Leveraged Loan Reading

Leveraged Loans: Is High Leverage Risk Priced in?
0F
1
David P.Newton
School of Management, University of Bath
[email protected]
Steven Ongena
University of Zurich, Swiss Finance Institute, KU Leuven and CEPR
[email protected]
Ru Xie
[email protected]
Binru Zhao
[email protected]
First draft: 21 December 2020
This draft: 21 November 2021
1
We thank Simone Giansante, Jarrad Harford, Christoph Herpfer, Winifred Huang, Clive Lennox, William L.
Megginson, Partha S. Mohanram, Philip Molyneux, Felix Noth, Neslihan Ozkan, Alessio Reghezza, Elizaveta
Sizova, Junyang Yin, participants at the EFMA 2021, IYFSC 2021 and IFABS Oxoford 2021 for helpful
comments. We thank Froloshka Kornelia for excellent data assistance in constructing our data sample. Errors and
omissions remain the responsibility of the authors. Steven Ongena acknowledges financial support from ERC
ADG 2016 - GA 740272 lending.
Abstract
We investigate the impact of the 2014 Interagency Clarification on leverage risk premium for
bank and nonbank originated loans. Using a novel dataset from 2007 to 2019, we show that
leveraged loan spreads have declined rapidly for nonbank-facilities relative to bank-facilities
since the introduction of the 2014 Interagency Clarification. The decline in leveraged loan
spreads is significant for highly leveraged borrowers, especially when involving term loans.
We further demonstrate that a higher degree of information asymmetry, driven by an increase
in covenant-lite loan issuance and weaker investor protection, are strongly associated with the
narrowed leverage risk premium. In addition, adverse selection and moral hazard associated
with the high level of Collateralized Loan Obligations issuance significantly contributes the
decline of leverage risk premium for nonbank leveraged loans. Our results suggest the necessity
to enhance the current regulation of leveraged lending, especially in the shadow banking sector.
JEL classification: G21, D82, G34
Key words: Leverage Risk, Syndicated Loan Pricing, Leveraged Loan, Risk Shifting
1. Introduction
The market for leveraged loans- a type of syndicated loan that is extended to borrowers
that have considerable amounts of debt or high credit risk- has grown significantly in recent
years 2. Prompted by an increase in participation of unregulated investors and a deterioration in
credit standards, in March of 2013 , the Office of the Comptroller of the Currency (OCC), the
Board of Governors of the Federal Reserve (Fed), and the Federal Deposit Insurance
Corporation (FDIC) issued the so-called Interagency Guidance on Leveraged Lending in
March 2013. Subsequently, in November 2014, the “frequently asked questions for
implementing the March 2013 guidance” (“the Clarification”) was issued to clarify regulators’
expectations regarding stronger risk management. Although the guidance and its clarification
aim to ensure safe and sound leveraged lending and achieve both macropreudential and
mircoprudential objectives, the guidance only applies to banks that are regulated by either the
OCC, the Fed or the FDIC. Nonbank lenders have been the main beneficiaries of the regulation,
with increased market shares in the leveraged finance market. Kim et al., (2018) find that
although the guidance together with the Clarification were effective in reducing regulated
bank-originated leveraged lending activity, it also triggered a migration of leveraged lending
to unregulated nonbank lenders. Nonbanks (or shadow banks) do (usually) not have access to
central bank liquidity and are denied access to deposit guarantee scheme, which made them
more vulnerable to shocks and can accelerate a systemic crisis (Plantin 2014; Fahri and Tirole
2017; Martinez-Miera and Repullo 2018; Chretien and Lyonnet 2018). Recently, at the Open
Session of the meeting of the Financial Stability Oversight Council, the US Secretary of the
Treasury Janet L. Yellen raises concern on the vulnerabilities in nonbank financial
intermediation as one of the major challenges to financial stability 3. Given these concerns, the
2
In this article, we follow LPC and define leveraged loan as a loan that is extended to borrowers rated BB+ or
lower or it is not rated or rated ‘BBB-‘ or higher but has (1) a spread of LIBOR +125 or higher and (2) is
secured by a first or second lien.
3
March 31, 2021, https://home.treasury.gov/news/press-releases/jy0092
increase in participation of unregulated nonbanks raises a question of whether the absence of
regulation on nonbank lenders in the leveraged loan market leads to laxer lending standards for
nonbank-originated leveraged loans and eventually leads to the potential instability of financial
system.
Another striking development in the leveraged loan market is the surge of so-called
covenant-lite loans issuance. In the U.S. the fraction of covenant-lite leveraged loans increased
rapidly from 30% in 2012 to 55% in 2014. According to Abuzov et al., (2020), the strong
competition between regulated banks and non-regulated banks in the leveraged loan market
has contributed to an increase in covenant-lite structures with reduced investor protection. After
the Clarification, unregulated nonbanks issued significantly more covenant-lite loans than
regulated banks to attract borrowers switching from regulated bank lenders. Apparently, the
different regulation attitudes towards banks and nonbanks lead to the laxer lending standard on
the non-price terms and weaker investor protection. The next question that arises is whether
the lax lending standard due to the absence of regulation is also reflected in loan spreads, i.e.,
whether borrowers’ leverage risk is really reflected in loan spreads. In addition, in the U.S. as
the mid of 2019, more than 50% of leveraged loans have been are securitized and distributed
in the form of Collateralized Loan Obligations (CLOs). The increasing number of unregulated
nonbank lenders in the market after the Clarification and the demand from hunting for high
yield from investors in a low-interest rate environment both contribute to the boom period of
CLOs issuance from 2014-2019. In this paper, we investigate the implications of the guidance
and the Clarification on leveraged loan spreads for bank vs. nonbank originated loans. We
compare the all-in-spread-drawn (AISD) between bank-originated leveraged loan, and
nonbank-originated leveraged loans and examine whether leverage risk has been effectively
priced in the AISD post the regulation.
Our primary empirical analysis compares the all-in-spread-drawn (AISD) between

