12-The Real Effects of Credit Ratings - Cleaned

THE JOURNAL OF FINANCE • VOL. LXXII, NO.
1 • FEBRUARY 2017
The Real Effects of Credit Ratings: The Sovereign

Ceiling Channel
HEITOR ALMEIDA, IGOR CUNHA, MIGUEL A. FERREIRA,
and FELIPE RESTREPO∗
ABSTRACT
We show that sovereign debt impairments can have a significant effect on financial
markets and real economies through a credit ratings channel. Specifically, we find
that firms reduce their investment and reliance on credit markets due to a rising cost
of debt capital following a sovereign rating downgrade. We identify these effects by
exploiting exogenous variation in corporate ratings due to rating agencies’ sovereign
ceiling policies, which require that firms’ ratings remain at or below the sovereign
rating of their country of domicile.
SOVEREIGN DEBT IMPAIRMENTS have become a significant problem for developed

countries in the aftermath of the 2007 to 2009 global financial crisis and the
European sovereign debt crisis. France and the United States were downgraded
from an AAA credit rating for the first time in history, and other developed
countries including Greece, Ireland, Italy, the Netherlands, Portugal, and Spain
also experienced rating downgrades. How do sovereign debt impairments affect
financial markets and real economic activity?
We examine this question by exploring the consequences of sovereign rating
downgrades for firms’ cost of capital, investment, and financing decisions. Our
∗ Heitor Almeida is at the University of Illinois at Urbana Champaign. Igor Cunha is at the
Nova School of Business and Economics. Miguel A. Ferreira is at the Nova School of Business and
Economics, CEPR, and ECGI. Felipe Restrepo is at the Ivey Business School at Western Univer-
sity. We thank Michael Roberts (the Editor), the Associate Editor, an anonymous referee, Viral
Acharya, Marco Bonomo, Murillo Campello, Sergey Chernenko, Paolo Colla, Jose Faias, Clifford
Holderness, Victoria Ivashina, Anastasia Kartasheva, Darren Kisgen, Spyridon Lagaras, Sebastien
Michenaud, Dean Paxson, Mitchell Petersen, Jun Qian, Henri Servaes, Kelly Shue, Rui Silva, Phil
Strahan, Jérôme Taillard, David Thesmar, Yuhai Xuan, Mike Weisbach, participants at the Adam
Smith Workshop for Corporate Finance, Annual Conference on Corporate Finance at Washington
University, European Finance Association Annual Meeting, Financial Intermediation Research
Society Conference, NBER Summer Institute Workshop on Corporate Finance, Portuguese Fi-
nance Network, Lubrafin, and seminar participants at the Bank of Portugal, Chicago Booth School
of Business, Federal Reserve Bank of New York, Hong Kong University of Science and Technol-
ogy, INSPER, London School of Economics, Manchester Business School, National University of
Singapore, Rice University, University of Houston, University of North Carolina, and University
of Washington for helpful comments. Ferreira acknowledges financial support from the European
Research Council and Fundação para a Ciência e Tecnologia. We have read the Journal of Finance’s
disclosure policy and have no conflicts of interest to disclose.
DOI: 10.1111/jofi.12434
249
250 The Journal of FinanceR
identification strategy exploits the variation in corporate ratings that is due

to rating agencies’ sovereign ceiling policies. These policies require that firms’
ratings remain at or below the sovereign rating of their country of domicile.
While rating agencies have been gradually moving away from a policy of never
rating a private borrower above the sovereign, corporate ratings that “pierce”
the sovereign ceiling are still not common (Standard & Poor’s Rating Services
(2012)).1
We show that the sovereign ceiling leads to an asymmetric change in cor-
porate ratings following a sovereign downgrade. Firms with a rating equal to
or above their sovereign prior to the downgrade (bound firms) are significantly
more likely to be downgraded after a sovereign downgrade than firms rated
below their sovereign (nonbound firms). One key advantage of our empirical
strategy is that nonbound firms have similar but lower credit quality than
bound firms. Thus, alternative explanations based on changes in fundamen-
tals and credit risk are not likely to explain the discontinuous change in ratings
around the sovereign ceiling. The asymmetric effect of sovereign downgrades
on firm ratings is thus likely to be due to the sovereign ceiling, and not to
changes in fundamentals.
We trace the financial and real consequences of this asymmetric effect of
sovereign downgrades on bound firms. Specifically, we find that bound firms
cut investment more than nonbound firms in the aftermath of a sovereign
downgrade. We also find some evidence that bound firms reduce net debt is-
suance and increase equity issuance more than nonbound firms following a
downgrade, although this evidence is not statistically as strong as the evidence
for investment. Finally, we find that sovereign downgrades also affect corpo-
rate bond markets, as the yields of bound firms increase more than the yields
of nonbound firms following a sovereign downgrade. The effect on the cost of
debt capital is statistically and economically significant.
Credit ratings are a major concern for corporate managers because of the
frictions associated with ratings (Kisgen 2006, 2007)). First, ratings affect a
firm’s access to the bond and commercial paper markets, because rating levels
determine whether institutional investors such as banks or pension funds are
allowed to invest in a firm’s securities. Second, ratings affect the capital require-
ments applied to banks and insurance companies when they invest in specific
firms. Third, rating downgrades can trigger events such as bond covenant vio-
lations, increases in bond coupons or loan interest rates, and forced bond repur-
chases.2 Finally, ratings can impact customer and employee relationships as
well as business operations, including a firm’s ability to enter into or maintain
1 CFO Magazine summarizes the key implication of the sovereign ceiling as follows: “If a com-
pany is a better credit risk than its home country, it might still have trouble getting a credit rating
agency to recognize that fact” (see “Corporate, sovereign debt ratings closely linked: S&P,” CFO
Magazine, April 29, 2013).
2 For example, performance-sensitive debt may incorporate explicit or implicit performance
pricing provisions that depend on credit ratings (Manso, Strulovici, and Tchistyi (2010)). Manso
(2013) shows that, in this setting, rating downgrades can significantly amplify adverse shocks to
firm fundamentals because of feedback effects between ratings and firm behavior.
The Real Effects of Credit Ratings 251
long-term contracts. Because of these effects, firms appear to react to rating

downgrades by reducing debt issuance and leverage (Kisgen (2009)).
We provide evidence on how rating downgrades matter in our context by
focusing on ratings-based regulation. Basel II bank capital requirements are
a nonlinear function of ratings. Because of this nonlinearity, some sovereign
downgrades are more likely to cause changes in capital requirements applied to
financial institutions. We find evidence that the financial and real consequences
of sovereign downgrades are stronger for downgrades that matter most for
capital requirements. While these results suggest that the regulation channel
plays a role, they need to be interpreted with caution due to the small sample
size and lack of statistical power.3
When financial markets operate normally, the consequences of sovereign
downgrades may not spill over into firm decisions and real economic activity.
For example, firms may be able to substitute equity for debt issuance. But
periods of sovereign downgrades are far from normal. Local financial markets
are likely to be in trouble, so it is difficult for firms to substitute equity for debt.
Sovereign downgrades also tend to happen during periods of global financial
turmoil, when even firms with access to global markets may find it difficult to
raise alternative sources of finance. Thus, the impact of sovereign downgrades
is often amplified by adverse market conditions.4
Our benchmark empirical specification employs the Abadie and Imbens
(2011) bias-corrected matching estimator of the Average Effect of the Treatment
on the Treated (ATT). We first isolate firms at the sovereign bound (treated
firms). Then, from the population of firms below the sovereign bound (non-
treated firms), we look for control firms that best match treated firms along
multiple dimensions (country, year, size, investment, Tobin’s Q, cash flow, cash,
leverage, foreign sales, government ownership, and exposure to government
spending) before the treatment (sovereign downgrades).5
We find economically significant effects of sovereign rating downgrades.
First, treated firms’ investment decreases from 26.6% to 17.7% of capital in
the year of the downgrade, implying a 8.9 percentage point reduction. In con-
trast, control firms reduce investment by 2.6 percentage points, resulting in
a difference-in-difference estimator of −6.4 percentage points and an ATT of
−8.9 percentage points. The differential effects on investment are statistically
significant. Second, treated firms’ net debt issuance decreases from 7.5% to
3 Our sample of bound (treated) firms includes 73 firms, so we have limited ability to split the
sample according to variables such as the likelihood of a change in capital requirements.

