Boyarchenko, Kovner, Shachar, It's What You Say and What You Buy: A Holistic Evaluation of the Corporate Credit Facilities (NY Fed Staff Report 935 (revised Jan 2022); J. Financial Economics 2022, 2022-01)
Summary
Federal Reserve Bank of New York Staff Reports, no. 935, dated July 2020 and revised January 2022, by Nina Boyarchenko, Anna Kovner and Or Shachar. The paper evaluates the impact of the Federal Reserve corporate credit facilities (PMCCF and SMCCF) on corporate bond markets. The abstract says conditions in primary markets improve once the facilities are announced, particularly for issuers that need to refinance before 2022. It reports a differential impact on eligible issues in the secondary market and a significant effect of direct bond purchases, but less so for purchases through ETFs. The authors say they find evidence that dealers link the primary and secondary market recovery.
Summary drafted by a model from the document's text below and checked by script against that text before publication. It is a navigation aid, not a reading of what the document proves. Where AI is used
Cite this document
“Boyarchenko, Kovner, Shachar, It's What You Say and What You Buy: A Holistic Evaluation of the Corporate Credit Facilities (NY Fed Staff Report 935 (revised Jan 2022); J. Financial Economics 2022, 2022-01).”
Original at www.newyorkfed.org. Archived at Pandemic Darlings, https://pandemicdarlings.com/source-documents/boyarchenko-kovner-shachar-it-s-what-you-say-and-what-you-buy-a-holist-de24fce49a/
, SHA-256 f0ef3906d3090c09e08b91803453e023337fbf1763bacc42aeb0f033cae46bbe.Full text
NO. 935
JULY 2020 It’s What You Say and What You Buy:
A Holistic Evaluation of the Corporate
REVISED
JANUARY 2022
Credit Facilities
Nina Boyarchenko | Anna Kovner | Or Shachar
It’s What You Say and What You Buy: A Holistic Evaluation of the Corporate Credit Facilities
Nina Boyarchenko, Anna Kovner, and Or Shachar
Federal Reserve Bank of New York Staff Reports, no. 935
July 2020; revised January 2022
JEL classification: G12, G18, G19
Abstract
We evaluate the impact of the Federal Reserve corporate credit facilities (PMCCF and SMCCF) on
corporate bond markets. Conditions in primary markets improve once the facilities are announced,
particularly for issuers that need to refinance before 2022. Issuance accelerates before spreads normalize.
The secondary market points to a causal role for the facilities, with a differential impact on eligible issues
and a significant effect of direct bond purchases, but less so for purchases through ETFs. We find
evidence that dealers link the primary and secondary market recovery, with facilities affecting dealers’
willingness to underwrite issuances and intermediate in secondary markets.
Key words: corporate credit facilities, bond liquidity, credit spreads, purchase effects
_________________
Boyarchenko, Kovner, Shachar: Federal Reserve Bank of New York (emails:
nina.boyarchenko@ny.frb.org, anna.kovner@ny.frb.org, or.shachar@ny.frb.org). Zachary Kuloszewski,
Beryl Larson and Hunter Wieman provided excellent research assistance. The authors thank Darrell
Duffie, Andrew Haughwout, Zhiguo He, Joshua Frost, Ralph Koijen, Annette Vissing-Jorgensen, and
audiences at the SaMMF Workshop: Liquidity in Fixed-Income Markets, John Hopkins Carey, WE-ARE,
UW Madison, NBER Summer Institute, and Uni.LU-ESM seminar for comments on previous drafts of
the paper. A special thanks to Alié Diagne from FINRA for clarifications about Corporate TRACE
reporting.
This paper presents preliminary findings and is being distributed to economists and other interested
readers solely to stimulate discussion and elicit comments. The views expressed in this paper are those of
the author(s) and do not necessarily reflect the position of the Federal Reserve Bank of New York or the
Federal Reserve System. Any errors or omissions are the responsibility of the author(s).
To view the authors’ disclosure statements, visit
https://www.newyorkfed.org/research/staff_reports/sr935.html.
1 Introduction
The corporate bond market experienced historic turmoil in March 2020. As investors shed
risky assets in response to the COVID-19 pandemic and associated shutdowns, U.S. investment-
grade corporate bond issuance slowed to levels not seen since the global financial crisis. On
March 23, 2020, as part of an extensive set of measures to support the U.S. economy, the
Federal Reserve announced its first ever corporate credit facilities (CCFs) in order to support
the supply of capital market credit to the non-financial sector. The facilities were designed
with a two-pronged approach, facilitating access to primary markets through direct lending
in the Primary Market Corporate Credit Facility (PMCCF) and acting in secondary mar-
kets through purchases of individual bonds and exchange-traded funds (ETFs) through the
Secondary Market Corporate Credit Facility (SMCCF).
In this paper, we document improvements in access to primary credit markets for non-
financial corporations and show how they arise from facilities-related improvements in sec-
ondary market conditions and underwriter and dealer behavior. Figure 1 illustrates the
pandemic-related slowdown and post-facility recovery of primary bond markets, comparing
patterns of short maturity investment grade bonds in 2020 to the previous decade’s histori-
cal median. An anemic market for shorter maturity investment grade bonds accelerated only
following the announcement of the facilities. By the end of June, investment grade issuance
was more than triple issuance by the same point in 2019. However, primary market pricing
took much longer to normalize.
Simple issuance statistics mask substantial variation in the types of issuers. Adding issuer
controls, we find facility-eligible (investment grade) issuers, even those without debt coming
due, disproportionately access primary markets after the announcement, with a four-fold
increase in the probability of issuance during the facility implementation period between
March and June (announcement date through initiation of direct bond market purchases).
While the probability of issuance by investment grade firms returns to pre-pandemic levels
1
Figure 1. Announcement of facilities restarts primary market for credit. This fig-
ure plots the cumulative amount of fixed coupon bonds with less than 5 year maturity issued
by investment grade non-financial issuers in 2020 and the median issuance from 2010 to 2019,
as well as the weighted average offering spread for investment grade issues. Offering spread
computed relative to nearest-maturity on-the-run Treasury yield. Dashed line at February
22 (start of pandemic onset period). Event lines at March 22 (initial CCF announcement)
and June 29 (PMCCF operational).
(a) Cumulative issuance (b) Offering spread
Cumulative amount issued ($ Billion)
60
Median 2010-2019 2020 300 Median 2010-2019 2020
50
Offering spread (bps)
250
40
200
30
150
20 100
10 50
0 1
2
3
4
5
6 0 1
2
3
4
5
6
7
8
9
10
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27 11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47 31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
48
49
50
Week of year
51
Week of year
after the facility is operational, the probability of issuance by high yield firms is actually
elevated relative to pre-pandemic levels, highlighting the slower dissemination of issuance
opportunities to riskier borrowers.
Primary market prices take longer to recover. After the facility announcement, offering
spreads for borrowers with refinancing needs remain substantially higher – up to 375 bps
more for high yield issuers – than before the pandemic onset, even controlling for the default
risk of the issuer. Eligible issuers disproportionately benefit after the announcement with
offering spreads that increased much less than did those of high yield issuers, although
some of that appears to reflect a differential relationship between default probabilities and
offering spreads post pandemic onset. After the facility is operational, the offering spreads
for eligible borrowers return to their pre-pandemic levels, while offering spreads for high yield
issuers remain elevated. This highlights the slow pace of normalization of primary market
conditions, especially for issuers not eligible for the facilities. Compensation earned by dealers
is consistent with this – Gross spreads paid to underwriters do not disproportionately fall
for eligible issuers until the facility is fully operational.
Given the many official sector interventions that occur around the time of the facilities,
2
we turn to secondary market to establish more causal linkages through an event study ap-
proach with narrow time windows. Conditions in the secondary market pass-through to the
primary market in two ways: directly through the benchmarking of primary market pricing
to secondary market prices of similar bonds, and indirectly by increasing the willingness of
dealers to underwrite bond issuance and increasing the demand from long-term investors.
We estimate a dramatic improvement in average duration-matched secondary market
spreads of almost 100 basis points in the three days after the initial announcement of the
facilities with an additional differential impact of almost 70 basis points for eligible bonds.
When the facility term sheet was revised and additional information provided, spreads fell
by more than an additional 60 basis points. In order to attribute these spread improvements
to the CCFs given other announcements, we estimate differences in the spread impact on
eligible and non-eligible securities and for eligible and non-eligible issuers looking at changes
around narrow windows around the announcement event dates. We further identify the causal
impact of CCFs by using bonds of the same issuer but with different eligibility.
Facility purchases had a statistically significant impact as well, although the magnitudes
are much smaller – spreads on bonds bought directly by the facility improved differentially by
6 basis points around each purchase date. The impact of direct cash bond purchases appears
to be much higher than that of purchases through ETFs. This might not be surprising given
that the empirical evidence on the transmission between corporate bond ETFs and the cash
market focuses mainly on the impact of ETFs’ redemptions on the cash market during ETF
stress (e.g., Dannhauser and Hoseinzade, 2022, and Falato et al., 2021), and the extent of
transmission under less stressed market conditions (such as those present in May and June
2020) remains an open question. When there is news about the closure of the facilities, we
estimate a small (2 to 6 basis points) spread increase for eligible issues, driven mostly by the
price of risk. We do not consistently find statistically significant impacts of exit on ineligible
bonds, although we find some evidence that the announcement resulted in a repricing within
issuers of eligible and ineligible bonds.
3
We delve into the possible reasons for the impact of purchases and conclude that dealer
behavior matters, consistent with O’Hara and Zhou (2021)’s results around the initial facil-
ities announcement. In the pandemic onset, dealers were less likely to buy SMCCF-eligible
collateral, perhaps as their customers’ sell-off of lower credit quality assets accelerated. We
add to O’Hara and Zhou (2021) by looking at dealers’ eligibility to trade with the SMCCF
once the facility is operational. We find evidence of differential changes for dealers that were
eligible to trade with the facilities, suggesting that the actual trades of the facility were
necessary to change participation. Eligible dealers were both more likely to buy and bought
significantly more dollars of bonds from their customers, especially of bonds that have been
announced to be eligible for the facility. The magnitude of the impact is economically sig-
nificant – 7-8 percent more, even after controlling for fixed differences in average amounts
purchased by dealers and of given issuers.
This eligible dealer response to purchases suggests that dealers’ willingness to underwrite
and intermediate may have played a role in the slower speed of primary market recovery.
We document that dealers not affiliated with U.S. bank holding companies required a higher
offering day trading profit from underwriting SMCCF-eligible securities during both the pan-
demic onset and the facility implementation period, earning greater compensation for bearing
underwriting risk during this period. For U.S. bank-affiliated dealers, instead, offering-day
underwriting profits increase for securities not eligible for direct SMCCF purchases, suggest-
ing that these underwriters are particularly reliant on well functioning secondary markets.
Moreover, the average time-to-profitability for U.S. bank-affiliated underwriters increases
during the pandemic onset period especially for SMCCF-eligible securities, and only goes
back to pre-pandemic levels after the facility becomes operational.
D’Amico et al. (2020), Kargar et al. (2021), Haddad et al. (2021), Nozawa and Qiu
(2021), and O’Hara and Zhou (2021) all study the disruptions in the secondary corporate
bond market and the improvement in secondary corporate bond market functioning following
the facilities announcement. To our knowledge, our paper is the first to document a pass-
4
through of U.S. secondary market interventions to improvements in the primary market
as well as the dealers’ role in that transmission. Unlike Acharya and Steffen (2020), we
find that issuance increased across the credit spectrum since the facilities announcement,
and not just for issuers at the top of the credit spectrum. We also show the key role of
dealers in improvements in the primary market, arguing that the reluctance of non-bank
dealers to underwrite securities during the spring of 2020 may have contributed to the slow
normalization of primary market spreads.
Our study adds to research on the impact of the CCFs in secondary markets along several
dimensions: First, we use granular bond-dealer-level data to study multiple dimensions of the
functioning of both primary and secondary markets, and we show that the improvements are
not uniform across issuers and intermediaries. Second, we utilize the design of the facility,
in particular, the eligibility criteria for direct purchases by the secondary market facility, to
isolate the causal effect of the announcement from the overall improvements in market condi-
tions. Gilchrist et al. (2020) confirm our findings of a significant differential improvement in
secondary market spreads and liquidity for facility-eligible bonds. Third, we use the volume
of purchases by the facility, both indirectly through ETF purchases and directly through
cash bond purchases, to disentangle the effect of actual purchases. Fourth, we shed light on
the role of banks’ balance sheet constraints played in March dislocations and subsequent
recovery by studying changes in the liquidity provision by different types of dealers in both
the primary and the secondary markets.
Our paper is also related to the literature studying the impact of the European Central
Bank’s (ECB) Corporate Sector Purchase Programme (CSPP) on corporate bond markets in
the European Union. Grosse-Rueschkamp et al. (2019) and Todorov (2020) document that
the announcement on the CSPP reduces bond yields of firms with eligible bonds. (De Santis
and Zaghini, 2019) find that this differential improvement in funding and trading conditions
for eligible bonds incentivized issuers to modify characteristics of their issuance to match
eligibility criteria. Relative to this literature, we show that, although secondary market func-
5
tioning improved on the CCF announcement differentially more for facility-eligible bonds,
liquidity improvements were not localized to eligible bonds, and issuers do not seem to tailor
characteristics of newly issued bonds to facility eligibility criteria. This is perhaps not sur-
prising: while the CSPP is a monetary policy tool, in the United States, the purpose of the
CCFs is instead to improve the functioning of the private corporate bond market.
More broadly, this paper is related to the literature on the effect that intermediary con-
straints play in market liquidity provision (see e.g. Gromb and Vayanos, 2002; Brunnermeier
and Pedersen, 2009; Gromb and Vayanos, 2010). In the corporate bond market in particular,
Adrian et al. (2017) document that the relationship between balance sheet constraints and
liquidity provision in the secondary corporate bond market changes after the implementation
of post-crisis banking regulation. We contribute to this literature by measuring the extent
to which liquidity facility announcements and purchases pass through facility counterparties
to the both the primary and secondary corporate bond markets.
Finally, this paper contributes to the literature studying the value of underwriter re-
lationships in the primary market (e.g., Fang, 2005, Yasuda, 2005, Andres et al., 2014).
Consistent with existing literature (e.g., Fernando et al., 2012, Dick-Nielsen et al., 2021), we
show that when these relationships are tested in the face of distress episodes, underwriting
by dealers affiliated with U.S. BHCs is more stable than that by small dealers especially in
longer-maturity and high-yield bonds.
The paper proceeds as follows: In Section 2, we describe the facilities and related an-
nouncements. Section 3 describes the data used in this paper. In Section 4, we study the
impact of the facilities on primary market functioning, and on secondary market credit
spreads and liquidity in Section 5. Section 6 explores how changes in market intermedia-
tion lead to improvements in both primary and secondary markets. Finally, we conclude in
Section 7. Additional results and technical details can be found in the Appendix.
6
2 Corporate Credit Facilities
On March 23, 2020 the Board of Governors of the Federal Reserve System announced a
number of interventions to respond to the economic and market dislocations of the pan-
demic and related shutdowns. With respect to capital markets corporate credit, pursuant to
the Board’s authorization, the Federal Reserve Bank of New York established the Primary
Market Corporate Credit Facility (PMCCF) and the Secondary Market Corporate Credit
Facility (SMCCF).1 We summarize the key dates of announcements related to the corpo-
rate credit facilities in Appendix Tables A.1 (PMCCF timeline) and A.2 (SMCCF timeline).
The facilities were designed to work together to support market functioning for corporate
bonds and syndicated loans, with an overarching goal of facilitating credit provision to the
non-financial corporate sector of the U.S. economy. The announcement included term sheets
for both facilities that outlined key terms and applicability. Key features of the term sheets
included the following eligibility conditions for issuers: rated investment-grade by at least
one nationally recognized statistical rating organization (NRSRO) and, if rated by multiple
NRSROs, investment-grade rated by at least two of them, headquartered in the United States
and with material operations in the United States. Issuers must not receive direct financial
assistance under pending federal legislation. The SMCCF would purchase bonds up to a 5
year maturity, while the PMCCF would purchase new debt with up to a 4 year maturity. In
addition, the SMCCF announced that it would purchase eligible bond portfolios in the form
of exchange traded funds (ETFs).
At the same time the Federal Reserve announced a number of actions including: 1) pur-
chasing Treasuries and Agency securities, 2) establishing the Term Asset-Backed Securities
Both facilities were authorized by the Board under the authority of Section 13(3) of the Federal Reserve
1
Act, with approval of the Treasury. To implement these facilities, the New York Fed formed a special purpose
vehicle (SPV) and the Treasury, using funds appropriated to the Exchange Stabilization Fund (ESF) through
the CARES Act, made an equity investment in the SPV. Under the PMCCF and the SMCCF, the New York
Fed will lend to the SPV, and the SPV will use the proceeds of such loans to purchase eligible assets. The
New York Fed’s loans to the SPV will be secured by all the assets of the SPV, including the Treasury’s
equity investment in the SPV.
7
Loan Facility (TALF), 3) establishing the Primary Dealer Credit Facility (PDCF), 4) ex-
panding the Money Market Mutual Fund Liquidity Facility (MMLF) to include a wider range
of securities, including municipal variable rate demand notes (VRDNs) and bank certificates
of deposit, and 5) expanding the Commercial Paper Funding Facility (CPFF) to include
high-quality, tax-exempt commercial paper as eligible securities and reduced the pricing of
the facility.
We also identify dates on which the Federal Reserve shared additional details about key
terms or execution of the corporate credit program, focusing on statements which affect the
eligibility of certain issues or issuers. On April 9, 2020, the size of the combined facilities
became larger with an increase in Treasury capital from $10 billion to $75 billion. Updated
term sheets added concentration limits and clarified the definition of eligibility to include
firms that were rated investment-grade as of March 22, 2020 but no lower than BB- when
purchased by facility (“fallen angels”). The SMCCF also extended purchase eligibility to high
yield ETFs.2 The SMCCF began purchasing ETFs on May 12, 2020 and cash bonds on June
16, 2020. The PMCCF was launched on June 29, 2020, concurrent with an update to the
PMCCF term sheet.
On November 19, it was announced that the facilities would not be extended past the
December 31, 2020 date. The facilities were closed on December 31, 2020, and had exited
all holdings by August 31, 2021. A more in-depth discussion of the operational details and
design of the facilities can be found in Boyarchenko et al. (2021).
2
The updated term sheets also made certain changes to the eligibility requirements. They specified that
issuers must not be insured depository institutions, depository institution holding companies, or subsidiaries
of such holding companies; nor must they have received specific support pursuant to the CARES Act or any
subsequent federal legislation. In addition the term sheets added a requirement that the issuer must satisfy
the conflict of interest requirements under section 4019 of the CARES Act.
8
3 Data
3.1 Primary market issuance
We use data from Mergent FISD to measure the volume and offering terms for new bond
issuances, underwriter information and credit rating. We exclude bonds issued in foreign
currency, bonds issued as either Yankee or Canadian bonds, convertible and asset backed
bonds, as well as bonds that remain unrated more than 2 weeks after the initial offering
date. We focus on bonds issued by non-financial issuers.3 Finally, we only retain senior and
senior secured bonds issued by issuers domiciled in the U.S. We measure the volume of
issuance in dollar terms. Since the market convention is to price offering yields as spreads
to nearest-maturity on-the-run Treasury yield, our measure of primary market spreads is
the spread of the offering yield to the corresponding nearest-maturity on-the-run Treasury
yield, as described in Appendix A.4. For the subset of securities for which it is available
in Mergent FISD, we also study the gross spread on the issuance, which is the difference
between the price that the issuer receives for its securities and the price that investors pay
for it. We normalize the reported gross spread by the face value to bring the gross spread to
basis points.
In addition to characteristics from Mergent FISD, we obtain one year expected default
frequencies (EDFs) from Moody’s KMV,4 available at the bond-day level. EDFs measure the
probability of a firm’s bond experiencing a credit event (failure to make a scheduled principal
or interest payment) over the following year, constructed from a Merton (1974)-style model.
EDFs thus provide a timely measure of the credit worthiness of both the firm as a whole and
the firm’s individual bonds.
Finally, we proxy for exposure to potential primary market credit stress faced by issuers
Results for financial issuers are available on request.
3
See https://www.moodysanalytics.com/-/media/products/edf-expected-default-frequency-overview.
4
pdf.
9
due to the COVID-19 epidemic. At the issuer level, we consider firms that had corporate
bonds maturing within the next 2 years (through March 2022) as those most likely to have
refinancing needs and thus most likely to be affected by any credit shortages. Table 1, Panel
A reports the summary statistics for primary market offering terms for the full 2020 sample.
3.2 Dealers’ primary market net profits
We use data from the regulatory version of Corporate TRACE (Trade Reporting and Compli-
ance Engine) to calculate profits earned by dealers who provide underwriting services. These
data include price, uncapped trade size, trade direction, the identity of registered FINRA
members in the trade, and other trade terms. Registered FINRA dealers are identified by a
designated Market Participant Identifier, and non-FINRA members are identified either as C
(for client), or as A (for a non-member affiliate). Starting in March 2010, the TRACE data
also includes a trading market indicator that flags whether the trade is a secondary market
trade or primary market trade executed at market price (S1), or if the trade is a primary
market trade that qualifies as a List or Fixed Offering Price Transaction or a Takedown
Transaction (P1).5 In the regulatory version of TRACE cancelled and corrected records are
linked with a control number, so we keep the most up to date record of the trade. We also
address multiple reporting of interdealer trades, as well as trades that were executed through
a non-exempt Alternative Trading System as described in Appendix A.1.
The majority of plain vanilla U.S. corporate bonds are brought to the market as fixed
or list price offerings. Therefore, to calculate profits of underwriting syndicate member deal-
ers who allocate the offering among investors in the secondary market, we condition on
reportable primary market trades (P1) on the day of issuance, where the dealer is the seller.
By definition, P1 trades are expected to occur on offering date. 6 We augment the Mergent
5
Note that “S1” trades are subject to 15-minute reporting requirement, while “P1” trades are subject to
T + 1 reporting.