bank-originated leveraged loans and nonbank-originated leveraged loans during 2007 – 2019 4
and finds that AISD between nonbanks and banks have been narrowed since the 2014
Clarification. From 2007 – 2013 the premium between bank-originated and nonbank-
originated loans follows a clear parallel trend with an average spread of 71 basis points.
However, the premium for nonbank-originated loans distinctly narrowed post the regulation
from 2014 – 2019, with an average spread of 49 basis points (see Figure 1). The decline in
leveraged loan spreads between banks and nonbanks raises concerns that the nonbank lenders
relax their lending policies to compete with banks and increase their leveraged loan market
shares. While leverage risk premiums have declined since 2014, overall borrower risk may
have declined along with spreads. To address this concern, we estimate the relation between
AISD and firm leverage risk after controlling for loan and borrower characteristics. We add an
interaction term between borrower leverage, a dummy variable for nonbank lenders, and a
dummy variable for the post-regulation period of November 2014 to December 2019. The
estimated coefficients on the interaction term are negative and highly significant, confirming
that the nonbank-originated leveraged loan premium for a given level of leverage has declined
since the issuance of the 2014 FAQ.
As a type of syndicated loan, leveraged loan suffers information asymmetry issues by
design, which can lead to both moral hazard and adverse selection. Prior studies suggest that
asymmetric information between lead and participant banks affects loan spreads because
participants want to mitigate the friction by requiring higher premiums (Bosch, 2007; Sufi,
2007; Vashnie, 2009). To mitigate information asymmetry, financial covenants play a key role
in monitoring borrower performance and provide lenders the right to renegotiate their loan
contracts, which can significantly reduce moral hazard/adverse selection (Rajan and Winton,
1995; Bradley and Roberts, 2015; Griffin et al., 2019). However, the strong competition
4
Our sample stops before the outbreak of the pandemic.
between banks and nonbanks on non-price terms after the regulation appears to have
accelerated the issuance of covenant-lite leveraged loans. Since the introduction of the
Clarification in 2014, covenant-lite loan issuance for unregulated nonbank lenders has started
picking up speed. The fraction of leveraged loans issued by nonbank lenders that are covenant-
lite rose from 54% in 2012 to 70% in 2019 (see Figure 4). When regulated banking institutions
slowed the issuance of covenant-lite loans after the 2014 Clarification on the U.S. leveraged
loan market (Abuzov et al., 2020), borrowers switched to unregulated nonbanks loans with
relatively fewer covenants (Schenck and Shi, 2017; Abuzov et al., 2020).
Relaxing investor protection in covenant-lite loans led by competition between banks
and nonbanks can intensify information asymmetry issues associated with leveraged loan
pricing. To investigate the role of information asymmetry on leveraged loan pricing, we
conduct a sub-sample analysis by splitting the sample into multiple groups. In our first sub-
sample analysis, we split the sample of leveraged loans into two groups: loans with covenant
provisions and covenant-lite loans. The results show that more severe information asymmetry
associated with covenant-lite loans leads to a stronger and more significant decline in nonbank-
originated leverage risk premium. Our further sub-group analysis with groups of with and
without performance pricing confirmed that leveraged loans without performance pricing are
associated with a higher and significant decline in loan spreads (than those with performance
pricing), especially for nonbank-originated and highly leveraged loans. The results indicate
that information asymmetry plays an important role in the underestimation of leverage risk.
In addition, after the introduction of the guidance and the Clarification, a large portion
of leveraged loans has been securitized and distributed in the form of CLOs. Since such
securitization, allows for the transfer of loan default risk to investors, provides originating
lenders less incentive to maintain high lending standards before securitization and to monitor
borrowers after securitization, which gives rise to both adverse selection and moral hazard.
Prior literature suggests that fa reduced share in the informed party’s ownership of leveraged
loans can aggravate the cost of asymmetric information (Leland and Pyle, 1977; Ivashina,
2009). Further studies show evidence that securitization activity leads to a lax screening of
mortgages (Mian and Sufi, 2009; Keys et al., 2010; Purnanandam, 2011, Nadauld and Sherlund,
2013) and increases the risk appetite of the issuing bank (Haensel and Krahnen, 2007). A recent
study from Bord and Santos (2015) investigates the effects of securitization of corporate loans
and find that institutional loans, which use lax standards to underwrite through CLOs, suffer
higher risk than non-securitized loans originated by the same banks. Two closely related works
to ours are Ivashina and Sun (2011) and Nadauld and Weisbach (2012). Ivashina and Sun (2011)
finds evidence that the institutional demand pressure for leveraged loans generated by
Collateralized Debt Obligations (CDOs) is negatively related to the spreads of these loans.
Nadauld and Weisbach (2012) suggest that the spread of loan facilities that are eventually
securitized through CLOs is lower than the spread of loan facilities that are not securitized. To
investigate the impact of CLO issuance on nonbank-originated leveraged loan pricing, we add
an interaction term linking CLO issuance, nonbank lenders and borrowers’ leverage risk. The
estimated coefficients on this interaction term are negative and highly significant, indicating
that the information asymmetry issues are driven by a high level of CLO issuance post the 2014
Clarification is strongly linked to the decline of leverage risk premium in nonbank-originated
leveraged loans from 2014 – 2019.
Our paper contributes to the literature in a number of ways. Few papers have examined
the pricing of leveraged loans (Angbazo et al., 1998; Lim et al., 2014). For example, Lim et al.,
(2014) argue that nonbank facilities are priced with premiums relative to bank-only facilities
in the same loan package and the nonbank premium is larger when borrowers face financial
limits. In this paper, we investigate the impact of the guidance and the Clarification by
comparing different responses of regulated banks and unregulated nonbanks in terms of loan
pricing. Unlike previous studies focusing on the effects of the Clarification on banks’ lending
activities (Schenck and Shi, 2017; Kim et al., 2018; Calem et al., 2020) and non-price terms
(Abuzov et al., 2020), we directly investigate the pricing of leverage risk after the guidance
and the Clarification. We show that although the declining trend of bank-originated leveraged
loan spreads has been reversed due to the 2014 Clarification, the risk premium for nonbank
facilities between bank facilities has been narrowed due to the more pronounced decline of
nonbank loan spread. We identify two possible mechanisms associated with the decline of
highly leveraged loan spread. First, information asymmetry arises with non-performance linked
pricing and covenant-lite issuance leading to a decline in the leverage risk premium post the
2014 Clarification. Second, the high level of CLO issuance since 2014 has reduced the share
in the informed party’s ownership of leveraged loan, aggravating the cost of asymmetric
information and leading to further decline in leverage risk premium of highly leveraged loans
from 2014 – 2019.
The remainder of the paper is organized as follows. Section 2 presents the data and
background. Section 3 provides empirical evidence on whether leveraged loan pricing
mechanism effectively reflects borrowers’ high leverage ratios and aggressive business
expansion strategies in loan spreads. Section 4 provides the potential mechanisms to explain
why borrowers’ leverage risk is not reflected in loan spreads. Section 5 provides robustness
checks to confirm our findings. Section 6 concludes.
2. Data
2.1 Sample Construction
We obtain our sample of leveraged loans from Refinitiv Eikon and WRDS-Thomson
Reuters DealScan LPC for the period 2007-2019. Leveraged loan coverage at Refinitiv Eikon
is provided by Refinitiv Loan Pricing Corporation (LPC), which features the most
comprehensive and accurate real-time and historical syndicated loans data. We follow LPC and
define leveraged loan as a loan that is extended to borrowers rated BB+ or lower or it is not
rated or rated ‘BBB-‘ or higher but has (1) a spread of LIBOR +125 or higher and (2) is secured
by a first or second lien. Focusing on leveraged loans allows us to investigate whether the loan
pricing mechanism effectively reflects borrowers’ high leverage ratios.
To construct the sample, we include all leveraged loan facilities that are denominated
in U.S. dollars and made to U.S. public firms with primary syndication location in the U.S.
covered in Refinitiv Eikon between 2007 and end of 2019, a total of 12,875 facilities. We only
include loan facilities with floating-rate interest payments in the sample. We require that the
data on AISD be non-missing. The AISD is calculated as the sum of the spread over LIBOR,
upfront fee, and annual fee, and is provided by LPC directly. We additionally restrict the sample
to the most common types of facilities, including term loan A, term loan B-F, revolvers and
others. Finally, we exclude facilities issued to financial firms (SIC Code 6000-6999). In the
literature, WRDS Dealscan has been widely used for syndicated loan studies. Although both
Refinitiv Eikon and Dealscan share the same data source, coverage is slightly different. In order
to check data consistency and to extend data availability, we construct a link table connecting
the two databases on leveraged loans of Refinitiv Eikon and Dealscan with the unique identifier
of LPC tranche. Linking Refinitiv Eikon and Dealscan provides us with a broader and more
accurate coverage of leveraged loan facility characteristics including: size and maturity, loan
purpose, arrangers, type of facilities as well as information on whether the facility is senior,
secured, covenant-lite and has performance-based pricing.
To obtain borrower-specific characteristics, we match the borrower and/or borrower’s
parent name to the Compustat firm following Chava and Roberts (2008). The current Dealscan
Compustat link table only contains matches through the end of 2017. We extend the current
version of the link table to the end of 2019 by using the 6-digit CUSIP number provided from
both Refinitiv Eikon and Compustat. We also manually confirm the matching between
DealScan and Compustat. We exclude observations with missing borrowers’ financial
statement information data on the end of fiscal year prior to the current loan issuance year.
Furthermore, we refine the sample by dropping all the non-positive equity observations. The
final sample contains 6,944 leveraged loan facilities in 4,459 deals to 1,631 U.S. nonfinancial
firms.
2.2 Definition of Nonbank and Bank Lenders
Following Elliott et al., (2019) we identify a lender as a nonbank if it is categorized as
“Insurance Company”, “Corporation”, “Finance Company”, “Investment Bank”, “Mutual
Fund”, “ Trust Company”, “Leasing Company”, “Pension Fund”, “Distressed (Vulture) Fund”,
“Prime Fund”, “CDO”, “Hedge Fund”, and any other institutional investor. In addition, lead
lenders normally act as the manager for the loan with primary responsibility for ex ante due
diligence and for ex post monitoring of the borrowers, which provide information for
participant lenders (Ivashina, 2009). Therefore, we define a nonbank-originated leveraged loan
facility if it has at least one U.S. nonbank lead arranger. We follow Bharath et al., (2011) to
classify lead lenders for each loan. We classify a lender as a lead lender if the
“LeadArrangerCredit” field in the Dealscan indicates “Yes” or if the “LenderRole” field in the
Dealscan indicates one of the following: administrative agent, agent, lead arranger, arranger,
or lead bank. In our sample, the nonbank-originated leveraged loans account for about 29% of
all leveraged loan sample. This is because nonbanks are less likely to be lead arrangers relative
to commercial banks.
2.3 Overview of Sample

Table 1 presents summary statistics for the key variables in our sample. To reduce the
effects of outliers, all our continuous variables are winsorized at the 1% and 99% levels. Panel
A of Table 1 summarizes the facility type in our leveraged loan sample. Nonbanks are more
likely to issue term loans facilities (59% vs. 41%) and banks are more likely to issue revolver
facilities (54% vs. 36%). Panel B of Table 1 shows the summary statistics for leveraged loan
facilities. The average AISD for the leveraged loans originated by banks in our sample is
284.15 basis points and the spread of leveraged loans originated by nonbank is 345.05 basis
point, which is much higher than bank-originated loan spreads due to borrowers’ greater credit
risk. Only 46% of the leveraged loan facilities include covenants in the loan agreements to
monitor risk and financial performance of borrowers and to avoid its deterioration over the life
of the loan. 21% of the leveraged loan facilities in our sample are with performance-related
pricing provisions, in which case, the spread is adjustable based on pre-defined financial
criteria. Panel C of Table 1 presents summary statistics on borrower characteristics of the year
prior to the loan origination. Our main proxy of leverage risk is estimated as a firm’s total
liabilities net of cash divided by the book value of total assets with the average value of 59.50%
in our sample. The borrower of nonbank-originated loans has larger average leverage risk than
the borrower of bank-originated loans (65.25% vs. 57.21%). We experiment with two
alternative measures of a firm’s leverage ratio in the robustness check, following Lemmon et
al., (2008) and DeAngelo and Roll (2015). In the first measure, we take total debt divided by
the book value of total assets and, in the second measure, we use long term debt relative to the
book value of total assets as a measure of the borrower’s leverage risk in the long run.
Although we show the summary statistics of our sample, time-varying covariates can
change or be influenced by the post-clarification period, leading to endogeneity problems. We
provide statistical tests of mean differences for borrowers and loan characteristics prior-post
Clarification in the bank and nonbank group, respectively (OA Table 2). As shown, the control
variables are, on average, in most cases statistically different, both before and after the
Clarification. It is worthy to note that for the unregulated nonbank loan group, the mean value
of AISD experienced a more pronounced contraction from 372.13 basis points before the
Clarification to 312.28 basis points after the Clarification (59.85 basis points in total), in
contrast the mean value of AISD for the regulated bank loan group only drops 37.81 basis
points.
< INSERT TABLE 1 HERE >
3. Empirical section
3.1 Baseline results
While figure 1 shows that the spread between nonbank-originated facilities and bank-
originated facilities has narrowed since the 2014 FAQ, overall borrower risk may have declined
along with spreads. In this section, we conduct multivariate analysis to better examine whether
leverage loan pricing mechanism effectively reflect borrowers’ high leverage ratios by
investigating following empirical model:
𝑌𝑌𝑖𝑖𝑖𝑖 = 𝛽𝛽1 𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿_1𝑖𝑖𝑖𝑖−1 + 𝛽𝛽2 𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿_1𝑖𝑖𝑖𝑖−1 ∗ 𝑁𝑁𝑁𝑁𝑛𝑛𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 ∗ 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 + 𝛽𝛽3 𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝑒𝑒1 𝑖𝑖𝑖𝑖−1 ∗ 𝑁𝑁𝑁𝑁𝑁𝑁bank +
𝛽𝛽4 𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝑒𝑒1 𝑖𝑖𝑖𝑖−1 ∗ 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 + 𝛽𝛽5 𝑁𝑁𝑁𝑁𝑁𝑁bank ∗ 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 + 𝑋𝑋𝑖𝑖𝑖𝑖−1 + 𝛼𝛼𝑖𝑖 + 𝛼𝛼𝑡𝑡 + 𝜖𝜖𝑖𝑖𝑖𝑖 (1)
𝑌𝑌𝑖𝑖𝑖𝑖 is the AISD of leveraged loan facility 𝑖𝑖 in fiscal year t. 𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿_1𝑖𝑖𝑖𝑖 is the
borrower’s total liabilities net of cash divided by the book value of total assets prior to the loan
issuance date. 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 is a dummy equal to 1 if the loan year is either at or after the issuance of
the Clarification in Nov 2014. 𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁 is a dummy variable that equals to one if one facility
has at least one U.S. nonbanks lead arranger, and zero otherwise. 𝑋𝑋𝑖𝑖𝑖𝑖−1 is a set of control
variables including: log of loan amount (LN_amount), log of loan maturity (LN_Maturity), an
indicator that takes the value of one if the facility is secured, and zero otherwise (Secured
indicator), an indicator that takes the value of one if performance pricing provisions are
included in the facility (Performance pricing indicator), an indicator that takes the value one if
the loan has covenants and zero otherwise (Covenants indicator); the log of borrower’s total
assets in the fiscal year prior to the loan issuance date (LN_TA), and the industry adjusted
return on total assets (Ind_adj ROA) at the end of the fiscal year prior to the loan issuance date.
Table 2 reports the baseline OLS regression results of Eq (1) with double-clustered
standard errors by firm and year to account for heteroscedasticity. We include facility-purpose
fixed effects, industry fixed effects and year fixed effects in all the regression models.
Leveraged loan in our sample includes mainly two categories of loan types: revolver and term
loan. Term loan spreads are expected to be higher than revolver reflecting longer maturities
and greater credit risk (Angbazo et al., 1998; Harjoto et al., 2006). Accordingly, we estimate
separate regression models for revolvers and term loans to identify whether any of the
differences we observe in pricing in the aggregate sample varies between the two types of loans.
Earlier research suggests that leverage risk is positively priced in syndicated loan spread
(Angbazo et al., 1998; Lim et al., 2014). In this paper, we also find a positive relation between
borrower leverage ratio (Leverage_1) and AISD. In model (i) with the whole sample of
leveraged loans, for a borrower with average leverage ratio, a 1-standard-deviation increase in
leverage ratio is associated with an increase of 29% of AISD in nonbank-originated facilities
from 2007 - 2014. Our main interest is the size, sign and statistical significance of the
coefficients on the triple interaction term Leverage_1 *Nonbank* Post, which captures the
difference, in the pre- and post-Nov 2014 periods, on leverage risk premium of nonbank-
originated loans. We find the coefficients on Leverage_1 *Nonbank* Post in the regression
model is sizeable, negative and statistically significant, indicating that the positive leverage
risk premium for a given level of leverage has significantly declined since the issuance of the
Clarification in 2014. The results also suggest a large economic magnitude of the coefficient
on the triple interaction term of Leverage_1*Nonbank*Post for a borrower with average