4 Gande and Parsley (2005) show that sovereign downgrades have spillover effects on the credit
spreads of other countries. In our sample, most sovereign downgrades happened in the aftermath of
the Asian and Russian crises and the burst of the Internet bubble (end of the 1990s and beginning
of the 2000s), and following the financial crisis of 2007 to 2009. Thus, it may also be costly for firms
to issue debt in other countries.
5 While we match perfectly on country-year, it is difficult to find industry matches in smaller
countries. Thus, our benchmark results use a sample of control firms that is not matched on
industry. We obtain similar estimates when we use a smaller sample for which we can find control
firms that match treated firms according to the Fama-French 12-industry classification.
2.4% of assets, implying a 5.1 percentage point reduction. For control firms,
net debt issuance falls only 2.3 percentage points. While the difference-in-
difference estimator is statistically insignificant, the ATT is −5.5 percentage
points, which is again statistically significant. There is also some evidence that
treated firms increase equity issuance more than control firms in the years
after the sovereign downgrade. Finally, bond yields of treated firms increase by
approximately 34 basis points more than the yields of control firms over the pe-
riod of three months after a sovereign downgrade relative to the three months
before. This differential effect is more pronounced as the postevent window
widens. For example, the differential effect is 61 basis points six months after
the downgrade.
We conduct several robustness tests and find that the differential effect of
sovereign downgrades on investment (ATT) ranges from 7% to 14% of capital,
which is consistent with the benchmark results. This effect represents about
25% to 50% of the precrisis average investment for treated firms. Other papers
that relate ratings and regulatory frictions to investment find effects of simi-
lar magnitude. For example, Lemmon and Roberts (2010) find that junk-rated
firms’ net investment falls by about 33% following the introduction of regula-
tions restricting the flow of institutional capital to these firms, and Tang (2009)
finds that firms that are upgraded due to Moody’s 1982 ratings refinement
increase investment by about 40%.
Overall, we interpret our results as follows. Treated firms find it more expen-
sive to raise debt in the aftermath of a sovereign downgrade, which leads them
to replace debt with equity and to reduce investment. This difference across
treated and control firms arises only following the sovereign downgrade, as
there is no evidence of significant preexisting differential trends in outcome
variables.
The key assumption of our identification strategy is that sovereign down-
grades are not related to differences in investment across treatment and con-
trol groups, other than through changes in ratings. This assumption would be
violated if treated firms had unobservable characteristics that predict greater
sensitivity to sovereign debt crises, even in the absence of downgrades.6 To
further validate our exclusion restriction, we conduct a series of placebo tests
that include recessions, the 2007 to 2009 financial crisis, and currency crises
that are not accompanied by sovereign downgrades. We do not find significant
differences in investment between treatment and control groups in any of the
placebo tests.
An additional concern is that treated firms may have greater exposure to the
government than control firms. While our baseline specification controls for
government ownership and exposure to government spending, these controls
6 For instance, our treatment group could have many bank-dependent firms. Chava and Pur-
nanandam (2011) find that bank-dependent firms are more affected during banking crises than
firms with access to public debt markets. Carvalho, Ferreira, and Matos (2015) and Chodorow-
Reich (2014) find that borrowers with precrisis relationships with less healthy lenders were more
affected by the 2007–2009 financial crisis compared to borrowers of healthier lenders.
may not be sufficient. For example, the group of treated firms may include
“national champions” that are expected to receive support from the govern-
ment. More broadly, treated firms may be more exposed to the government’s
health (even without being a national champion). We investigate this possi-
bility in several ways. First, we examine short-term effects of the sovereign
downgrade on after-tax profitability of treated and control firms. Second, we
conduct robustness tests in which we exclude firms with government owner-
ship or utilities from the treatment group. Third, we conduct a placebo test in
which treated firms are those with a rating one notch below the sovereign, since
these firms might also be expected to receive government support. Finally, we
examine whether security (stock and bond) prices of treated and control firms
are differentially affected by changes in the prices of government securities.
The results of these tests suggest that exposure to the government is similar
across treated and control firms. Thus, our results are likely to be due to the
sovereign downgrade itself, and not to differential exposure to government
shocks. Nevertheless, we cannot completely rule out the possibility that the
results may be driven in part by the government exposure channel, since this
channel may matter only in the aftermath of a sovereign downgrade.
Our paper makes three contributions. First, we provide evidence of a link
between sovereign debt impairments and the real economy. Sovereign debt
impairments can lead to sovereign downgrades that induce corporate rating
downgrades because of the ceiling rule. These sovereign-driven corporate down-
grades affect credit markets and real economic activity. Second, we contribute
to the literature on the effects of credit ratings on firm outcomes. This literature
shows that credit ratings matter for capital structure decisions (Kisgen (2006))
and cost of capital (Kisgen and Strahan (2010), Baghai, Servaes, and Tamayo
(2014)), as well as for firms’ real decisions (Sufi (2009), Tang (2009), Lemmon
and Roberts (2010), Chernenko and Sunderam (2012), and Harford and Uysal
(2014)).7 However, these studies are subject to omitted variables concerns be-
cause changes in ratings are correlated with changes in firm fundamentals.
Our results, which support the argument that ratings affect firm investment
and financial policy, appear to be driven specifically by changes in ratings, and
not by changes in fundamentals and crowding-out effects (e.g., Graham, Leary,
and Roberts (2014)). Finally, we contribute to the recent literature that studies
credit supply effects on large and high-credit-quality firms (Adrian, Colla, and
Shin (2013) and Becker and Ivashina (2014a)). We show that a negative shock
to credit supply induced by sovereign downgrades can have real effects even
for firms with the highest credit quality due to the ceiling rule.
Our study is subject to the standard limitations of quasi-natural experiments.
First, we can only observe the consequences of sovereign debt impairments due
to rating changes on bound firms—the estimates do not tell us much about the
importance of sovereign debt for firms of different characteristics or during
7 There is also a literature on the relation between sovereign and corporate credit risk (e.g.,
Durbin and Ng (2005), Borensztein, Cowan, and Valenzuela (2013), Augustin et al. (2014), and
Bedendo and Colla (2015)).
periods of less turmoil. Second, our effects should not be interpreted as the av-
erage effect of sovereign downgrades on real economic activity—since there are
only a few bound firms for each country, the aggregate magnitude of our effect
is smaller than the effects that we report in the paper.8 Sovereign downgrades
can also have other effects on economic activity that we do not measure. For in-
stance, Adelino and Ferreira (2016) find that sovereign downgrades reduce the
supply of bank lending, and Becker and Ivashina (2014b) find that sovereign
debt crises can affect firms through financial repression. Our findings can also
apply to a country as a whole: a sovereign downgrade can lead to a reduction
in the supply of capital for the country due to a ratings channel (i.e., over and
above the deterioration in fundamentals). This reduction in capital supply can
in turn affect real economic activity.
I. Methodology and Data

In this section, we first describe our experimental design and the match-
ing estimator that we use. We then describe the data and present summary
statistics.
A. Sovereign Downgrades and Ceilings: Institutional Background

Credit rating agencies play a crucial role in providing information about the
ability and willingness of issuers, including governments and private firms, to
meet their financial obligations. The three major agencies—Standard & Poor’s
(S&P), Moody’s, and Fitch—assign ratings depending on the maturity (short
term or long term) and currency denomination of an issue (foreign currency or
local currency). We focus on foreign currency long-term issuer ratings, which
are most likely to be bound by the sovereign rating. We prefer the S&P rating
history over those of the other agencies because S&P tends to be more active in
making rating revisions and leads other agencies in rerating (Kaminsky and
Schmukler (2002), Brooks et al. (2004)). Rating announcements by S&P also
appear to have a larger own-country stock market impact and to not be fully
anticipated by the market (Reisen and von Maltzan (1999)).
Until 1997, rating agencies followed a strict policy of not granting a pri-
vate company a foreign currency rating higher than the sovereign rating. In
April of 1997, S&P first relaxed its sovereign ceiling rule in three dollarized
economies: Argentina, Panama, and Uruguay.9 Although rating agencies have
moved away from strict enforcement of the sovereign ceiling over the last two
decades, corporate ratings that pierce the ceiling are still not common. Boren-
sztein, Cowan, and Valenzuela (2013) show that sovereign ratings continue
8 Since bound firms are typically large firms, the reduction in aggregate investment can be sig-
nificant. Bound firms are responsible for 18% of the aggregate investment in a country, on average.
Thus, a 25% decrease in their investment implies a 4.5% reduction in aggregate investment in a
country.
9 Fitch and Moody’s followed suit in 1998 and 2001.
to represent a strong upper bound and an important determinant of ratings