6
In the TRACE data, however, there are P1 trades that are reported either before or after the offering
day. Such trades are result of either a entry error in TRACE or a reference data issue. For P1 trades that
are reported after offering date, it is also possible that it is a result of an issuance that was upsized. Yet, for
10
FISD offering date with the offering date from Bloomberg, and keep P1 records if their offer-
ing date equals either the one recorded in Mergent FISD or in Bloomberg. As for entry data
errors, we exclude P1 trades where the reported price is not equal to the offering price.7
We construct two metrics of dealer compensation to underwrite new issuances: offering
day trading profits (losses) and time to first trading profit. We calculate the offering day
profit as the sum of two components. The first is the profit at the dealer-bond-level from
P1 trades that is calculated as the difference between the offering amount (divided by the
number of book runners) and the allocated amount times the offering price. The second is
the profit for S1 trades on the offering day, computed as the difference between the quantity
sold times the sell price and the quantity bought times the buy price. After the offering day,
we cumulate the daily profit from S1 trades at the dealer-bond-level of underwriting dealers
up to one trading month (21 trading days) post-issuance. The number of days before the
cumulative profit (including profits from both P1 and S1 trades on the offering day) becomes
positive for the first time is the time to first trading profit.
3.3 Secondary market corporate bond data
For the analysis of traded secondary market prices and quantities we again use the regulatory
version of Corporate TRACE. We focus on the universe of corporate bonds with issue and
issuer information in Mergent FISD, allowing us to control for bond specific characteristics.
As in the primary market analysis, we exclude bonds issued in foreign currency, bonds issued
as either Yankee or Canadian bonds, convertible and asset backed bonds, as well as bonds
that remain unrated more than 2 weeks after the initial offering date and focus on bonds
issued by non-financial issuers. We apply the same cleaning steps as described in Section
3.2 and in Appendix A.1, and we keep trades that are flagged with secondary (S1) market
indicator. Using this set of transactions, we construct two bond-day level measures of priced
U.S. corporate bonds this would be rare.
7
Recent papers that use P1 trades, Bessembinder et al., 2021 and Goldstein et al. (2021), consider P1
trades within a [-1,+1]-window around the offering date. Our results are robust to such alternative treatment
of data errors.
11
spreads: duration-matched spreads as in Gilchrist and Zakrajšek (2012) and the component
of duration-matched spreads adjusted for bond and issuer characteristics and expected de-
fault frequencies from Moody’s KMV (“default-adjusted” spreads). Details of both of these
calculations are available in Appendix A.2. Table 1, Panel B reports the summary statistics
for duration-matched spreads, default-adjusted spreads, 1y expected default frequencies, and
bid-ask spreads for the full 2020 sample.
In addition to bond and issuer characteristics from Mergent FISD, for time-varying bond
characteristics, such as the amount outstanding and the credit rating of the bond, we use
information contemporaneous with the trading date. We coalesce bond-level ratings by mul-
tiple rating agencies into a single number based on the plurality rule: if a bond is rated by
more than one agency, we use the rating agreed upon by at least two rating agencies and use
the lowest available rating otherwise. Following the facility term sheets, we define a bond
as being eligible for direct purchases by the facility if the bond is investment-grade rated, is
not issued by a bank or a bank subsidiary, and has less than 5 years remaining to maturity.
Starting on April 9, we also include bonds that were investment grade rated as of March
22 and were subsequently downgraded to no lower than BB-. At the issuer-level, instead of
using bond-level credit ratings to determine CCF eligibility, we use issuer-level credit ratings
(see details in Appendix A.3).
3.4 Secondary market bid-ask spread data
Given that many corporate bonds are not traded frequently, we utilize CMA Datavision
Bonds data that collects aggregated levels for quoted bonds that are based on over-the-
counter communications between top-tier, credit-focused buy-side trading desks, including
hedge funds, asset managers and proprietary and correlation desks at investment banks, and
their counterparties.
The aggregation process begins on the buy-side desk by collecting a sample of quotes
from which to perform an aggregation. The aggregation set is determined by looking within
12
a minimum window set to 60 minutes from the last observed quote. If at least three observable
quotes are found in this window, all the quotes are used to determine aggregation set. If not,
quotes are added in time priority order to the aggregation set until there are at least three
quotes available from the sample. For the very illiquid names, two quotes may be used. All
quotes in the aggregation set are time weighted and a median calculation is performed to
determine the “contribution”. CMA applies a further level of aggregation from the contributed
buy-side quotes sent to CMA Datavision in the above process to calculate the final published
consensus level. For some of the aggregated quotes we also observe the associated trade size
(which gives more credence to the validity of the quote; the trade sizes are ≥$1 million
for high-yield bonds, and ≥ $5 million for investment-grade). Since our analysis is at the
bond-day-level, we use the last ticks available of the bid and the ask for the day.
4 Effects on primary market issuance
The goal of the corporate credit facilities is to stabilize market functioning in order to support
the provision of credit to U.S. non-financial corporations. Therefore, we begin by studying
the evolution of the access to primary credit markets over the course of 2020 to see if the
facilities have achieved that goal, focusing on the dollar amount of corporate bonds issued
since the start of the year and the offering spreads paid by the issuers.
The relatively low frequency of corporate bond market primary issuance prevents us from
undertaking evaluating the facility impact with a narrow time window event study approach.
Instead, we focus on three sub periods: (1) the month preceding the facility announcement
(“Pandemic Onset”) (2) the period between announcement and implementation of the facility
on June 29 when the PMCCF was operational along with ETF purchases (“Facility Imple-
mentation”)8 and (3) the operational period of the facility where the SMCCF was purchasing
corporate bonds directly and the PMCCF was open (“Facility Operational”).
8
Results are similar if we define the operational period as starting on May 12, the date at which the
facilities began purchasing ETFs.
13
A key measure of primary market functioning is the composition of issuers that are able
to issue. The simple graphs shown in Figure 1 underestimate the deterioration in primary
market conditions because only higher quality borrowers are able to issue during periods
of stress. We thus evaluate the effect that the announcement and implementation of the
facilities had on both the extensive margin of issuance – which issuers had access to the
primary market – and the intensive margin – conditional on having access to the primary
market, at which terms were issuers able to issue.
4.1 Extensive margin: who issues?
Figure 2 plots the cumulative amount issued by non-financial corporations in the primary
corporate bond market since January 1, 2020, together with the median cumulative amount
issued by the same week in each of the preceding 10 years (2010 – 2019). In March 2020,
prior to the facilities’ announcement, issuance by investment-grade rated issuers in matu-
rities below 5 year was lagging relative to the pre-pandemic median pace of issuance. The
announcement of the facilities spurred a dramatic increase in the pace of investment-grade
issuance, for both below five year (Figure 2a) and above five year (Figure 2b) maturities.
High-yield-rated issuance ceased entirely for almost a month, and resumed after the facility
announcement albeit more slowly (Figure 2c). Thus, although the facilities announcement
improves primary market conditions across the board, the improvement for issuers not eligi-
ble for the facilities was more gradual.
We look at this more formally in the first two columns of Table 2 which report estimates
from a probit regression for the probability of issuer f issuing in week t as a function of
the issuer’s credit rating, the weighted average time-to-maturity of the issuer’s outstanding
14
debt, and sub-period dummies:
3
X 3
X
Pt (Issuancef ) = α + βi Subperiodit + φcr IGf t + γi Subperiodit × IGf t + φ
⃗ m Xf t + ϵf t ,
i=1 i=1
(1)
where Subperiodit is a dummy equal to 1 if the issuance occurs during the pandemic on-set
period (i = 1), or during the facility implementation period (i = 2), or during the facility
operational period (i = 3), and IGf t is a dummy equal to 1 if the issuer is investment grade
rated as of the time of issuance. Xf t is a vector of issuer characteristics, which includes
the weighted average maturity and total amount outstanding of existing bonds of the issuer
and a fixed effect for the issuer’s 2 digit NAIC industry. The probit regression also includes
month-of-year fixed effects to control for seasonalities in issuance.
In addition to the full sample of issuers, in Table 3 we follow the intuition of Almeida
et al. (2011) and split the sample into two types of corporate bond issuers. One set of issuers
has outstanding bonds maturing within two years (i.e. issuers that at the beginning of the
pandemic had debt maturing between March 2020 and March 2022) and are thus more likely
to face refinancing risk during the pandemic. We call the second set of issuers “opportunistic”,
as they do not have an imminent need to refinance outstanding bonds. For simplicity we do
not consider the effect on a possibly larger universe of potential issuers that may want to
raise capital but do not have any outstanding bonds.
Beginning in the first column of Table 3 with issuers that need to refinance, the onset
of the pandemic is marked by a dramatic reduction in the probability of high yield issuance
from a pre-pandemic average weekly probability of issuance of 39 basis points for high yield
issuers to essentially 0 during the pandemic onset period.9 Instead, even after controlling for
expected default frequencies, the probability of issuance for investment grade issuers slowed
only modestly (by 10 bps) from a pre-pandemic average of 74 bps. All issuers are more likely
9
To make the probit estimates in Tables 2 and 3 interpretable, we compute marginal probabilities, holding
the non-categorical variables at their realized values.
15
to issue after the facility announcement, with the probability of high yield issuers with a need
to refinance issuing increasing to 88 bps per week, and the probability of investment grade
issuers with a need to refinance issuing increasing to 221 bps per week during the facility
implementation period.
The overall issuance impact in the facility implementation period is positive and differ-
entially so for eligible issuers. Surprisingly, we find that the issuance impact of the facility
announcement is strongest on eligible opportunistic issuers (i.e. investment grade). Eligible
opportunistic issuers are disproportionately accessing primary markets after the announce-
ment, with a more than three-fold increase in the probability of issuance during the facility
implementation period (from 42 bps pre-pandemic to 136 bps). Finally, while the probabil-
ity of issuance by investment grade firms returns to pre-pandemic levels during the facility
operational period, the probability of issuance by high yield firms with refinancing needs
is actually elevated relative to pre-pandemic levels, highlighting the slow dissemination of
issuance opportunities to riskier borrowers.
It is important to note that increased issuance does not necessarily translate into increased
real activity, such as investment. Firms can issue new bonds while simultaneously calling
existing bonds or re-paying credit from other sources, thereby re-optimizing their overall
debt costs without changing materially their overall debt liabilities. In addition, firms can
use bond issuance to build liquidity buffers in anticipation of future cash-flow shocks, self-
insuring against future distress. We leave the study of the uses of funds raised through the
unprecedented corporate bond issuance since March and how those uses compare to the uses
of funds raised in “normal” recessions for future research.
4.2 Intensive margin: at what terms?
Beyond the extensive margin of which issuers are able to access the primary market for
corporate bonds, key measures of market functioning are the price and quantity at which
issuers are able to issue debt. In the remaining columns of Tables 2 and 3, we estimate how
16
offering terms changed over 2020 for firms that were actually able to issue. While we did
not estimate a disproportionate impact of the facility announcement on issuance probability,
eligible issuers are issuing much greater amounts of debt after the facility announcement,
but even before the PMCCF was fully operational to provide a full backstop. The differential
increase in offering amount is even higher after controlling for issuer risk.
In particular, for bond b of firm f issued in week t, we estimate
Off. termb(f ),t = α + βd Sub-period dummyt + βIG IGb,t + βd,IG Sub-period dummyt × IGb,t
(2)
+ λMills ratiof,t + φ
⃗ m Xf t + ⃗γ Yb,t + ϵb(f ),t
where the Mills ratio is the predicted Mills ratio from the corresponding first stage in Ta-
ble 2 and 3, Xf t are issuer characteristics including the weighted average maturity and
total amount outstanding of existing bonds of the issuer and a fixed effect for the issuer’s
2 digit NAIC industry, and Yf t are bond characteristics including dummies for callable,
shelf-registered and secured bonds. In each regression, we control for the other three offering
terms; thus, for example, the regression for the log offering amount controls for the offering
spread, offering maturity and gross spread. Note that there was no high yield issuance by
firms with maturing debt in the pandemic onset period thus we do not estimate a coeffi-
cient on the interaction of IG and Pandemic onset. In the facility implementation period,
investment grade firms with refinancing needs were able to issue bonds with larger offering
amounts, on average nearly doubling the offering amount relative to the offering amounts in
the pre-pandemic period.
Turning to maturity, in Figure 2aand 2b, we see that the acceleration in the pace of
issuance triggered by the announcement of the CCF is not concentrated in the five year
or less maturities that are eligible for purchases by the secondary market facility. Instead,
issuance in the more than 5 years maturity category in 2020 is nearly double that issued
17
during the same period in 2019 ($630 billion in 2020 vs $320 billion in 2019). Similarly, the
issuance of bonds with maturity of five years or below in 2020 is also nearly double that of
the same period in 2019 ($52 billion in 2020 vs $25 billion in 2019). These results suggest
that issuers are not issuing debt specifically targeting SMCCF purchase eligibility, as was the
experience with the European Central Bank’s Corporate Sector Purchase Program (CSPP)
(see e.g. De Santis and Zaghini, 2019). If anything, we estimate in Table 2 that eligible issuers
are extending maturities once the facility becomes operational, even though only short term
debt is eligible for the facilities.
There are a number of reasons as to why we would not expect issuers to modify their
offering maturity choices to issue only facility eligible debt. First, as shown in the next section,
while there were larger reductions in credit spreads on shorter maturity bonds immediately
following the announcement of the facilities, these were accompanied by large declines in the
risk-free rate, all along the curve. Thus issuers were able to take advantage of historically low
interest rates (not just spreads) – by issuing longer maturity bonds, they were able to lock in
those low funding rates for longer periods of time. Indeed, we find that the largest maturity
extension effects are concentrated among the opportunistic issuers – that is, those issuers
that had no debt maturing within two years – suggesting that firms actively re-optimized
their capital structure to take advantage of historically low rates. Second, the CCFs were
announced with a termination date (September 2020 that was subsequently extended to
December 2020), limiting the period of time over which issuers could accrue any potential
benefits from tailoring issuance to facility eligibility requirements.
A key measure of market conditions is the offering price. After the facility announcement,
offering spreads for borrowers with refinancing needs remain substantially higher – up to
500 bps more for high yield issuers – than before the pandemic onset, even controlling for
the default risk (EDF) of the issuer. Eligible issuers disproportionately benefit after the
announcement with offering spreads that increased much less than did those of high yield
issuers, although some of that appears to reflect a differential relationship between EDF and
18
offering spreads post pandemic onset. By the facility operational period, the offering spreads
for eligible borrowers return to their pre-pandemic levels, while offering spreads for high yield
issuers remain elevated. This highlights the slow pace of normalization of primary market
conditions, especially for issuers not eligible for the facilities.
Finally in the last 2 columns, we look at gross spreads paid to underwriters which are
one measure of the compensation required by dealers to underwrite bond offerings. 10 These
do not disproportionately fall for eligible issuers, except marginally statistically significant
during the facility implementation period. The amount appears small at around 12 basis
points, however gross spreads are generally small in the first place, meaning that this is a
sizable economic impact.11
We look further into the impact of underwriters by hand-matching each lead underwriter
reported by Mergent FISD to the corporate parent and, if applicable, to the registered el-
igible seller. We split underwriters between those that are broker dealers affiliated with a
US-headquartered bank and those that are not, reflecting the different regulatory environ-
ment and possible constraints of different underwriters as well as variation in the timing of
dealer registration for the facility. We find that issues underwritten by eligible sellers post
registration for the facility have offering spreads that are dramatically lower - more than 70
basis points. The impact on issuance terms of different types of underwriters is shown in
detail in the Appendix (Table A.6).
5 Effect of facilities on secondary markets
The previous section documents the improvements in access and slow normalization of pri-
mary market conditions following the announcement of the corporate credit facilities. We
now turn to examining potential mechanisms through which the normalization occurs, de-
Underwriters also profit through underwriting fees which tend to be smaller for investment grade issuance
10
but larger for high yield.
11
Table 1a shows that the average gross spread in our sample is 6 bps and the median gross spread is 5
bps.
19
spite a lack of take-up of the PMCCF, by evaluating the evolution of conditions in the
secondary market. Conditions in the secondary market pass-through to the primary market
in two ways: directly through the benchmarking of primary market pricing to secondary
market prices of similar bonds, and indirectly by increasing the willingness of dealers to
underwrite bond issuance. In addition, daily prices in the secondary market allows for more
causal interpretations as more narrow event windows can be examined.
5.1 Secondary market conditions over the course of 2020
We begin by examining the long sweep of the evolution of credit and bid-ask spreads over
2020, which allows us to take into account the potentially long-lasting impact of the facilities
announcement on the secondary market. Our main object of interest is the cumulative change
in each metric relative to the corresponding peak during the week of March 16 - 20, 2020.
This approach also has the benefit of creating an apples-to-apples comparison to secondary
market conditions prior to the start of the COVID-19-related market disruptions in March.
In particular, for each metric M for bond b trade date t, we estimate the following regression
∆Mb,t = αt + βb,t SMCCF eligibleb,t + ⃗γt Bond characteristicsb,t + ϵb,t , (3)
where ∆Mb,t is the cumulative change in metric M relative to the peak in metric M for bond b
during the week of March 16 - 20, 2020. Specification (3) thus estimates the improvements in
secondary market pricing and functioning for each individual bond, as a function of bond and
issuer characteristics. A negative estimate of βb,t indicates that secondary market conditions
for bonds eligible for direct purchases by the facility have improved more relative to secondary
market conditions for bonds not eligible for direct purchases. In addition to the eligibility
dummy, we control for standard bond characteristics: log age, log amount outstanding, log
offering amount, shelf registration dummy, callable dummy, and secured dummy. We estimate
specification (3) as a repeated panel for each trading date in the sample.
20
Figure 3 reports the estimated average effect and the differential effect on SMCCF eligible
bonds. Both duration-matched and default-adjusted spreads increased over February and
March, with the increases only arrested by the announcement of the facilities on March
22. Following the facility announcement, credit spreads retraced, with spreads on SMCCF
eligible bonds retracing more than spreads on the average bond. Indeed, the net benefit to
the SMCCF eligible bonds only disappears at the closure of the facility at the end of 2020.
In contrast, bid-ask spreads on SMCCF eligible bonds increased less than those of the
average bond ahead of the facilities announcement. The facility announcements and, in
particular, the expansion of the facility on April 9, eliminates the differential liquidity the
SMCCF eligible bonds had in the face of the pandemic. Instead, the bid-ask spreads for the
average bond decline rapidly after the facility announcement on March 23.
5.2 Effect of facilities announcements
We next isolate the direct effect of the facilities by examining a narrow window around key
announcement dates. To test formally the facility announcement effects, we calculate the
daily changes of metric M for bond b of firm f 3-days around the event date t, either the
initial facility announcement or the facility expansion announcement. We then estimate the
following empirical model:
∆Mb(f ),t−1,t+3 = α + βSMCCF-eligibleb(f ),t + ⃗γt Bond characteristicsb,t + ϵb(f ),t , (4)
where ∆Mb(f ),t−1,t+3 is the changes of duration-matched spreads, default-adjusted spreads,
1-year EDF, or bid-ask spread; SMCCF-eligible is a dummy variable equals one if the bond
meets the facility conditions of rating and maturity; Bond characteristics include log age, log
amount outstanding, log offering amount, shelf registration dummy, callable dummy, and se-
cured dummy. The β coefficient on SMCCF-eligible identifies the marginal improvements in
secondary market conditions for bonds that are eligible for direct purchases by the SMCCF
21
over and above the improvement in secondary market conditions for bonds that are ineligible
for direct purchases. Issuers with multiple bonds may have both eligible and ineligible bonds,
as bonds issued by the same issuer can have different maturity dates and different individual
bond ratings. Thus, when adding issuer fixed effects to the baseline specification, we iden-
tify the marginal improvements in secondary market conditions for bonds that are eligible
for direct purchases by the SMCCF over and above the improvement in secondary market
conditions for bonds issued by the same issuer that are ineligible for direct purchases. For
most of our exercises, the specification with issuer fixed effects has similar results as the
specification without issuer fixed effects. In our discussion, we thus focus on the specification
without issuer fixed effects and note if the results with issuer fixed effects differ materially.
Table 4 reports the results of this regression, focusing on the initial announcement. As
we show in Figure 3, up until March 23, spreads had been steadily climbing as investors
responded to the pandemic. Starting in panel (a), in Column (1) we look at the more than one
thousand bonds that traded around the announcement and find that on average, duration-
matched spreads fell by more than 90 basis points after the announcement (based on three
times the coefficient on average daily 3 day changes of -31). If the only effect of the facilities
were the direct effect on eligible bonds, we would not expect prices for all bonds to move
dramatically. That said, highly rated bonds eligible for the SMCCF fell by 50 percent more
than other bonds (an additional 52 basis points), implying a total effect of more than 140
basis points. Spreads on the riskiest bonds eligible for the SMCCF – those rated BBB+,
BBB or BBB- and thus just above the high yield cut-off – fell the most, implying a total
effect of more than 165 bps. In order to better identify the impact of the facility, we add
issuer fixed effects in column (2) and still find an economically large, statistically significant
coefficient. That is, spreads on bonds of the same issuer that mature before September
2025 had retraced more than spreads on bonds that have more than five years of maturity
remaining, by almost 50 basis points, or 17 basis points on average for the three day event
window. The spread impact also appears to be larger for eligible issuers in industries affected
22
by the pandemic, but the difference is not statistically significant. Finally, in column (4) we
narrow in on the issuers with rollover risk that we examined in Table 3 – BBB issuers with
any debt maturing within two years that are eligible for the facility see a differential 75 basis
point fall in spreads.
We turn now to the impact of the facilities on secondary market liquidity, shown in the
last 4 columns of Table 4, Panel (a). Starting with the average effect αt , improvements in
average bid-ask spreads we observed in Figure 3 appear to be concentrated in eligible issuers.