leverage ratio in our sample in the period from Nov 2014 – Dec 2019. A 1-standard-deviation
increase in leverage ratio only results in a 9% increase in AISD (12%-20%+17%) driven by a
strong declining effect of 20%. The results indicate a significant drop in AISD for a given
leverage risk from Nov 2014 – Dec 2019. In model (ii) and (iii) of Table 3, we present estimates
of Eq.(1) for the subsamples of term loans and revolvers, respectively. The results demonstrate
a stronger underestimation of leverage risk from 2014 - 2019 of leverage risk in the subsample
of term loans with both higher significance and economic magnitude in model (ii). Specifically,
a 1-standard-deviation increase in leverage ratio is associated with an increase of 21.3% of
AISD for nonbank-originated term loans from 2007 – Nov 2014. However, during Nov 2014-
Dec 2019 period, a 1-standard-deviation increase in leverage ratio is associated with a 3%
increase in AISD for term loans (10%-21%+11.3%).
To control for other potential effects on leveraged loan spread, we include variables on
loan and borrower specific characteristics. In line with Dennis, et al. (2000) we find that loan
spreads decline with maturity. Prior studies show that loan spreads are higher on secured
facilities because lenders require collateral on high-risk loans, and the pledged assets do not
diminish default and recovery risk sufficiently to result in lower spreads (Ivashina, 2009; Lim
et al., 2014). A facility with performance pricing provision and/or covenant protection tends to
have lower spread. On the firm characteristic side, larger borrower with better profitability
measured by industry adjusted ROA (Ind_adj ROA) is associated with lower loan spread,
although the coefficients are insignificant. Our main results hold-up well after including all the
control variables, loan and year fixed effects. The coefficients on the interaction term of
Leverage_1*Nonbank*Post remain negative and significant at the 1% level in all estimates.
Overall, our results suggest that although leverage risk is positively priced in AISD, the
leverage risk premium of nonbank-originated loan facilities declined significantly from Nov
2014 – 2019 relative to bank-originated loan facilities. Furthermore, we find that the decline
has been more pronounced in term loans compared with revolvers.
In order to alleviate the endogeneity issues within our empirical setup we present the
treatment effects from Equ (1) in Figure 2. Following previous literature (e.g., Defusco, 2018;
Iftekhar et al., 2020; O’Malley, 2021) we omit the year -1 as the reference period. We find that
before 2014 the coefficient plots are insignificant, indicating that the leverage risk premium for
bank loans and nonbank loans follows parallel trends. However, from 2014 – 2019 the
cofficeint plots become significant different from zero, which means there is a discontinuity
around 2014. The effect is economically meaningful and persistent until the end of our sample
period. Figure 2 shows a clear discontinuity post 2014, suggesting a statistically significant and
negative long-run effect on the leverage risk premium between nonbanks and banks. The
finding provides us a strong support to establish the causal identification through our
difference-difference (diff-in diff) setup.
<INSERT Figure 2 HERE>
3.2. Propensity score matching-difference-in-difference
If the debt choice between borrowing from banks or nonbanks is not random, the diff-
in-diff set up may suffer from endogeneity and selection bias. To further address this concern,
we apply the propensity score matching approach to improve the comparability between
nonbank loan and bank loan spreads. Specifically, we choose a pair of borrowers with similar
pre-loan financial characteristics, but one from the nonbank group, and one from the group of
banks. The predictive probability (propensity score) of borrowing from a nonbank lender is
obtained from the Probit model. To select the matching variables, we are motivated by previous
literature (Lim et al., 2014 and Biswas et al., 2020), and use various borrowers’ financial
characteristics ( Leverage (%), LN_Cash, LN_TA, LN_DLTT, and Industry-adjusted ROA). All
the matching variables are lagged to help alleviate endogeneity concerns. The propensity score
matching model is estimated as follow:
′
𝜌𝜌𝑖𝑖 = Pr�𝐷𝐷𝑖𝑖 = 1�𝑋𝑋𝑖𝑖𝑖𝑖𝑖𝑖 � = 𝛿𝛿(𝑋𝑋𝑖𝑖𝑖𝑖𝑖𝑖 𝛽𝛽 + 𝜀𝜀𝑖𝑖 ) (2)
Where 𝐷𝐷𝑖𝑖 is a dummy variable, which equals one if the lead lender is a nonbank and
′
equals zero if the lead lender is a bank. 𝑋𝑋𝑖𝑖𝑖𝑖𝑖𝑖 is a vector of firm characteristics one year before
the loan issuance date. 𝛿𝛿 is a standard normal cumulative distribution function. Specifically,
we implement the nearest neighbour matching to construct the counterfactual outcome. We
match each nonbank borrower with the three nearest neighbours with replacement. Prior
literature suggest that limited access to bank funding is the reason why firms borrow from
nonbanks. Lim et al., (2014) find that non-bank facilities are more expensive relative to bank-only
facilities in the same loan package when borrowers face limited access to bank funding. Further,
Chernenko et al., (2020) confirm that nonbank loans carry 190 basis points higher interest rates.
Our estimation results from Equ(2) are presented in Table 3. As displayed, most of the
covariates are significant at 1%. Firms with a higher leverage ratio, more cash, and more long-
term debt are more likely to borrow from nonbanks, and a firm with higher industry-adjusted
ROA is less likely to borrow from a nonbank. The results are in line with Lim et al., (2014)
and Chernenko et al., (2020). Table 4 confirms that after matching the bank and nonbank
groups are well balanced with no statistical differences in terms of selected matching variables.

After obtaining a closely matched sample, we present the PSM diff-in-diff estimations
in Table 5. Specifically, we re-estimate Equ(1) based on the new matched sample in Table 4.
As shown, the nonbank loan spread in all the regression specifications displays a sizeable and
statistically significant decline after 2014. The loan spread decline in term loan subsample is
more pronounced than revolvers providing further evidence on the reliability of the baseline
estimates.
3.3. Predictive margin on nonbank loans V.S. bank loans
To further investigate the dynamic change of AISD from 2009-2018, we illustrate the
predictive margins of loan spread for bank and nonbank groups in Figure 3. We find a declining
trend of the predictive margins on leveraged loan spreads for both banks and nonbanks.
However, the decline has been more sharply for nonbanks after the 2014 Clarification, and the
predictive margins between nonbank groups and bank groups have been significantly narrowed
since the 2014 Clarification. Economically, in 2019 the predictive margin for nonbank loan
spread drops 13% relative to 2014, but only slightly decreases 5% for bank loans. It is
interesting to note that the effect of the regulation has been persistent and increasing. This result
provides further visual evidence on the treatment effects of the Clarification on nonbank vs
bank groups.
<INSERT Figure 3 HERE>
4. Investigating the channels: information asymmetry
We provide both regression and visual evidence that nonbanks’ leverage risk premium
declined sharply after 2014 relative to the leverage risk premium for bank loans. This raises a
further research question: what could be the underlying mechanism working behind the
narrowed leverage risk premium? In this section, we identify and investigate the underlying
channels that give rise to the narrowed AISD with respect to leverage risk.
4.1 Covenant-lite loans and loan spread
Syndicated loans suffer information asymmetry issues between lead banks and
participants by design. Acting as the mandated manager for the loan, the lead bank is granted
primary responsibility for ex ante due diligence and ex post monitoring of the borrower.
Participants and investors rely on the lead bank for collecting borrower information. Financial
covenants play a key role in monitoring borrower performance and provide lenders the right to
renegotiate their loan contracts (Wang and Xia et al., 2014; Griffin et al., 2019), which can
significantly mitigate the moral hazard. With increasing concerns about the surge of covenant-
lite loans after the global financial crisis, banking regulators are motivated to issue the guidance
and the 2014 Clarification to eliminate the issues on “the absence of meaningful maintenance
covenants in loan agreements” and “the participation of unregulated investors” (Berlin et al.,
2020). However, the guidance and its Clarification did not successfully control the absence of
financial covenants in the leveraged loan agreements with the average percentage of covenant-
lite loans reaching 68% during 2014 – 2019. This is because the Clarification only effectively
affected the lending standards of regulated banks, unregulated nonbanks have strong incentive
to provide covenant-lite loan agreements to gain market share. As Abuzov et al., (2020) find,
borrowers are more likely to switch to borrowing from unregulated non-bank lenders and this
switch is subsequently associated with less covenant protection. Figure 4 compares the parallel
trend of covenant-lite shares for banks and nonbanks from 2007 to 2019, respectively. The
figure clearly shows that the fraction of covenant-lite loans for both nonbanks and banks
increased rapidly post the Clarification driven by the strong competition in the leveraged loan
market. The figure also presents a strong divergent post-trend after the Clarification. The
unregulated nonbank dramatically increased the issuance of covenant-lite loans from 55% in
2013 to 70% in 2019; while the average covenant-lite loans issued by regulated banks remains
relatively constant from 2014 - 2019. After the Clarification, the fraction of covenant-lite loans
issued by banks is far below the level of nonbank covenant-lite loan issuance.
< INSERT Figure 4 HERE >
The relaxation of investor protection in covenant-lite loans intensifies information
asymmetry associated with leveraged loan pricing. To investigate the issue of information
asymmetry on leveraged loan pricing, we conduct a sub-sample analysis by splitting the sample
into facilities with covenant and covenant-lite groups. In Panel A of Table 6 we present
regression results on the subsample of covenant-lite leveraged loans. The results demonstrate
a strong negative and significant relation between Leverage_1*Nonbank *Post and AISD in all
the estimations with respect to leveraged loan (Column (1)), term loans (Column (2)) and
revolvers (Column (3)). However, in Panel B, the coefficients of triple interaction term
Leverage_1*Nonbank *Post in all the estimations become insignificant, indicating no clear
evidence of underestimation of leverage risk premium in the subgroup of loan facilities with
covenant protections from 2014-2019. The results confirm that more severe information
asymmetry associated with covenant-lite loans leads to a stronger and more significant decline
in leverage risk premium. In addition, to further compare the statistical difference between
covenant-lite loans and covenant loans, a quadruple interaction term is constructed by linking
leverage ratio, nonbank dummy variable, post dummy variable, and covenant-lite dummy
variable (Panel A of Table OA4). The coefficients on the quadruple interaction term are
statistically significant and negative, confirming a decline of the leverage risk premium for
nonbank covenant-lite loans post 2014.