assigned to firms.10
Why do rating agencies use the sovereign rating as an upper bound when
they rate corporate issuers? Agencies use two key factors in rating issuers:
the issuer’s inherent likelihood of making repayment, and, in the case of for-
eign currency ratings, the issuer’s profile after taking into account the risk
that capital and exchange controls might be imposed that would hinder the
ability of the issuer to meet its financial obligations in foreign currency. In
general, rating agencies grant an issuer a rating above the sovereign only if it
is able to demonstrate strong resilience and low default dependence vis-a-vis
the sovereign, as well as some degree of insulation from the domestic economic
and financial disruptions typically associated with sovereign distress. Firms
with foreign assets, high export earnings, and foreign parents are more likely
to be rated above their sovereign.
Interestingly, S&P recently updated its methodology to address some of the
limitations of the previous approach. The Standard & Poor’s Rating Services
(2013) methodology applies a sovereign foreign currency default stress scenario
(stress test) with respect to the country or countries in which the firm has
economic exposures and when the potential rating exceeds the rating on the
sovereign (in general, the reference sovereign rating is a weighted average
of the sovereign ratings of the countries for which the company has material
exposures). Firms that pass the stress test can be rated up to two or four
notches above the sovereign rating, depending on whether S&P views their
sector’s sensitivity to country risk as high or moderate, respectively. As a result
of this new methodology, Standard & Poor’s Rating Services (2013) expects some
corporations to receive upgrades. This suggests that S&P issued conservative
ratings to some firms due to the sovereign ceiling before the recent revision of
the methodology.
B. Identification Strategy
The main challenge in tracing the effect of sovereign downgrades on firm
outcomes is the inherent endogeneity between a sovereign’s credit quality and
the creditworthiness of firms in that country. We explicitly address this concern
in our empirical strategy by examining the differential effect of sovereign rating
changes on firms that are bound by the sovereign ceiling (bound firms) and on
other firms in the same country that are not bound by the sovereign ceiling
(nonbound firms).
Figure 1 presents the distribution of corporate ratings across sovereign rating
classes. The figure shows that only a few corporations are rated above the
sovereign ceiling and only to a limited degree. In our sample, 88.2% of firms
receive a rating lower than the sovereign, 8.4% receive the same rating, and
just 3.4% receive a rating higher than the sovereign.
10 For example, Standard & Poor’s Rating Services (2012) reports only 54 nonfinancial corpora-
tions worldwide with ratings that exceed the sovereign rating as of October 2012.
Figure 1. Sovereign ceiling rule. This figure shows the relation between sovereign and cor-
porate credit ratings. The sample consists of WRDS-Factset Fundamentals Annual Fiscal (North
America and International) nonfinancial firms over the 1990 to 2013 period. Panel A shows the fre-
quency of S&P long-term foreign-currency corporate credit ratings by the sovereign rating of their
country of domicile. Observations for countries with AAA ratings are excluded as, by definition, the
sovereign ceiling policy does not represent a constraint for corporations when the sovereign has
the maximum attainable rating. Panel B plots the distribution of the difference between corporate
ratings and sovereign ratings.
Figure 2. Frequency of corporate downgrades around a sovereign downgrade by dis-

tance from sovereign. This figure shows the fraction of firms in each group whose rating is
downgraded in the month before, the month of, and the month after a sovereign downgrade. Ob-
servations are grouped according to the predowngrade difference between the corporate rating
and sovereign rating. The sample consists of WRDS-Factset Fundamentals Annual Fiscal (North
America and International) nonfinancial firms over the 1990 to 2013 period.
Figure 2 illustrates the key empirical regularity that we use to identify

causal effects of ratings on firm outcomes. The figure shows that the proba-
bility a corporate issuer will obtain a rating downgrade within the month of a
sovereign downgrade is discontinuous exactly at the sovereign bound (where
the difference between a firm’s rating and its sovereign is equal to zero). The
middle panel in Figure 2 shows that, conditional on a sovereign downgrade,
bound firms have a 59% chance of obtaining a rating downgrade within the
month, compared to 9% and 4% for firms that are, respectively, one and two
notches below the sovereign rating. The left panel shows that this disparity
in the response of corporate ratings is not observed in the month before the
sovereign downgrade. The right panel shows the frequency of corporate down-
grades in the month after the sovereign downgrade. Firms with a rating one
notch above the sovereign rating have the highest frequency of a downgrade in
the month after the sovereign downgrade, which is consistent with the ceiling
rule. We conclude that bound firms have a significantly higher probability of a
downgrade than nonbound firms following a sovereign downgrade.
The discontinuity in ratings downgrades across bound and nonbound firms
is not likely caused by factors other the sovereign ceiling rule. For example, a
deterioration in macroeconomic fundamentals could only generate this discon-
tinuity if credit risk increases for bound firms but stays constant for similar
firms right below the bound. In fact, if there were any differential macro ef-
fects, better-quality firms (our treatment group) should be less affected than
poorer-quality firms (our control group).
Our evidence thus suggests that credit rating agencies continue to apply
the sovereign ceiling rule in the event of a sovereign downgrade. Thus, the
differential effect on firm outcomes across bound firms and nonbound firms in
the event of a sovereign downgrade should stem from the change in ratings,
and not from differences in fundamentals.
C. Matching Approach
We test whether firms that are downgraded as a consequence of the sovereign
ceiling change their investment and financial decisions. The treatment group
includes bound firms (those with a rating equal to or above the sovereign rating
of the firm’s country of domicile in the year prior to the sovereign downgrade).
The nontreatment group includes nonbound firms (those with a predowngrade
rating below the sovereign rating). Our analysis needs to account for the fact
that treated and nontreated firms potentially have different observable char-
acteristics.
One way to tackle this issue is to estimate differences between plausibly coun-
terfactual outcomes and those that are observed in the data. The strategy we
apply in our main tests is nonparametric, combining the sovereign downgrade
episode with the use of matching estimators. The idea of this estimator is to first
isolate treated firms and then, from the population of nontreated firms, find
observations that best match the treated ones along multiple dimensions. In
this framework, the set of counterfactuals is restricted to the matched controls,
that is, in the absence of the treatment (in our context, sovereign downgrades),
the treatment group would behave similarly to the control group.
We employ the Abadie and Imbens (2011) estimator, as implemented by
Abadie et al. (2004). The Abadie–Imbens matching estimator minimizes the
(Mahalanobis) distance between a vector of observed covariates across treated
and nontreated firms to find matched control firms. We select one matched
control observation for each treated observation. The estimator allows control
firms to serve as matches more than once, which (compared to matching with-
out replacement) reduces the estimation bias but increases the variance. The
Abadie–Imbens estimator produces exact matches on categorical variables, but
less exact matches on continuous variables (though they should be close). The
procedure corrects this issue by applying a bias-correction component to the es-
timates. The categorical variables include year, country, and whether a firm has
a credit rating.11 The noncategorical variables include firm size, investment,
Tobin’s Q, cash flow, cash, leverage, foreign sales, government ownership, and
exposure to government spending.
Inferences regarding the effect of sovereign downgrades are based on the
Abadie–Imbens matching estimator of the ATT. We perform difference-in-
differences estimations by comparing changes in the outcome variables be-
tween treatment and control groups around the sovereign downgrade. Stan-
dard errors are clustered by country event (i.e., sovereign downgrade) to
account for within-event residual correlation.
11 We also include industry (Fama-French 12-industry classification) as a covariate in an alter-
native control group.

D. Sample and Variable Construction

The sample of firms is taken from the WRDS-Factset Fundamentals Annual
Fiscal (North America and International) database. It includes firms from 80
countries from 1990 to 2013. We exclude financial firms (SIC codes 6000-6999)
because these firms tend to have significantly different investment and finan-
cial policies. We drop any observation with negative total assets. We obtain
firm accounting and market variables from Factset and sovereign and corpo-
rate credit ratings (foreign currency long-term issuer ratings) from Bloomberg.
We match firms in Factset to Bloomberg using ISIN, SEDOL, CUSIP, or com-
pany name. The initial sample includes 583,219 firm-year observations and
55,584 different firms. Only a small fraction of these firms have a rating (35,526
firm-year observations for 3,991 unique firms).
In our experiments, the outcome variables are the annual changes in firm
investment, debt issuance, and equity issuance around a sovereign downgrade.
Specifically, Investment is defined as the ratio of annual capital expenditures
(Factset item FF_CAPEX_FIX) to lagged net property, plant, and equipment
(Factset item FF_PPE_NET); Net Debt Issuance is computed from the state-
ment of cash flows as the ratio of net debt issuance (Factset item FF_DEBT_CF)
to lagged total assets (Factset item FF_ASSETS); and Net Equity Issuance is
also computed from the statement of cash flows as the ratio of sales of common
and preferred stock (Factset item FF_STK_SALE_CF) minus repurchases of
common and preferred stock (Factset item FF_STK_PURCH_CF) to lagged to-
tal assets. In some tests, we also consider ROA (return on assets), defined as
the ratio of operating income (Factset item OPER_INC) minus income taxes
(Factset item FF_INC_TAX) to lagged total assets; Long-Term Leverage, de-
fined as the ratio of long-term debt (Factset item FF_DEBT_LT) to total assets;
Leverage, defined as the ratio of total debt (Factset item FF_DEBT) to total as-
sets; and Cash, defined as the ratio of cash and short-term investment (Factset
item FF_CASH_ST) to total assets.
Table I reports the number of treated firm-year observations by country
and year. Table IA.I in the Internet Appendix presents the full list of treated
firms as well as their country of domicile, treatment year, and rating at the
beginning and end of the treatment year.12 The treatment group comprises
73 firms from a total of 13 different countries. The treated firms come from
both developed markets (Ireland, Italy, Japan, Portugal, Spain, and the United
States) and emerging markets (Argentina, Brazil, Hungary, Indonesia, Mexico,
Philippines, and Thailand). Some of these countries had multiple downgrades
over the sample period, such as Italy with five downgrades, Argentina with four,
Japan with three, and Portugal and Thailand with two, and thus the treated
sample spans 24 sovereign downgrades. The median sovereign downgrade is
one notch and the average is two notches. There are 14 downgrades during
the post-2007 period corresponding to the global financial crisis and eurozone
12 The Internet Appendix is available in the online version of the article on the Journal of
Finance website.
Table I
Sample of Sovereign Rating Downgrades
This table presents the sample of sovereign credit rating downgrades and the number of treated
observations (i.e., firm-year observations with credit rating equal to or above the sovereign rating
in the year before a sovereign downgrade) using S&P long-term foreign currency issuer ratings.
Sovereign Rating
Number of
Country Downgrade Year Before Downgrade After Downgrade Observations
Argentina 2001 BB− SD 4