We see an immediate differential response in the prices of eligible bonds and improvement
in secondary market liquidity but not for all bonds, with bid-ask spreads for higher rated
eligible bonds falling over the three day event window by more than 45 cents and the spreads
for BBB eligible bonds falling by more than 30 cents. As with the credit spreads, in order to
better identify the impact of the facility, we add issuer fixed effects in column (5) and still
find economically large, statistically significant coefficients. For riskier eligible issuers with
debt maturing within 2 years, however, bid-ask spreads continue increasing even after the
facilities announcement. The corporate credit appear to provide a “buyer of last resort” in the
secondary market. That is, the announcement of the facilities on its own is sufficient to reduce
fire sales incentives in the market though maybe not for issuers with immediate refinancing
needs. For the subset of bonds that traded, bid-ask spreads were reduced presumably as
dealers were more willing to make markets and demand improved (Section 6 explores the
dealer response in more detail).
Decomposing credit spreads into default-adjusted spreads and the expected default fre-
quency (EDF) allows to separate the impact of the facilities on the price of risk from the
quantity of risk. Table 4b shows that the largest impact is eligible bonds for the default-
adjusted spreads as well, suggesting that the improvements in spreads we saw in Table 4a
and in Figure 3 are primarily due to a reduction in the default risk premium priced in cor-
porate bond spreads. While Table 4b shows that the announcement of the facilities also
led to a decrease in one year expected default frequencies, this is not differentially true for
23
eligible bonds. Focusing first on the specification without issuer fixed effects, we instead see
that overall EDFs fall more for the bonds of issuers in industries affected by COVID, but
if anything less for eligible bonds. This may reflect that the overall effect of the economic
implications of the COVID-response is stronger for non-investment-grade bonds which are
closer to the default boundary. An exception is for issuers with debt maturing within two
years, where the EDFs fall differently by -3.2 basis points for higher rated eligible bonds,
and by -4.2 basis points for BBB rated eligible bonds.
Taken together, Table 4 shows that the main effect of the facilities announcement was to
lower the price of risk, something that may explain the impact on primary market issuance
quantities. Indeed, the estimated coefficient for eligible issues shows a larger differential
decrease in default-adjusted spreads, bigger than in credit spreads themselves, consistent
with larger decreases in credit risk premia. The fall in the expected default probabilities
and improvements in pricing for non-eligible bonds indicates that the announcement of the
facilities also acted to improve market participants’ beliefs about the prospects for the U.S.
economy, primarily through reducing the credit risk premium but also through reducing
expected default frequencies, something that may again increase issuance quantities. Finally,
an impact on improved access to capital markets finance is suggested by the large differential
impact on issuers with maturing bonds, which differentially decreased both risk premia and
EDFs.
One challenge to interpreting the results is that there are many other reasons bonds of
different ratings and maturity may see different price patterns. This suggests that some of
the estimated coefficients for the differential impact on eligible bonds may be biased down, if
the other announcements increased the prospects for economic growth by stabilizing financial
markets, as this should differentially improve the prospects for lower rated firms. While the
impact of changes in risk should be captured by looking at the impact of changes in EDFs,
this may be seen in the changing price of risk as well. It is harder to know what the effect
of the changing term slope should be. While the reversal of the inversion of the term spread
24
that occurred in March is consistent with an impact of the facilities on eligible bonds (short
maturity), similar reversals of term spread inversions have also been seen at the start of the
financial crisis and again after the collapse of Lehman Brothers.
Table 5 looks at the impact of the subsequent announcement on April 9 which clarified
the facility eligibility. Newly eligible bonds experience additional falls in spreads when their
eligibility is clarified, despite EDFs which continued to increase. This clarification should
particularly impact industries disproportionately impacted by COVID, as it clarified that
fallen angels would remain eligible for the facilities. Accordingly spreads fall more for the
most affected borrowers, particularly the BBB-rated eligible ones. This exercise is further
supportive of a causal role for the facilities, given how we can tie the impact of the an-
nouncement directly to the eligible bonds, and the lack of other announcements on April
9.
5.3 Effect of purchases
Given the strong response to the facilities’ announcement, it is natural to wonder if purchases
contributed to maintaining the improvements in market conditions. To help answer this
question, we examine the response of secondary market spreads and liquidity to the facility
purchases. Note that this is a different way of looking at the Facility Operational time period
which we explored in the primary market setting in Section 4 as the direct purchases all occur
in that time period.
In particular, we modify the announcement date regression (4) and estimate the following
specification in stacked windows around the purchases:
∆Mb(f ),t−1,t+3 = α + βPurchasedb(f ),t + ⃗γt Bond characteristicsb,t + ϵb(f ),t (5)
∆Mb(f ),t−1,t+3 = α + βETF Indirectly purchasedb(f ),t + βbond Directly purchasedb(f ),t (6)
+ ⃗γt Bond characteristicsb,t + ϵb(f ),t
25
where ∆Mb(f ),t−1,t+3 is the change in metric M for bond b of firm f around purchase date t;
Purchasedb(f ),t is a 0/1 indicator for bond b(f ) bought on date t through either direct (cash-
bond) purchases or indirect (ETF) purchases; Indirectly purchasedb(f ),t is a 0/1 indicator for
bond b(f ) bought on date t indirectly through ETF purchases; and Directly purchasedb(f ),t
is a 0/1 indicator for bond b(f ) bought on date t directly through cash-bond purchases. As
above, bond characteristics include log age, log amount outstanding, log offering amount,
shelf registration dummy, callable dummy, and secured dummy. That is, we estimate the
purchase effect by comparing whether market conditions have improved differentially for
bonds purchased directly in the cash market and for bonds purchased indirectly through
ETFs purchases. We include date fixed effects in these specifications to control for differences
in the timing of ETF purchases.
Table 6 shows that despite the very large announcement date effects, individual purchases
also matter. On average, facility purchases decrease spreads by about 6 basis points, or
approximately 3% percent of the average duration-matched spread in 2020. This fall in
spreads is over and above the average fall in spreads of almost 5 basis points that happens
on purchase dates, some of which may also be due to the presence of the facilities in the
market. The spread impact seems to be occurring through changes in the price of risk as there
is no differential change in EDFs for purchased bonds, but a similar, statistically significant
fall in the default adjusted spreads of purchased bonds (see panels (b) and (c)).
Since the SMCCF purchased bonds both directly and indirectly through ETFs, it is
important to understand the differential impact of cash bond and ETF purchases. Since
the ETF purchases occur earlier in calendar time than the loose bond purchases, we add
date fixed effects to control for differences in the path of the pandemic and other changes
over time. Controlling for date fixed effects, it appears that most of the direct effect of the
purchases is through direct bond purchases. We estimate a negative statistically significant
impact on spreads for cash bond purchases but not ETFs. Curiously this seems to impact the
EDFs in addition to the price of risk. The effect of purchases on bid-ask spreads is relatively
26
small, slightly more than 1 basis point, and seems to be seen across both types of purchases,
results that are not always statistically significant.
In summary, we document an ongoing effect of the facility through purchases, particularly
on spreads although less so on bid-ask spreads. This continuing support in the secondary
markets was particularly important given the slow pace of normalization of the primary
markets we documented in Section 4. Bonds purchased through eligible ETFs have smaller
differential improvements. This may be because purchases of ETF shares from dealers may
not necessarily correspond to new ETF share creations; without new ETF share creation,
which necessitates purchases of portfolios of the underlying bonds, transactions in the ETF
market have limited pass-through to the corporate bond market. In Appendix Table A.9, we
look at the impact of the share of the bond issue that is purchased and find similar results.
ETF purchases may, however, still support the primary market as more than 60 percent of
ETFs purchase new issuances within one day of issuance.
5.4 Effect of closure announcement
On November 19, 2020 Treasury Secretary Steven Mnunchin sent a letter to Chairman of the
Federal Reserve Board of Governors Jerome Powell, stating “With respect to the facilities
that used CARES Act funding (PMCCF, SMCCF, MLF, MSLP, and TALF), I was person-
ally involved in drafting the relevant part of the legislation and believe the Congressional
intent as outlined in Section 4029 was to have the authority to originate new loans or pur-
chase new assets (either directly or indirectly) expire on December 31, 2020. As such, I am
requesting that the Federal Reserve return the unused funds to the Treasury”. The letter was
acknowledged by Chair Powell on the following day. Subsequent to that letter, the PMCCF
and SMCCF ceased purchasing assets on December 31, 2020. We examine the impact of the
letter from Secretary Mnunchin as well as the impact of the end of year facility.
To the extent that the facilities served as a backstop and markets had resumed func-
tioning, it is unclear what effect these announcements should have. Table 7 examines the
27
impact of the November 19 announcement. While overall corporate bond spreads fall in the
three days after the announcement, we find a statistically significant increase in spreads for
SMCCF eligible bonds. The effect is small, however, less than 2 basis points, and only one
basis point looking within issuers. There is no statistically significant effect at the end of the
year when purchases cease. As with the initial announcement effect, the effect of the Trea-
sury announcement appears to come mostly through the price of risk, with default adjusted
spreads increasing similarly. There is little effect on issuers with debt maturing within two
years.
Liquidity as measured by bid-ask spreads also appears mostly unaffected, with no statis-
tically significant impact on the announcement date. If anything, bid-ask spreads for SMCCF
eligible issues appear to fall by 1 basis point, although the effect is not statistically significant
on the Treasury announcement dates.
These non-effects are consistent with the facilities no longer having an important role as
a backstop by the end of 2020. The result is perhaps not surprising, given the historically
low credit and bid-ask spreads in bond markets by 2020 indicative of ample intermedia-
tion and trading. However, we cannot rule out an alternative interpretation that given one
intervention, bond markets would expect future shocks to be met with future interventions.
6 Dealer intermediation and the facilities
Having established the impact of the facilities on primary and secondary markets, in this
section we examine the relevance of the intermediary constraints channel in these improve-
ments. Similar to the approach in Section 4, we look at the impact on intermediation over
the same three sub periods, comparing the beginning of 2020 to the pandemic onset, the
post-announcement implementation period and the operational period.
From the perspective of intermediary asset pricing, the facilities act by providing a “buyer
of last resort”, relaxing balance sheet constraints of marginal intermediaries in the corporate
28
bond market. Improvements in balance sheet constraints of the marginal intermediary lead
to a reduction in the intermediary’s effective risk aversion, explaining the change in will-
ingness to underwrite offerings, and substantial improvement in credit spreads and liquidity
conditions. In addition to differentiating between issues that are eligible for the facilities
(SMCCF eligible), as we did in the previous section, we also consider intermediation ac-
tivity across different classifications of dealers. Comparing dealers that are affiliated with
U.S. bank-holding companies (U.S. BHCs) to those not affiliated with U.S. BHCs allows us
to examine the role that balance sheet constraints played in March 2020 dislocations and
subsequent normalization in corporate bond markets.
We also look at dealers that were eligible to sell to the facility. While 41 dealers ultimately
became eligible to sell to the facility, eligibility required the submission of legal documenta-
tion to the New York Fed. This was submitted over a few weeks or even months depending on
the dealer, which adds some time variation in dates at which dealers became eligible sellers,
but means that we focus on the secondary market given the narrow time window. Table A.4
provides the list of eligible sellers to the facility, together with an indicator of whether these
institutions underwrite corporate bonds in our sample.
6.1 Extensive margin: Does intermediation become more likely?
We begin by examining the probability of intermediation by different types of dealers in the
market over 2020. Starting with the question of whether regulatory constraints played a role
in corporate bond market dislocations, we estimate a probit regression for the probability of
dealer d buying bond b on day t from non-dealer customers as a function of the eligibility of
the bond for direct purchases by the facility, the sub-period, whether the dealer is affiliated
29
with a U.S. bank holding company (BHC) and standard bond characteristics:
3
X 3
X
Pt Buyd,b = α0 + αi Subperiodit + φ0 SMCCF-eligibleb,t + φi Subperiodit × SMCCF-eligibleb,t
i=1 i=1
(7)
3
X
+ β0 BHC-affiliatedd,t + βi Subperiodit × BHC-affiliatedd,t
i=1
+ η0 BHC-affiliatedd,t × SMCCF-eligibleb,t
3
X
+ ηi Subperiodit × SMCCF-eligibleb,t × BHC-affiliatedd,t
i=1
+ ⃗γt Bond characteristicsd,b,t + ϵb,t .
Subperiodit is a dummy equal to 1 if the issuance occurs during the pandemic on-set period
(i = 1), or during the facility implementation period (i = 2), or during the facility opera-
tional period (i = 3), SMCCF-eligibleb,t is a dummy equal to 1 if the bond is eligible for
direct purchases by the SMCCF, and BHC-affiliatedd,t is a dummy equal to 1 if the dealer
is U.S. BHC-affiliated on date t. While all dealers are driven by risk management consider-
ations, bank-affiliated dealers are more likely to reflect regulatory balance sheet constraints
associated with consolidated regulation of the bank holding company. The probit regression
also includes week fixed effects to control for seasonalities in trading.
Table 8 summarizes the results. Column (1) illustrates the dynamics of the dash for cash
in the overall market, with the probability of a dealer buying a particular bond from non-
dealer customers on a particular day rising by 30 percent (from 1.2% in the pre-pandemic
period to 1.5% in the pandemic onset period). Column (2) shows that on average stand-
alone dealers are less active in the market, with a pre-pandemic probability of 0.8% of
buying from non-dealer customers relative to the BHC-affiliated dealer average probability
of 2%. The positive impact of the facility announcement is clear, as the probability of dealer
purchases from customers starts decreasing after announcement, highlighting the attenuation
30
of selling pressure for customers, so that by the time the facility becomes fully operational,
the probability of buying from customers is even lower than the pre-pandemic levels.
Turning next to the question of why the facilities had differential effects on eligible collat-
eral, in columns (3) and (4) we add the interaction effects with the SMCCF eligible dummy,
and uncover differential behavior by BHC-affiliated and stand-alone dealers once the facil-
ity is announced. Bank-affiliated dealers reduced their buying activity in the market for
SMCCF-eligible collateral following the announcement of the facilities, with the probability
of buying SMCCF-eligible collateral dropping from 2.2% during the pandemic on-set period
to 2% during the facility implementation period. In contrast, stand-alone dealers continued
buying these safer assets during the facility implementation period, with the probability of
buying SMCCF-eligible collateral remaining at 1.2% during both the pandemic on-set period
and the facility implementation period. This points to a potential “flight to intermediation
safety” on the part of stand-alone dealers as the overall selling pressures from non-dealer
customers were abating at the same time. These results are robust to the inclusion of dealer
fixed effects, as shown in column (4).
The results in Table 8 suggest that bank regulatory constraints are unlikely to have
played a role in the corporate market dislocations in March 2020 as BHC-affiliated dealers
did not differentially decrease intermediation activity during the pandemic onset period and
returned to the market for non-SMCCF eligible collateral more quickly than did stand-alone
dealers. An alternative hypothesis is that, by standing as a “buyer of last resort”, the facilities
relaxed balance sheet constraints more broadly for counterparties to the facilities. We follow
the approach in equation (7), replacing the bank-affiliated dummy with an indicator for
facilities’ counterparties (dubbed “eligible sellers”). Since dealers that may ultimately become
eligible may be systematically different, we include both a dummy variable for all dealers
that ever become eligible as well as a dummy for those dealers who register after the facilities
are fully operational.12
12
Table A.4 in the Appendix lists the SMCCF counterparties together with their registration dates.
31
Table 9 summarizes the results. Column (2) illustrates the dynamics of the dash for
cash, and the expansion in dealer intermediation required to facilitate increased demand for
trading. The probability of dealers buying from non-dealers increases from 1.2% to 1.5%
(a coefficient of 0.10). Dealers who ultimately become eligible sellers are generally the most
active buyers in this market, and they are buying more from their customers in the pandemic
onset. The positive impact of the facility announcement is clear, as the probability of dealer
purchase from customers increases after announcement, an effect that lessens by the time
the facility becomes operational.
In columns (3)-(4) we add interactions with eligible collateral and find a more complicated
story. In the pandemic onset, dealers were more likely to buy SMCCF-eligible collateral,
perhaps as their customers sell off of higher credit quality assets accelerated. Eligible dealers
were particularly less likely to buy these IG assets in the pandemic onset (3.4% probability of
buying ineligible collateral vs 2.9% probability of buying eligible collateral) and in the time
period overall (2.7% probability of buying ineligible collateral vs 2.5% probability of buying
eligible collateral). The real difference is seen when the facility is operational – once eligible
dealers were registered, they were differentially likely to buy SMCCF eligible collateral,
suggesting that the operations of the facility were necessary to change their participation.
This result is robust to the addition of individual dealer fixed effects (instead of just a fixed
effect for eligibility), which we add in the final column of the table.
6.2 Intensive margin: Does the amount of intermediation increase?
In addition to the extensive margin of the probability of buying from non-dealer customers,
constrained dealers can adjust on the intensive margin, buying more or less of the same
securities over time. We investigate the intensive margin by estimating how the quantity of
32
bonds bought from non-dealer customers changes across sub-periods:
3
X
Amount boughtd,b,t = α0 + αi Subperiodit + φ0 SMCCF-eligibleb,t (8)
i=1
3
X
+ φi Subperiodit × SMCCF-eligibleb,t
i=1
3
X
+ β0 BHC-affiliatedd,t + βi Subperiodit × BHC-affiliatedd,t
i=1
+ η0 BHC-affiliatedd,t × SMCCF-eligibleb,t
3
X
+ ηi Subperiodit × SMCCF-eligibleb,t × BHC-affiliatedd,t
i=1
+ λMills ratiod,b,t + ⃗γt Bond characteristicsb,t + ϵd,b,t ,
where the Mills ratio is the Mills ratio from the corresponding probit in Table 8.
The results in column (5) of Table 8 point to the importance of the facility on dealer
actions. Consistent with the extensive margin results above, BHC-affiliated dealers reduce
the extent of their buying of SMCCF-eligible collateral already in the pandemic onset period
and continue buying at a reduced intensity once the facility is announced and operational.
In contrast, the stand-alone dealers increase differentially their purchases of SMCCF-eligible
collateral after the facility is announced, once again illustrating the flight to intermediation
safety we documented above. These effects are economically, as well as statistically signif-
icant. Relative to the pre-pandemic period, bank-affiliated dealers reduced their purchase
volumes of SMCCF-eligible collateral by $1.3 million par during the facility implementation
period and by $0.5 million par during the facility operational period. At the same time,
standalone dealers increased purchases of the average SMCCF-eligible bond by $0.7 million
par during the facility implementation period and by $0.2 million par during the facility
operational period.
Column (5) of Table 9 highlights the role that being a counterparty to the facilities plays
in the intensive margin of intermediation. Column (5) illustrates that, while dealers that
33
become facility counterparties buy fewer SMCCF-eligible securities on average – by $2.3
million par less on an average day – they buy even fewer SMCCF-eligible securities during
the pandemic onset and the facility implementation periods. Thus, dealers that eventually
become counterparties to the facilities do not appear to build up their inventory of eligible
securities ahead of time and instead serve as intermediaries for securities that the non-
eligible sellers shy away from. During the facility operational period, the amount bought by
eligible sellers of SMCCF-eligible securities increases but such increases are driven by the
later registrants to the facility, which tended to be smaller dealers and electronic platforms.
Furthermore, the differential activity by late registrants in SMCCF-eligible securities persists
even after we include dealer fixed effects, suggesting a direct impact of facility registration
on intermediation decisions at the intensive margin.
Taken together, the results in this section suggest that, for larger dealers, the announce-
ment of the facilities was sufficient to incentivize greater intermediation activity, at both the
extensive and the intensive margin, especially in securities not eligible for direct purchases
by the SMCCF. For such dealers, the SMCCF thus acts through an increase in the over-
all intermediation capacity, which these dealers use to intermediate in securities that are
less attractive to smaller, stand-alone dealers. For smaller dealers, the announcement of the
facilities alone was not sufficient to incentivize intermediation in riskier, non-eligible securi-
ties, and becoming a registered counterparty to the facility and the start of active facility
purchases more broadly was needed to restore such dealers’ confidence in the market.
6.3 Primary market spillovers
Finally, we turn to the underwriting profits earned by dealers. For each dealer in an un-
derwriting syndicate, we compute the net dollar trading profit earned on the offering day
from secondary market re-trading of under- and overallocated issues.13 We estimate a linear
regression for the net profit of dealer d underwriting bond b on date t as a function of the
13
Note that trading profit does not include the gross fees earned, if any.
34
eligibility of the bond for direct purchases by the facility, the sub-period, whether the dealer
is affiliated with a U.S. bank holding company (BHC) and standard bond characteristics:
3
X
Net profitd,b,t = α0 + αi Subperiodit + φ0 SMCCF-eligibleb,t (9)
i=1
3
X
+ φi Subperiodit × SMCCF-eligibleb,t
i=1
3
X
+ β0 BHC-affiliatedd,t + βi Subperiodit × BHC-affiliatedd,t
i=1
+ η0 BHC-affiliatedd,t × SMCCF-eligibleb,t
3
X
+ ηi Subperiodit × SMCCF-eligibleb,t × BHC-affiliatedd,t
i=1
+ ⃗γt Bond characteristicsb,t + ϵd,b,t .
Consistent with the asynchronous recovery between the primary and secondary markets,
column (1) of Table 10 shows that average underwriter compensation increased during the
facility implementation period, rising from a negative $222 million during the pre-pandemic
period to a negative $9 million during the facility implementation period on average. While
the average profitability of underwriting is not statistically significantly different during the
pandemic onset period, column (1) of Table 11 shows that the number of trading days
until the dealer generates a positive profit rises during the pandemic on-set period, so that
the underwriter bears underwriting risk longer. Column (2) of Table 10 and of Table 11
highlight the extraordinary dislocation in the primary market for short maturity (below 5
year), investment grade bonds – those bonds that would become eligible for direct purchases
by the facilities – during the onset of the pandemic, with the net underwriting profit on the
offering day declining to negative $246 million on average and the average time to profitability
increasing to more than 14 days from a pre-pandemic average of less than 5.
Turning to the role that dealer constraints play, we see in columns (3)-(5) of Table 10 that
bank-affiliated dealers in particular had lower profits when underwriting SMCCF-eligible
35
collateral during both the pandemic onset and the facility implementation period, while
dealers not affiliated with U.S. BHCs saw rising profits. This is consistent with the secondary
market “flight to intermediation safety” we documented above. Dealers not affiliated with U.S.
BHC were both more willing to intermediate in the secondary market for SMCCF-eligible
collateral and demanded more compensation for underwriting risk during the pandemic on-
set and especially during the facilities implementation period.
