (0.026) (0.018)

Lender surprise bins Yes Yes Yes

Xit Yes Yes Yes Fixed Effects:

__Lender Yes Yes Yes

__Industry × Quarter Yes Yes Yes

R2 0.7138 0.7809 0.3421

Obs. 23,072 23,093 22,355

45

Table 5. Lender Short-termism and Lender Switching This table presents loan package-quarter level linear probability model estimates of LenderSwitch, an indicator that identifies borrowers that switch lenders on a loan issued in the next five years, on NegativeSlack, MBLender, NegativeSlack × MBLender, and control variables. We semiparametrically control for lender earnings surprise using bin indicators for large miss, just miss, and large beat, and we interact each of these with NegativeSlack to ensure that the coefficient on NegativeSlack × MBLender is identified from differences in lender behavior to borrowers with NegativeSlack when lenders just beat earnings benchmarks and when they do not. Other controls, Xit, include loan characteristics (i.e., amount, spread, maturity, collateral indicator, term loan indicator, tranche B indicator, loan purprose indicators, number of tranches, number of covenants, and refinancing indicator), borrower characteristics (i.e., leverage, Whited-Wu index, return-on-assets, M/B, loss firm indicator), and lender characteristics (i.e., percentage nonperforming loans, loan loss reserves, provision for loan losses, tier 1 capital ratio, total assets, number of loans). We incrementally include borrower, lender, quarter, lender-by-borrower, and industry-by-quarter fixed effects to eventually isolate within-pair variation in lenders’ treatment of borrowers conditional on short-termism behavior. Heteroskedasticity-robust standard errors are clustered by lender, and presented in parentheses. ***, **, and * denote results significant at the 1%, 5%, and 10% levels.

Dependent variable: LenderSwitch

(1) (2) (3) (4) NegativeSlackit 0.040*** 0.010* -0.011** -0.014*** (0.008) (0.006) (0.004) (0.005) MBLenderjt 0.055*** -0.000 -0.007** -0.009*** (0.005) (0.004) (0.003) (0.003) NegativeSlackit × MBLenderjt 0.014* 0.018** 0.012** 0.012** (0.008) (0.007) (0.005) (0.006)

Lender surprise bins Yes Yes Yes Yes Xit No No No Yes Fixed Effects:

__Borrower No Yes No No __Lender No Yes No No __Quarter No Yes No No __Lender × Borrower No No Yes Yes __Industry × Quarter No No Yes Yes R2 0.0082 0.5426 0.8304 0.8285

Obs. 90,125 90,062 89,576 67,571

46

Table 6. Lender Short-termism and Renegotiation This table presents loan package-quarter level linear probability model estimates of ΔlnSpread(%) (Panel A) and ΔlnSpread(%) (Panel B), which measure the percentage change in loan spreads and amounts, respectively, on renegotiated loans, on NegativeSlack, MBLender, NegativeSlack × MBLender, and control variables. We semiparametrically control for lender earnings surprise using bin indicators for large miss, just miss, and large beat, and we interact each of these with NegativeSlack to ensure that the coefficient on NegativeSlack × MBLender is identified from differences in lender behavior to borrowers with NegativeSlack when lenders just beat earnings benchmarks and when they do not. Other controls, Xit, include . We incrementally include borrower, lender, quarter, lender-by-borrower, and industry-by-quarter fixed effects to eventually isolate within-pair variation in lenders’ treatment of borrowers conditional on short-termism behavior. Heteroskedasticity-robust standard errors are clustered by lender, and presented in parentheses. ***, **, and * denote results significant at the 1%, 5%, and 10% levels.

Panel A. Lender Short-termism and Loan Spread Changes

Dependent variable: ΔlnSpread(%)it

(1) (2) (3) (4)

NegativeSlackit 0.046* 0.024** 0.007 0.003 (0.027) (0.011) (0.011) (0.011) MBLenderjt -0.090* 0.001 -0.014 0.002 (0.047) (0.008) (0.010) (0.009) NegativeSlackit × MBLenderjt 0.076*** 0.040*** 0.032** 0.029* (0.028) (0.013) (0.015) (0.016)

Lender surprise bins Yes Yes Yes Yes Xit No No No Yes Fixed Effects:

__Borrower No Yes No No __Lender No Yes No No __Quarter No Yes No No __Lender × Borrower No No Yes Yes __Industry × Quarter No No Yes Yes R2 0.0045 0.3783 0.6848 0.7926

Obs. 81,168 81,113 80,618 65,843

47

Panel B. Lender Short-termism and Loan Amount Changes

Dependent variable: ΔlnAmount(%)it

(1) (2) (3) (4)