2008 B+ B− 3
2012 B B− 1
2013 B− CCC+ 2
Brazil 2002 BB− B+ 5
Hungary 2012 BB+ BB 1
Indonesia 1998 BB+ CCC+ 4
Ireland 2011 A BBB+ 4
Italy 2004 AA AA− 1
2006 AA− A+ 2
2011 A+ A 2
2012 A BBB+ 2
2013 BBB+ BBB 7
Japan 2001 AAA AA 1
2002 AA AA− 4
2011 AA AA− 13
Mexico 2009 BBB+ BBB 4
Philippines 2005 BB BB− 2
Portugal 2010 A+ A− 1
2011 A− BBB− 1
Spain 2012 AA− BBB− 2
Thailand 1997 A BBB 1
1998 BBB BBB− 2
U.S. 2011 AAA AA+ 4
Total 73
sovereign debt crises, but there are also a sizable number of downgrades in
earlier periods. There were additional sovereign downgrades during our sample
period (e.g., France), but we rely only on those for which we can identify treated
firms in the downgraded country.
The firm-level covariates include firm size, investment, Tobin’s Q, cash flow,
cash, leverage, and foreign sales, where Size is defined as the logarithm of
total assets; Tobin’s Q is defined as the ratio of total assets plus market cap-
italization (Factset item FF_MKT_VAL) minus common equity (Factset item
FF_COM_EQ) to total assets; Cash Flow is defined as the ratio of annual oper-
ating income (Factset item FF_OPER_INC) plus depreciation and amortization
(Factset item FF_DEP_AMORT_EXP) to lagged total assets; and Foreign Sales
is the ratio of foreign sales to total sales (Factset item FF_FOR_SALES_PCT).
The covariates also include Government Ownership, defined as the total (di-
rect and indirect) number of shares held by the government as a percentage
of shares outstanding (using Factset ownership data), and Government Expo-

sure, defined as the percentage of output purchased, directly or indirectly, by
the government at the three-digit SIC level.13 The matching estimator uses
the pretreatment (year prior to the sovereign downgrade) value of the covari-
ates. To minimize the impact of outliers on these comparisons, we winsorize
variables at the top and bottom 1% level.
We also match firms on year and their country of domicile. Accordingly, we
impose the condition that the control firm should exactly match the country
and year of the treated firm so that we compare outcomes among firms in the
same country and year. We also require control firms to have a credit rating,
as treated firms are necessarily rated.14
Our benchmark specification does not include industry as a covariate in
the matching because matching on industry in addition to country and year
would significantly reduce the sample size, especially in smaller countries. We
also report estimates using an alternative group of control firms that matches
the Fama-French 12-industry classification of treated firms. In this alterna-
tive specification, we find a match within the same Fama-French 12-industry
classification for 40 of the 73 treated observations. We drop the remaining 33
observations from this analysis.
E. Example
One of the firms in our treatment group comes from the energy sector: EDP
Energias de Portugal. S&P downgraded Portugal’s sovereign rating on March
25, 2011, from A− to BBB, and then again on March 28, 2011, to BBB−. As
a consequence, EDP was downgraded on March 28, 2011, from A− to BBB.
The effect of the sovereign downgrade on the firm’s rating was explained by
Miguel Viana, Head of its Investor Relations Office, in the 2011 year-end results
conference call:
In terms of credit ratings, EDP recently suffered with downgrades by S&P
and Moody’s, penalized by the maximum notch differential allowed be-
tween EDP and Portugal Sovereign, so right now EDP is one notch above
Portugal by S&P and two notches above Portugal by Moody’s. Neverthe-
less, we consider that these by-the-book credit agencies methodologies are
unable to reflect EDP’s distinct credit profile, namely the geographical
13 We thank Frederico Belo for providing us the government exposure variable used in Belo,
Gala, and Li (2013).

14 We implement the matching estimator using the Stata command nnmatch. The algorithm
does not automatically force the match to be exact, but instead gives a weight of 1,000 (instead of
one) for the categorical variables for which we request an exact match. For example, the algorithm
may find an observation in a different country-year that is closer to our treated observation based
on the other covariates. This happens when the distance for the other covariates in the same
country-year is so large that it cancels the effects of the large weight for country-year. In our
application, we drop treated firms for which we are unable to find a match in the same country-
year (15 observations out of 88 potential observations on treated firms). We thus end up with a
sample of 73 treated firms.
diversification, the high quality of our generation fleet, our resilient

EBITDA, and the fact that our operations in Portugal have low sensi-
tivity to the economic cycle.
The effect of the sovereign downgrade on EDP’s investment and financial

policy was explained by Antonio Mexia, Chief Executive Officer, in the 2011
and 2012 year-end results conference call:
We are reducing CAPEX not only because of the evolution of the energy
market but also to improve financials. The CAPEX fell 19% to less than
2.2 billion euros, especially because of the lower additions in the U.S.
market. In the disposals program we reached 440 million euros in cash
proceeds. I would also like to mention the fact that CAPEX were 2 billion
euros, 7% lower on year-on-year basis, namely due to fewer expansion
projects in wind power especially in the U.S. market, and by the fact that
we went down the road once again in what concerns the deleveraging
through disposals.
Although EDP signed a credit line of 2 billion euros in November 3, 2010,

with a spread of 90 basis points over Euribor, the company saw the spread
increase significantly as the facility terms included a credit rating grid (i.e.,
the spread to be paid at each rating level). The managers’ comments indicate
that the link between the corporate rating and sovereign rating is due to ceiling
policies that are unrelated to firm fundamentals. The EDP example shows how
a rating downgrade can affect a firm’s cost of capital, investment, and financial
policy.
F. Summary Statistics
Panel A of Table II compares means and medians of the covariates between
the 73 treated firm-years and the remaining 21,618 nontreated firm-years (i.e.,
firms that are not assigned to the treatment group in the year prior to the
sovereign downgrade). We restrict the group of nontreated firms to countries
that have least one sovereign downgrade over the sample period. The treated
firms are bigger and have higher investment rate, Tobin’s Q, cash flow, and
leverage, as well as more government ownership than nontreated firms, on
average. These differences are expected, given that we are relying on obser-
vational data. The goal of the matching estimator techniques is to control for
these distributional differences, which could affect posttreatment outcomes.
Panel A of Table II also shows medians and means of the covariates for
the matched control firms. The Abadie–Imbens matching estimator identifies
a match for each firm in the treatment group. We thus have 73 firm-year
observations in both groups, but because matching is done with replacement,
we have 53 unique firm-year observations in the control group. The Pearson
χ 2 statistic tests for differences in the medians of the covariates between the
treatment and control groups.
Table II
Summary Statistics: Median and Mean Tests
This table presents the pretreatment median and mean of nontreated, treated and control groups. The sample consists of WRDS-Factset Fundamentals
Annual Fiscal (North America and International) nonfinancial firms over the 1990 to 2013 period. Treated firms have a credit rating equal to or above
the sovereign rating in the year before a sovereign downgrade. Nontreated firms are all other firms in the sample. The group of nontreated firms is
restricted to countries that have least one sovereign downgrade over the sample period. Control firms are matched firms using the Abadie and Imbens
matching estimator. The covariates are country, year, size, investment, Tobin’s Q, cash flow, cash, leverage, foreign sales, government ownership, and
exposure to government spending (pretreatment values). In Panel B, the covariates also include industry (Fama-French 12-industry classification).
The sample consists of 73 treated and control observations in Panel A, and 40 treated and control observations in Panel B. The Pearson’s χ 2 statistic
tests the difference in medians between treated and control firms. The Kolmogorov-Smirnov statistic tests the difference in distribution of treated and
control firms. Standard deviations of means are reported in parentheses. ***,**, and * indicate significance at the 1%, 5%, and 10% level, respectively.
Median Mean Kolmogorov-