7 Conclusion
With corporate bonds an ever growing fraction of U.S. non-financial corporate debt, un-
derstanding the impact of 13(3) facilities and policy choices on businesses’ ability to fund
themselves through the primary market for corporate debt gains in importance. Yet most of
the literature to date, including evaluations of both monetary policy and extraordinary inter-
ventions, has focused on the impact of policy changes on credit spreads in secondary markets.
In this paper, we document that, although improvements in the primary market do follow
decreases in secondary market credit spreads, the normalization in primary market offering
spreads is much slower, highlighting the importance of evaluating the impact of policy inter-
ventions in a holistic manner. Generally, other official sector interventions such as lower risk
free rates and fiscal spending should disproportionally benefit riskier (non-eligible) issuers,
suggesting that our results of a disproportional impact on eligible issuers may underestimate
the direct impact of the facilities.
We also highlight the key role of dealers who are both making markets in secondary
markets and underwriting debt securities. Underwriters appear to be demanding higher
profits in the wake of the pandemic, with underwriter profits on new issuance growing sharply,
and falling only as the facility began purchasing bonds. The effect on dealer intermediation in
secondary markets is complex, and we find a key role for the actual operations of the facility,
with differential purchases by the subset of dealers eligible to transact with the facility.
36
However, we do not find evidence to support balance sheet constraints of BHC-affiliate
dealers as an important contributor to the March 2020 corporate bond market dislocations.
37
References
Acharya, V. V. and S. Steffen (2020): “The risk of being a fallen angel and the corporate
dash for cash in the midst of COVID,” The Review of Corporate Finance Studies, 9, 430–
471.
Adrian, T., N. Boyarchenko, and O. Shachar (2017): “Dealer balance sheets and
bond liquidity provision,” Journal of Monetary Economics, 89, 92–109.
Almeida, H., M. Campello, B. Laranjeira, S. Weisbenner, et al. (2011): “Cor-
porate Debt Maturity and the Real Effects of the 2007 Credit Crisis,” Critical Finance
Review, 1, 3–58.
Andres, C., A. Betzer, and P. Limbach (2014): “Underwriter reputation and the quality
of certification: Evidence from high-yield bonds,” Journal of Banking & Finance, 40, 97–
115.
Bessembinder, H., S. E. Jacobsen, W. F. Maxwell, and K. Venkataraman (2021):
“Syndicate Structure, Overallocation, and Secondary Market Outcomes in Corporate Bond
Offerings,” SMU Cox School of Business Research Paper.
Boyarchenko, N., C. Cox, R. K. Crump, A. Danzig, A. Kovner, O. Shachar, and
P. Steiner (2021): “COVID Response: The Primary and Secondary Corporate Credit
Facilities,” Economic Policy Review.
Brunnermeier, M. K. and L. H. Pedersen (2009): “Market Liquidity and Funding
Liquidity,” Review of Financial Studies, 22, 2201–2238.
D’Amico, S., V. Kurakula, and S. Lee (2020): “Impacts of the Fed Corporate Credit
Facilities through the Lenses of ETFs and CDX,” Working paper 2020-14, Federal Reserve
Bank of Chicago.
Dannhauser, C. D. and S. Hoseinzade (2022): “The unintended consequences of cor-
porate bond ETFs: Evidence from the Taper Tantrum,” The Review of Financial Studies,
35, 51–90.
De Santis, R. A. and A. Zaghini (2019): “Unconventional monetary policy and corporate
bond issuance,” Working Paper Series 2329, ECB.
Dick-Nielsen, J., M. S. Nielsen, and S. L. von Rüden (2021): “The value of bond
underwriter relationships,” Journal of Corporate Finance, 68, 101930.
Falato, A., I. Goldstein, and A. Hortaçsu (2021): “Financial fragility in the COVID-
19 crisis: The case of investment funds in corporate bond markets,” Journal of Monetary
Economics, 123, 35–52.
Fang, L. H. (2005): “Investment bank reputation and the price and quality of underwriting
services,” The Journal of Finance, 60, 2729–2761.
38
Fernando, C. S., A. D. May, and W. L. Megginson (2012): “The value of investment
banking relationships: Evidence from the collapse of Lehman Brothers,” The Journal of
Finance, 67, 235–270.
Gilchrist, S., B. Wei, V. Z. Yue, and E. Zakrajšek (2020): “The Fed Takes on
Corporate Credit Risk: An Analysis of the Efficacy of the SMCCF,” Working paper N.
27809, National Bureau of Economic Research.
Gilchrist, S. and E. Zakrajšek (2012): “Credit spreads and business cycle fluctuations,”
American Economic Review, 102, 1692–1720.
Goldstein, M. A., E. S. Hotchkiss, and S. S. Nikolova (2021): “Dealer behavior and
the trading of newly issued corporate bonds,” Available at SSRN 1022356.
Gromb, D. and D. Vayanos (2002): “Equilibrium and welfare in markets with financially
constrained arbitrageurs,” Journal of Financial Economics, 66, 361–407.
——— (2010): “A model of financial market liquidity based on intermediary capital,” Journal
of the European Economic Association, 8, 456–466.
Grosse-Rueschkamp, B., S. Steffen, and D. Streitz (2019): “A capital structure
channel of monetary policy,” Journal of Financial Economics, 133, 357–378.
Haddad, V., A. Moreira, and T. Muir (2021): “When selling becomes viral: Disruptions
in debt markets in the COVID-19 crisis and the Fed’s response,” The Review of Financial
Studies, 34, 5309–5351.
Kargar, M., B. Lester, D. Lindsay, S. Liu, P.-O. Weill, and D. Zúñiga (2021):
“Corporate bond liquidity during the COVID-19 crisis,” The Review of Financial Studies,
34, 5352–5401.
Merton, R. C. (1974): “On the pricing of corporate debt: The risk structure of interest
rates,” The Journal of Finance, 29, 449–470.
Nozawa, Y. and Y. Qiu (2021): “Corporate bond market reactions to quantitative easing
during the COVID-19 pandemic,” Journal of Banking & Finance, 133, 106–153.
O’Hara, M. and X. A. Zhou (2021): “Anatomy of a liquidity crisis: Corporate bonds in
the COVID-19 crisis,” Journal of Financial Economics, 142, 46–68.
Todorov, K. (2020): “Quantify the quantitative easing: Impact on bonds and corporate
debt issuance,” Journal of Financial Economics, 135, 340 – 358.
Yasuda, A. (2005): “Do bank relationships affect the firm’s underwriter choice in the
corporate-bond underwriting market?” The Journal of Finance, 60, 1259–1292.
39
Table 1: Summary statistics. This table reports the summary statistics for the secondary-
market sample and the offering terms sample used in the paper (January 1, 2020 – December 31,
2020). Secondary market spreads, 1y EDFs, bid-ask spreads, offering spreads and gross spreads are
reported in basis point terms. Offering amount in USD million (log offering amount in log USD
million) and offering maturity is reported in years.
(a) Primary market
Mean Median P25 P75 Std. Dev.
Offering amount 260 50 10 300 505
Log offering amount 4 4 2 6 2
Offering maturity 9 6 3 10 8
Offering spread 112 90 57 145 108
Gross spread 6 5 3 7 5
(b) Secondary market
Mean Median P25 P75 Std. Dev.
Duration-matched spread 232 141 82 259 322
Default-adjusted spread 88 1 -39 110 277
1y EDF 83 18 10 58 288
Bid-ask spread 117 96 57 160 88
40
Table 2: Access to primary credit markets. This table reports the estimated coefficients from
the probit regression of the probability of issuer f issuing a fixed rate bond in week t (columns
1 and 2), together with the estimated coefficients from the regression of offering terms of bond b
issued by issuer f in week t on bond characteristics and sub-period dummies. “IG” is a dummy
equal to 1 if the issuer is investment grade rated in week t. “Pandemic onset” is a dummy equal
to 1 if issuance occurs between February 22, 2020 and March 21, 2020; “Facility implementation”
is a dummy equal to 1 if issuance occurs between March 22, 2020 and June 29, 2020; “Facility
operational” is a dummy equal to 1 if issuance occurs between June 30, 2020 and December 31,
2020. “R-sqr” is the pseudo R2 for the probit regressions, and the adjusted R2 for the intensive
margin regression. All regressions include the other offering terms, standard bond characteristics
and month-of-year fixed effects. Standard errors clustered at the issuer level reported in parentheses
below point estimates. Reference period is January 1, 2017 – February 21, 2020. *** significant at
1% level; ** significant at 5% level; * significant at 10% level.
Probit Amount Maturity Offering spread Gross spread
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
Constant -4.69 -3.82 -32.25 -12.71 329.04 655.44 -1652.47 -9775.88 184.62 91.24
(0.14)∗∗∗ (0.18)∗∗∗ (23.62) (26.37) (265.32) (328.97)∗∗ (4100.39) (3820.17)∗∗ (82.52)∗∗ (84.07)
Pandemic onset -0.38 -0.50 -2.98 -1.29 26.73 73.00 -275.26 -1291.55 9.47 3.05
(0.32) (0.37) (1.89) (2.81) (21.38) (35.21)∗∗ (337.66) (418.22)∗∗∗ (6.69) (9.23)
Facility implementation 0.08 0.12 1.01 0.13 -12.02 -26.89 237.96 558.56 -3.38 -1.74
(0.11) (0.12) (0.62) (0.70) (5.50)∗∗ (9.05)∗∗∗ (119.75)∗∗ (114.01)∗∗∗ (1.77)∗ (2.03)
Facility operational -0.16 -0.10 -1.35 -0.55 4.77 7.50 87.87 -80.74 7.55 2.99
(0.10) (0.10) (0.81)∗ (0.59) (9.01) (7.28) (144.90) (96.71) (2.91)∗∗∗ (2.27)
IG 0.11 0.11 1.00 0.57 -4.88 -9.99 -102.37 83.51 -7.79 -5.87
(0.05)∗∗ (0.06)∗ (0.59)∗ (0.63) (6.74) (7.81) (99.01) (88.23) (2.11)∗∗∗ (1.91)∗∗∗
IG × Pandemic onset 0.41 0.16 3.38 0.96 -26.24 -23.98 362.54 442.09 -10.50 2.04
(0.33) (0.35) (2.04)∗ (0.90) (22.64) (11.62)∗∗ (361.02) (139.08)∗∗∗ (7.16) (3.11)
IG × Facility implementation 0.36 0.39 2.38 1.69 -19.75 -49.80 42.79 649.97 -11.84 -6.26
(0.11)∗∗∗ (0.13)∗∗∗ (1.77) (2.17) (19.72) (27.32)∗ (323.55) (340.51)∗ (6.18)∗ (7.11)
IG × Facility operational -0.05 -0.09 -0.14 -0.05 7.03 14.93 -154.55 -284.44 0.66 0.28
(0.11) (0.14) (0.27) (0.54) (3.26)∗∗ (7.04)∗∗ (56.30)∗∗∗ (80.13)∗∗∗ (1.19) (1.68)
Log EDF -0.04 -0.22 5.08 -61.13 0.87
(0.02)∗∗ (0.22) (2.78)∗ (30.64)∗∗ (0.68)
Pandemic onset × Log EDF -0.15 -0.21 20.78 -405.67 2.23
(0.08)∗ (0.85) (10.75)∗ (129.95)∗∗∗ (2.79)
Facility implementation × Log EDF 0.04 0.17 -7.16 97.84 -0.65
(0.03) (0.22) (2.70)∗∗∗ (37.73)∗∗ (0.66)
Facility operational × Log EDF 0.01 0.05 -2.66 37.36 -0.55
(0.05) (0.10) (1.46)∗ (19.49)∗ (0.28)∗
Mills ratio 8.67 4.08 -77.49 -155.45 406.67 2302.56 -40.17 -18.73
(5.42) (6.15) (61.10) (76.64)∗∗ (946.10) (894.60)∗∗ (19.04)∗∗ (19.67)
Issuer characteristics ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Bond characteristics ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
R.-sqr. 0.12 0.10 0.41 0.45 0.14 0.16 0.56 0.67 0.59 0.64
N. of obs 158900 112822 910 876 910 876 910 876 856 826
N. of clusters 923 789 236 223 236 223 236 223 229 215
41
Table 3: Access to primary credit markets by refinancing need. This table reports the
estimated coefficients from the probit regression of the probability of issuer f issuing a fixed rate
bond in week t (columns 1 and 2), together with the estimated coefficients from the regression
of offering terms of bond b issued by issuer f in week t on bond characteristics and sub-period
dummies. “IG” is a dummy equal to 1 if the issuer is investment grade rated in week t. “Pandemic
onset” is a dummy equal to 1 if issuance occurs between February 22, 2020 and March 21, 2020;
“Facility implementation” is a dummy equal to 1 if issuance occurs between March 22, 2020 and
June 29, 2020; “Facility operational” is a dummy equal to 1 if issuance occurs between June 30, 2020
and December 31, 2020. “Refi” are issuers with debt maturity within 2 years; “Opp.” are those that
don’t have any debt maturing within 2 years. “R-sqr” is the pseudo R2 for the probit regressions,
and the adjusted R2 for the intensive margin regression. All regressions include the other offering
terms, standard bond characteristics and month-of-year fixed effects. Standard errors clustered at
the issuer level reported in parentheses below point estimates. Reference period is January 1, 2017 –
February 21, 2020. *** significant at 1% level; ** significant at 5% level; * significant at 10% level.
Probit Amount Maturity Offering spread Gross spread
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
Refi Opp. Refi Opp. Refi Opp. Refi Opp. Refi Opp.
Constant -4.52 -3.89 4.41 -22.56 367.21 907.80 -4675.09 -9371.30 5.25 85.15
(0.33)∗∗∗ (0.20)∗∗∗ (40.26) (26.78) (345.48) (458.03)∗∗ (5534.00) (4175.18)∗∗ (84.42) (95.05)
Pandemic onset -3.50 -0.37 0.92 -1.93 41.98 78.34 -1170.07 -1067.40 0.14 1.72
(0.27)∗∗∗ (0.34) (6.13) (2.13) (53.90) (36.51)∗∗ (850.67) (340.90)∗∗∗ (13.48) (7.75)
Facility implementation 0.49 0.04 -0.50 0.18 -51.26 -18.50 819.18 310.27 4.55 -0.41
(0.23)∗∗ (0.15) (4.02) (0.34) (35.31) (5.55)∗∗∗ (559.32) (65.48)∗∗∗ (8.84) (1.23)
Facility operational 0.17 -0.14 0.18 -1.13 -25.69 22.76 357.37 -228.04 4.87 3.37
(0.21) (0.11) (1.37) (0.77) (13.65)∗ (13.44)∗ (207.25)∗ (137.49)∗ (3.74) (3.22)
IG 0.23 0.11 0.37 0.74 -15.29 -15.55 57.30 90.20 1.14 -6.52
(0.17) (0.06)∗ (1.91) (0.65) (17.21) (10.84) (262.98) (100.60) (4.53) (2.33)∗∗∗
IG × Pandemic onset 2.74 0.13 1.49 -21.28 402.67 3.65
(0.22)∗∗∗ (0.37) (0.80)∗ (14.26) (122.34)∗∗∗ (2.91)
IG × Facility implementation 0.15 0.43 0.26 2.51 -3.11 -77.93 21.84 749.22 -4.24 -6.16
(0.26) (0.17)∗∗ (1.31) (2.37) (12.90) (40.53)∗ (219.18) (392.69)∗ (2.85) (8.55)
IG × Facility operational -0.45 -0.02 -0.22 0.34 40.19 2.97 -574.23 -114.51 -7.64 -0.52
(0.28) (0.16) (3.75) (0.25) (32.63) (4.45) (537.06) (37.03)∗∗∗ (8.12) (1.05)
Log EDF -0.06 -0.04 -0.09 -0.27 3.89 7.47 -40.84 -56.57 0.57 0.59
(0.04) (0.02)∗∗ (0.46) (0.23) (4.07) (4.00)∗ (64.85) (35.07) (1.02) (0.78)
Pandemic onset × Log EDF -0.30 -0.11 0.41 -0.33 16.80 26.21 -490.33 -361.04 -0.32 2.99
(0.11)∗∗∗ (0.07) (2.45) (0.65) (21.42) (11.81)∗∗ (339.02) (107.76)∗∗∗ (5.35) (2.50)
Facility implementation × Log EDF 0.11 0.02 0.01 0.09 -10.98 -4.68 160.11 35.66 -0.34 -0.11
(0.06)∗∗ (0.04) (0.92) (0.11) (8.09) (1.91)∗∗ (133.45) (27.21) (1.99) (0.34)
Facility operational × Log EDF 0.03 0.00 -0.12 0.09 -4.48 -1.27 52.28 -1.19 -1.52 0.03
(0.09) (0.06) (0.22) (0.07) (2.40)∗ (1.87) (40.74) (15.57) (0.51)∗∗∗ (0.26)
Mills ratio -0.01 6.30 -85.00 -212.92 1070.83 2227.63 0.39 -17.05
(9.12) (6.18) (77.82) (105.76)∗∗ (1261.56) (968.73)∗∗ (18.98) (22.00)
Issuer characteristics ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Bond characteristics ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
R.-sqr. 0.11 0.10 0.45 0.48 0.06 0.23 0.75 0.66 0.65 0.66
N. of obs 27035 85787 401 473 401 473 401 473 365 458
N. of clusters 370 737 93 175 93 175 93 175 88 171
42
Table 4: Facility announcement effects. This table reports the estimated coefficients from
the regression of average daily changes in duration-matched credit spreads, bid-ask spreads, and
components of credit spreads in 3 days around the announcement of the facilities on March 22,
2020. “AAA/AA/A eligible” (“BBB eligible”) is a dummy equal to 1 if the bond matures with 5
years, is not issued by a bank or a bank-subsidiary and is rated at least A- (rated BBB+, BBB
or BBB-) on a plurality rating basis. “Debt maturing” is a dummy equal to 1 if the issuer has any
bonds maturing within the next 2 years. Standard errors clustered at the issuer level reported in
parentheses below point estimates. *** significant at 1% level; ** significant at 5% level; * significant
at 10% level.
(a) Credit and bid-ask spreads
Duration-matched spread Bid-ask spread
(1) (2) (3) (4) (5) (6)
Constant -30.66 -14.44 -29.69 1.45 22.34 1.35
(11.24)∗∗∗ (13.49) (11.77)∗∗ (20.67) (14.76) (20.54)
AAA/AA/A eligible -17.22 -14.89 -17.11 -16.01 -18.24 -15.47
(5.76)∗∗∗ (4.94)∗∗∗ (5.79)∗∗∗ (3.12)∗∗∗ (2.74)∗∗∗ (3.39)∗∗∗
BBB eligible -24.57 -17.68 -6.34 -12.37 -8.91 -23.58
(6.18)∗∗∗ (6.12)∗∗∗ (7.35) (6.57)∗ (3.72)∗∗ (7.26)∗∗∗
Debt maturing w/in 2 year -2.45 1.03
(4.22) (4.37)
AAA/AA/A eligible × Debt maturing w/in 2 year 0.78 -1.03
(9.03) (5.05)
BBB eligible × Debt maturing w/in 2 year -25.22 19.07
(11.07)∗∗ (9.10)∗∗
Issuer FE ✓ ✓
Adj. R-sqr. 0.03 0.15 0.04 0.01 0.44 0.01
N. of obs 1062 878 1062 538 435 538
N. of clusters 372 188 372 209 106 209
(b) Components of credit spreads
Default-adjusted spread 1 year EDF
(1) (2) (3) (4) (5) (6)
Constant -27.31 -18.24 -25.91 -7.12 -7.78 -8.69
(12.02)∗∗ (16.74) (12.42)∗∗ (2.93)∗∗ (3.87)∗∗ (3.17)∗∗∗
AAA/AA/A eligible -18.59 -15.52 -18.39 4.35 0.29 5.87
(5.93)∗∗∗ (5.26)∗∗∗ (6.18)∗∗∗ (0.98)∗∗∗ (0.27) (1.48)∗∗∗
BBB eligible -26.06 -17.66 -8.58 1.29 -1.75 3.51
(6.22)∗∗∗ (6.06)∗∗∗ (7.40) (1.41) (1.01)∗ (1.93)∗
Debt maturing w/in 2 year -4.02 3.35
(4.30) (1.90)∗
AAA/AA/A eligible × Debt maturing w/in 2 year 1.18 -3.16
(9.47) (1.72)∗
BBB eligible × Debt maturing w/in 2 year -23.52 -4.18
(11.19)∗∗ (2.53)∗
Issuer FE ✓ ✓
Adj. R-sqr. 0.05 0.15 0.06 0.00 0.71 0.00
N. of obs 1041 859 1041 1041 859 1041
N. of clusters 367 185 367 367 185 367
43
Table 5: Facility expansion effects. This table reports the estimated coefficients from the regres-
sion of average daily changes in duration-matched credit spreads, bid-ask spreads, and components
of credit spreads in 3 days around the expansion of the facilities on April 9, 2020. “AAA/AA/A
eligible” (“BBB eligible”)(“Fallen angel eligible”) is a dummy equal to 1 if the bond matures with
5 years, is not issued by a bank or a bank-subsidiary and is rated at least A- (rated BBB+, BBB
or BBB-) (investment grade rated as of March 22 and was subsequently downgraded no lower than
BB-) on a plurality rating basis. “Debt maturing” is a dummy equal to 1 if the issuer has any bonds
maturing within the next 2 years. Standard errors clustered at the issuer level reported in paren-
theses below point estimates. *** significant at 1% level; ** significant at 5% level; * significant at
10% level.