In addition to financial covenants, performance pricing is also a widely used clause in
loan packages to align incentive between creditors and borrowing firms and to mitigate
information asymmetry issues. Performance pricing provisions, which defines performance
levels based on certain criteria and their corresponding interest spreads, include both interest –
increasing and interest – decreasing performance pricings. Although performance pricing
provision features some characteristics similar to financial covenant, earlier studies find that
they protect creditors in different ways contingent on the movement of borrower performance
after loan syndication (Asquith et al., 2005; Roberts and Sufi 2009, and Manso et al., 2010).
For example, Manso et al., (2010) argue that performance pricing features are used as a
screening device to mitigate information asymmetry and show that firms using performance
pricing are more likely to improve their credit ratings subsequently. In Table 7, we present the
estimation results on two subgroups of leveraged loans with and without performance pricing
provisions. In Panel A we present regression results on the subsample of leveraged loans
without performance pricing provisions. The results show a strong negative and significant
relation between Leverage_1*Nonbank *Post and AISD in all the estimations with respect to
leveraged loan ( Column (1)), term loans (Column (2)) and revolvers (Column (3)). However,
in the subgroup of leveraged loans with performance pricing provisions in Panel B, the
coefficients of Leverage_1*Nonbank*Post become insignificant in all of estimations,
indicating that there is no clear evidence of underestimation of leverage risk premium in for
loan facilities with performance pricing. The results are in line with Table 7 and confirmed that
information asymmetry issues strongly contributed to the underestimation of nonbank facilities’
leverage risk premium from 2014 - 2019. In Panel B of Table OA 4, We also construct a
quadruple interaction term by linking leverage ratio, nonbank dummy variable, post dummy
variable and performance pricing dummy variable to compare the statistically different
between coefficients in the two groups. The coefficients on the quadruple interaction term are
statistically significant and negative, confirming a decline of the leverage risk premium for
nonbank loans with without performance pricing after 2014.
4.2 CLO issuance and Loan spread
In the post-global financial crisis, CLOs issuance has surged in an environment of
continuing very low-interest rates driven by increased investor demand for higher yields. A
large portion of leveraged loans is securitized and structured into tranches to accommodate
different levels of risk appetite from investors, especially after 2014. From 2007 – 2013 the
average annual CLOs issuance in the US was $ 39.28 billion. However, the average annual
CLOs issuance from 2014 – 2019 reached $110.42 billion. 5 The rapid development in the
CLOs market makes it easier for originating lenders to securitise and sell these loans. Therefore,
it attracts an increasing number of bank and nonbank lenders entering the leveraged loan
market. The substantial growth in CLOs issuance from 2014, which accounts for approximately
half of the leveraged loan market, significantly contributed to the boom of the leverage loan
market. Further, although the guidance and the Clarification provide banks operating in a safe
and sound manner in leveraged lending, the supervision is not covered to either bank or
nonbank securitization process of creating CLOs. The Clarification effectively reduces banks’
leveraged lending activity, but nonbanks increase their leveraged lending activity following it
(Kim et al., 2018). Unlike banks, nonbanks have a stronger incentive to hold riskier term loans,
specifically, institutional loans, which are more likely to be securitized through CLOs instead
5
Our data on annual CLOs issuance and outstanding in the US are obtained from U.S. Federal Reserve & S&P Global Market Intelligence
of holding them (Marsh and Lee, 2019). Thus, from the supply side, the increasing number of
nonbank lenders in the leveraged loan market also contributes to the CLOs boom since 2014.
Overall, the CLOs market experienced a boom period after 2014 with size of CLO outstanding
from 2014-2019 doubled (see Appendix Figure 1).
Since securitization through CLOs issuance effectively allows the transfer of loan
default risk to investors, originating lenders have fewer incentives to maintain high lending
standards before securitization and monitor borrowers after securitization, which gives rise to
both adverse selection and moral hazard. Earlier literature provide evidence that the
securitization activity leads to lax screening for mortgage (e.g., Mian and Sufi, 2009; Keys et
al., 2010; Purnanandam, 2011; Nadauld and Sherlund, 2013). More recently, Bord and Santos
(2015) find that securitization activity leads to adverse selection in the quality of CLO collateral.
Specifically, their results show that the performance of loans sold to the CLOs is poorer than
the similar unsecuritized loan originated by the same bank. Prior studies also document a
negative relationship between syndicated loan securitization and loan spread (e.g., Ivashina and
Sun, 2011; Nadauld and Weisbach, 2012).
To investigate the impact of CLO issuance on leveraged loan pricing, we add a triple
interaction term linking CLO issuance, a dummy variable for nonbank lead lenders and
borrowers’ leverage risk. Table 6 reports the estimation results on leverage risk, nonbank, CLO
issuance and loan spread (AISD) based on the propensity score matched sample. In column (1)
of Table 8, we present the results with the whole sample of leveraged loan. The estimated
coefficient on the interaction term of Leverage_1*Nonbank*CLO is negative and highly
significant, indicating that the information asymmetry issues associated with the high level of
CLO issuance since 2014 has strongly explained the decline in spreads of nonbank-originated
highly leveraged loans from 2014 – 2019. For a given leverage level a 1-standard-deviation
increase in CLO issuance results in a decline in leverage premium for nonbank facilities of
35%. In columns (2) and (3) we estimate the subsamples of term loans and revolvers,
respectively. The results demonstrate a strong negative relationship between the triple
interaction term of Leverage_1*Nonbank*CLO and AISD in both subsamples of term loans
and revolvers. Specifically, for a given leverage level a 1-standard-deviation increase in CLO
issuance will result in a decline in AISD of 40% for term loans and 32% for revolvers
respectively. We find that the decline is pronounced for term loans because term loans are more
likely to be securitized than revolvers. which is in line with Nadauld and Weisbach, (2012).
The results confirm that the decline in AISD during 2014 - 2019 is associated with the CLOs
issuance boom since the issuance of the Clarification in 2014. Consistent with our results in
Table 4, the effect of CLO issuance on AISD is stronger for term loans with larger coefficients
and significance compared with revolvers. This is due to the high proportion of institutional
tranches, which are designed to be securitized and distributed to institutional investors,
including in the term loan facilities.
5. Robustness checks
This subsection provides several robustness checks for the baseline results from
estimating Eq(1). To conserve space, we provide the tables in the Online Appendix. All the
results are estimated based on the propensity score matched sample.
5.1 Is the decline driven by low interest rate expectation?
A potential concern with our results is whether the recent decline in the leveraged loan
spread is driven by the low interest rate expectations. Since the 2009 recession, the Federal
Reserve (Fed) has maintained an accommodative monetary policy with historically low interest
rates. From late 2013, Fed began normalizing the stance of monetary policy and has gradually
increased the pace of tightening. The rising expectation of interest rate hikes coupled with
improving economic fundamentals, which improves investor demand for high yield leveraged
loan. In Table OA5 we present the regression results with an additional control variable of
projected short-term interest rate (Interest Rate Forecast) to identify the effect of interest rate
expectations on leveraged loan spread. We find a strong negative effect of interest rate
expectation on leveraged loan spreads. The main results on the interaction term of
Leverage_1*Nonbank*Post remain negative and significant after including the control variable
of projected interest rate. The results suggest that the narrowed spread of high leverage risk in
nonbank-originated loan facilities from 2014 – 2019 has not been driven by low interest rate
expectations. Instead, since late 2013 due to improving economic fundamentals and monetary
policy normalisations interest rate expectations has been improved.
5.2 Is the decline driven by borrower high growth potential?
A further potential concern with our results is whether the narrowed spread of leverage
risk is caused by the high growth potential of the borrowers. To address this issue, in the Panel
A of Table OA6 we present estimation results based on the matched sample with the subsample
of high growth potential borrowers measured by borrowers’ enterprise value relative to EBIT.
A high multiple of firm’s enterprise value relative to EBIT represents high prospects for future
revenues and growth. The estimation results in the Panel A of Table OA6 show no evidence
that a borrower’s growth potential is linked with the narrowed spread of nonbank-originated
leveraged loan. We also present estimation results with the subsample of low growth potential
borrowers. The estimation results in Panel B of Table OA 6 suggest that the decline of leverage
risk premium is statistically significant. Overall, our results confirm that the narrowed leverage
risk premium is not driven by high growth potential.
5.3 Change different leverage risk proxies

Last but not least, we apply two alternative leverage risk measures defined as: a
borrower's total debt relative to total assets and a borrower’s total long-term debt relative to
total assets, respectively. The estimation results based on the propensity score matched sample
in Table OA7 with full leveraged loan sample, and the subsamples of term loans and revolvers
are in line with our baseline results in Table 2 and the PSM diff-in-diff results in Table 5. The
results confirm our main findings that nonbank facilities’ leverage risk premium has narrowed
compared with bank facilities from 2014 – 2019.
6. Conclusion
The introduction of the 2013 guidance and its 2014 Clarification attempts to mitigate
the increasing concerns on the tremendous growth in the volume of leveraged loans, especially
in the absence of meaningful maintenance covenants in loan agreements and regulated
investors. However, the regulation results in increasing competition between bank and
nonbanks and shifting risky loans from traditional banks to nonbank lenders. Following the
2014 Clarification, the covenant-lite loan issuance has appeared to be picking up speed because
borrowers switched to unregulated nonbank loans with relatively fewer covenants and weaker
investor protection. Further, the leveraged loan securitisation is energised by the growth in
institutional investor participation in the leveraged loan market, which lead to moral hazard
and adverse selection issues. In this paper, we investigate the impact of the Clarification on
leveraged lending market from a different angle compared with previous literature (e.g. Kim
et al., 2018; Calem et al., 2020 Abuzov et al., 2020). We directly foucus on leveraged loan
pricing after the Clarification and show that a higher degree of information asymmetry driven
by an increase in covenant-lite loans and weaker investor protections are strongly associated
with the narrowed leverage risk premium in the period 2014 – 2019. In addition, adverse
selection and moral hazard associated with the high level of CLO issuance strongly explain the
decline of nonbank leveraged loan spreads.
Our sample ends before the outbreak of Covid-19 pandemic. The results viewed in the
context of the leveraged loan market before the Covid-19 pandemic help us to understand why
leveraged loans are vulnerable to the economic downturn and help us to rethink the pricing
mechanism of leveraged loans on the absence of enough regulations, especially for nonbank
financial institutions. Currently, nonbank financial institutions are subject to very limited
regulatory restrictions on leveraged loan issuance. On July 17, 2020, in the conference of “A
decade of Dodd-Frank,” Yellen stated that “We need to change the structure of the Financial
Stability Oversight Council (FSOC) and build up its powers to be able to deal more effectively
with all the problems that exist in the shadow banking sector.” In this context, we believe our
paper brings important policy indication on prudential regulation of leveraged loan market and
how to increase the safety and soundness of the financial institutions. Our paper also opens
several avenues for future research in the post-Covid-19 era. One question is the extent, to
which, the link between adverse selection associated from leveraged loan securitization and
decline in the loan spread is detrimental to financial stability especially after the shock of the
Covid-19 pandemic. Also, it is important to identify whether reduced investor protection,
increased information asymmetry and larger presence of nonbanks increase the complexity of
the leveraged loan market and impede the healthy development of the financial system.
References
Abuzov, R., Henpeck and Steri, R., 2020. Do banks compete on non-price terms? Evidence
from loan covenants. In EFA 2020 Annual Conference Working Paper.
Angbazo, L.A., Mei, J. and Saunders, A., 1998. Credit spreads in the market for highly
leveraged transaction loans. Journal of Banking & Finance, 22(10-11), pp.1249-1282.
Asquith, P., Beatty, A. and Weber, J., 2005. Performance pricing in bank debt
contracts. Journal of Accounting and Economics, 40(1-3), pp.101-128.
Bharath, S.T., Dahiya, S., Saunders, A. and Srinivasan, A., 2011. Lending relationships and
loan contract terms. Review of Financial Studies, 24(4), pp.1141-1203.
Biswas, S., Ozkan, N. and Yin, J., 2020. Non-bank loans, corporate investment, and firm
performance. Available at SSRN 3327539.
Bord, V.M. and Santos, J.A., 2015. Does securitization of corporate loans lead to riskier
lending? Journal of Money, Credit and Banking, 47(2-3), pp.415-444.
Bosch, O., 2007, December. Information asymmetry and the pricing of private debt-evidence
from European syndicated loans. In EFA 2007 Ljubljana Meetings Paper.
Bradley, M. and Roberts, M.R., 2015. The structure and pricing of corporate debt covenants.
Quarterly Journal of Finance, 5(02), pp.1-37.