Pearson χ 2 Smirnov
Nontreated Treated Control p-value Nontreated Treated Control p-value
Panel A: Matched Sample
Size 5.25 9.58 9.64 1.00 5.20 9.01 8.87 0.50

(0.01) (0.15) (0.16)
Investment 11.91 15.95 11.71 0.32 27.92 26.62 20.01 0.25
(0.42) (5.01) (2.81)
Q 1.02 1.19 1.01 0.02** 2.87 1.42 1.13 0.00***
(0.07) (0.08) (0.05)
Cash Flow 5.04 9.66 7.75 0.10* −8.99 11.02 7.48 0.04**
(0.54) (0.89) (0.78)
Cash 9.81 8.46 7.19 0.18 15.75 11.30 8.61 0.14
The Real Effects of Credit Ratings
(0.11) (1.21) (0.86)

Leverage 23.07 33.39 36.96 0.74 30.07 33.32 36.62 0.64
(0.24) (2.15) (2.03)
Foreign Sales 0.00 0.00 23.19 0.19 11.72 21.46 27.34 0.28
(0.14) (3.15) (3.35)
Gov. Ownership 0.00 0.00 0.00 0.47 0.55 4.39 3.12 1.00
(0.03) (1.37) (1.12)
Gov. Exposure 12.12 11.80 11.80 0.62 14.69 15.02 13.44 0.77
(0.07) (1.18) (0.78)
263
(Continued)
264
Table II—Continued
Median Mean Kolmogorov-

Pearson χ 2 Smirnov
Nontreated Treated Control p-value Nontreated Treated Control p-value
Panel B: Matched Sample with Industry Match (Fama-French 12-industry classification)
Size 5.25 9.79 9.75 0.91 5.20 9.26 9.08 0.90

(0.01) (0.19) (0.21)
Investment 11.91 15.07 16.45 0.43 27.92 23.49 21.02 0.38
(0.42) (7.74) (3.38)
Q 1.02 1.26 1.06 0.02** 2.87 1.36 1.17 0.00***
(0.07) (0.07) (0.06)
Cash Flow 5.04 9.24 6.11 0.14 −8.99 11.18 6.32 0.00***
(0.54) (1.24) (0.96)
Cash 9.81 10.37 7.46 0.31 15.75 11.89 10.59 0.26
(0.11) (1.50) (1.48)
Leverage 23.07 31.99 33.36 0.74 30.07 30.86 33.43 0.45
The Journal of FinanceR
(0.24) (2.97) (2.00)

Foreign Sales 0.00 16.64 29.31 0.74 11.72 23.70 26.87 0.93
(0.14) (4.15) (4.19)
Gov. Ownership 0.00 0.00 0.00 0.96 0.55 5.81 4.67 1.00
(0.03) (2.32) (2.10)
Gov. Exposure 12.12 11.80 11.80 0.15 14.69 14.26 11.96 0.41
(0.07) (1.81) (0.92)
After the matching procedure, there are no statistically significant differ-

ences in the predowngrade median values of the covariates across treatment
and control groups, with the exception of cash flow and Q. The median cash
flow and Q are higher for firms in the treatment versus the control group. The
difference in cash flow cannot explain our findings, as we would expect firms
with higher cash flow to be less affected by a sovereign downgrade than firms
with lower cash flow. In contrast, the difference in Q can explain our findings,
as firms with higher Q may be more sensitive to shocks than firms with lower
Q.15
The last column in Table II compares the entire distributions of the matching
covariates between the treatment and control groups using the Kolmogorov–
Smirnov test. While treated firms differ significantly from nontreated firms,
these differences disappear when we compare the group of treated firms to the
group of matched control firms. Similarly to the median tests, treated firms
have higher average Q and cash flow than control firms. Table IA.II in the
Internet Appendix shows the distribution support of the covariates across the
three groups.
A concern is that treated firms could be more affected following a sovereign
downgrade because of higher exposure to the government. Treated firms have
average government ownership of 4% versus 3% for control firms, but the dif-
ference is statistically insignificant. The difference in exposure to government
spending is also insignificant. The distribution of government ownership and
exposure to government spending is also similar between treatment and control
groups.16
Panel B of Table II shows medians and means of the covariates for the treated,
nontreated, and matched control firms using the alternative matched sample
with exact industry matching (Fama-French 12-industry classification). This
alternative matching procedure generates similar groups to those in Panel A.
And again, there are no statistically significant differences in the predowngrade
median and mean values of the covariates across treatment and control groups,
with the exception of cash flow and Q.
II. Effect on Corporate Ratings

We examine whether sovereign downgrades have a differential effect on cor-
porate ratings for bound firms (treatment group) and nonbound firms (control
group). We expect treated firms to be more affected than otherwise similar firms
at the time of a sovereign downgrade through the sovereign ceiling channel.
Spillovers or common macro shocks associated with the sovereign downgrade,
15 In Section I.E, we conduct a robustness test in which we use only Q and size (and categorical
variables) as matching covariates. In this alternative sample, there is no difference in Q across

treatment and control firms.
16 We also check whether treated and control firms differ in terms of debt maturity structure or
debt rollover risk (e.g., Almeida et al. (2012)). There are no significant differences in the distribution
of the ratio of long-term debt to total assets (Long-Term Leverage) between treatment and control
groups (see Table IA.II in the Internet Appendix).
Table III
Difference-in-Differences in Corporate Ratings around a Sovereign
Downgrade
This table presents difference-in-differences matching estimators for corporate ratings around a
sovereign downgrade. Corporate ratings are converted to a numerical scale with 22 corresponding
to the highest rating (AAA) and one to the lowest (default). Treated firms have a credit rating
equal to or above the sovereign rating in the year before a sovereign downgrade. Control firms are
matched firms using the Abadie and Imbens matching estimator. The covariates are country, year,
size, investment, Tobin’s Q, cash flow, cash, leverage, foreign sales, government ownership, and
exposure to government spending (pretreatment values). In Panel B, the covariates also include
industry (Fama-French 12-industry classification). The sample consists of 73 treated and control
observations in Panel A, and 40 treated and control observations in Panel B. Robust standard
errors clustered by country-event are reported in parentheses. ***,**, and * indicate significance
at the 1%, 5%, and 10% level, respectively.
Year before Downgrade Year of Downgrade Difference
Treated Firms 16.11*** 14.75*** −1.36***

(1.09) (1.18) (0.41)
Control Firms 12.60*** 11.94*** −0.65
(1.01) (1.21) (0.47)
Difference 3.51*** 2.81***
(0.38) (0.51)
Difference in Differences −0.71*
(0.38)
Matching Estimator −0.88**
(0.35)
Treated Firms 16.95*** 16.13*** −0.83

(1.08) (1.18) (2.24)
Control Firms 12.83*** 12.75*** −0.08
(1.08) (1.10) (2.04)
Difference 4.13*** 3.38***
(0.33) (0.47)
(0.35)
(0.56)
however, should affect treated and control firms equally or, if differently, they
should affect control firms more than treated firms.
Table III presents difference-in-differences matching estimators for corporate
ratings. To perform this test we map the ratings into 22 numerical values (see
Table IA.III in the Internet Appendix for details), where 22 corresponds to the
highest rating (AAA) and one to the lowest (default). Panel A of Table III reports
the average rating for treated and control firms in the year before the sovereign
downgrade and in the year of the sovereign downgrade. Not surprisingly, we
see that the predowngrade rating is significantly higher for treated firms than
for control firms. The average treated firm has a rating value of 16 (i.e., A−),
and the average control firm has a rating value of 13 (i.e., BBB−).
We find that sovereign downgrades have a much stronger effect on treated
firm ratings, with a reduction of 1.4 notches, than on control firms ratings,
which are reduced by only 0.7 notches. These estimates suggest that ratings
decline 0.7 notches more for bound firms than for otherwise similar firms that
are not bound by the sovereign ceiling. The effect of the sovereign downgrade
on treated firm ratings is nearly one-to-one, while control firm ratings are not
significantly affected by the sovereign downgrade. The median sovereign rating
downgrade is one notch and the average is two notches.
Panel A of Table III also reports the differential change in ratings that
is produced by the matching estimator of the ATT. The estimate is equal to
−0.9 notches, indicating a significant asymmetry in the reaction of treatment
and control group ratings to a sovereign downgrade. Panel B presents similar
results using the matched sample with industry match. The ATT estimate is
stronger at −1.4 notches.
Figure 3 plots the evolution of corporate ratings over the two years before
and after the sovereign downgrade for the treatment and control groups. The
ratings of the two groups follow parallel trends before the sovereign downgrade.
Furthermore, the ratings fall significantly more for the treatment group in the
year of the downgrade (which occurs between year −1 and year 0) than for the
control group.
We perform other tests to study the relation between sovereign and corporate
ratings. In particular, we test whether bound firms have a more “pessimistic”
rating than nonbound firms. Tables IA.IV and IA.V in the Internet Appendix
show that ratings are more pessimistic for bound firms than nonbound firms
with the same actual ratings. This finding is consistent with the notion that
the sovereign ceiling represents a meaningful friction and not just an unbiased
and accurate assessment of a firm’s creditworthiness.
III. Effect on Investment and Financial Policy