(a) Credit and bid-ask spreads
Duration-matched spread Bid-ask spread
(1) (2) (3) (4) (5) (6)
Constant -22.37 -8.27 -24.57 13.32 3.88 14.07
(10.14)∗∗ (10.60) (10.58)∗∗ (8.79) (12.70) (8.85)
AAA/AA/A eligible 13.59 -0.83 12.81 -3.67 -0.70 -2.68
(3.55)∗∗∗ (2.54) (3.07)∗∗∗ (2.22)∗ (2.90) (2.21)
BBB eligible 6.21 2.84 -2.43 -2.62 -2.66 -3.56
(7.23) (7.50) (6.58) (2.76) (2.31) (5.11)
Fallen angels eligible -47.85 -11.29 -16.79 -34.33 -18.42 -62.34
(21.79)∗∗ (3.20)∗∗∗ (2.16)∗∗∗ (7.36)∗∗∗ (0.83)∗∗∗ (1.48)∗∗∗
Debt maturing w/in 2 year 5.23 -2.73
(3.06)∗ (2.25)
AAA/AA/A eligible × Debt maturing w/in 2 year -0.85 -0.40
(5.62) (3.63)
BBB eligible × Debt maturing w/in 2 year 11.45 2.14
(11.97) (5.97)
Fallen angels eligible × Debt maturing w/in 2 year -38.85 33.84
(27.55) (8.48)∗∗∗
Issuer FE ✓ ✓
Adj. R-sqr. 0.02 0.36 0.02 0.03 0.24 0.03
N. of obs 1075 886 1075 829 689 829
N. of clusters 391 202 391 291 151 291
(b) Components of credit spreads
Default-adjusted spread 1 year EDF
(1) (2) (3) (4) (5) (6)
Constant -21.12 -8.40 -23.58 -0.23 2.05 -0.14
(10.35)∗∗ (10.72) (10.81)∗∗ (1.58) (1.74) (1.58)
AAA/AA/A eligible 13.08 -0.42 12.81 0.22 -0.14 0.09
(3.60)∗∗∗ (2.62) (3.19)∗∗∗ (0.59) (0.15) (0.69)
BBB eligible 6.01 3.27 -1.78 1.39 0.43 0.64
(7.45) (7.62) (6.87) (0.78)∗ (0.39) (0.95)
Fallen angels eligible -47.90 -9.69 -16.57 13.86 14.70 -2.68
(21.76)∗∗ (3.19)∗∗∗ (2.17)∗∗∗ (5.37)∗∗ (0.10)∗∗∗ (0.73)∗∗∗
Debt maturing w/in 2 year 5.78 -0.04
(3.09)∗ (1.37)
AAA/AA/A eligible × Debt maturing w/in 2 year -1.78 0.21
(5.68) (1.53)
BBB eligible × Debt maturing w/in 2 year 9.86 1.15
(12.25) (1.77)
Fallen angels eligible × Debt maturing w/in 2 year -39.40 19.30
(27.58) (4.09)∗∗∗
Issuer FE ✓ ✓
Adj. R-sqr. 0.02 0.35 0.02 0.01 0.77 0.00
N. of obs 44
1036 854 1036 1041 858 1041
N. of clusters 377 195 377 380 197 380
Table 6: Purchase effects. This table reports the estimated coefficients from average daily
changes in duration-matched credit spreads, bid-ask spreads, and components of credit spreads in 3
days around facility purchases. “Purchased” is the 0/1 indicator for the bond being purchased either
directly (as a cash bond) or indirectly (through ETF purchases). “ETF” is the 0/1 indicator for the
bond purchased indirectly through ETFs; “Cash bond” is the 0/1 indicator for the bond purchased
directly. Standard errors clustered at the issuer level reported in parentheses below point estimates.
*** significant at 1% level; ** significant at 5% level; * significant at 10% level.
(a) Duration-matched spreads (b) Default-adjusted spreads
(1) (2) (3) (4) (5) (1) (2) (3) (4) (5)
Constant -1.87 -2.23 -2.23 -2.10 -4.54 Constant -1.16 -1.46 -1.46 -0.84 -3.53
(0.40)∗∗∗ (0.39)∗∗∗ (0.39)∗∗∗ (0.49)∗∗∗ (2.94) (0.68)∗ (0.70)∗∗ (0.70)∗∗ (1.13) (2.88)
Purchased -2.00 -0.08 Purchased -2.03 -0.28
(0.13)∗∗∗ (0.34) (0.14)∗∗∗ (0.41)
ETF -0.07 0.22 -0.00 ETF -0.29 0.01 -0.11
(0.35) (0.36) (0.43) (0.43) (0.43) (0.45)
Cash bond -0.58 -0.81 -0.93 Cash bond -0.47 -0.69 -0.82
(0.28)∗∗ (0.27)∗∗∗ (0.26)∗∗∗ (0.28)∗ (0.26)∗∗∗ (0.26)∗∗∗
Date FE ✓ ✓ ✓ ✓ Date FE ✓ ✓ ✓ ✓
Issuer FE ✓ ✓ Issuer FE ✓ ✓
45 Issue FE ✓ Issue FE ✓
Adj. R-sqr. 0.01 0.07 0.07 0.09 0.13 Adj. R-sqr. 0.00 0.05 0.05 0.06 0.09
N. of obs 179150 179150 179150 179127 179065 N. of obs 176170 176170 176170 176153 176097
N. of clusters 673 673 673 650 648 N. of clusters 633 633 633 616 614
(c) 1 year EDF (d) Bid-ask spread
(1) (2) (3) (4) (5) (1) (2) (3) (4) (5)
Constant -0.24 -0.37 -0.37 -0.63 -1.19 Constant 0.10 0.13 0.13 0.11 0.31
(0.23) (0.20)∗ (0.20)∗ (0.23)∗∗∗ (0.71)∗ (0.25) (0.25) (0.25) (0.31) (0.35)
Purchased 0.11 0.42 Purchased -0.41 -0.46
(0.09) (0.29) (0.09)∗∗∗ (0.32)
ETF 0.44 0.45 0.45 ETF -0.45 -0.56 -0.35
(0.30) (0.35) (0.40) (0.33) (0.37) (0.47)
Cash bond -0.31 -0.39 -0.37 Cash bond -0.62 -0.54 -0.54
(0.14)∗∗ (0.14)∗∗∗ (0.14)∗∗∗ (0.36)∗ (0.35) (0.34)
Date FE ✓ ✓ ✓ ✓ Date FE ✓ ✓ ✓ ✓
Issuer FE ✓ ✓ Issuer FE ✓ ✓
Issue FE ✓ Issue FE ✓
Adj. R-sqr. 0.00 0.02 0.02 0.03 0.02 Adj. R-sqr. 0.00 0.00 0.00 0.01 0.00
N. of obs 176400 176400 176400 176382 176324 N. of obs 227320 227320 227320 227314 227296
N. of clusters 634 634 634 616 614 N. of clusters 543 543 543 537 537
Table 7: Facility closure announcement effects. This table reports the estimated coefficients
from the regression of average daily changes in duration-matched credit spreads, bid-ask spreads,
and components of credit spreads in 3 days around Secretary Mnuchin’s letter requesting the return
of CARES Act funds on November 19, 2020 and in 3 days around the closure of the facilities on
December 31, 2020. “SMCCF eligible” is a dummy equal to 1 if the bond matures with 5 years,
is not issued by a bank or a bank-subsidiary and was investment grade rated as of March 22 and
was subsequently downgraded no lower than BB- on a plurality rating basis. “Debt maturing” is a
dummy equal to 1 if the issuer has any bonds maturing within the next 2 years. Standard errors
clustered at the issuer level reported in parentheses below point estimates. *** significant at 1%
level; ** significant at 5% level; * significant at 10% level.
(a) Credit and bid-ask spreads
Duration-matched spread Bid-ask spread
November 19 December 29 November 19 December 29
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)
Constant -4.89 -5.55 -5.64 -1.90 -2.89 -1.58 -0.84 -2.99 -0.57 5.25 4.89 4.90
(3.86) (2.82)∗∗ (3.99) (2.53) (2.16) (2.78) (2.76) (2.81) (2.82) (5.06) (5.54) (5.26)
SMCCF eligible 1.66 0.77 1.08 -0.41 0.15 -1.16 0.62 0.02 0.28 -1.24 -0.02 -1.62
(0.87)∗ (0.35)∗∗ (1.44) (0.60) (0.63) (1.04) (0.60) (0.73) (0.78) (0.89) (1.61) (1.17)
Debt maturing w/in 2 year 1.32 -0.47 -0.51 0.52
(1.02) (0.73) (0.88) (1.56)
Debt maturing w/in 2 year× eligible 0.70 1.29 0.64 0.43
(1.60) (1.52) (1.16) (1.77)
Issuer FE ✓ ✓ ✓ ✓
Adj. R-sqr. 0.02 0.52 0.03 0.05 0.21 0.05 0.00 0.31 -0.00 0.01 0.11 0.01
N. of obs 1159 969 1159 865 672 865 1198 1033 1198 806 668 806
N. of clusters 417 227 417 354 161 354 402 237 402 298 160 298
(b) Components of credit spreads
Default-adjusted spread 1 year EDF
November 19 December 29 November 19 December 29
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)
Constant -4.48 -5.57 -5.32 -0.87 -2.28 -0.64 1.64 -0.50 1.27 1.81 0.22 2.20
(3.82) (2.84)∗ (3.89) (2.38) (2.00) (2.60) (3.39) (0.76) (3.68) (1.04)∗ (0.68) (1.21)∗
SMCCF eligible 1.60 0.81 1.12 -0.42 0.14 -1.15 0.44 -1.46 1.67 0.20 0.08 -0.08
(0.83)∗ (0.34)∗∗ (1.43) (0.61) (0.65) (1.05) (0.95) (0.86)∗ (0.69)∗∗ (0.27) (0.09) (0.26)
Debt maturing w/in 2 year 1.41 -0.34 0.64 -0.55
(0.98) (0.72) (0.98) (0.49)
Debt maturing w/in 2 year× eligible 0.53 1.21 -2.24 0.57
(1.56) (1.53) (1.54) (0.62)
Issuer FE ✓ ✓ ✓ ✓
Adj. R-sqr. 0.02 0.54 0.02 0.04 0.20 0.04 0.01 0.89 0.01 0.05 0.86 0.05
N. of obs 1143 956 1143 856 665 856 1143 956 1143 858 667 858
N. of clusters 410 223 410 350 159 350 410 223 410 350 159 350
46
Table 8: Dealer constraints and secondary market intermediation. This table reports the
estimated probit model coefficients for the probability of dealer d buying bond b on date t from non-
dealer customers (columns 1-4), as well as the corresponding intensive margin (column 5). “SMCCF
eligible” is a dummy equal to 1 if the bond matures with 5 years, is not issued by a bank or a
bank-subsidiary and was investment grade rated as of March 22 and was subsequently downgraded
no lower than BB- on a plurality rating basis. “Bank affiliated” is a dummy equal to 1 if the dealer
is affiliated with a U.S. BHC. “Pandemic onset” is a dummy equal to 1 if trade occurs between
February 22, 2020 and March 21, 2020; “Facility implementation” is a dummy equal to 1 if trade
occurs between March 22, 2020 and June 29, 2020; “Facility operational” is a dummy equal to 1 if
trade occurs between June 30, 2020 and December 31, 2020. Reference period is January 1, 2019
– February 21, 2020. “R-sqr” is the pseudo R2 for the probit regressions, and the adjusted R2 for
the intensive margin regression. All regressions control for standard bond characteristics. Standard
errors clustered at the issuer level reported in parentheses below point estimates. *** significant at
1% level; ** significant at 5% level; * significant at 10% level.
(1) (2) (3) (4) (5)
Constant -2.09 -2.22 -2.22 -2.64 -30.90
(0.03)∗∗∗ (0.03)∗∗∗ (0.03)∗∗∗ (0.07)∗∗∗ (6.22)∗∗∗
Pandemic onset 0.11 0.12 0.11 0.12 1.64
(0.01)∗∗∗ (0.01)∗∗∗ (0.01)∗∗∗ (0.01)∗∗∗ (0.32)∗∗∗
Facility implementation 0.08 0.10 0.09 0.09 1.14
(0.01)∗∗∗ (0.01)∗∗∗ (0.01)∗∗∗ (0.01)∗∗∗ (0.23)∗∗∗
Facility operational 0.01 0.02 0.02 0.01 -0.04
(0.01) (0.01)∗∗∗ (0.01)∗∗∗ (0.01) (0.02)∗
Bank-affiliated 0.35 0.36
(0.01)∗∗∗ (0.01)∗∗∗
Bank-affiliated × Pandemic onset -0.02 0.00 -0.00 -0.03
(0.01)∗∗∗ (0.01) (0.01) (0.06)
Bank-affiliated × Facility implementation -0.03 -0.01 -0.01 -0.01
(0.01)∗∗∗ (0.01) (0.01)∗ (0.08)
Bank-affiliated × Facility operational -0.02 -0.01 -0.02 -0.46
(0.01)∗∗∗ (0.01)∗∗ (0.01)∗∗∗ (0.08)∗∗∗
SMCCF eligible 0.01 0.01 0.09
(0.02) (0.02) (0.03)∗∗∗
SMCCF eligible × Pandemic onset 0.02 0.03 0.49
(0.01)∗∗ (0.01)∗∗∗ (0.09)∗∗∗
SMCCF eligible × Facility implementation 0.04 0.05 0.73
(0.01)∗∗∗ (0.01)∗∗∗ (0.13)∗∗∗
SMCCF eligible × Facility operational 0.00 0.01 0.18
(0.01) (0.01) (0.04)∗∗∗
Bank-affiliated × SMCCF eligible -0.05 0.01 -0.25
(0.01)∗∗∗ (0.01) (0.05)∗∗∗
Bank-affiliated × SMCCF eligible -0.07 -0.08 -1.01
× Pandemic onset (0.01)∗∗∗ (0.01)∗∗∗ (0.20)∗∗∗
Bank-affiliated × SMCCF eligible -0.09 -0.10 -1.30
× Facility implementation (0.01)∗∗∗ (0.01)∗∗∗ (0.27)∗∗∗
Bank-affiliated × SMCCF eligible -0.04 -0.05 -0.46
× Facility operational (0.01)∗∗∗ (0.01)∗∗∗ (0.12)∗∗∗
Mills ratio 16.14
(3.05)∗∗∗
Bond chars ✓ ✓ ✓ ✓ ✓
Dealer FE ✓ ✓
R.-sqr. 0.0247 0.04 0.04 0.16 0.12
N. of obs 86,330,397 86,330,397 86,330,397 80,183,928 1,065,724
N. of clusters 834 834 834 834 832
Table 9: Facility counterparties and secondary market intermediation. This table reports
the estimated probit model coefficients for the probability of dealer d buying bond b on date t from
non-dealer customers. “SMCCF eligible” is a dummy equal to 1 if the bond matures with 5 years, is
not issued by a bank or a bank-subsidiary and was investment grade rated as of March 22 and was
subsequently downgraded no lower than BB- on a plurality rating basis. “Eligible seller” is a dummy
equal to 1 if the dealer becomes a facility counterparty during facility purchases; “registration date
post operational” is a dummy equal to 1 if the dealer’s registration date with the facility is post
June 29, 2020. “Pandemic onset” is a dummy equal to 1 if trade occurs between February 22, 2020
and March 21, 2020; “Facility implementation” is a dummy equal to 1 if trade occurs between March
22, 2020 and June 29, 2020; “Facility operational” is a dummy equal to 1 if trade occurs between
June 30, 2020 and December 31, 2020. Reference period is January 1, 2019 – February 21, 2020.
“R-sqr” is the pseudo R2 for the probit regressions, and the adjusted R2 for the intensive margin
regression. All regressions control for standard bond characteristics. Standard errors clustered at the
issuer level reported in parentheses below point estimates. *** significant at 1% level; ** significant
at 5% level; * significant at 10% level.
(1) (2) (3) (4) (5)
Constant -2.23 -2.24 -2.24 -2.63 -30.87
(0.03)∗∗∗ (0.03)∗∗∗ (0.03)∗∗∗ (0.07)∗∗∗ (6.24)∗∗∗
Pandemic onset 0.10 0.09 0.09 0.10 1.33
(0.01)∗∗∗ (0.01)∗∗∗ (0.01)∗∗∗ (0.01)∗∗∗ (0.26)∗∗∗
Facility implementation 0.05 0.05 0.05 0.05 0.69
(0.01)∗∗∗ (0.01)∗∗∗ (0.01)∗∗∗ (0.01)∗∗∗ (0.14)∗∗∗
Facility operational -0.01 -0.01 -0.01 -0.01 -0.29
(0.01) (0.01) (0.01) (0.01) (0.05)∗∗∗
Eligible seller 0.47 0.49 0.49
(0.01)∗∗∗ (0.02)∗∗∗ (0.02)∗∗∗
Eligible seller × Pandemic onset 0.02 0.04 0.04 0.03 -0.52
(0.01)∗∗ (0.01)∗∗∗ (0.01)∗∗∗ (0.03) (0.21)∗∗
Eligible seller × Facility implementation 0.05 0.05 0.05 0.03 -0.25
(0.01)∗∗∗ (0.01)∗∗∗ (0.01)∗∗∗ (0.03) (0.25)
Eligible seller × Facility operational 0.02 0.01 0.02 -0.02 -1.36
(0.01)∗ (0.01) (0.01)∗∗ (0.03) (0.22)∗∗∗
SMCCF eligible 0.03 0.03 0.03 0.41
(0.02) (0.02) (0.03) (0.09)∗∗∗
SMCCF eligible × Pandemic onset 0.03 0.03 0.04 0.59
(0.01)∗∗ (0.01)∗∗ (0.01)∗∗∗ (0.11)∗∗∗
SMCCF eligible × Facility implementation 0.01 0.01 0.02 0.29
(0.01) (0.01) (0.02) (0.05)∗∗∗
SMCCF eligible × Facility operational -0.02 -0.02 -0.01 -0.17
(0.01) (0.01) (0.02) (0.04)∗∗∗
Eligible seller × SMCCF eligible -0.05 -0.05 -0.05 -1.02
(0.02)∗∗∗ (0.02)∗∗∗ (0.02)∗∗ (0.16)∗∗∗
Eligible seller × SMCCF eligible -0.08 -0.08 -0.10 -1.29
× Pandemic onset (0.02)∗∗∗ (0.02)∗∗∗ (0.02)∗∗∗ (0.25)∗∗∗
Eligible seller × SMCCF eligible -0.02 -0.02 -0.02 -0.12
× Facility implementation (0.02) (0.02) (0.02) (0.09)
Eligible seller × SMCCF eligible 0.01 -0.02 -0.01 -0.04
× Facility operational (0.02) (0.02) (0.02) (0.06)
Eligible seller -0.04 0.09 1.50
× registration date post operational × Facility operational (0.01)∗∗∗ (0.01)∗∗∗ (0.25)∗∗∗
Eligible seller × SMCCF eligible 0.10 0.11 1.77
× registration date post operational × Facility operational (0.01)∗∗∗ (0.02)∗∗∗ (0.28)∗∗∗
Mills ratio 16.19
(3.07)∗∗∗
Bond chars ✓ ✓ ✓ ✓ ✓
Dealer FE ✓ ✓
R.-sqr. 0.05 0.05 0.05 0.17 0.12
N. of obs 4886,330,397 86,330,397 86,330,397 80,183,928 1,065,724
N. of clusters 834 834 834 834 832
Table 10: Dealer constraints and offering day underwriting profits. This table reports the
estimated coefficients from the regression of profits of dealer d underwriting bond b on date t on
bond characteristics, sub-period dummies and bank characteristics. “SMCCF eligible” is a dummy
equal to 1 if the bond matures with 5 years, is not issued by a bank or a bank-subsidiary and
was investment grade rated as of March 22 and was subsequently downgraded no lower than BB-
on a plurality rating basis. “Bank affiliated” is a dummy equal to 1 if the dealer is affiliated with
a U.S. BHC. “Pandemic onset” is a dummy equal to 1 if issued between February 22, 2020 and
March 21, 2020; “Facility implementation” is a dummy equal to 1 if issued between March 22, 2020
and June 29, 2020; “Facility operational” is a dummy equal to 1 if issued between June 30, 2020
and December 31, 2020. Reference period is January 1, 2017 – February 21, 2020. All regressions
control for standard bond characteristics. Standard errors clustered at the issuer level reported in
parentheses below point estimates. *** significant at 1% level; ** significant at 5% level; * significant
at 10% level.
(1) (2) (3) (4) (5)
Constant 1.91 1.74 21.03 19.21 8.58
(0.30)∗∗∗ (0.28)∗∗∗ (4.00)∗∗∗ (5.54)∗∗∗ (4.41)∗
Pandemic onset 0.07 0.09 2.94 -1.04 0.14
(0.14) (0.14) (2.97) (2.27) (2.30)
Facility implementation 0.13 0.12 -0.35 -1.47 -0.84
(0.06)∗∗ (0.06)∗∗ (2.30) (2.33) (2.60)
Facility operational -0.00 0.00 0.80 -1.48 -0.21
(0.07) (0.08) (1.71) (1.80) (1.83)
SMCCF eligible 0.20 -0.51 -3.25 -3.49
(0.04)∗∗∗ (1.64) (1.61)∗∗ (1.79)∗
SMCCF eligible × Pandemic onset -0.29 6.28 11.63 11.49
(0.15)∗ (3.21)∗ (2.71)∗∗∗ (3.71)∗∗∗
SMCCF eligible × Facility implementation 0.14 11.55 11.97 12.78
(0.12) (3.94)∗∗∗ (4.88)∗∗ (4.65)∗∗∗
SMCCF eligible × Facility operational -0.15 -2.51 -0.14 -1.11
(0.08)∗ (3.12) (3.36) (3.51)
Bank-affiliated -14.46
(1.69)∗∗∗
Bank-affiliated × Pandemic onset -3.29 4.46 7.11
(4.19) (3.05) (3.09)∗∗
Bank-affiliated × Facility implementation 2.22 5.09 4.74
(2.98) (2.83)∗ (3.18)
Bank-affiliated × Facility operational -2.80 -0.31 -1.07
(2.50) (2.49) (2.66)
Bank-affiliated × SMCCF eligible -1.00 2.55 3.31
(2.66) (2.62) (2.74)
Bank-affiliated × SMCCF eligible × Pandemic onset -20.59 -28.11 -30.40
(4.65)∗∗∗ (5.13)∗∗∗ (5.51)∗∗∗
Bank-affiliated × SMCCF eligible × Facility implementation -19.60 -19.43 -21.26
(7.62)∗∗ (8.17)∗∗ (7.89)∗∗∗
Bank-affiliated × SMCCF eligible × Facility operational 4.54 0.94 0.59
(5.23) (5.49) (5.85)
Bond chars ✓ ✓ ✓ ✓ ✓
Dealer FE ✓ ✓
Issuer FE ✓
Adj. R.-sqr. -0.00 -0.00 0.03 0.10 0.07
N. of obs 8857 8857 8857 8837 8814
N. of clusters 485 485 485 485 462
49
Table 11: Dealer constraints and time to profitability. This table reports the estimated
coefficients from the censored regression of number of days to profitability dealer d underwriting
bond b on date t on bond characteristics, sub-period dummies and bank characteristics. “SMCCF
eligible” is a dummy equal to 1 if the bond matures with 5 years, is not issued by a bank or a
bank-subsidiary and was investment grade rated as of March 22 and was subsequently downgraded
no lower than BB- on a plurality rating basis. “Bank affiliated” is a dummy equal to 1 if the dealer
is affiliated with a U.S. BHC. “Pandemic onset” is a dummy equal to 1 if issued between February
22, 2020 and March 21, 2020; “Facility implementation” is a dummy equal to 1 if issued between
March 22, 2020 and June 29, 2020; “Facility operational” is a dummy equal to 1 if issued between
June 30, 2020 and December 31, 2020. Reference period is January 1, 2017 – February 21, 2020. All
regressions control for standard bond characteristics. Standard errors clustered at the issuer level
reported in parentheses below point estimates. *** significant at 1% level; ** significant at 5% level;
* significant at 10% level.