NegativeSlackit -0.022*** -0.014** -0.013** -0.015** (0.008) (0.006) (0.005) (0.007) MBLenderjt -0.005 -0.008 -0.003 0.001 (0.007) (0.006) (0.005) (0.005) NegativeSlackit × MBLenderjt 0.023** 0.011** 0.008** 0.009* (0.011) (0.005) (0.004) (0.005)

Lender surprise bins Yes Yes Yes Yes Xit No No No Yes Fixed Effects:

__Borrower No Yes No No __Lender No Yes No No __Quarter No Yes No No __Lender × Borrower No No Yes Yes __Industry × Quarter No No Yes Yes R2 0.0008 0.2999 0.5762 0.6594

Obs. 90,125 90,062 89,556 58,508

48

Table 7. Lender Short-termism and Future Bankruptcy Panel A of this table presents loan package-quarter level linear probability model estimates of Bankruptcyt+1→t+5, an indicator that identifies borrowers that declare bankruptcy within the next five years, on NegativeSlack, MBLender, NegativeSlack × MBLender, and control variables. Panel B of this table presents loan package-quarter level linear probability model estimates of Bankruptcyt+1→t+5 on Violation, MBLender, Violation × MBLender, and control variables in the sample of loan package-quarters with NegativeSlack. We semiparametrically control for lender earnings surprise using bin indicators for large miss, just miss, and large beat, and we interact each of these with NegativeSlack. Other controls, Xit, include loan characteristics (i.e., amount, spread, maturity, collateral indicator, term loan indicator, tranche B indicator, loan purprose indicators, number of tranches, number of covenants, and refinancing indicator), borrower characteristics (i.e., leverage, Whited-Wu index, return-on-assets, M/B, loss firm indicator), and lender characteristics (i.e., percentage nonperforming loans, loan loss reserves, provision for loan losses, tier 1 capital ratio, total assets, number of loans). We incrementally include borrower, lender, quarter, lender-by-borrower, and industry-by-quarter fixed effects to eventually isolate within-pair variation in lenders’ treatment of borrowers conditional on short-termism behavior. Heteroskedasticity-robust standard errors are clustered by lender, and presented in parentheses. ***, **, and * denote results significant at the 1%, 5%, and 10% levels.

Panel A. Lender Short-termism, Negative Slack, and Future Bankruptcy

Dependent variable: Bankruptcyit+1→t+5

(1) (2) (3) (4)

NegativeSlackit 0.081*** 0.077*** 0.073*** 0.053** (0.028) (0.028) (0.026) (0.025) MBLenderjt 0.006 0.005 0.003 0.005 (0.004) (0.004) (0.003) (0.004) NegativeSlackit × MBLenderjt -0.028** -0.025* -0.026** -0.030* (0.014) (0.014) (0.013) (0.016)

Lender surprise bins Yes Yes Yes Yes Xit No No No Yes Fixed Effects:

__Borrower No Yes No No __Lender No Yes No No __Quarter No Yes No No __Lender × Borrower No No Yes Yes __Industry × Quarter No No Yes Yes R2 0.0201 0.0576 0.1708 0.2122

Obs. 90,848 90,820 90,705 59,865

49

Panel B. Lender Short-termism, Technical Default, and Future Bankruptcy

Dependent variable: Bankruptcyit+1→t+5

(1) (2) (3) (4)

Violationit 0.117*** 0.109*** 0.079** 0.088** (0.034) (0.032) (0.031) (0.042) MBLenderjt -0.014 0.005 0.003 0.003 (0.014) (0.007) (0.009) (0.011) Violationit × MBLenderjt -0.087*** -0.082*** -0.067** -0.055* (0.028) (0.030) (0.029) (0.031)

Lender surprise bins Yes Yes Yes Yes Xit No No No Yes Fixed Effects:

__Borrower No Yes No No __Lender No Yes No No __Quarter No Yes No No __Lender × Borrower No No Yes Yes __Industry × Quarter No No Yes Yes R2 0.0053 0.0844 0.2840 0.3759