In this section, we examine the investment and financial policy of treated
and matched control firms around sovereign downgrades.
A. Investment Policy
Panel A of Table IV presents difference-in-differences matching estimators
for investment rates around sovereign downgrades as measured by annual cap-
ital expenditures as a percentage of capital (Investment). The table presents
the investment rates in the year before the sovereign downgrade and in the
year of the sovereign downgrade. Firms in the treatment group (those with pre-
downgrade rating equal to or above the sovereign rating) are compared with
close counterfactuals (matched control firms). We find that treated firms reduce
investment significantly more than control firms at the time of a sovereign
downgrade. For firms in the treatment group, average investment drops from
Figure 3. Corporate ratings around a sovereign downgrade. This figure shows the evolu-
tion of corporate ratings of treatment and control groups around a sovereign downgrade (which
occurs between year −1 and year 0). Corporate ratings are converted to a numerical scale with
22 corresponding to the highest rating (AAA) and one to the lowest (default). Treated firms have
a credit rating equal to or above the sovereign rating in the year before a sovereign downgrade.
Control firms are matched firms using the Abadie and Imbens matching estimator. The covariates
are country, year, size, investment, Tobin’s Q, cash flow, cash, leverage, foreign sales, government
ownership, and exposure to government spending (pretreatment values). In Panel B, the covariates
also include industry (Fama-French 12-industry classification).
26.6% to 17.7% of capital, a reduction of 8.9 percentage points. For control

firms, investment decreases only slightly from 19.2% to 16.6% of capital, a
reduction of 2.6 percentage points. Investment is therefore reduced 6.4 per-
centage points more for treated firms than control firms, which is statistically
and economically significant.
Panel A of Table IV also reports the differential change in investment
that is produced by the bias-corrected matching estimator of the ATT. The
Table IV
Difference-in-Differences in Investment around a Sovereign
Downgrade
This table presents difference-in-differences matching estimators for investment rate around a
sovereign downgrade. Investment rate is the ratio of annual capital expenditures to lagged net
property, plant, and equipment. Treated firms have a credit rating equal to or above the sovereign
rating in the year before a sovereign downgrade. Control firms are matched firms using the Abadie
and Imbens matching estimator. The covariates are country, year, size, investment, Tobin’s Q, cash
flow, cash, leverage, foreign sales, government ownership, and exposure to government spending
(pretreatment values). In Panel B, the covariates also include industry (Fama-French 12-industry
classification). The sample consists of 73 treated and control observations in Panel A, and 40
treated and control observations in Panel B. Robust standard errors clustered by country-event
are reported in parentheses. ***,**, and * indicate significance at the 1%, 5%, and 10% level,
respectively.
Treated Firms 26.62*** 17.70*** −8.92

(7.28) (1.83) (6.34)
Control Firms 19.15*** 16.57*** −2.58
(5.06) (3.23) (3.95)
Difference 7.48** 1.13
(3.56) (2.70)
(3.71)
(4.32)
Treated Firms 23.49** 16.15*** −7.33

(8.83) (2.27) (9.12)
Control Firms 21.02*** 18.65*** −2.37
(4.46) (3.33) (5.57)
Difference 2.47 −2.49
(6.57) (1.83)
Difference in Differences −4.96
(6.82)
(3.10)
ATT is equal to −8.9 percentage points, which corresponds to 33% of the

predowngrade investment level for treated firms. The ATT estimate is signif-
icantly higher (in absolute terms) than the difference-in-differences estimate
because it introduces a bias correction to account for differences in the distri-
bution of covariates between the treatment and control groups. For example,
control firms have lower average cash flow than treated firms. Since firms with
lower cash flow are expected to have a larger reduction in investment, the
difference-in-differences estimate is biased downwards (in absolute terms). If
control firms were to have a cash flow distribution similar to that of treated
firms, then the differential reduction in investment across treated and control
firms would likely increase.
Panel B of Table IV reports the above estimates using the alternative match-
ing procedure in which the control firms match the industries of the treated
firms. The differential change in investment produced by the matching estima-
tor of the ATT is −6.7 percentage points. This effect is lower than that in Panel
A but still statistically and economically significant: the ATT corresponds to
about 25% of the predowngrade investment level for treated firms.
A concern about inferences based on the treatment effects framework is
whether the processes generating the treatment and control group outcomes
would have followed parallel trends in the absence of the treatment. Differences
in the posttreatment period can only be attributed to the treatment when this
assumption holds. While not a direct test of the parallel trends assumption, it
is standard in the literature to examine the evolution of the outcome variable
(Investment) in the years preceding the treatment separately for the treatment
and control groups. If the trends are not parallel prior to the event, it is unlikely
that postevent differences can be attributed to the treatment.
Figure 4 plots the evolution of investment rates in the two years before and
after the sovereign downgrade. The investment processes of the two groups
follow similar trends before the downgrade. Investment falls dramatically for
the treatment group in the year of the downgrade and only slightly for the con-
trol group. In the two years following the downgrade, the investment processes
again follow similar dynamics. Thus, we identify a unique effect on investment
at the time of the sovereign downgrade.17
B. Placebo Tests
A potential concern regarding our difference-in-differences approach is
whether macro factors other than sovereign downgrades affecting both treat-
ment and control groups can explain the differential behavior in the posttreat-
ment period. For example, treated firms can face greater exposure to adverse
economic conditions and credit supply shocks, which may happen at the same
time as sovereign downgrades. One appealing feature of our identification strat-
egy is that it is difficult to find a story in which higher-quality (treated) firms
are more affected than lower-quality (control) firms.
To strengthen the interpretation of the results, in this section we replicate
the experiment that we run above for sovereign downgrades but using placebo
periods. In particular, we use sovereign and corporate rating information to
sort firms into treatment and nontreatment groups, and covariates to produce
a matched control group of firms. We then compare treated versus control firm
investment behavior during periods without a sovereign downgrade. We con-
sider three placebo periods: (1) recession periods, (2) the 2007 to 2009 financial
17 We further check the parallel trend assumption for investment rates by computing the differ-
ence between treatment and control groups between year −2 and year −1. The ATT estimate is −0.2
percentage points and statistically insignificant, which supports the notion that the investment
policies of our treatment and control groups only diverge from each other after the downgrade.
Figure 4. Investment around a sovereign downgrade. This figure shows the evolution of
the investment rate of treatment and control groups around a sovereign downgrade (which occurs
between year −1 and year 0). Investment rate is the ratio of annual capital expenditures to lagged
net property, plant, and equipment. Treated firms have a credit rating equal to or above the
sovereign rating in the year before a sovereign downgrade. Control firms are matched firms using
the Abadie and Imbens matching estimator. The covariates are country, year, industry (two-digit
SIC in Panel A and Fama-French 12-industry classification in Panel B), size, investment, Tobin’s
Q, cash flow, cash, leverage, foreign sales, government ownership, and exposure to government
spending (pretreatment values).
crisis, and (3) currency crises. These falsification tests can help rule out the
possibility that treated firms are more sensitive to demand and credit supply
shocks than control firms.
Panel A of Table V presents results of the placebo test using recession pe-
riods without a sovereign downgrade. We identify recession periods using the
Organisation for Economic Co-operation and Development (OECD) recession
indicators for each country, which come from the Federal Reserve Economic
Table V
Placebo Tests: Difference-in-Differences in Investment
This table presents difference-in-differences matching estimators for investment rate. Investment
rate is the ratio of annual capital expenditures to lagged net property, plant, and equipment. Panel
A presents the estimates around the first year of a recession not accompanied by a sovereign
downgrade. Panel B presents the estimates around the 2007 to 2009 financial crisis. Countries
downgraded during the crisis are excluded. Panel C presents the estimates around a currency
crisis not accompanied by a sovereign downgrade. Treated firms have a credit rating equal to or
above the sovereign rating in the year before a sovereign downgrade. Control firms are matched
firms using the Abadie and Imbens matching estimator. The covariates are country, year, size,
investment, Tobin’s Q, cash flow, cash, leverage, foreign sales, government ownership, and expo-
sure to government spending (pretreatment values). The sample consists of 53 treated and control
observations in Panel A, 56 treated and control observations in Panel B, and 53 treated and obser-
vations in Panel C. Robust standard errors clustered by country-event are reported in parentheses.
***,**, and * indicate significance at the 1%, 5%, and 10% level, respectively.
Panel A: Recession without a Sovereign Downgrade