Calem, P., Correa, R. and Lee, S.J., 2020. Prudential policies and their impact on credit in the
United States. Journal of Financial Intermediation, 42, p.100826.
Chava, S. and Roberts, M.R., 2008. How does financing impact investment? The role of debt
covenants. Journal of Finance, 63(5), pp.2085-2121.
Chretien, E., and V. Lyonnet. 2018. Traditional and shadow banks. Working Paper, Ohio State
University.
DeAngelo, H. and Roll, R., 2015. How stable are corporate capital structures? Journal of
Finance, 70(1), pp.373-418.
DeFusco, A.A., 2018. Homeowner borrowing and housing collateral: New evidence from
expiring price controls. The Journal of Finance, 73(2), pp.523-573.
Dennis, S., Nandy, D., & Sharpe, I. (2000). The determinants of contract terms in bank
revolving credit agreements. Journal of Financial and Quantitative Analysis, 35(1), pp.87-110.
Elliott, D., Meisenzahl, R.R., Peydró, J.L. and Turner, B.C., 2019. Nonbanks, banks, and
monetary policy: US loan-level evidence since the 1990s. Available at SSRN 3475427.
Fahri, E., and J. Tirole. 2017. Shadowbanking and the four pillars of traditional financial
intermediation.Working Paper, Harvard University.

Griffin, T.P., Nini, G. and Smith, D.C., 2019. Losing control? The 20-year decline in loan
covenant restrictions.
Haensel, Dennis and Krahnen, Jan Pieter., 2007. Does credit securitization reduce bank risk?
Evidence from the European CDO market. Available at SSRN 967430.
Harjoto, M., Mullineaux, D.J. and Yi, H.C., 2006. A comparison of syndicated loan pricing at
investment and commercial banks. Financial Management, 35(4), pp.49-70.
Hasan, I., Manfredonia, S. and Noth, F., 2021. Cultural resilience, religion, and economic
recovery: Evidence from the 2005 hurricane season (No. 9/2021). IWH Discussion Papers.
Ivashina, V. and Sun, Z., 2011. Institutional demand pressure and the cost of corporate
loans. Journal of Financial Economics, 99(3), pp.500-522.
Ivashina, V., 2009. Asymmetric information effects on loan spreads. Journal of Financial
Economics, 92(2), pp.300-319
Keys, B.J., Mukherjee, T., Seru, A. and Vig, V., 2010. Did securitization lead to lax screening?
Evidence from subprime loans. Quarterly Journal of Economics, 125(1), pp.307-362.
Kim, S., Plosser, M.C. and Santos, J.A., 2018. Macroprudential policy and the revolving door
of risk: Lessons from leveraged lending guidance. Journal of Financial Intermediation, 34,
pp.17-31.
Lemmon, M.L., Roberts, M.R., Zender, J.F., 2008. Back to the beginning: persistence and the
cross-section of corporate capital structure. Journal of Finance 63 (4), 1575–1608.
Leland, E., Pyle, D., 1977. Information asymmetries, financial structure, and financial
intermediation. Journal of Finance 32, 371–387.
Lim, J., Minton, B.A. and Weisbach, M.S., 2014. Syndicated loan spreads and the composition
of the syndicate. Journal of Financial Economics, 111(1), pp.45-69.
Manso, G., Strulovici, B. and Tchistyi, A., 2010. Performance-sensitive debt. Review of
Financial Studies, 23(5), pp.1819-1854.
Marsh, W.B. and Lee, S.J., 2019. What’s Driving Leveraged Loan Spreads?. Macro Bulletin,
pp.1-5.
Martinez-Miera, D., and R. Repullo. 2018. Markets, banks and shadow banks. Working Paper,
CEMFI.
Mian, A. and Sufi, A., 2009. The consequences of mortgage credit expansion: Evidence from
the US mortgage default crisis. Quarterly Journal of Economics, 124(4), pp.1449-1496.
Nadauld, T.D. and Sherlund, S.M., 2013. The impact of securitization on the expansion of
subprime credit. Journal of Financial Economics, 107(2), pp.454-476.
Nadauld, T.D. and Weisbach, M.S., 2012. Did securitization affect the cost of corporate
debt? Journal of Financial Economics, 105(2), pp.332-352.
O'Malley, T., 2021. The impact of repossession risk on mortgage default. Journal of
Finance, 76(2), pp.623-650.
Plantin, G. 2014. Shadow banking and bank capital regulation. Review of Financial Studies
28:146–75.
Purnanandam, A., 2011. Originate-to-distribute model and the subprime mortgage crisis.
Review of Financial Studies, 24(6), pp.1881-1915.
Rajan, R. and Winton, A., 1995. Covenants and collateral as incentives to monitor. Journal of
Finance, 50(4), pp.1113-1146.
Roberts, M.R. and Sufi, A., 2009. Renegotiation of financial contracts: Evidence from private
credit agreements. Journal of Financial Economics, 93(2), pp.159-184.
Schenck, N. and Shi, L., 2017. Leveraged lending regulation and loan syndicate structure: a
shift to shadow banking? Unpublished manuscript.
Sufi, A., 2007. Information asymmetry and financing arrangements: Evidence from syndicated
loans. Journal of Finance, 62(2), pp.629-668.
Wang, Y. and Xia, H., 2014. Do lenders still monitor when they can securitize loans?. Review
of Financial Studies, 27(8), pp.2354-2391.

Figure 1:
Average AISD among nonbank-originated loans (red line) and bank-originated banks (blue line) from 2007-2019.
Figure 2: Parallel trend
This figure reports coefficient plots and 95% confidence intervals on the triple interaction among leverage, nonbank (=1 for
loan facilities issued by nonbank lead arrangers) and post dummies (=1 for loan issuance year is after 2014). We also include
tranche-purpose, industry fixed, and year fixed effects in our regressions. Vertical line separates pre-IGLL period and post-
IGLL in 2014. We drop pre_1 time as the reference period in our analysis by following previous literature (e.g., Defusco, 2018;
Iftekhar et al., 2020; O’Malley, 2021)
Figure 3: Predictive margins
This figure reports predictive margins and 95% confidence intervals on the triple interaction among leverage, nonbank (=1 for
loan facilities issued by nonbank lead arrangers) and post dummies (=1 for loan issuance year is after 2014). We also include
tranche-purpose, industry fixed, and year fixed effects in our regressions. Vertical line separates pre-IGLL period and post-
IGLL in 2014.
Figure 4: Trends in the fraction of covenant lite lending: nonbank lenders versus bank lenders
This figure plots average fraction of covenant lite lending among nonbank and bank groups around the introduction of the
Clarification (2014). Nonbanks are the red line and banks are the blue line.
Table 1: Summary statistics
This table presents the sample facility types (Panel A), and summary statistics for selected facility characteristics (Panel B)
and borrower characteristics (Panel C). Mean values are reported for the full sample of leveraged loan facilities, for the
subsample of bank-originated loan facilities and nonbank-originated facilities. Panel B includes selected borrowing firm
characteristics, which are computed as of the year prior to the loan transaction. The sample of loan facilities is from the LPC’s
Dealscan and Refinitiv Eikon, originated between 2007 and 2019 to US-based non-financial firms. Amount is the size of
facility in $ millions; Maturity is the maturity of the facility in months; Secured indicator is an indicator that takes the value
of one if the facility is secured, and zero otherwise; Performance pricing indicator is an indicator that takes the value of one if
performance pricing provisions are included int the facility, and zero otherwise Covenant is an indicator that takes the value
of one if the loan has covenants, and zero otherwise; institutional loan is an indicator that takes a value of one if the facility is
an institutional tranche, and zero otherwise; AISD is the basis point spread over LIBOR plus the annual fee and the up-front
fee spread; Total asset is the total assets of the borrower at the end of the fiscal year prior to the current loan in $ millions;
Total debt is the total debt of the borrower at the end of the fiscal year prior to the current loan in $ millions; Long-term debt
is the long term debt of the borrower at the end of the fiscal year prior to the current loan in $ millions; Leverage is the
borrower's book leverage ratio at the end of fiscal year prior to the current loan, estimated as total liabilities net of cash divided
by the book value of total assets; Leverage_2 is the borrower's book leverage ratio at the end of the fiscal year prior to the
current loan, estimated as total debts divided by the book value of total assets; Leverage_3 is the borrower's book leverage
ratio at the end of fiscal year prior to the current loan, estimated as long term debt relative to the book value of total assets. All
continuous variables are winsorized at the 1% and 99% levels.
All leveraged loan facilities Bank Non-Bank
Variable N Mean Std. dev. N Mean Std. dev. N Mean Std. dev.
Panel A: Facility type
% Revolver 6,944 0.49 0.50 4,954 0.54 0.50 1,990 0.36 0.48
% Term loan 6,944 0.47 0.50 4,954 0.42 0.49 1,990 0.59 0.49
% Other 6,944 0.04 0.19 4,954 0.04 0.18 1,990 0.04 0.20
Panel B: Facility Characteristics