(1) (2) (3) (4) (5)
Constant 18.12 18.74 12.53 7.65 2.77
(1.80)∗∗∗ (1.82)∗∗∗ (1.81)∗∗∗ (1.63)∗∗∗ (1.45)∗
Pandemic onset 0.98 0.89 -0.21 1.10 1.52
(0.59)∗ (0.58) (0.98) (0.64)∗ (0.74)∗∗
Facility implementation 0.45 0.43 -0.26 0.30 0.45
(0.27)∗ (0.28) (0.52) (0.47) (0.58)
Facility operational -0.08 -0.07 -0.57 0.41 0.37
(0.32) (0.34) (0.39) (0.38) (0.44)
SMCCF eligible -0.62 -0.43 0.53 1.04
(0.19)∗∗∗ (0.46) (0.40) (0.44)∗∗
SMCCF eligible × Pandemic onset 6.79 -2.46 -1.72 -5.70
(0.56)∗∗∗ (1.05)∗∗ (0.82)∗∗ (1.20)∗∗∗
SMCCF eligible × Facility implementation 0.31 -1.01 -1.34 -2.34
(0.44) (0.75) (0.77)∗ (1.00)∗∗
SMCCF eligible × Facility operational 0.21 0.92 0.00 -0.28
(0.62) (1.03) (0.82) (0.95)
Bank-affiliated 4.50
(0.33)∗∗∗
Bank-affiliated × Pandemic onset 1.35 -1.49 -1.91
(1.39) (0.96) (1.04)∗
Bank-affiliated × Facility implementation 0.41 -0.67 -0.87
(0.81) (0.67) (0.76)
Bank-affiliated × Facility operational 1.34 0.25 0.19
(0.64)∗∗ (0.64) (0.72)
Bank-affiliated × SMCCF eligible 0.36 -0.77 -0.97
(0.71) (0.64) (0.67)
Bank-affiliated × SMCCF eligible × Pandemic onset 61.74 57.82 61.10
(2.86)∗∗∗ (2.53)∗∗∗ (2.52)∗∗∗
Bank-affiliated × SMCCF eligible × Facility implementation 2.71 2.81 3.62
(1.42)∗ (1.37)∗∗ (1.48)∗∗
Bank-affiliated × SMCCF eligible × Facility operational -1.36 -0.11 0.12
(1.60) (1.46) (1.57)
Bond chars ✓ ✓ ✓ ✓ ✓
Dealer FE ✓ ✓
Issuer FE ✓
Pseudo R.-sqr. 0.00 0.00 0.01 0.03 0.04
N. of obs 8857 8857 8857 8857 8857
N. of clusters 485 485 485 485 485
50
Figure 2. Primary market issuance improved since CCF announcement. This
figure plots the year-to-date cumulative USD amount of fixed coupon bonds issued by non-
financial issuers in 2020 and the corresponding median cumulative issuance from 2010 to
2019. Event lines at March 22 (initial CCF announcement) and June 29 (PMCCF opera-
tional).
(a) Investment grade, below 5 year (b) Investment grade, above 5 year
60
Cumulative amount issued ($ Billion) Cumulative amount issued ($ Billion)
Median 2010-2019 2020 600
Median 2010-2019 2020
50
500
40
400
30 300
20 200
10 100
0 1
2
3
4
5
6
7 0 1
2
3
4
5
6
7
8
9
10
11
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30 12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51 35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
Week of year Week of year
(c) High yield
60
Cumulative amount issued ($ Billion)
Median 2010-2019 2020
50
40
30
20
10
0 1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
Week of year
51
Figure 3. Spreads have retraced from March 22 highs. This figure plots the esti-
mated coefficients from the regression of cumulative bond-level changes in duration-matched
spreads, default-adjusted spreads, 1 year expected default frequency and bid-ask spread on
SMCCF eligibility dummy and standard bond and issuer characteristics. 95% confidence
bands based on standard errors clustered at the issuer level reported as shaded areas around
the point estimate. Regressions estimated as repeated cross-sections for each trading date in
the sample. Event lines at: March 22 (initial CCF announcement); April 9 (first term sheet
update); May 12 (commencement of ETF purchases); June 16 (commencement of cash bond
purchases); June 29 (PMCCF operational); November 19 (facility closure announcement).
(a) Duration-matched spreads (b) Default-adjusted spreads
200 200
Estimated coefficient (bps) Estimated coefficient (bps)
0 0
-200 -200
-400 -400
-600 -600
8Feb
15Feb
22Feb
29Feb
7Mar
14Mar
21Mar
28Mar 8Feb
15Feb
22Feb
29Feb
7Mar
14Mar
21Mar
28Mar
4Apr
11Apr
18Apr
25Apr
2May
9May
16May
23May 4Apr
11Apr
18Apr
25Apr
2May
9May
16May
23May
30May
6Jun
13Jun
20Jun
27Jun
4Jul
11Jul
18Jul 30May
6Jun
13Jun
20Jun
27Jun
4Jul
11Jul
18Jul
25Jul
1Aug
8Aug
15Aug
22Aug
29Aug
5Sep
12Sep 25Jul
1Aug
8Aug
15Aug
22Aug
29Aug
5Sep
12Sep
19Sep
26Sep
3Oct
10Oct
17Oct
24Oct
31Oct
7Nov 19Sep
26Sep
3Oct
10Oct
17Oct
24Oct
31Oct
7Nov
14Nov
21Nov
28Nov
5Dec
12Dec
19Dec
26Dec
2Jan 14Nov
21Nov
28Nov
5Dec
12Dec
19Dec
26Dec
2Jan
Constant SM-CCF eligible Constant SM-CCF eligible
(c) 1 year EDF (d) Bid-ask spreads
Estimated coefficient (cent)
100
Estimated coefficient (bps)
100
0
0
-100
-100
-200
-200 8Feb
15Feb
22Feb -300 8Feb
15Feb
22Feb
29Feb
7Mar
14Mar
21Mar
28Mar
4Apr
11Apr
18Apr 29Feb
7Mar
14Mar
21Mar
28Mar
4Apr
11Apr
18Apr
25Apr
2May
9May
16May
23May
30May
6Jun
13Jun 25Apr
2May
9May
16May
23May
30May
6Jun
13Jun
20Jun
27Jun
4Jul
11Jul
18Jul
25Jul
1Aug
8Aug 20Jun
27Jun
4Jul
11Jul
18Jul
25Jul
1Aug
8Aug
15Aug
22Aug
29Aug
5Sep
12Sep
19Sep
26Sep
3Oct 15Aug
22Aug
29Aug
5Sep
12Sep
19Sep
26Sep
3Oct
10Oct
17Oct
24Oct
31Oct
7Nov
14Nov
21Nov
28Nov 10Oct
17Oct
24Oct
31Oct
7Nov
14Nov
21Nov
28Nov
5Dec
12Dec
19Dec
26Dec
2Jan 5Dec
12Dec
19Dec
26Dec
2Jan
Constant SM-CCF eligible Constant SM-CCF eligible
52
A Technical appendix
A.1 TRACE data cleaning
In our analysis, we use TRACE data provided by FINRA at the end of each business day.
Starting in July 2002, each registered FINRA member that is a party to a reportable trans-
action in a TRACE-eligible security has a reporting obligation. The reporting is done in
real-time. The set of TRACE-eligible securities has changed throughout the years. We start
our sample in 2005, when all investment-grade and high-yield U.S. corporate bonds were
included in the TRACE-eligible securities definition (except for 144A). A trade report in-
cludes the security identifier, date, time, size (par value), and price of the transaction. A
report also identifies the member firm’s side of the transaction (buy or sell), their capacity
as a principal or agent, and the other parties to the transaction. The required reporting time
varies between categories of TRACE-eligible securities. Member firms must report a sec-
ondary corporate bond transaction as soon as practicable, no later than within 15 minutes
of the time of execution. There a few issues that needs to be addressed:
1. Correction and Cancellations. A trade record that is corrected or cancelled at a
later time because of misreporting remains on the tape, and additional records indicate
its current status.
What do we do? We keep the most recent status of each trade record based on the
system control number and the record type.
2. Interdealer Trades. The reporting requirements require all registered broker-dealers
(BDs) to report to TRACE. Hence, a trade between two BDs is reported twice, while
a trade between a client and a BD is reported once.
What do we do? To keep one record of each trade, we keep the sell side of an
interdealer trade.
3. Non-Member Affiliates. While BDs are identified in trade records, clients’ identities
are masked, and all clients are reported as “C”. Effective on November 2, 2015, firms
are required to identify transactions with non-member affiliates , entering “A” instead
of “C” if the affiliate is a non-FINRA member.
The reporting rule amendment also requires firms to use an indicator to identify cer-
tain trades that typically are not economically distinct and, as such, would not pro-
vide investors useful information for pricing, valuation or risk evaluation purposes if
disseminated publicly. Specifically, FINRA is requiring firms to identify trades with
non-member affiliates that occur within the same day and at the same price as a trade
between the firm and another contra-party in the same security. Thus, firms are re-
quired to use “non-member affiliate—principal transaction indicator” when reporting a
transaction to TRACE in which both the member and its non-member affiliate act in a
principal capacity, and where such trade occurs within the same day, at the same price
and in the same security as a transaction between the member and another counter-
party. A firm is not required to append the indicator if it does not reasonably expect
1
to engage in a same day, same price transaction in the same security with another
counterparty as with a non-member affiliate.
What do we do? We exclude records where the field SPCL_PRCSG_CD is non-
missing. In addition, for volume calculations, we break down dealer-to-client (DC) and
dealer-to-affiliate (DA) trading activity. We exclude non-member affiliate trades with
the same price and the same size that happen within 60 seconds of each other.
4. Trades on Electronic Platforms. With the growth of electronic trading platforms,
we see more transactions being executed through such platforms. Electronic platforms
may or may not have a reporting obligation. The reporting obligation of an electronic
platform is dependent on whether the platform is a party to the trade, and a registered
alternative trading system (ATS) with the SEC. An ATS platform is a party to all
transactions executed through its system, and therefore has a reporting obligation. An
electronic platform that is not an ATS is not necessarily a party to all trades executed
through its system so may not always have a reporting obligation.
Trades on an electronic platform which also has a reporting obligation increases the
number of observations in the TRACE data. For example, a trade between two member
firms on an electronic platform with a reporting obligation results in four observations
in the TRACE data: a sell by the first member firm to the platform, a purchase by
the platform from the first member firm, a sell by the platform to the second member
firm, and a purchase by the second member firm from the platform. This needs to
be addressed to avoid an upward-bias of trading activity, and a downward bias of
price-based liquidity measures.
What do we do? Depending on the analysis, one might want to flag such trades.
We use the counterparties identities and FINRA’s TRACE ATS identifiers list to flag
such trades. We also construct an additional trade size variable that reset to 0 if
the seller is an ATS platform. For trading volume calculations, for example, we use the
ATS-adjusted volume variable. If we do not account for multiple trade reports, then we
would include some trades more than once depending on whether the counterparties are
FINRA members and whether an electronic platform also had a reporting obligation.
This would result in an overestimation of the trading activity on electronic platforms
with a reporting obligation (e.g., non-6732 ATSs), and an inaccurate comparison of the
trading activity between platforms with different reporting obligations (e.g., 6732 ATSs
and non-6732 ATSs). Overall, the filter that we apply to the TRACE data ensures that
we include each trade only once in our sample.
A.2 Spreads calculation
We begin by computing duration-matched spreads at the bond-trade level. As in Gilchrist
and Zakrajšek (2012), define the Treasury-implied yield yb,t
f
on bond b on trade date t as
2T s 2T Cb
X Cb X
2 100
Zt + 100Zt (T ) = s + 2T ,
2 2 f
yb,t y f
s=1 s=1 1+ 2 1+ b,t
2
2
where T is the time-to-maturity of the bond, Cb is the coupon on the bond, and Zt (s) is the
Treasury zero-coupon bond price for time-to-maturity s. The trade-level duration-matched
spread on bond b on trade date t is then
f
zb,k,t = yb,k,t − yb,t ,
where yb,k,t is the yield on bond b priced in trade k on trade date t. We aggregate to the
bond-trade day level by averaging using trading volume weights:
P
k∈K zb,k,t Vb,k,t
zb,t = P b,t ,
k∈Kb,t Vb,k,t
where Kb,t is the set of all trades in bond b in on trading day t and Vb,k,t is the volume of the
k th trade in bond b on trade date t.
Duration-matched spreads measure the spread differential between corporate bonds and
Treasuries with similar duration, capturing risk premia for both the differential credit and
liquidity risk between Treasuries and corporate bonds. To separate these two components,
similar to Gilchrist and Zakrajšek (2012), we estimate the duration-matched spread that
would be predicted based on bond and issuer characteristics using the following regression
log zb,t = α + βEDFb,t + ⃗γ Fb,t + ϵb,t ,
where EDFb,t is the one year expected default probability for bond b on day t estimated
by Moody’s KMV, and Fb,t is a vector of bond and issuer characteristics: log duration, log
amount outstanding, log age of the bond, log coupon rate, a dummy for call provision, and
a 3-digit NAICS industry fixed effect. When bond-level EDFs are not available, we use the
issuer-level EDF instead and include a dummy variable for whether bond- or issuer-level is
used in the specification.
We estimate this regression separately for each credit rating category, allowing different
credit ratings to have a different relationship between expected duration-matched spreads
and bond characteristics. Table A.3 reports the estimated coefficients for the above regression
for the full sample January 1, 2005 – June 30, 2020. The default-adjusted spread for bond b
on date t is then calculated as the difference between the priced and the predicted duration-
matched spread on bond b on date t
σ2
db,t = zb,t − exp α + βEDFb,t + ⃗γ Fb,t + ,
2
where σ 2 is the estimated variance of the idiosyncratic error ϵb,t .
A.3 Credit ratings
For secondary market functioning, we classify bonds into investment-grade and speculative
grade (high yield) categories based on the issue-level credit ratings reported in Mergent
FISD. We coalesce bond-level ratings by multiple rating agencies into a single number based
on the plurality rule: if a bond is rated by more than one agency, we use the rating agreed
3
upon by at least two rating agencies and use the lowest available rating otherwise. For our
purposes, a bond is identified as investment-grade if its plurality rating is BBB- or higher
on the S&P ratings scale, or equivalent, and as high yield if its plurality rating is between
BB+ and C, inclusive, on the S&P ratings scale, or equivalent. In our sample, few bonds
that were investment-grade as of March 22, 2020, and have subsequently been downgraded
to BB+/BB/BB-; to keep our definitions consistent with facility eligibility, we include those
bonds in the investment-grade category. Bonds that were investment-grade as of March 22,
2020, and have subsequently been downgraded to below BB- on the S&P scale or equivalent
but remain rated are included in the high yield category.
Similarly, for primary market functioning, we classify issuers into investment-grade and
speculative grade (high yield) categories based on the issuer-level plurality rating, with S&P,
Moody’s and Fitch issuer-level ratings collected from Thompson Reuters Eikon.
A.4 Nearest maturity Treasury spreads
Primary market issuances are priced as a spread to nearest-maturity on-the-run Treasury
yields. In particular, we use the following maturity matches in computing the offering spread
to the on-the-run Treasury:
• For bonds with less that 4.5 month maturity, spread to the 3 month Treasury bill
• For bonds with maturity of 4.5 months or more and less than 9 months, spread to the
6 month Treasury bill
• For bonds with maturity of 9 months or more and less than 1.5 years, spread to the 1
year Treasury note
• For bonds with maturity of [1.5, 2.5) years, spread to the 2 year Treasury note
• For bonds with maturity of [2.5, 4) years, spread to the 3 year Treasury note
• For bonds with maturity of [4, 6) years, spread to the 5 year Treasury note
• For bonds with maturity of [6, 8.5) years, spread to the 7 year Treasury note
• For bonds with maturity of [8.5, 20) years, spread to the 10 year Treasury bond
• For bonds with 20 years maturity or more, spread to the 20 year Treasury bond
Note that we exclude bonds with more than 40 years maturity (including perpetual bond)
from the offering spread summary statistics.
4
Table A.1: PMCCF Timeline of Major Events. This table summarizes the major events as
of the time of writing for the Primary Market Corporate Credit Facility (PMCCF).
Date Event
March 22, 2020 PMCCF approved unanimously by the Board of Governors and the Secretary
of the Treasurya
March 23, 2020 Public announcement and initial Term Sheet publishedb
Keys:c
• The PMCCF and SMCCF are designed to work together to support the
flow of credit to large investment-grade U.S. corporations so that they can
maintain business operations and capacity during the period of dislocations
relative to COVID-19.
April 9, 2020 Updated Term Sheet Publishedd
Key Facts:
• Treasury capital increased from $10B to $50B
• Extended eligibility to firms that were rated IG as of March 22, 2020 and
downgraded to no lower than BB- at the time of accessing the facility (“fallen
angels”)
• The PMCCF will buy bonds and syndicated loans with maturities up to
four years via two different mechanisms:
1. As the sole investor in newly issued corporate bonds
2. As a participant in a loan or bond syndication at issuance. Facility may
purchase no more than 25 percent of any loan syndication or bond issuance.
June 29, 2020 Launch datee
Updated Term Sheet Publishedf
Key Facts:
• Pricing of individual corporate bonds will be issuer specific, informed by
market conditions, plus a 100 bps fee, and subject to minimum and maximum
yield spreads over comparable U.S. Treasury Securities
• Pricing of syndicated loans will be the same as that of other syndicate
members, plus a 100 bps fee
November 19, 2020 Letter from Secretary Mnuchin requesting that the Federal Reserve return
the unused funds from CARES Act-funded facilities (PMCCF, SMCCF, MLF,
MSLP, and TALF)g
December 31, 2020 Authority to purchase eligible assets expired
Continued on next page
5
Table A.1 – Continued from previous page
Date Event
a Federal Reserve Board: https://www.federalreserve.gov/publications/files/
primary-market-corporate-credit-facility-3-29-20.pdf
b Federal Reserve Board: https://www.federalreserve.gov/newsevents/pressreleases/files/
monetary20200323b1.pdf
c Federal Reserve Board: https://www.federalreserve.gov/newsevents/pressreleases/
monetary20200323b.htm
d Federal Reserve Board: https://www.federalreserve.gov/newsevents/pressreleases/files/
monetary20200409a5.pdf
e Federal Reserve Bank of New York: https://www.newyorkfed.org/newsevents/news/markets/
2020/20200629
f Federal Reserve Board: https://www.federalreserve.gov/newsevents/pressreleases/
monetary20200629a.htm
g U.S. Department of the Treasury: https://home.treasury.gov/news/press-releases/sm1190
6
Table A.2: SMCCF Timeline of Major Events. This table summarizes the major events as
of the time of writing for the Secondary Market Corporate Credit Facility (SMCCF).
Date Event
March 23, 2020 Initial Term Sheet publisheda
Key Facts:
• The PMCCF and SMCCF are designed to work together to support the flow of
credit to large investment-grade U.S. corporations so that they can maintain business
operations and capacity during the period of dislocation related to COVID-19.b
• The SMCCF can purchase ETFs or individual corporate bonds
April 9, 2020 Updated Term Sheet publishedc
Key Facts:
• Treasury capital increased from $10B to $25B
• Extended eligibility to bonds issued by firms that were rated IG as of March 22,
2020 and no lower than BB- when purchased by facility (“fallen angels”).
• Extended eligibility to high yield ETFs, with a “preponderance” in investment-grade
ETFs
• Concentration limits apply (max 1.5% of CCFs; max 10% of issuers’ bonds)
May 12, 2020 Began purchasing ETFsd
June 15, 2020 Updated Term Sheet publishede
Updated FAQs releasedf
Key Facts:
• The SMCCF will purchase corporate bonds to construct a corporate bond portfolio
that tracks a broad market index developed for the SMCCF
• The facility can purchase a broad market index of individual bonds from corpora-
tions that satisfy a few simple criteria: maturity of under 5 years, domiciled in the
U.S., not an insured depository institution, and meets the issuer rating requirements
for Eligible Individual Corporate Bonds
• Individual issuer weights will form the basis of sector weights, with each issuer
mapped to one of twelve sectors. Purchases of corporate bonds will track as closely
as possible the sector weights in the index.