Year before Recession Year of Recession Difference
Treated Firms 20.65*** 21.43*** 0.78
(1.91) (2.05) (2.46)
Control Firms 17.09*** 18.98*** 1.89
(2.68) (3.71) (5.41)
Difference 3.56 2.45
(2.16) (3.31)
(2.03)
Matching Estimator 0.56
(1.84)
Panel B: 2007 to 2009 Financial Crisis
2008 2009 Difference
Treated Firms 20.79*** 16.57*** −4.21
(3.14) (2.15) (3.75)
Control Firms 20.11*** 14.51*** −5.61**
(2.50) (1.00) (2.66)
(3.62) (1.78)
Difference in Differences 1.39
(2.39)
(3.05)
Panel C: Currency Crisis without a Sovereign Downgrade
Year before Year of
Currency Crisis Currency Crisis Difference
Treated Firms 23.15*** 23.78*** 0.63
(2.72) (2.42) (3.72)
Control Firms 20.91*** 21.95*** 1.03
(3.54) (2.47) (4.26)
(2.31) (3.93)
(3.63)
(0.38)
Data (FRED) database. For each country, we exclude recession years in which
the country is downgraded.18 There are 53 treated and control firms in this
placebo test. Treated and control firms have virtually identical investment dy-
namics before recessions. More important, there is no difference in investment
rate between the two groups of firms in the posttreatment period. The ATT
estimate is 0.6 percentage points and statistically insignificant. Simply put,
our treatment-control differences do not appear in recession periods that are
not accompanied by a sovereign downgrade.
Panel B of Table V presents results of the placebo test using the 2007 to 2009
financial crisis. We exclude countries that experienced a downgrade during the
crisis. This crisis was characterized by a large shock to the supply of capital
to firms. We find that the control group cuts investment by 1.4 percentage
points more than the treatment group in the aftermath of the crisis, which is
consistent with the idea that treated firms are less affected than control firms
in periods of financial turmoil. The ATT estimate is 1.9 percentage points but
statistically insignificant.
Panel C of Table V presents results of the placebo test using currency crises
not accompanied by a sovereign downgrade. This placebo addresses the concern
that currency crises affect treatment and control groups differently. For exam-
ple, treated firms may be more affected by a currency depreciation because they
use more foreign currency debt than control firms. If this is the case, we should
find differential effects between treatment and control groups during currency
crises. The currency crises indicators for each country come from Reinhart and
Rogoff (2009). 19 We find no difference in the investment rates of the two groups
at the time of currency crises. The ATT estimate is 0.5 percentage points and
statistically insignificant.
Panels A to C of Figure IA.1 in the Internet Appendix plot the evolution
of investment rates in the two years before and after recessions, the 2007 to
2009 financial crisis, and currency crises, respectively. The investment rates
for treated firms are higher than those for control firms, but the investment
processes follow similar dynamics around the placebo periods. There is no
evidence that treated firms are more affected than control firms around placebo
periods.
An additional concern is that treated firms may have greater exposure to
government fiscal crises than control firms. The group of treated firms may
include “national champions” that might be expected to receive support from
the government and for which a sovereign downgrade reduces the value of these
18 The recession indicator is available for 38 countries at a monthly frequency and we adopt the
“From the Period following the Peak through the Trough” definition. We aggregate the monthly
series into an annual series and classify a country as being in a recession in a given year if it has
more than six months of recession.
19 The currency crisis indicators are available up to 2010 on Carmen Reinhart’s website at
http://www.carmenreinhart.com/. We update the currency crisis indicators for the 2011 to 2013
period.
explicit or implicit subsidies.20 Additionally, treated firms might be subject

to higher corporate taxes at the time of a sovereign downgrade associated
with a sovereign debt crisis. Another alternative is that treated firms could
sell more goods or services to the government than control firms. We believe
that these alternative hypotheses are unlikely to explain our results, since the
matched control group includes firms of similar but lower credit quality in the
same country as those in the treatment group. Control firms also have similar
exposure to government spending (which is one of our matching variables).
However, to further examine the possibility that differential exposure to a
government factor drives our results, we perform several tests.
First, we repeat our investment tests after excluding firms with government
ownership from the treatment group (13 observations). Panel A of Table VI
presents the results. We obtain slightly higher differential effects on invest-
ment (ATT is −13 percentage points). Thus, if anything, government owner-
ship seems to protect firms against the adverse effects of a downgrade. We also
repeat our tests after excluding utilities (SIC codes 4900-4999) from the treat-
ment group, since these firms are arguably the most likely to have direct links
to the government and to receive support. The number of treated firms falls to
41 firm-year observations, but the main results hold. Panel B of Table VI shows
that the matching ATT is −13.8 percentage points when we exclude utilities.
Table IA.VI in the Internet Appendix shows that the estimates are qualitatively
similar to those in Table VI when we use the alternative matching procedure
in which the control firms match the industries of the treated firms.
Second, Panel C of Table VI presents results of a placebo test that compares
firms with a rating one notch below the sovereign (treated firms) to matched
control firms with a rating more than one notch below the sovereign. There
are 40 treated and control firms in this placebo test. We find no difference in
investment rate between the two groups of firms around sovereign downgrades.
The absence of a differential effect shows that our results do not simply capture
highly rated firms, but rather firms that are bound by the sovereign ceiling.
Third, we study the implications for security prices. If treated firms are
systematically more exposed to a government factor relative to control firms,
then the sensitivity of treated firms’ security prices to changes in the price
of government securities should be higher. The advantage of exploring this
implication is that it should hold even if there is some unobservable factor
through which firms are exposed to the government.
We explore this possibility by examining the sensitivity of both corporate
yields and stock returns to government yields for bound and nonbound firms.
Panel A of Table VII shows the estimates of regressions of corporate bond yields
on a dummy variable that takes a value of one if a firm has a rating equal
20 For example, Moody’s announcements around Japan’s downgrade mentioned bank and gov-
ernment support for the country’s major firms, such as Toyota: “The ratings of these corporates
incorporate one or two notches of uplift to their stand-alone credit profiles, reflecting our expec-
tation for strong support from the major domestic banks and the government to many Japanese
corporates, including these six issuers.”
Table VI
Government Support Tests: Difference-in-Differences in Investment
around a Sovereign Downgrade
property, plant, and equipment. Panel A presents results for a subsample that excludes firms
with government ownership. Panel B presents results for a subsample that excludes firms with
government ownership and utilities (SIC codes 4900-4999). Panel C presents results for a placebo
test in which treated firms have a credit rating one notch below the sovereign and control firms
have a credit rating more than one notch below the sovereign. Treated firms have a credit rating
equal to or above the sovereign rating in the year before a sovereign downgrade. Control firms
are matched firms using the Abadie and Imbens matching estimator. The covariates are country,
year, size, investment, Tobin’s Q, cash flow, cash, leverage, foreign sales, government ownership,
and exposure to government spending (pretreatment values). The sample consists of 60 treated and
control observations in Panel A, 41 treated and control observations in Panel B, and 40 treated and
control observations in Panel C. Robust standard errors clustered by country-event are reported
in parentheses. ***,**, and * indicate significance at the 1%, 5%, and 10% level, respectively.