All-in-drawn spread (bps) 6,782 301.52 149.03 4,848 284.15 137.12 1,934 345.05 167.66
Amount ($ million) 6,944 470.86 676.94 4,954 417.92 603.72 1,990 602.65 817.10
Maturity (months) 6,882 57.20 19.36 4,920 55.35 19.22 1,962 61.84 18.95
Secured indicator 6,944 0.72 0.45 4,954 0.68 0.47 1,990 0.81 0.39
Performance pricing indicator 6,944 0.21 0.41 4,954 0.22 0.41 1,990 0.19 0.39
Covenants indicator 6,944 0.46 0.50 4,954 0.47 0.50 1,990 0.42 0.49
Panel C: Borrower Characteristics
Total asset ($M) 6,795 5,131.27 13,747.56 4,859 4,196.95 12,065.37 1,936 7,476.25 17,041.56
Total debt ($M) 6,700 2,061.81 5,384.01 4,777 1,555.88 4,100.30 1,923 3,318.61 7,552.50
Long-term debt ($M） 6,876 1,964.01 5,078.49 4,905 1,482.073 3,846.66 1,971 3,163.38 7,152.28
Leverage(%) 6,744 59.50 26.68 4,825 57.21 26.34 1,919 65.25 26.70
Leverage_2(%) 6,618 38.74 23.62 4,730 36.57 22.96 1,888 44.19 24.36
Leverage_3(%) 6,792 36.03 23.14 4,856 33.92 22.47 1,936 41.33 23.92
Table 2: Pricing of Leverage Risk in Leveraged Loans
This table presents the regression results of Equation (1). The sample of loan facilities is from the DealScan and Eikon database,
originated between 2007 and 2019 to US-based non-financial firms. The dependent variable is the AISD (all-in-spread-drawn),
and the analysis is conducted at the loan facility level. The coefficient of the interaction term linking Leverage, Nonbank and
Post suggests that leverage risk premium of nonbank facilities is underestimated after 2014 and the underestimation is highly
significant for both term loan and revolving credit facilities. All specifications include facility-purpose fixed effects, industry
fixed effects, and year fixed effects. Standard errors are double-clustered by both firm and year. ***, **, and * correspond to
statistical significance at the 1%, 5%, and 10% level, respectively.
Leveraged Loans Term Loans Revolvers

Sample
Coefficient Coefficient Coefficient
Dependent variable=AISD
(t-statistic) (t-statistic) (t-statistic)
Leverage *Nonbank* Post -1.279*** -1.321*** -0.748**
(-4.23) (-3.39) (-2.23)
Leverage *Nonbank 0.776*** 0.529*** 0.589***
(7.45) (3.87) (6.06)

Leverage*Post -0.010 0.354 -0.238**
(-0.06) (1.36) (-2.04)
Nonbank*Post 77.71*** 89.43*** 35.76*
(4.09) (3.75) (1.73)
Leverage_1 0.669*** 0.607*** 0.474***
(5.79) (3.43) (5.29)

LN_Amount -14.82*** -15.39*** -29.49***
(-7.03) (-4.02) (-17.00)
LN_Maturity -31.35*** -25.25** -45.40***
(-5.45) (-2.52) (-6.09)
Performance -63.02*** -81.26*** -13.80***
(-12.83) (-11.30) (-3.28)

Secured 52.98*** 99.65*** 10.99***
(11.58) (13.64) (3.14)
Covenant -16.94*** -20.02*** -1.88
(-3.74) (-3.09) (-0.50)
LN_TA -12.79*** -24.85*** 0.758
(-5.78) (-7.13) (0.48)

Ind_adj ROA -0.053 -0.227 0.124*
(-0.26) (-0.65) (1.81)
Year FE Yes Yes Yes
Industry FE Yes Yes Yes
Purpose FE Yes Yes Yes
Obs 6,495 3,094 3,227

Adj R^2 0.257 0.281 0.368
Table 3: Propensity score estimation: Probit model
This table represents the regression results of probit model based on propensity score estimation. We use a probit model to
estimate the propensity that a borrower will borrow from non-bank lead arrangers. Leverage is the borrower’s book leverage
ratio at the end of fiscal year prior to the current loan, estimated as total liabilities net of cash divided by the book value of
total assets; Industry-adjusted ROA is the borrower's ROA in excess of the median of the corresponding two-digit SIC industry
ROA at the end of fiscal year prior to the current loan. LN_Cash is the natural logarithm borrower’s cash at the end of fiscal
year prior to the current loan; LN_TA is the natural logarithm borrowers’ total assets at the end of fiscal year prior to the
current loan. LN_DLTT is the natural logarithm borrowers’ long-term debt at the end of fiscal year prior to the current loan.
***, **, and * correspond to statistical significance at the 1%, 5%, and 10% level, respectively.
(1) (2) (3) (4) (5) (6)

Variable: US_nonbank Coefficient Coefficient Coefficient Coefficient Coefficient Coefficient
(z-statistic) (z-statistic) (z-statistic) (z-statistic) (z-statistic) (z-statistic)
.0067*** 0.0062***
Leverage (%)
(11.12) (7.95)
0.097*** 0.0995***
LN_Cash
(11.99) (8.01)
0.126*** -0.0309
LN_TA
(11.81) (0.224)
0.124*** 0.0693***
LN_DLTT
(14.06) (4.04)
0.019 -0.1825***
Industry-adjusted ROA
(1.38) (-2.45)
Obs. 6,744 6,764 6,795 6,468 6,752 6,261
Log Likelihood -3,965 -3977.4891 -3988.7147 -3769.838 -4031.3067 -3575.3443
Table 4: Test for Balance
This table presents the test of balance property, before and after the matching in the sample of leveraged loan borrowers. For
each observation in the treatment group, we find a control observation using the nearest neighbour method. Leverage is the
borrower’s book leverage ratio at the end of fiscal year prior to the current loan, estimated as total liabilities net of cash divided
by the book value of total assets; Industry-adjusted ROA is the borrower's ROA in excess of the median of the corresponding
two-digit SIC industry ROA at the end of fiscal year prior to the current loan. LN_Cash is the natural logarithm borrower’s
cash at the end of fiscal year prior to the current loan; LN_TA is the natural logarithm borrowers’ total assets at the end of
fiscal year prior to the current loan. LN_DLTT is the natural logarithm borrowers’ long-term debt at the end of fiscal year
prior to the current loan. ***, **, and * correspond to statistical significance at the 1%, 5%, and 10% level, respectively.
Unmatched Sample Matched Sample

Variable Nonbank Bank Pre-Match (% bias) Nonbank Bank After-Match (%bias)
Leverage 66.588 58.954 30.8*** 66.547 66.177 1.5
Industry-adjusted ROA -0.08032 -0.08192 1 -0.08044 -0.08574 3.4
LN_Cash 4.668 3.9143 37.8*** 4.6664 4.6972 -1.5
LN_TA 7.892 7.3506 38.2*** 7.8916 7.8889 0.2
LN_DLTT 6.812 5.9894 42.2*** 6.8107 6.8019 0.5

Table 5: Difference-in-differences - PSM results
This table presents the regression results of Equation (1) based on the matched sample through PSM. The dependent variable
is the AISD (all-in-spread-drawn), and the analysis is conducted at the loan facility level. The coefficient of the interaction
term linking Leverage, Nonbank and Post suggests that leverage risk premium of nonbank facilities is underestimated after
2014 and the underestimation is highly significant for both term loan and revolving credit facilities. All specifications include
facility-purpose fixed effects, industry fixed effects, and year fixed effects. Standard errors are double-clustered by both firm
and year. ***, **, and * correspond to statistical significance at the 1%, 5%, and 10% level, respectively.
Sample
Dependent var.=AISD

-1.073*** -1.505*** -0.764**

Leverage *Nonbank* Post
(-2.75) (-2.87) (-2.05)
0.773*** 0.584*** 0.590***
Leverage *Nonbank (6.56) (3.85) (4.99)
0.136 0.942** -0.108
Leverage*Post (0.52) (2.20) (-0.62)
61.15** 86.94*** 41.34*
Nonbank*Post (2.48) (2.66) (1.74)
0.296* 0.191 0.245*
Leverage (1.68) (0.76) (1.74)
-10.86*** -17.10*** -27.81***
LN_Amount (-3.57) (-3.30) (-9.97)
-30.38*** -20.77 -44.09***
LN_Maturity (-3.85) (-1.53) (-4.22)
-70.61* -78.24*** -20.56***
Performance (-10.57) (-8.46) (-3.35)
48.82*** 91.29*** 2.829
Secured (7.57) (8.87) (0.53)
-10.84* -16.14* 7.320
Covenant (-1.7) (-1.89) (1.27)
-18.32*** -28.77*** -1.369
LN_TA (-5.85) (-6.35) (-0.53)
-198.3*** -94.47***
Ind_adj ROA -167.3***
(-3.61) (-3.79)
(-4.80)
Year FE Yes Yes Yes
Obs 3,243 1,702 1,446
Adj R^2 0.2735 0.2987 0.38

Table 6: Covenant-lite loans and Leverage risk premium.
This table presents the regression results on coefficient estimates with subsamples of covenant – lite leveraged loans and loans
with covenant provision. The dependent variable is AISD; numbers in parentheses are t-stats. Panel A reports the coefficient
estimates for the subsample of covenant – lite leveraged loans, and Panel B reports the estimation results for the subsample of
leveraged loan with covenant provision. We also controlled loan facility characteristics (LN_Amount, Performance, Secured,
Covenant) and borrower characteristics (Leverage, LN_TA, Ind_adj ROA). All specifications include tranche-purpose fixed effects,
industry fixed effects, and year fixed effects. Standard errors are double-clustered by both firm and year. ***, **, and *
correspond to statistical significance at the 1%, 5%, and 10% level, respectively.

Sample
A. Covenant-lite
Leverage *Nonbank* Post -1.668*** -2.140*** -1.420***

(-3.27) (-3.27) (-2.99)
(4.78) (2.93) (3.36)

Leverage*Post 0.180 1.153** -0.0379
(0.49) (2.20) (-0.15)
Nonbank* Post 91.60*** 122.5*** 80.52**
(2.88) (3.08) (2.58)
Loan Facility Controls Yes Yes Yes
Borrower Characteristics Controls Yes Yes Yes
Year FE Yes Yes Yes

Obs 1774 986 715
Adj R^2 0.2695 0.3054 0.3623
B. with Covenant
Leverage_1 *Nonbank* Post 0.0192 -0.810 0.492

(-0.79) (-1.10) (-0.3)
Leverage *Nonbank 0.743*** 0.527** 0.643***

(4.63) (2.55) (3.93)
Leverage*Post 0.117 0.952 -0.149
(0.33) (1.34) (-0.65)
Nonbank* Post -3.863 42.66 -37.76
(-0.09) (0.73) (-1.04)

Year FE Yes Yes Yes

Obs 1469 716 731
Adj R^2 0.2900 0.3094 0.4072
Table 7: Performance pricing and Leverage risk premium
This table presents the regression results on coefficient estimates with subsamples of covenant-lite leveraged loans and loans
with covenant provision. The dependent variable is AISD; numbers in parentheses are t-stats. Panel A reports the coefficient
estimates for the subsample of covenant – lite leveraged loans, and Panel B reports the estimation results for the subsample of
leveraged loan with covenant provision. We also controlled loan facility characteristics (LN_Amount, Performance, Secured,
Covenant) and borrower characteristics (Leverage, LN_TA, Ind_adj ROA). All specifications include tranche-purpose fixed effects,
industry fixed effects, and year fixed effects. Standard errors are double-clustered by both firm and year. ***, **, and *
correspond to statistical significance at the 1%, 5%, and 10% level, respectively.