June 16, 2020 Began purchasing individual corporate bondsg
November 19, 2020 Letter from Secretary Mnuchin requesting that the Federal Reserve return the unused
funds from CARES Act-funded facilities (PMCCF, SMCCF, MLF, MSLP, and TALF)h
December 31, 2020 Authority to purchase eligible assets expired
June 3, 2021 Announces the start of ETF salesi
Updated FAQs releasedj
Key Facts:
• Sales to commence on June 7, 2021
• Sales gradual and orderly, with aim to minimize any adverse impact on market
functioning
Continued on next page
7
Table A.2 – Continued from previous page
Date Event
July 8, 2021 Announces the start of bond salesk
Updated FAQs releasedl
Key Facts:
• Sales to commence on July 12, 2021
• Sales gradual and orderly, with aim to minimize any adverse impact on market
functioning
August 31, 2021 All holdings of corporate bonds and ETFs have either matured or have been sold.
a
Federal Reserve Board: https://www.federalreserve.gov/newsevents/pressreleases/files/
monetary20200323b2.pdf
b
Federal Reserve Board: https://www.federalreserve.gov/newsevents/pressreleases/
monetary20200323b.htm
c
Federal Reserve Board: https://www.federalreserve.gov/newsevents/pressreleases/files/
monetary20200409a2.pdf
d
Federal Reserve Bank of New York: https://www.newyorkfed.org/newsevents/news/markets/2020/
20200511
e
Federal Reserve Board: https://www.federalreserve.gov/newsevents/pressreleases/files/
monetary20200615a1.pdf
f
Federal Reserve Bank of New York: https://www.newyorkfed.org/markets/
primary-and-secondary-market-faq/archive/corporate-credit-facility-faq-200615
g
Federal Reserve Bank of New York: https://www.newyorkfed.org/newsevents/news/markets/2020/
20200615
h
U.S. Department of the Treasury: https://home.treasury.gov/news/press-releases/sm1190
i
Federal Reserve Bank of New York: https://www.newyorkfed.org/newsevents/news/markets/2021/
20210603
j
Federal Reserve Bank of New York: https://www.newyorkfed.org/markets/
primary-and-secondary-market-faq/archive/corporate-credit-facility-faq-210603
k
Federal Reserve Bank of New York: https://www.newyorkfed.org/newsevents/news/markets/2021/
20210708
l
Federal Reserve Bank of New York: https://www.newyorkfed.org/markets/
primary-and-secondary-market-faq/corporate-credit-facility-faq
8
Table A.3: Estimated relationship between duration-matched spreads and
characteristics. This table reports the estimated coefficients from the regression of log
duration-matched spreads on bond-level 1 year expected default frequency (EDF) and bond issuer
characteristics. Standard errors clustered at the issuer level reported in parentheses below the point
estimates. *** significant at 1% level; ** significant at 5% level; * significant at 10% level.
AAA,AA A+,A,A- BBB+, BBB BBB- BB+, BB, BB- B+ and Lower All
Constant -5.19∗∗∗ -5.39∗∗∗ -4.92∗∗∗ -4.25∗∗∗ -3.99∗∗∗ -5.09∗∗∗ -5.05∗∗∗
(0.02) (0.01) (0.01) (0.01) (0.01) (0.01) (0.00)
Log duration 0.34∗∗∗ 0.44∗∗∗ 0.45∗∗∗ 0.48∗∗∗ 0.42∗∗∗ 0.06∗∗∗ 0.29∗∗∗
(0.01) (0.01) (0.01) (0.01) (0.01) (0.01) (0.00)
Log coupon 0.53∗∗∗ 0.46∗∗∗ 0.43∗∗∗ 0.51∗∗∗ 0.53∗∗∗ 1.22∗∗∗ 0.77∗∗∗
(0.03) (0.01) (0.01) (0.02) (0.03) (0.01) (0.01)
Log amount outstanding -0.07∗∗∗ -0.05∗∗∗ -0.06∗∗∗ -0.11∗∗∗ -0.09∗∗∗ -0.05∗∗∗ -0.06∗∗∗
(0.01) (0.01) (0.01) (0.01) (0.01) (0.00) (0.00)
Log age -0.06∗∗∗ -0.05∗∗∗ -0.05∗∗∗ -0.02∗∗∗ -0.03∗∗∗ -0.08∗∗∗ -0.08∗∗∗
(0.01) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00)
Callable -0.28∗∗∗ -0.21∗∗∗ -0.17∗∗∗ -0.01 -0.10∗∗∗ -0.25∗∗∗ -0.19∗∗∗
(0.03) (0.02) (0.02) (0.02) (0.03) (0.02) (0.01)
EDF1y × Firm EDF dummy 0.03∗∗∗ 0.05∗∗∗ 0.04∗∗∗ 0.23∗∗∗ 0.08∗∗∗ 0.08∗∗∗ 0.07∗∗∗
(0.00) (0.01) (0.01) (0.02) (0.01) (0.00) (0.00)
EDF1y × Bond EDF dummy -0.07∗ -0.04∗∗∗ 0.08∗∗∗ 0.10∗∗∗ 0.08∗∗∗ 0.06∗∗∗ 0.08∗∗∗
(0.04) (0.02) (0.02) (0.02) (0.01) (0.00) (0.00)
N. obs. 794,284 3,296,510 3,476,717 1,285,831 1,070,938 3,715,628 13,639,908
N. clusters 4,085 20,170 25,738 12,791 12,247 54,234 114,110
Adj. R2 0.30 0.31 0.32 0.34 0.24 0.42 0.44
9
Table A.4: List of eligible sellers. This table reports the SMCCF eligible sellers together
with the seller registration date with the facility. An eligible seller is considered to be an under-
writer if any subsidiary of the corporate parent of the eligible seller is reported as a lead un-
derwriter in any corporate bond issuance in either 2019 or 2020 in Mergent FISD. Source: Fed-
eral Reserve Bank of New York, https://www.newyorkfed.org/medialibrary/media/markets/
secondary-market-corporate-credit-facility-eligible-sellers.
Eligible seller Registration date Underwriter?
BMO Capital Markets Corp. May 7, 2020 Y
Cantor Fitzgerald & Co. May 7, 2020 Y
Jefferies LLC May 7, 2020 Y
NatWest Markets Securities Inc. May 7, 2020 N
UBS Securities LLC May 7, 2020 Y
Wells Fargo Securities, LLC May 7, 2020 Y
Goldman Sachs & Co. LLC May 8, 2020 Y
Barclays Capital Inc. May 11, 2020 Y
BofA Securities, Inc. May 11, 2020 Y
Morgan Stanley & Co. LLC May 11, 2020 Y
BNP Paribas Securities Corp. May 12, 2020 Y
Mizuho Securities U.S.A LLC May 12, 2020 Y
TD Securities (U.S.A) LLC May 12, 2020 Y
Amherst Pierpont Securities LLC May 13, 2020 N
Deutsche Bank Securities Inc. May 13, 2020 Y
Citigroup Global Markets Inc. May 14, 2020 Y
Daiwa Capital Markets America Inc. May 14, 2020 N
HSBC Securities (U.S.A) Inc. May 15, 2020 Y
J.P. Morgan Securities LLC May 22, 2020 Y
RBC Capital Markets, LLC May 22, 2020 Y
Scotia Capital (U.S.A) Inc. June 10, 2020 Y
Credit Suisse Securities (U.S.A) LLC June 11, 2020 Y
SG Americas Securities, LLC June 26, 2020 Y
Academy Securities, Inc September 9, 2020 Y
Jane Street Execution Services, LLC September 9, 2020 N
Loop Capital Markets LLC September 9, 2020 Y
MarketAxess Corporation September 9, 2020 N
R. Seelaus&Co., LLC September 9, 2020 Y
SumRidge Partners, LLC September 9, 2020 N
Tradeweb Direct LLC September 9, 2020 N
FHN Financial Securities Corp October 23, 2020 Y
Flow Traders U.S. Institutional Trading LLC October 23, 2020 N
Guzman & Company October 23, 2020 Y
Imperial Capital, LLC October 23, 2020 N
Mischler Financial Group, Inc. October 23, 2020 N
Continued on next page
10
Table A.4 – Continued from previous page
Eligible seller Registration date Underwriter?
MUFG Securities Americas Inc. October 23, 2020 Y
Samuel A. Ramirez & Co., Inc. October 23, 2020 Y
CastleOak Securities, L.P., Inc. November 6, 2020 N
Great Pacific Securities November 6, 2020 N
SMBC Nikko Securities America, Inc. November 6, 2020 Y
U.S. Bancorp Investments, Inc. November 6, 2020 Y
11
Table A.5: Probability of issuance for different cut-offs. This table reports the estimated
coefficients from the probit regression of issuer f issuing a fixed rate bond in week t. “IG” is a dummy
equal to 1 if the issuer is investment grade rated in week t. “WAM” is the amount-outstanding
weighted average maturity of bonds outstanding of the issuer. “Pandemic onset” is a dummy equal
to 1 if issuance occurs between February 22, 2020 and March 21, 2020; “Facility implementation”
is a dummy equal to 1 if issuance occurs between March 22, 2020 and June 29, 2020; “Facility
operational” is a dummy equal to 1 if issuance occurs between June 30, 2020 and December 31,
2020. “Opportunistic issuers” are those that don’t have any debt maturing within 1 year (panel a)/3
years (panel b). All regressions include month fixed effects. Reference period is January 1, 2017 –
February 21, 2020. *** significant at 1% level; ** significant at 5% level; * significant at 10% level.
(a) 1 year cut-off (b) 3 year cut-off
Non-financial Financial Non-financial Financial
(1) (2) (3) (4) (5) (6) (7) (8) (1) (2) (3) (4) (5) (6) (7) (8)
Mat. w/in 1 year Mat. w/in 1 year Opportunistic Opportunistic Mat. w/in 1 year Mat. w/in 1 year Opportunistic Opportunistic Mat. w/in 3 years Mat. w/in 3 years Opportunistic Opportunistic Mat. w/in 3 years Mat. w/in 3 years Opportunistic Opportunistic
12 issued issued
Constant -6.26 -6.34 -5.61 -5.76 -0.27 -1.26 -5.54 -5.46 Constant -6.13 -6.06 -5.54 -5.74 -0.94 -2.31 -5.96 -5.79
(0.59)∗∗∗ (0.75)∗∗∗ (0.26)∗∗∗ (0.28)∗∗∗ (0.81) (0.85) (0.49)∗∗∗ (0.50)∗∗∗ (0.40)∗∗∗ (0.44)∗∗∗ (0.31)∗∗∗ (0.33)∗∗∗ (0.89) (0.78)∗∗∗ (0.64)∗∗∗ (0.67)∗∗∗
Feb 22, 2020 – Mar 21, 2020 -12.37 -12.56 -0.96 -1.12 0.61 0.84 -13.21 -13.35 Feb 22, 2020 – Mar 21, 2020 -13.10 -13.82 -0.77 -0.92 0.49 1.14 -11.93 -12.07
(0.75)∗∗∗ (0.65)∗∗∗ (1.01) (1.04) (0.38) (0.53) (0.45)∗∗∗ (0.45)∗∗∗ (0.37)∗∗∗ (0.54)∗∗∗ (1.02) (1.06) (0.29)∗ (0.42)∗∗∗ (0.86)∗∗∗ (0.56)∗∗∗
Mar 22, 2020 – Jun 29, 2020 0.29 0.51 0.61 0.63 0.63 1.11 -0.24 0.03 Mar 22, 2020 – Jun 29, 2020 1.16 1.35 0.18 0.07 0.60 1.36 0.52 0.68
(1.10) (1.15) (0.33)∗ (0.35)∗ (0.38)∗ (0.44)∗∗ (0.99) (0.98) (0.45)∗∗ (0.45)∗∗∗ (0.46) (0.51) (0.31)∗ (0.34)∗∗∗ (1.07) (1.07)
Jun 30, 2020 – Dec 31, 2020 0.50 0.21 -0.27 -0.15 -0.07 0.43 0.62 0.82 Jun 30, 2020 – Dec 31, 2020 0.31 0.37 -0.36 -0.27 -0.12 0.63 0.87 0.98
(0.78) (0.98) (0.34) (0.35) (0.59) (0.69) (0.57) (0.57) (0.50) (0.54) (0.39) (0.40) (0.56) (0.67) (0.72) (0.75)
Jan 1, 2021 – -11.91 -11.78 -12.94 -13.88 0.01 0.62 -13.04 -13.23 Jan 1, 2021 – -12.77 -13.45 0.00 0.00 -0.11 0.51 -11.77 -12.03
(0.47)∗∗∗ (0.74)∗∗∗ (0.42)∗∗∗ (2.70)∗∗∗ (0.70) (0.86) (.) (0.73)∗∗∗ (0.36)∗∗∗ (2.17)∗∗∗ (.) (.) (0.65) (0.80) (0.53)∗∗∗ (0.55)∗∗∗
WAM -0.04 -0.02 -0.06 -0.04 -0.23 -0.13 -0.06 -0.04 WAM -0.05 -0.03 -0.06 -0.04 -0.22 -0.10 -0.06 -0.04
(0.01)∗∗∗ (0.01)∗ (0.01)∗∗∗ (0.01)∗∗∗ (0.11)∗∗ (0.07)∗ (0.01)∗∗∗ (0.01)∗∗∗ (0.01)∗∗∗ (0.01)∗∗∗ (0.01)∗∗∗ (0.01)∗∗∗ (0.11)∗ (0.06)∗ (0.01)∗∗∗ (0.02)∗∗∗
IG 1.38 0.95 0.78 0.53 -1.29 -1.07 1.14 0.96 IG 1.34 0.84 0.63 0.49 -1.04 -0.53 1.23 1.00
(0.41)∗∗∗ (0.50)∗ (0.16)∗∗∗ (0.18)∗∗∗ (0.34)∗∗∗ (0.65) (0.36)∗∗∗ (0.36)∗∗∗ (0.29)∗∗∗ (0.35)∗∗ (0.18)∗∗∗ (0.19)∗∗ (0.37)∗∗∗ (0.62) (0.44)∗∗∗ (0.46)∗∗
IG × Feb 22, 2020 – Mar 21, 2020 11.95 11.38 1.30 0.71 -0.93 -0.94 13.50 13.66 IG × Feb 22, 2020 – Mar 21, 2020 13.20 13.34 1.05 0.36 -0.77 -1.05 12.31 12.42
(0.97)∗∗∗ (0.88)∗∗∗ (1.05) (1.05) (0.53)∗ (0.67) (.) (.) (.) (.) (1.07) (1.08) (0.44)∗ (0.51)∗∗ (0.99)∗∗∗ (0.77)∗∗∗
IG × Mar 22, 2020 – Jun 29, 2020 0.80 0.70 0.71 0.78 0.02 -0.40 0.58 0.34 IG × Mar 22, 2020 – Jun 29, 2020 0.07 0.29 1.15 1.09 -0.32 -0.94 0.47 0.22
(1.13) (1.28) (0.34)∗∗ (0.38)∗∗ (0.53) (0.59) (1.02) (1.01) (0.46) (0.52) (0.47)∗∗ (0.54)∗∗ (0.46) (0.43)∗∗ (1.10) (1.10)
IG × Jun 30, 2020 – Dec 31, 2020 -1.71 -2.18 0.09 -0.17 0.66 -0.00 -0.97 -1.11 IG × Jun 30, 2020 – Dec 31, 2020 -0.96 -1.28 0.22 -0.05 0.38 -0.46 -0.70 -0.79
(0.91)∗ (0.87)∗∗ (0.39) (0.41) (0.75) (0.94) (0.69) (0.69) (0.57)∗ (0.57)∗∗ (0.45) (0.49) (0.73) (0.87) (0.80) (0.85)
IG × Jan 1, 2021 – 10.12 9.16 11.29 11.67 0.57 0.01 11.50 11.78 IG × Jan 1, 2021 – 11.81 12.29 0.00 0.00 0.48 -0.14 10.58 10.80
(0.79)∗∗∗ (0.90)∗∗∗ (.) (1.89)∗∗∗ (0.83) (0.98) (0.74)∗∗∗ (.) (.) (2.03)∗∗∗ (.) (.) (0.79) (0.91) (1.01)∗∗∗ (0.98)∗∗∗
Log EDF -0.30 -0.15 -0.26 -0.02 Log EDF -0.20 -0.14 -0.25 0.02
(0.12)∗∗ (0.04)∗∗∗ (0.22) (0.07) (0.07)∗∗∗ (0.05)∗∗∗ (0.18) (0.08)
Feb 22, 2020 – Mar 21, 2020 × Log EDF -0.23 -0.30 0.02 0.13 Feb 22, 2020 – Mar 21, 2020 × Log EDF -0.23 -0.31 0.22 -0.02
(0.23) (0.13)∗∗ (0.36) (0.29) (0.17) (0.14)∗∗ (0.24) (0.69)
Mar 22, 2020 – Jun 29, 2020 × Log EDF 0.13 0.09 0.27 -0.10 Mar 22, 2020 – Jun 29, 2020 × Log EDF 0.23 -0.08 0.16 -0.03
(0.18) (0.07) (0.33) (0.20) (0.08)∗∗∗ (0.11) (0.27) (0.15)
Jun 30, 2020 – Dec 31, 2020 × Log EDF -0.31 -0.07 -0.01 0.05 Jun 30, 2020 – Dec 31, 2020 × Log EDF -0.10 -0.10 -0.00 0.06
(0.30) (0.11) (0.33) (0.20) (0.14) (0.14) (0.24) (0.25)
Jan 1, 2021 – × Log EDF -0.39 -0.40 0.30 0.28 Jan 1, 2021 – × Log EDF -0.18 0.00 0.23 0.37
(0.25) (0.47) (0.24) (0.12)∗∗ (0.32) (.) (0.20) (0.10)∗∗∗
Pseudo R.-sqr. 0.08 0.08 0.06 0.05 0.13 0.08 0.04 0.03 Pseudo R.-sqr. 0.06 0.05 0.05 0.05 0.10 0.05 0.06 0.04
N. of obs 21689 14816 156894 106412 14525 11028 41812 30249 N. of obs 59221 41940 111297 74479 25174 19585 31163 21692
N. of clusters 396 310 960 801 157 136 291 254 N. of clusters 543 452 863 717 206 187 242 197
Table A.6: Offering terms and underwriters. This table reports the estimated coefficients
from the regression of offering terms of bond b issued by issuer f in week t. “IG” is a dummy equal
to 1 if the issuer is investment grade rated in week t. “Opportunistic issuers” are those that don’t
have any debt maturing within 2 years. “At least 1 bank eligible seller” is an indicator equal to
1 if any underwriter of the bond is affiliated with a U.S. BHC and becomes an eligible seller at
some point during the life of the facility. “At least 1 bank eligible seller” is an indicator equal to 1
if any underwriter of the bond is not affiliated with a U.S. BHC and becomes an eligible seller at
some point during the life of the facility. All regressions include week fixed effects. Standard errors
clustered at the issuer level reported in parentheses below point estimates. *** significant at 1%
level; ** significant at 5% level; * significant at 10% level.
(a) Full sample
Amount Maturity Offering spread Gross spread
(1) (2) (3) (4) (5) (6) (7) (8)
Constant 2.49 5.44 3.55 -22.23 -459.23 -80.18 24.48 25.87
(3.18) (1.72)∗∗∗ (37.46) (36.65) (805.25) (703.07) (10.78)∗∗ (7.72)∗∗∗
IG 0.15 0.21 5.11 7.09 -165.79 -173.33 -4.22 -4.56
(0.14) (0.13) (2.11)∗∗ (2.00)∗∗∗ (32.32)∗∗∗ (25.69)∗∗∗ (0.79)∗∗∗ (0.61)∗∗∗
At least 1 bank eligible seller -0.16 -0.17 0.03 0.32 1.60 -5.85 0.29 0.36
(0.13) (0.13) (2.93) (2.91) (18.91) (15.97) (0.88) (0.84)
At least 1 bank eligible seller × post registration -0.08 0.03 -6.34 -5.52 -40.07 -53.57 0.99 1.10
(0.21) (0.24) (4.16) (4.35) (35.75) (34.03) (0.56)∗ (0.55)∗∗
At least 1 non-bank eligible seller -0.04 -0.05 -0.76 -0.26 -24.99 -18.35 -0.05 -0.03
(0.09) (0.09) (1.34) (1.45) (11.06)∗∗ (12.27) (0.22) (0.24)
At least 1 non-bank eligible seller × post registration 0.42 0.59 -4.82 -3.95 -43.56 -72.22 -0.37 -0.40
(0.24)∗ (0.23)∗∗ (4.15) (5.02) (31.87) (38.59)∗ (0.63) (0.75)
Log EDF -0.04 -1.18 33.45 0.20
(0.04) (0.57)∗∗ (9.05)∗∗∗ (0.11)∗
Mills ratio 0.81 -0.08 -3.11 1.35 156.18 77.45 -3.53 -3.71
(0.75) (0.41) (8.83) (8.50) (187.10) (164.86) (2.46) (1.79)∗∗
Issuer characteristics ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Bond characteristics ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Week FE ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Adj. R.-sqr. 0.47 0.50 0.11 0.11 0.60 0.68 0.64 0.67
N. of obs 883 848 883 848 883 848 827 796
N. of clusters 234 221 234 221 234 221 226 212
(b) By refinancing need
Amount Maturity Offering spread Gross spread
(1) (2) (3) (4) (5) (6) (7) (8)
Refi Opp. Refi Opp. Refi Opp. Refi Opp.