Obs. 23,920 23,891 23,500 15,825

50

Table 8. Lender Short-termism and Borrower Investment This table presents loan package-quarter level linear probability model estimates of CutR&D (Panel A) and CutCAPEX (Panel B), which are indicators that identify borrowers that cut R&D and investment, respectively, in the year t+1, on NegativeSlack, MBLender, NegativeSlack × MBLender, and control variables. We semiparametrically control for lender earnings surprise using bin indicators for large miss, just miss, and large beat, and we interact each of these with NegativeSlack to ensure that the coefficient on NegativeSlack × MBLender is identified from differences in lender behavior to borrowers with NegativeSlack when lenders just beat earnings benchmarks and when they do not. Other controls, Xit, include loan characteristics (i.e., amount, spread, maturity, collateral indicator, term loan indicator, tranche B indicator, loan purprose indicators, number of tranches, number of covenants, and refinancing indicator), borrower characteristics (i.e., leverage, Whited-Wu index, return-on-assets, M/B, loss firm indicator), and lender characteristics (i.e., percentage nonperforming loans, loan loss reserves, provision for loan losses, tier 1 capital ratio, total assets, number of loans). We incrementally include borrower, lender, quarter, lender-by-borrower, and industry-by-quarter fixed effects to eventually isolate within-pair variation in lenders’ treatment of borrowers conditional on short-termism behavior. Heteroskedasticity-robust standard errors are clustered by lender, and presented in parentheses. ***, **, and * denote results significant at the 1%, 5%, and 10% levels.

Panel A. Lender Short-termism and Cutting Research and Development Expenses

Dependent variable: CutR&Dit+1

(1) (2) (3) (4)

NegativeSlackit 0.070*** 0.006 0.001 0.009* (0.011) (0.006) (0.004) (0.005) MBLenderjt -0.040*** -0.007** -0.004 -0.001 (0.009) (0.003) (0.003) (0.004) NegativeSlackit × MBLenderjt 0.024** 0.017*** 0.012*** 0.008* (0.012) (0.006) (0.004) (0.005)

Lender surprise bins Yes Yes Yes Yes Xit No No No Yes Fixed Effects:

__Borrower No Yes No No __Lender No Yes No No __Quarter No Yes No No __Lender × Borrower No No Yes Yes __Industry × Quarter No No Yes Yes R2 0.0055 0.6942 0.8616 0.8572 E[CutR&D | NegativeSlack] 0.129

Obs. 90,125 90,621 89,576 60,096

51

Panel B. Lender Short-termism and Cutting Capital Expenditures

Dependent variable: CutCAPEXit+1

(1) (2) (3) (4)

NegativeSlackit 0.010* 0.012* 0.011* 0.007 (0.006) (0.007) (0.007) (0.006) MBLenderjt 0.021 -0.012 -0.014 -0.014 (0.014) (0.010) (0.009) (0.008) NegativeSlackit × MBLenderjt 0.049*** 0.017** 0.015** 0.014* (0.013) (0.008) (0.007) (0.008)

Lender surprise bins Yes Yes Yes Yes Xit No No No Yes Fixed Effects:

__Borrower No Yes No No __Lender No Yes No No __Quarter No Yes No No __Lender × Borrower No No Yes Yes __Industry × Quarter No No Yes Yes R2 0.0055 0.4415 0.7201 0.7199 E[CutCAPEX | NegativeSlack] 0.441

Obs. 90,125 90,062 89,576 62,714

52

Table 9. Lender Short-termism and the Market Reaction to Covenant Violations This table presents loan package-quarter level regression estimates of CMAR, the sum of the three-day cumulative market-adjusted announcement returns for the days on which negative slack and material covenant violation were announced, on Violation, MBLender, Violation × MBLender, and control variables. We semiparametrically control for lender earnings surprise using bin indicators for large miss, just miss, and large beat, and we interact each of these with Violation to ensure that the coefficient on Violation × MBLender is identified from differences in lender behavior to borrowers with Violation when lenders just beat earnings benchmarks and when they do not. Other controls, Xit, include loan characteristics (i.e., amount, spread, maturity, collateral indicator, term loan indicator, tranche B indicator, loan purprose indicators, number of tranches, number of covenants, and refinancing indicator), borrower characteristics (i.e., leverage, Whited-Wu index, return-on-assets, M/B, loss firm indicator), and lender characteristics (i.e., percentage nonperforming loans, loan loss reserves, provision for loan losses, tier 1 capital ratio, total assets, number of loans). We incrementally include borrower, lender, quarter, lender-by-borrower, and industry-by-quarter fixed effects to eventually isolate within-pair variation in lenders’ treatment of borrowers conditional on short-termism behavior. Heteroskedasticity-robust standard errors are clustered by lender, and presented in parentheses. ***, **, and * denote results significant at the 1%, 5%, and 10% levels.

Dependent variable: CMARit

(1) (2) (3) (4)

Violationit × MBLenderjt -0.724* -0.758* -0.920** -0.883** (0.391) (0.387) (0.394) (0.355)

Lender surprise bins Yes Yes Yes Yes Xit No No No Yes Fixed Effects:

__Borrower No Yes No No __Lender No Yes No No __Quarter No Yes No No __Lender × Borrower No No Yes Yes __Industry × Quarter No No Yes Yes R2 0.0518 0.3412 0.4323 0.4402 E[CMAR | Violation] -3.54%

Obs. 22,716 22,690 22,313 12,395