Sample
Panel A. without Performance Pricing
Provision
Leverage *Nonbank* Post -1.233*** -1.751*** -0.869**
(-3.86) (-3.05) (-1.91)

(5.50) (3.34) (3.55)
Leverage*Post 0.211 1.209*** -0.290
(0.68) (2.65) (-1.38)

Nonbank* Post 71.69*** 99.74*** 56.51**
(2.61) (2.88) (2.03)
Year FE Yes Yes Yes
Obs 2600 1463 1054
Adj R^2 0.2682 0.2916 0.3908

Panel B. with Performance Pricing Provision
Leverage_1 *Nonbank* Post -0.0844 0.218 -0.373

(-1.60) (-1.57) (-0.97)
Leverage *Nonbank 0.537*** 0.395 0.657***
(3.27) (1.49) (3.61)

Leverage*Post -0.268 -1.077* 0.438*
(-0.90) (-1.66) (1.73)
Nonbank* Post -3.678 -4.925 -13.98
(-0.09) (-0.06) (-0.47)

Year FE Yes Yes Yes

Obs 643 239 392
Adj R^2 0.2800 0.2694 0.3646
Table 8: CLO issuance and Leverage risk premium.
This table presents coefficient estimates from OLS regressions linking the leverage risk, nonbank and CLO issuance.
Definitions of all variables are provided in Appendix A. The dependent variable is AISD (bps). The coefficient of interaction
term Leverage_1*nonbank*CLO denotes if the leverage risk premium of nonbank facilities is impacted by CLO issuance. All
specifications include tranche-purpose fixed effects, industry fixed effects, and year fixed effects. Standard errors are double-
clustered by both firm and year. ***, **, and * correspond to statistical significance at the 1%, 5%, and 10% level, respectively.

Sample
Leverage* Nonbank*CLO -0.0154*** -0.0183*** -0.0108***
(-4.36) (-3.84) (-3.56)
Leverage * Nonbank 0.972*** 0.872*** 0.860***

(5.02) (3.23) (4.71)
Leverage*CLO 0.00438 0.0101** -0.00124
(1.39) (2.18) (-0.45)

Nonbank*CLO 0.861*** 0.988*** 0.460***
(4.50) (3.86) (2.80)
Year FE Yes Yes Yes
Obs 3243 1702 1446
Adj R^2 0.2793 0.3034 0.3863

Appendix
Appendix Figure 1:
U.S. CLOs Outstanding and issuance (in $B) and Covenant-lite share of outstanding, U.S. leveraged loans from 2007-2019.
OA1: Variable definitions and data sources
Variable Definition Source

AISD (bps) Basis point spread over LIBOR plus the annual fee and the up-front fee
spread Dealscan and Eikon
Nonbank Nonbank is a dummy variable that euqals to one if one facility has at Dealscan
least one U.S nonbanks lead arranger, and zero otherwise.
LN_Amount Natural log of the facility size. Dealscan and Eikon
LN_Maturity Natural log of the maturity of the facility in months Dealscan and Eikon
Secured indicator An indicator variable that takes a value of one if the facility is secured,
and zero otherwise. Dealscan and Eikon
Performance pricing indicator An indicator variable that takes a value of one if the facility has
performance pricing features, and zero otherwise. Dealscan and Eikon
Covenants indicator An indicator variable that takes a value of one if the loan has covenants,
and zero otherwise. Dealscan and Eikon
Nr.Covenants The number of financial covenants in the loan package Dealscan and Eikon
Short-term interest rates forecast refers to projected values of three- OECD Economic
Interest Rate Forecast
month money market rates in percentage. Outlook
Natural log of the total assets of the borrower at the end of fiscal year Compustat
LN_TA
prior to the current loan.
Leverage_1 The borrower's book leverage ratio at the end of fiscal year prior to the
current loan, calculated as ((LT – CHE)/TA Compustat
Leverage_2 The borrower's book leverage ratio at the end of fiscal year prior to the Compustat
current loan, calculated as DT/ TA
Leverage_3 The borrower's book leverage ratio at the end of fiscal year prior to the Compustat
current loan, calculated as LT debt/ TA
The borrower's ROA in excess of the median of the corresponding two-

Industry-adjusted ROA
digit SIC industry ROA at the end of fiscal year prior to the current Compustat
loan.
Enterprise value multiple The borrower’s enterprise value to EBITDA at the end of fiscal year
prior to the current loan. Compustat
The borrower’s long-term debt at the end of fiscal year prior to the
Long-term debt
current loan. Compustat
The borrower's total debt at the end of fiscal year prior to the current Compustat
Total debt
loan.
The borrower's total asset at the end of fiscal year prior to the current Compustat
Total asset
loan.
OA2: Descriptive statistics of nonbank and bank group prior to and after the introduction of IGLL.
This table represent the descriptive statistics of nonbank and bank group prior to and after the introduction of IGLL. Panel
compares the constitute of facility types in nonbank and bank group prior to and after the introduction of IGLL (% Revolver
and %Term loan). Panel B reports the difference of facility characteristics before and after the introduction of IGLL (AISD,
Loan Amount, Maturity, Secured indicator, Performance pricing indicator, Covenants indicator, Nr.Covenants). Panel C
reports the borrower characteristics at the end of fiscal year prior to the current loan (Total asset, Total debt, Long-term debt,
Leverage, Leverage_2, Leverage_3). T-test difference in means prior and after IGLL in both groups is also reported. ***, **,
* indicate statistical significance at 1%, 5% and 10% respectively.
Nonbank Bank
Sample
Pre 2014 FAQ After 2014 FAQ Diff (After-Pre) Pre 2014 FAQ After 2014 FAQ Diff (After-Pre)
Variables Mean Mean Mean t-statistics Mean Mean Mean t-statistics
Panel A: Facility type
% Revolver 0.38 0.341 -0.04* -1.689 0.55 0.528 -0.02 -1.31
% Term loan 0.57 0.63 0.06*** 2.78 0.42 0.44 0.02 1.31
Panel B: Facility
Characteristics
AISD 372.13 312.28 -59.85*** -7.9388 300.91 263.11 -37.80*** -9.623
Loan Amount ($ million) 449.67 790.20 340.53*** 9.45 344.35 510.62 166.27*** 9.72
Maturity 60.93 62.959 2.03*** 2.36 54.23 56.75 2.52*** 4.58
Secured indicator 0.84 0.785 -0.05*** -3.046 0.72 0.63 -0.09*** -6.751
Performance pricing indicator 0.26 0.12 -0.14*** -7.988 0.29 0.13 -0.15*** -13.043
Covenants indicator 0.49 0.34 -0.15*** -6.6852 0.50 0.43 -0.07*** -4.8777
Nr.Covenants 1.02 0.582 -0.44*** -8.8401 1.09 0.82 -0.27*** -8.209
Panel C: Borrower
Characteristics
Total asset ($M) 4,774.301 8,754.464 3,980.16*** 7.868 3,244.92 4,525.22 1,280.30*** 5.9773
Total debt ($M) 2,079.14 3,852.142 1,773.00*** 8.019 1,220.05 1,734.10 514.06*** 5.92
Long-term debt ($M) 1,971.46 3,716.90 1,745.45*** 8.404 1,184.41 1,654.56 470.14*** -5.64
Leverage (%) 0.66 0.64 -0.02* -1.727 0.58 0.56 -0.02** -2
Leverage_2(%) 0.44 0.438 -0.01 -0.4965 0.37 0.36 -0.01 -0.9652
Leverage_3(%) 0.41 0.42 0.00 0.29 0.34 0.34 0.00 0.1255
QA3: A list of contemporaneous bank regulation events.
Year Name Description
A supervisory guidance on leveraged lending, which applies to all national banks,

federal savings associations, and federal branches and agencies of foreign banks
22/03/2013 Leveraged Lending: Guidance on Leveraged Lending
(collectively, banks). This guidance was published in the Federal Register on March
22, 2013, and replaces similar guidance issued in April 2001 (2001 guidance).
A bulletin to clarify certain provisions of the market risk capital rule.1 This
Market Risk Capital Rule: Clarification of the Treatment
10/05/2013 clarification is applicable only to those institutions supervised by the OCC that are
of Certain Sovereign and Securitization Positions
subject to that rule.
Troubled Debt Restructurings: Guidance on Certain Issues A supervisory guidance on certain issues related to commercial and residential real
24/10/2013
Related to Troubled Debt Restructurings estate loans that have undergone troubled debt restructurings (TDRs).
This guidance replaces the previously issued OCC Bulletin 2004-25, “Uniform
Agreement on the Classification of Securities” (2004 Agreement) by applying the
29/10/2013 Classification of Securities: Interagency Guidance
agencies’ revised investment grade standards of creditworthiness, in place of credit
ratings, as the basis for classifying investment securities.
This bulletin provides guidance to national banks and federal savings associations
(collectively, banks) for assessing and managing risks associated with third-party
30/10/2013 Third-Party Relationships: Risk Management Guidance
relationships. A third-party relationship is any business arrangement between a bank
and another entity, by contract or otherwise.1
A guidance that the Office of the Comptroller of the Currency (OCC) uses when it
Use and Review of Independent Consultants in requires national banks, federal savings associations, or federal branches or agencies
12/11/2013
Enforcement Actions: Guidance for Bankers (collectively, banks) to employ independent consultants as part of an enforcement
action to address significant violations of law, fraud, or harm to consumers.
This guidance helps financial institutions identify potential risks in the social media to
Social Media - Consumer Compliance Risk Management
17/12/2013 ensure they are aware of their responsibilities to address these risks within their overall
Guidance: Final Supervisory Guidance
risk management program.
Risk Management of Home Equity Lines of Credit

A supervisory guidance on risk management practices for home equity lines of credit
01/07/2014 Approaching the End-of-Draw Periods: Interagency
(HELOC) approaching the end-of-draw (EOD) period.
Guidance
A guidance from the Office of the Comptroller of the Currency (OCC) to national
banks and federal savings associations (collectively, banks) on the application of
04/08/2014 Consumer Debt Sales: Risk Management Guidance consumer protection requirements and safe and sound banking practices to consumer
debt-sale arrangements with third parties (e.g., debt buyers) that intend to pursue
collection of the underlying obligations.
Leveraged Lending: Frequently Asked Questions for FAQ is designed to foster industry and examiner understanding of the 2013 guidance
07/11/2014 Implementing March 2013 Interagency Guidance on and to promote consistent application of the guidance in policy formulation,
Leveraged Lending implementation, and regulatory supervisory assessments.
Source: Office of the Comptroller of the Currency (OCC) website

OA4: Statistically difference between different subsamples
This table presents the regression results on the statistical difference between the coefficient in loans with covenant-lite and in
loans without covenants (Panel A) and the statistically difference between the coefficient in loans with performance pricing
and in loans without performance pricing (Panel B). We also include main effects of each quadruple interaction terms in our
regressions. In addition, we also control the loan facility characteristics and borrower characteristics. Year fixed effects,
industry fixed effects and purpose fixed effects are also included in all of specifications. Standard errors are double-clustered
by both firm and year. ***, **, and * correspond to statistical significance at the 1%, 5%, and 10% level, respectively.
Panel A:
Sample
Dependent var.=AISD