Constant 7.35 7.43 -100.29 10.00 1963.50 -770.35 12.83 17.35
(3.21)∗∗ (2.71)∗∗∗ (58.72)∗ (48.02) (498.92)∗∗∗ (680.06) (8.27) (10.85)
IG 0.02 0.05 10.73 5.67 -296.17 -108.15 -3.52 -4.77
(0.29) (0.16) (6.56) (2.90)∗ (58.41)∗∗∗ (33.86)∗∗∗ (0.81)∗∗∗ (0.80)∗∗∗
At least 1 bank eligible seller -0.19 -0.04 11.65 -2.52 70.56 -9.08 2.32 -0.94
(0.30) (0.19) (4.13)∗∗∗ (3.25) (38.14)∗ (21.28) (1.51) (1.40)
At least 1 bank eligible seller × post registration -0.35 0.20 -5.22 -5.20 -54.82 12.43 0.13 0.80
(0.44) (0.25) (6.28) (7.13) (36.96) (34.40) (0.91) (0.64)
At least 1 non-bank eligible seller -0.24 -0.06 1.17 -0.93 -8.82 -25.12 0.82 -0.58
(0.22) (0.10) (2.05) (1.97) (14.88) (20.11) (0.49)∗ (0.42)
At least 1 non-bank eligible seller × post registration 0.48 0.62 -1.92 -1.99 -66.77 -39.54 0.29 0.26
(0.59) (0.21)∗∗∗ (9.45) (5.73) (83.31) (39.96) (1.29) (0.88)
Log EDF -0.08 -0.03 -2.30 -1.13 49.44 27.66 0.29 0.02
(0.05) (0.04) (1.14)∗∗ (0.91) (9.04)∗∗∗ (10.93)∗∗ (0.13)∗∗ (0.19)
Mills ratio -0.57 -0.58 16.66 -10.05 -408.53 281.28 -0.50 -1.40
(0.73) (0.63) (12.52) (10.94) (109.98)∗∗∗ (157.43)∗ (1.89) (2.36)
Issuer characteristics ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Bond characteristics ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Week FE ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Adj. R.-sqr. 0.56 0.52 0.01 0.22 0.83 0.71 0.76 0.65
N. of obs 380 439 380 439 380 439 348 425
N. of clusters 90 171 90 171 90 171 85 166
13
Table A.7: Facility announcement effects across event horizons. This table reports the es-
timated coefficients from the regression of average daily changes in duration-matched credit spreads,
bid-ask spreads, and components of credit spreads in 1, 3 or 5 days around the announcement of
the facilities on March 22, 2020 and the expansion of the facilities on April 9, 2020. “AAA/AA/A
eligible” (“BBB eligible”) is a dummy equal to 1 if the bond matures with 5 years, is not issued by
a bank or a bank-subsidiary and is rated at least A- (rated BBB+, BBB or BBB-) on a plurality
rating basis. “Fallen angel eligible” is a dummy equal to 1 if the bond matures with 5 years, is not
issued by a bank or a bank-subsidiary and was investment grade rated as of March 22 and was sub-
sequently downgraded no lower than BB- on a plurality rating basis. All regressions include issuer
FEs. Standard errors clustered at the issuer level reported in parentheses below point estimates.
*** significant at 1% level; ** significant at 5% level; * significant at 10% level.
(a) Duration-matched spreads (b) Default-adjusted spreads
March 22 April 9 March 22 April 9
1D 3D 5D 1D 3D 5D 1D 3D 5D 1D 3D 5D
Constant -25.14 -14.58 -12.57 -45.10 -8.47 -3.04 Constant -29.01 -18.50 -14.37 -42.85 -8.60 -4.16
(25.21) (13.48) (7.66) (20.37)∗∗ (10.58) (10.64) (25.36) (16.72) (9.77) (20.70)∗∗ (10.71) (11.26)
AAA/AA/A eligible -36.31 -14.93 -9.70 2.61 -0.83 -2.25 AAA/AA/A eligible -37.70 -15.56 -10.01 3.08 -0.42 -1.99
(11.80)∗∗∗ (4.94)∗∗∗ (2.19)∗∗∗ (6.00) (2.53) (1.49) (12.52)∗∗∗ (5.25)∗∗∗ (2.31)∗∗∗ (6.20) (2.62) (1.52)
BBB eligible -26.84 -17.71 -16.75 -0.87 2.84 -2.61 BBB eligible -28.62 -17.69 -16.62 -0.10 3.27 -2.39
14 (16.61) (6.12)∗∗∗ (2.92)∗∗∗ (10.54) (7.50) (2.09) (16.62)∗ (6.06)∗∗∗ (2.87)∗∗∗ (10.71) (7.62) (2.16)
Fallen angels eligible -192.24 -11.31 -19.09 Fallen angels eligible -188.34 -9.70 -18.50
(5.26)∗∗∗ (3.18)∗∗∗ (2.23)∗∗∗ (5.05)∗∗∗ (3.17)∗∗∗ (2.21)∗∗∗
Issuer FE ✓ ✓ ✓ ✓ ✓ ✓ Issuer FE ✓ ✓ ✓ ✓ ✓ ✓
Adj. R-sqr. 0.27 0.16 0.21 0.49 0.37 0.39 Adj. R-sqr. 0.27 0.16 0.19 0.49 0.36 0.38
N. of obs 824 880 885 800 888 903 N. of obs 808 861 866 768 856 868
N. of clusters 178 189 190 182 203 208 N. of clusters 176 186 187 175 196 200
(c) 1y EDF (d) Bid-ask spread
March 22 April 9 March 22 April 9
1D 3D 5D 1D 3D 5D 1D 3D 5D 1D 3D 5D
Constant -12.36 -7.85 -3.09 2.54 2.14 2.14 Constant 6.32 6.52 4.26 11.37 3.66 0.76
(6.15)∗∗ (3.88)∗∗ (1.27)∗∗ (2.51) (1.74) (1.89) (25.74) (7.78) (4.56) (22.35) (7.86) (2.71)
AAA/AA/A eligible 0.56 0.29 0.09 -0.25 -0.14 -0.12 AAA/AA/A eligible -30.42 -13.08 -6.97 -11.88 -1.19 -2.85
(0.53) (0.27) (0.09) (0.28) (0.15) (0.16) (3.13)∗∗∗ (6.87)∗ (0.87)∗∗∗ (6.15)∗ (2.37) (1.53)∗
BBB eligible -3.15 -1.76 -0.65 0.21 0.43 0.20 BBB eligible -15.85 -12.89 -7.31 -19.47 0.07 -0.41
(1.76)∗ (1.01)∗ (0.47) (0.54) (0.39) (0.33) (20.77) (5.94)∗∗ (2.88)∗∗ (8.21)∗∗ (2.16) (1.74)
Fallen angels eligible 6.90 14.71 7.39 Fallen angels eligible -44.22 -16.57 -4.33
(0.17)∗∗∗ (0.10)∗∗∗ (0.08)∗∗∗ (1.01)∗∗∗ (0.32)∗∗∗ (0.20)∗∗∗
Issuer FE ✓ ✓ ✓ ✓ ✓ ✓ Issuer FE ✓ ✓ ✓ ✓ ✓ ✓
Adj. R-sqr. 0.78 0.71 0.80 0.86 0.79 0.55 Adj. R-sqr. 0.37 0.46 0.32 0.27 0.25 0.29
N. of obs 808 861 866 772 860 874 N. of obs 456 465 465 862 907 912
N. of clusters 176 186 187 177 198 203 N. of clusters 118 122 122 189 199 201
Table A.8: Facility closure announcement effects across event horizons. This table reports
the estimated coefficients from the regression of average daily changes in duration-matched credit
spreads, bid-ask spreads, and components of credit spreads in 1, 3 or 5 days around Secretary
Mnuchin’s letter requesting the return of CARES Act funds on November 19, 2020 and in 1, 3 or
5 days around the closure of the facilities on December 31, 2020. “SM-CCF eligible” is a dummy
equal to 1 if the bond matures with 5 years, is not issued by a bank or a bank-subsidiary and was
investment grade rated as of March 22 and was subsequently downgraded no lower than BB- on
a plurality rating basis. All regressions include issuer FEs. Standard errors clustered at the issuer
level reported in parentheses below point estimates. *** significant at 1% level; ** significant at 5%
level; * significant at 10% level.
(a) Duration-matched spreads (b) Default-adjusted spreads
Nov 19 Dec 29 Nov 19 Dec 29
1D 3D 5D 1D 3D 5D 1D 3D 5D 1D 3D 5D
Constant -50.37 -5.73 -0.75 -0.85 -3.39 0.60 Constant -51.34 -5.75 1.86 -1.67 -2.79 0.92
(37.86) (2.78)∗∗ (4.32) (7.39) (2.25) (1.30) (40.58) (2.80)∗∗ (3.83) (7.65) (2.09) (1.40)
SMCCF eligible 0.40 0.76 0.04 0.57 0.14 -0.44 SMCCF eligible 0.55 0.80 0.05 0.60 0.12 -0.46
15 (0.98) (0.35)∗∗ (0.53) (1.56) (0.63) (0.72) (1.00) (0.34)∗∗ (0.56) (1.57) (0.65) (0.73)
Issuer FE ✓ ✓ ✓ ✓ ✓ ✓ Issuer FE ✓ ✓ ✓ ✓ ✓ ✓
Adj. R-sqr. 0.41 0.52 0.50 0.19 0.21 0.57 Adj. R-sqr. 0.45 0.54 0.35 0.20 0.20 0.54
N. of obs 845 970 982 577 673 689 N. of obs 841 957 969 570 666 682
N. of clusters 202 227 230 141 161 164 N. of clusters 202 223 226 139 159 162
(c) 1y EDF (d) Bid-ask spread
Nov 19 Dec 29 Nov 19 Dec 29
1D 3D 5D 1D 3D 5D 1D 3D 5D 1D 3D 5D
Constant 1.63 -0.59 -0.18 -3.20 -0.32 -0.57 Constant 1.16 -1.79 -1.38 -5.59 5.38 2.73
(0.82)∗∗ (0.77) (0.56) (2.35) (0.96) (1.03) (6.16) (1.82) (1.61) (9.74) (4.53) (2.71)
SMCCF eligible -1.05 -1.46 -0.77 -0.34 0.06 -0.13 SMCCF eligible -0.13 -0.65 1.03 2.81 -1.01 -0.69
(0.56)∗ (0.86)∗ (0.49) (0.19)∗ (0.09) (0.09) (2.04) (0.68) (0.55)∗ (3.09) (1.18) (0.62)
Issuer FE ✓ ✓ ✓ ✓ ✓ ✓ Issuer FE ✓ ✓ ✓ ✓ ✓ ✓
Adj. R-sqr. 0.96 0.89 0.91 0.61 0.84 0.86 Adj. R-sqr. 0.31 0.29 0.28 0.03 0.14 0.11
N. of obs 841 957 969 572 668 684 N. of obs 1265 1439 1441 767 854 854
N. of clusters 202 223 226 139 159 162 N. of clusters 266 288 288 184 199 199
Table A.9: Purchase effects. This table reports the estimated coefficients from average daily
changes in duration-matched credit spreads, bid-ask spreads, and components of credit spreads in
3 days around facility purchases. “Pct purchased” is the percent of amount outstanding purchased.
Standard errors clustered at the issuer level reported in parentheses below point estimates. ***
significant at 1% level; ** significant at 5% level; * significant at 10% level.
(a) Duration-matched spreads (b) Default-adjusted spreads
(1) (2) (3) (4) (5) (1) (2) (3) (4) (5)
Constant -2.57 -2.25 -2.26 -2.06 -4.56 Constant -1.88 -1.52 -1.55 -0.85 -3.58
(0.39)∗∗∗ (0.39)∗∗∗ (0.39)∗∗∗ (0.51)∗∗∗ (2.94) (0.68)∗∗∗ (0.65)∗∗ (0.65)∗∗ (1.09) (2.88)
Pct purchased -1.90 -0.48 Pct purchased -2.10 -0.09
(0.93)∗∗ (0.83) (0.93)∗∗ (0.83)
ETF pct purchased 47.24 38.77 63.90 ETF pct purchased 73.15 52.02 96.32
(93.69) (80.84) (97.41) (97.74) (83.17) (102.45)
Cash bond pct purchased -0.58 -1.60 -2.18 Cash bond pct purchased -0.24 -1.14 -1.76
(0.84) (0.83)∗ (0.84)∗∗∗ (0.82) (0.81) (0.82)∗∗
Date FE ✓ ✓ ✓ ✓ Date FE ✓ ✓ ✓ ✓
Issuer FE ✓ ✓ Issuer FE ✓ ✓
16 Issue FE ✓ Issue FE ✓
Adj. R-sqr. 0.00 0.07 0.07 0.09 0.13 Adj. R-sqr. 0.00 0.05 0.05 0.06 0.09
N. of obs 179150 179150 179150 179127 179065 N. of obs 176170 176170 176170 176153 176097
N. of clusters 673 673 673 650 648 N. of clusters 633 633 633 616 614
(d) Bid-ask spread
(c) 1y EDF
(1) (2) (3) (4) (5)
(1) (2) (3) (4) (5)
Constant -0.20 -0.28 -0.24 -0.48 -1.03 Constant -0.05 0.04 0.06 -0.00 0.23
(0.24) (0.23) (0.23) (0.25)∗∗ (0.69) (0.24) (0.24) (0.24) (0.28) (0.33)
Pct purchased 0.53 -1.60 Pct purchased -3.03 -1.96
(0.31)∗ (0.62)∗∗∗ (1.03)∗∗∗ (1.02)∗
ETF pct purchased -108.15 -120.67 -161.40 ETF pct purchased -60.72 -63.33 -7.22
(70.27) (64.13)∗ (80.11)∗∗ (64.91) (67.18) (75.46)
Cash bond pct purchased -1.38 -1.68 -1.70 Cash bond pct purchased -1.81 -1.59 -1.61
(0.52)∗∗∗ (0.52)∗∗∗ (0.52)∗∗∗ (0.99)∗ (0.96)∗ (0.96)∗
Date FE ✓ ✓ ✓ ✓ Date FE ✓ ✓ ✓ ✓
Issuer FE ✓ ✓ Issuer FE ✓ ✓
Issue FE ✓ Issue FE ✓
Adj. R-sqr. 0.00 0.02 0.02 0.03 0.02 Adj. R-sqr. 0.00 0.00 0.00 0.01 0.00
N. of obs 176400 176400 176400 176382 176324 N. of obs 227320 227320 227320 227314 227296
N. of clusters 634 634 634 616 614
N. of clusters 543 543 543 537 537
Figure A.1. Daily CCF purchase volume. This figure plots the time series of daily
CCF purchase volume by asset class.
500
Daily Purchases ($ Million)
400
300
200
100
0
2Jun 9Jun 16Jun 23Jun 30Jun
7Jul 4Aug 1Sep 8Sep 6Oct 13Oct 20Oct 27Oct 3Nov 1Dec 8Dec
12May 19May 26May 14Jul 21Jul 28Jul 11Aug 18Aug 25Aug 15Sep 22Sep 29Sep 10Nov 17Nov 24Nov 15Dec 22Dec 29Dec
IG ETF Fallen Angels ETF HY ETF Fallen Angels Bonds IG Bonds
17
Table A.10: Purchase effects across event horizons. This table reports the estimated coeffi-
cients from average daily changes in duration-matched credit spreads, bid-ask spreads, and compo-
nents of credit spreads in 1,3 or 5 days around facility purchases. “Purchased” is the 0/1 indicator for
the bond being purchased either directly (as a cash bond) or indirectly (through ETF purchases).
“ETF” is the 0/1 indicator for the bond purchased indirectly through ETFs; “Cash bond” is the
0/1 indicator for the bond purchased directly. “Pct purchased” is the percent of amount outstand-
ing purchased. Standard errors clustered at the issuer level reported in parentheses below point
estimates. *** significant at 1% level; ** significant at 5% level; * significant at 10% level.
(a) Duration-matched spreads (b) Default-adjusted spreads
1D 3D 5D 1D 3D 5D
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)
Constant -3.27 -6.72 -3.72 -7.07 -1.87 -4.54 -1.99 -4.56 -1.78 -3.85 -1.90 -3.92 Constant -1.45 -5.03 -1.90 -5.38 -1.17 -3.53 -1.30 -3.58 -1.36 -3.22 -1.50 -3.32
(0.71)∗∗∗ (3.71)∗ (0.70)∗∗∗ (3.70)∗ (0.40)∗∗∗ (2.94) (0.40)∗∗∗ (2.94) (0.34)∗∗∗ (2.34)∗ (0.34)∗∗∗ (2.34)∗ (1.85) (3.53) (1.85) (3.51) (0.68)∗ (2.88) (0.69)∗ (2.88) (0.38)∗∗∗ (2.28) (0.39)∗∗∗ (2.28)
ETF -2.65 -1.05 -2.01 -0.00 -1.91 -0.25 ETF -2.71 -1.05 -2.04 -0.11 -1.93 -0.31
(0.19)∗∗∗ (0.78) (0.13)∗∗∗ (0.43) (0.12)∗∗∗ (0.32) (0.20)∗∗∗ (0.83) (0.14)∗∗∗ (0.45) (0.13)∗∗∗ (0.36)
Cash bond 0.40 -1.44 0.51 -0.93 0.69 -0.60 Cash bond 0.27 -1.29 0.46 -0.82 0.65 -0.52
(0.60) (0.52)∗∗∗ (0.29)∗ (0.26)∗∗∗ (0.21)∗∗∗ (0.18)∗∗∗ (0.58) (0.52)∗∗ (0.28) (0.26)∗∗∗ (0.21)∗∗∗ (0.17)∗∗∗
ETF pct purchased -817.75 256.02 -933.72 63.90 -873.79 -18.73 ETF pct purchased -858.99 247.13 -946.33 96.32 -876.30 41.46
18 (76.00)∗∗∗ (164.51) (61.96)∗∗∗ (97.41) (60.90)∗∗∗ (94.16) (81.91)∗∗∗ (166.89) (64.35)∗∗∗ (102.45) (62.23)∗∗∗ (99.77)
Cash bond pct purchased 0.76 -5.27 2.12 -2.18 2.03 -1.93 Cash bond pct purchased 0.18 -4.73 1.94 -1.76 1.88 -1.63
(1.63) (1.56)∗∗∗ (0.79)∗∗∗ (0.84)∗∗∗ (0.59)∗∗∗ (0.57)∗∗∗ (1.62) (1.54)∗∗∗ (0.78)∗∗ (0.82)∗∗ (0.59)∗∗∗ (0.55)∗∗∗
Issuer FE ✓ ✓ ✓ ✓ ✓ ✓ Issuer FE ✓ ✓ ✓ ✓ ✓ ✓
Date FE ✓ ✓ ✓ ✓ ✓ ✓ Date FE ✓ ✓ ✓ ✓ ✓ ✓
Issue FE ✓ ✓ ✓ ✓ ✓ ✓ Issue FE ✓ ✓ ✓ ✓ ✓ ✓
Adj. R-sqr. 0.00 0.09 0.00 0.09 0.01 0.13 0.01 0.13 0.01 0.17 0.01 0.17 Adj. R-sqr. 0.00 0.06 0.00 0.06 0.00 0.09 0.00 0.09 0.01 0.13 0.01 0.13
N. of obs 160462 160371 160462 160371 179150 179065 179150 179065 181743 181663 181743 181663 N. of obs 158137 158054 158137 158054 176170 176097 176170 176097 178600 178533 178600 178533
N. of clusters 645 627 645 627 673 648 673 648 679 654 679 654 N. of clusters 610 595 610 595 633 614 633 614 637 619 637 619
(c) 1y EDF (d) Bid-ask spread
1D 3D 5D 1D 3D 5D
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)
Constant -0.56 -1.73 -0.52 -1.70 -0.24 -1.19 -0.19 -1.03 -0.15 -0.58 -0.14 -0.53 Constant 0.49 -0.05 0.39 -0.19 0.11 0.31 0.05 0.23 0.04 0.33 0.01 0.29
(0.96) (0.95)∗ (0.96) (0.91)∗ (0.23) (0.71)∗ (0.24) (0.69) (0.16) (0.57) (0.17) (0.56) (0.29)∗ (0.77) (0.28) (0.75) (0.25) (0.35) (0.24) (0.33) (0.22) (0.16)∗∗ (0.22) (0.15)∗
ETF 0.29 -0.10 0.11 0.45 0.04 0.03 ETF -0.33 -0.61 -0.41 -0.35 -0.29 -0.18
(0.17)∗ (0.60) (0.09) (0.40) (0.06) (0.27) (0.15)∗∗ (0.70) (0.09)∗∗∗ (0.47) (0.08)∗∗∗ (0.29)
Cash bond 0.39 -0.52 0.19 -0.37 0.10 -0.32 Cash bond -0.42 0.20 -0.75 -0.54 -0.28 -0.05
(0.14)∗∗∗ (0.26)∗∗ (0.09)∗∗ (0.14)∗∗∗ (0.07) (0.10)∗∗∗ (0.62) (0.65) (0.37)∗∗ (0.34) (0.16)∗ (0.16)
ETF pct purchased 109.04 -273.75 -7.65 -161.40 5.32 -179.93 ETF pct purchased -36.12 -27.87 -158.58 -7.22 -133.91 21.63
(43.40)∗∗ (123.01)∗∗ (23.11) (80.11)∗∗ (19.35) (74.34)∗∗ (99.40) (145.54) (39.41)∗∗∗ (75.46) (32.05)∗∗∗ (55.54)
Cash bond pct purchased 1.12 -2.93 0.57 -1.70 0.33 -1.38 Cash bond pct purchased -2.08 -0.39 -2.26 -1.61 -0.75 -0.03
(0.40)∗∗∗ (0.93)∗∗∗ (0.30)∗ (0.52)∗∗∗ (0.20) (0.38)∗∗∗ (2.26) (2.41) (0.98)∗∗ (0.96)∗ (0.49) (0.52)
Issuer FE ✓ ✓ ✓ ✓ ✓ ✓ Issuer FE ✓ ✓ ✓ ✓ ✓ ✓
Date FE ✓ ✓ ✓ ✓ ✓ ✓ Date FE ✓ ✓ ✓ ✓ ✓ ✓
Issue FE ✓ ✓ ✓ ✓ ✓ ✓ Issue FE ✓ ✓ ✓ ✓ ✓ ✓
Adj. R-sqr. 0.00 0.02 0.00 0.02 0.00 0.02 0.00 0.02 0.00 0.02 0.00 0.02 Adj. R-sqr. 0.00 -0.00 0.00 -0.00 0.00 0.00 0.00 0.00 0.00 0.02 0.00 0.02
N. of obs 158350 158265 158350 158265 176400 176324 176400 176324 178832 178761 178832 178761 N. of obs 220774 220761 220774 220761 227320 227296 227320 227296 228349 228322 228349 228322
N. of clusters 610 595 610 595 634 614 634 614 638 619 638 619 N. of clusters 541 534 541 534 543 537 543 537 549 539 549 539
File and source
- File
- its-what-you-say-and-what-you-buy-ny-fed-sr-935.pdf
- Size
- 1,084,996 bytes
- SHA-256
- f0ef3906d3090c09e08b91803453e023337fbf1763bacc42aeb0f033cae46bbe
- Original
- www.newyorkfed.org