Leverage *Nonbank* Post*Covenant- -2.378*** -2.075*** -1.990***

lite (-4.64) (-3.07) (-3.97)
0.687*** 0.652** 0.532***
(3.52) (2.54) (2.99)
0.841*** 0.693*** 0.503***
Leverage *Nonbank* Covenant-lite
(5.13) (3.32) (3.04)
0.138 0.208 -0.0832
Leverage* Post* Covenant-lite
(0.65) (0.67) (-0.52)
94.06*** 69.76** 88.62***
Nonbank* Post* Covenant-lite
(3.41) (2.03) (2.95)
Year FE Yes Yes Yes
Obs 3243 1702 1446
Adj R^2 0.2697 0.297 0.374
Panel B:
Sample
Dependent var.=AISD

Leverage *Nonbank* Post*without-PP -1.391*** -1.210** -1.123**

0.339 0.568* 0.118
(1.59) (1.86) (0.55)
0.751*** 0.465*** 0.539***
Leverage *Nonbank*without-PP
(5.48) (2.77) (3.67)
-0.210 -0.243 -0.289**
Leverage* Post*without-PP
(-1.22) (-1.03) (-2.03)
65.89*** 41.99 65.98**
Nonbank* Post*without-PP
(2.60) (1.36) (2.38)
Year FE Yes Yes Yes
Obs 3243 1702 1446
Adj R^2 0.2718 0.2945 0.3773
OA5: Robust check: is the decline in leverage risk premium driven by low interest rate expectation?
This table presents the regression results on coefficient estimates with an addition control variable of projected short-term
interest rate. We also controlled loan facility characteristics (LN_Amount, Performance, Secured, Covenant) and borrower
characteristics (Leverage, LN_TA, Ind_adj ROA). The dependent variable is AISD; numbers in parentheses are t-stats. All
specifications include tranche-purpose fixed effects, industry fixed effects, and year fixed effects. Standard errors are double-
clustered by both firm and year. ***, **, and * correspond to statistical significance at the 1%, 5%, and 10% level, respectively.
Sample

-1.078*** -1.509*** -0.772**

(-2.77) (-2.90) (-2.06)
0.773*** 0.584*** 0.591***
Leverage *Nonbank
(6.55) (3.84) (4.98)
0.137 0.942** -0.107
Leverage*Post
(0.53) (2.21) (-0.62)
61.46** 87.26*** 41.84*
Nonbank*Post
(2.50) (2.68) (1.76)
0.297* 0.192 0.245*
Leverage
(1.68) (0.76) (1.74)
-4.104 -3.803 -6.202
Short-term interest rates
(-0.23) (-0.15) (-0.37)
-10.87*** -17.11*** -27.80***
LN_Amount
(-3.57) (-3.29) (-9.96)
-30.42*** -20.79 -44.24***
LN_Maturity
(-3.85) (-1.53) (-4.23)
-70.65*** -78.23*** -20.62***
Performance
(-10.58) (-8.46) (-3.36)
48.79*** 91.24*** 2.853
Secured
(7.56) (8.86) (0.54)
-10.82* -16.09* 7.268
Covenant
(-1.70) (-1.88) (1.26)
-18.33*** -28.77*** -1.401
LN_TA
(-5.84) (-6.35) (-0.54)
-167.4*** -198.5*** -94.43***
Ind_adj ROA
(-4.81) (-3.63) (-3.79)
Year FE Yes Yes Yes

Obs 3243 1702 1446
Adj R^2 0.2733 0.2983 0.3796
OA6: Robust check: is the decline of leverage risk driven by borrower high growth potential?
This table presents the regression results on coefficient estimates with subsamples of leveraged loans for borrowers with high
growth potential (Panel A) and leveraged loans for borrowers with low growth potential (Panel B). The dependent variable is
AISD; numbers in parentheses are t-stats. We also controlled loan facility characteristics (LN_Amount, Performance, Secured,
Covenant) and borrower characteristics (Leverage, LN_TA, Ind_adj ROA). All specifications include tranche-purpose fixed
effects, industry fixed effects, and year fixed effects. Standard errors are double-clustered by both firm and year. ***, **, and
* correspond to statistical significance at the 1%, 5%, and 10% level, respectively.
Sample Leveraged Loans Term Loans Revolvers

Panel A: Loan Borrowers with high growth potential
-0.421 -0.645 0.0537

(-0.89) (-1.00) (0.12)
Leverage *Nonbank 0.699*** 0.534** 0.494***

(4.23) (2.48) (3.22)
Leverage* Post -0.111 0.522 -0.114

(-0.33) (0.93) (-0.48)
25.85 53.19 -4.646

Nonbank* Post
(0.85) (1.27) (-0.17)
Year FE Yes Yes Yes
Obs 1515 771 700
Adj R^2 0.267 0.334 0.393
Panel B: Loan Borrowers with low growth potential
-1.596*** -1.859** -1.886***

Leverage*Nonbank* Post
(-2.64) (-2.40) (-3.18)
Leverage*Nonbank 0.832*** 0.517** 0.750***

(5.04) (2.54) (3.95)
Leverage* Post 0.369 1.079* -0.175

(0.99) (1.88) (-0.69)
Nonbank* Post 90.38** 105.6** 96.12***

(2.38) (2.18) (2.59)

Year FE Yes Yes Yes
Obs 1728 931 746
Adj R^2 0.289 0.309 0.379

OA7: Robust check: alternative proxies on leverage risk
This table presents the main regression results on coefficient estimates with subsamples of leveraged loans based on two
alternative leverage risk measures. The dependent variable is AISD; numbers in parentheses are t-stats. We also controlled
loan facility characteristics (LN_Amount, Performance, Secured, Covenant) and borrower characteristics (Leverage, LN_TA,
Ind_adj ROA). All specifications include tranche-purpose fixed effects, industry fixed effects, and year fixed effects. Standard
errors are double-clustered by both firm and year. ***, **, and * correspond to statistical significance at the 1%, 5%, and 10%
level, respectively.
Sample Leveraged Loans Term Loans Revolvers

Dependent var.=AISD
Leverage_2 *Nonbank* Post -1.786*** -2.118*** -1.156**

(-3.90) (-3.56) (-2.57)
Leverage_2 *Nonbank 1.057*** 0.812*** 0.854***

(6.79) (4.18) (5.43)
Leverage_2 *Post 0.714** 1.606*** 0.191

(2.34) (3.29) (0.88)
Nonbank* Post 74.95*** 86.78*** 42.36**

(3.64) (3.25) (2.13)
Leverage_3*Nonbank*Post -1.869*** -2.080*** -1.272***
(-3.94) (-3.34) (-2.76)
Leverage_3*Nonbank 1.106*** 0.811*** 0.890***

(6.86) (3.99) (5.74)
Leverage_3 *Post 0.701** 1.500*** 0.262

(2.17) (2.88) (1.18)
80.69***
Nonbank* Post 72.64*** 45.07**
(3.02)
(3.52) (2.25)
Year FE Yes Yes Yes
Obs 3172 3243 1664 1702 1418 1446
Adj R^2 0.2708 0.2690 0.2979 0.2949 0.3805 0.3785

Leveraged Loan Reading

Uploaded by

Copyright:

Available Formats

Leveraged Loan Reading

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Leveraged Loan Reading

Uploaded by

Copyright:

Available Formats

Leveraged Loans: Is High Leverage Risk Priced in?

School of Management, University of Bath

University of Zurich, Swiss Finance Institute, KU Leuven and CEPR

School of Management, University of Bath

School of Management, University of Bath

First draft: 21 December 2020

This draft: 21 November 2021

JEL classification: G21, D82, G34

years 2. Prompted by an increase in participation of unregulated investors and a deterioration in

Corporation (FDIC) issued the so-called Interagency Guidance on Leveraged Lending in

bank-originated leveraged lending activity, it also triggered a migration of leveraged lending

Treasury Janet L. Yellen raises concern on the vulnerabilities in nonbank financial

compare the all-in-spread-drawn (AISD) between bank-originated leveraged loan, and

priced in the AISD post the regulation.

Our primary empirical analysis compares the all-in-spread-drawn (AISD) between

since the issuance of the 2014 FAQ.

As a type of syndicated loan, leveraged loan suffers information asymmetry issues by

Relaxing investor protection in covenant-lite loans led by competition between banks

pricing. To investigate the role of information asymmetry on leveraged loan pricing, we

Clarification is strongly linked to the decline of leverage risk premium in nonbank-originated

leveraged loans from 2014 – 2019.

from 2014 – 2019.

background. Section 3 provides empirical evidence on whether leveraged loan pricing

checks to confirm our findings. Section 6 concludes.

2.1 Sample Construction

pricing mechanism effectively reflects borrowers’ high leverage ratios.

secured, covenant-lite and has performance-based pricing.

To obtain borrower-specific characteristics, we match the borrower and/or borrower’s

DealScan and Compustat. We exclude observations with missing borrowers’ financial

2.2 Definition of Nonbank and Bank Lenders

Following Elliott et al., (2019) we identify a lender as a nonbank if it is categorized as

“Insurance Company”, “Corporation”, “Finance Company”, “Investment Bank”, “Mutual

participant lenders (Ivashina, 2009). Therefore, we define a nonbank-originated leveraged loan

2.3 Overview of Sample

change or be influenced by the post-clarification period, leading to endogeneity problems. We

< INSERT TABLE 1 HERE >

3.1 Baseline results

investigating following empirical model:

leverage ratio is associated with an increase of 29% of AISD in nonbank-originated facilities

on the triple interaction term of Leverage_1*Nonbank*Post for a borrower with average

increase in leverage ratio only results in a 9% increase in AISD (12%-20%+17%) driven by a

a 1-standard-deviation increase in leverage ratio is associated with an increase of 21.3% of

Dec 2019 period, a 1-standard-deviation increase in leverage ratio is associated with a 3%

increase in AISD for term loans (10%-21%+11.3%).

Leverage_1*Nonbank*Post remain negative and significant at the 1% level in all estimates.

< INSERT TABLE 2 HERE >

difference-difference (diff-in diff) setup.

<INSERT Figure 2 HERE>

3.2. Propensity score matching-difference-in-difference

matching model is estimated as follow:

we implement the nearest neighbour matching to construct the counterfactual outcome. We

< INSERT TABLE 3 HERE >

< INSERT TABLE 4 HERE >

< INSERT TABLE 5 HERE >

3.3. Predictive margin on nonbank loans V.S. bank loans

<INSERT Figure 3 HERE>

4. Investigating the channels: information asymmetry

4.1 Covenant-lite loans and loan spread

< INSERT Figure 4 HERE >

on the triple interaction term of Leverage_1NonbankPost for a borrower with average

Leverage_1NonbankPost remain negative and significant at the 1% level in all estimates.

Leverage_1Nonbank Post in all the estimations become insignificant, indicating no clear

coefficients of Leverage_1NonbankPost become insignificant in all of estimations,

coefficient on the interaction term of Leverage_1NonbankCLO is negative and highly

interaction term of Leverage_1NonbankCLO and AISD in both subsamples of term loans