Panel A: Excluding Firms with Government Ownership
Treated Firms 29.78*** 19.49*** −10.30
(8.82) (2.19) (9.60)
Control Firms 19.96*** 18.02*** −1.94
(4.72) (3.88) (6.16)
Difference 9.83* 1.47
(5.06) (3.14)
(6.39)
(5.13)
Panel B: Excluding Utilities
Treated Firms 33.22*** 22.81*** −10.41
(10.48) (1.90) (11.09)
Control Firms 25.21*** 20.88*** −4.33
(8.19) (5.08) (10.01)
Difference 8.00** 1.93
(3.65) (4.07)
(5.10)
(6.94)
Panel C: Treated Firms with Rating One Notch Below Sovereign
Treated Firms 13.80*** 12.01*** −1.79
(0.86) (0.88) (1.16)
Control Firms 15.09*** 13.92*** −1.17
(1.81) (1.35) (2.92)
Difference −1.29 −1.92
(1.85) (1.28)
(1.51)
Matching Estimator −0.30
(0.99)
Table VII
Government Exposure Tests: Corporate Yields and Stock Returns
This table presents estimates of linear regressions of corporate bond yields (Panel A) and stock
returns (Panel B) at the monthly frequency. Bound is a dummy variable that takes the value of
one if a firm has a credit rating equal to or above the sovereign rating in year t − 1. Sovereign
Yield is the 10-year constant-maturity sovereign yield. Market Return is the local stock market
return. Government Exposure is the industry-level measure of exposure to government spending.
Bond issue controls (coefficients not shown) in Panel A are coupon rate, issue amount (in U.S.
dollars), and maturity. Treated firms have a credit rating equal to or above the sovereign rating
in the year before a sovereign downgrade. Control firms are matched firms using the Abadie and
Imbens matching estimator. The covariates are country, year, size, investment, Tobin’s Q, cash
flow, cash, leverage, foreign sales, government ownership, and exposure to government spending
(pretreatment values). The sample consists of treated and control firms for which yields on local
currency bond issues (Panel A) and stock returns (Panel B) are available in the 1990 to 2013
period. Robust standard errors clustered by firm are reported in parentheses. ***,**, and * indicate
significance at the 1%, 5%, and 10% level, respectively.
Panel A: Corporate Yield
(1) (2) (3)
Bound 0.504 2.161 0.472

(0.429) (2.332) (0.641)
Sovereign Yield 0.548*** 0.666** 0.712***
(0.187) (0.265) (0.083)
Bound × Sovereign Yield −0.240* −0.412 −0.036
(0.140) (0.391) (0.098)
Bond Issue Controls No No Yes
Time FE Yes Yes Yes
Country FE Yes No No
Firm FE No Yes Yes
Corporate Ratings FE No No Yes
Observations 54,202 54,202 53,109
R2 0.500 0.555 0.876
Panel B: Stock Return
(1) (2) (3) (4)
Bound 0.002 0.000 0.004 0.005*

(0.001) (0.001) (0.003) (0.003)
Market Return 0.825*** 0.825***
(0.071) (0.072)
Market Return × Bound −0.157** −0.159**
(0.065) (0.066)
Sovereign Yield −0.332* −0.073
(0.171) (0.144)
Sovereign Yield × Bound 0.098 −0.096
(0.233) (0.176)
Gov. Exposure 0.024 0.022
(0.018) (0.018)
Gov. Exposure × Bound −0.022 −0.022
(0.019) (0.019)
Time FE Yes Yes Yes Yes
Country FE Yes Yes Yes Yes
Observations 14,097 14,097 14,097 14,097
R2 0.339 0.228 0.228 0.339
to or above the sovereign rating in year t − 1 (Bound), the 10-year constant-

maturity sovereign bond yield (Sovereign Yield), and the interaction between
these two variables. The sample consists of monthly data on local currency
bond yields for the treated and matched control firms from 1990 to 2013 (31
treated firms and 11 control firms for which bond yields and their corresponding
10-year sovereign bond yields are available). We use the whole time series of
yields for each bond issue and thus we do not restrict the sample to sovereign
downgrades, as our purpose is to explore a channel other than the sovereign
ceiling that potentially links government and corporate outcomes. The base
model in column (1) includes time and country fixed effects, while columns (2)
and (3) add firm fixed effects, corporate ratings fixed effects, and bond issue
controls (coupon rate, issue amount in U.S. dollars, and maturity).
As expected, there is a strong and positive association between corporate and
government yields. The coefficient on Sovereign Yield is about 0.6 to 0.7. More
importantly, however, the coefficient on the interaction term Sovereign Yield ×
Bound is negative in all specifications and generally statistically insignificant,
which does not support the hypothesis that treated firm yields are systemati-
cally more sensitive to changes in government yields. If anything, the negative
coefficient would suggest that the market price of treated firms’ debt securities
is less sensitive to changes in government bond yields.
Panel B of Table VII presents estimates of a similar analysis to test whether
the stock returns of treatment and control firms have systematically different
government betas. We run a regression of a firm’s monthly stock return on
its local stock market return (Market Return), the monthly change in 10-year
constant-maturity sovereign bond yields (Sovereign Yield), and the interac-
tion of these two variables with Bound. The sample consists of monthly obser-
vations on stock returns for the treated and matched control firms from 1990
to 2013 (55 treated firms and 25 control firms for which stock returns are avail-
able). As in Panel A, we use the whole time series of stock returns for each firm.
The base regression in column (1) shows that, if anything, bound firms have a
statistically lower stock beta relative to control firms: 0.67 for the treatment
group versus 0.83 for the control group. More importantly for our purpose, the
magnitude of the “government beta” does not appear to differ across the two
groups, as shown by the statistically insignificant coefficient on the interaction
term Sovereign Yield × Bound in columns (2) and (4). We also explore the
sensitivity to the government using the industry-level measure of exposure to
government spending in the regressions in columns (3) and (4). The coefficient
on the interaction term Gov. Exposure × Bound is statistically insignificant,
which is consistent with results using sovereign bond yields. We conclude that
treated firms do not appear to be systematically more exposed to a government
factor than control firms.
Finally, we check whether there is a differential effect on the after-tax prof-
itability of treated firms versus control firms. If there is differential government
support to treated and control firms, we should observe short-term effects on
the profitability of these firms. We examine the differential effect on ROA after
taxes. Table IA.VII in the Internet Appendix shows that there is no differential
effect of sovereign downgrades on the profitability of treated firms in the year

of the downgrade.
C. Financial Policy
We next examine whether sovereign downgrades affect the financial policy
of treated and matched control firms differently. We expect treated firms to
reduce debt issuance, as they face a contraction in debt supply and an increase
in the cost of debt, and expect control firms to be less affected.
Table VIII examines the effect of a sovereign downgrade on net debt issuance
as a percentage of assets (Net Debt Issuance) in Panel A, and on net equity is-
suance as a percentage of assets (Net Equity Issuance) in Panel B. Immediately
following the downgrade, treated firms experience a sharp decrease in net debt
issuance from 7.5% of assets to 2.4%, a reduction of 5.1 percentage points. For
control firms, net debt issuance falls by 2.3 percentage points. The difference-
in-difference estimator is −2.8 percentage points but statistically insignificant.
The table also reports the differential change in net debt issuance that is pro-
duced by the bias-corrected matching estimator of the ATT, which is equal to
−5.5 percentage points. The effect is statistically significant at the 5% level.
Unlike debt issuance, there is no significant reduction in equity issuance after
the downgrade for the treatment group. The control group experiences a de-
crease in net equity issuance of 2.4 percentage points. The ATT of net equity
issuance is equal to 2.6 percentage points, which is statistically significant at
the 5% level. Table IA.VIII in the Internet Appendix reports the estimates using
the alternative matching procedure in which the control firms match the indus-
tries of the treated firms exactly. The differential effects on net debt and equity
issuance are qualitatively similar to those in the benchmark specification but
more imprecisely estimated due to the smaller sample size.
Panels A and B of Figure 5 plot the evolution of net debt and equity issuance
in the two years before and after the sovereign downgrade for treatment and
control firms. The security issuance patterns of firms in the two groups follow
similar trends before the downgrade although there is a spike in debt issuance
just before the sovereign downgrade. Panel A shows that net debt issuance
falls dramatically for the treatment group in the year of the downgrade and
only slightly for the control group. In the two years following the downgrade,
the debt issuance patterns again follow similar dynamics.21 Panel B shows
that net equity issuance falls significantly for the control group in the year of
the downgrade and only slightly for the treatment group. Furthermore, equity
issuance increases for the treatment group in the year after the sovereign
downgrade while it stays unchanged for the control group. Taken together,
this evidence suggests that treated firms experience a shock to the ability to
21 We also estimate the difference in net debt and equity issuance between treatment and control
groups between year −2 and year −1. The ATT estimates of net debt and equity issuance are 2.98
and −0.14 percentage points, respectively, and statistically insignificant, which supports the notion
that the change in financial policy is not part of long-term trends before the downgrade.
Table VIII
Difference-in-Differences in Debt and Equity Issuance around a
Sovereign Downgrade
This table presents difference-in-differences matching estimators for net debt and net equity is-
suance around a sovereign downgrade. Net debt issuance is the ratio of debt issuance to lagged total
assets. Net equity issuance is the ratio of sales of common and preferred stock minus repurchases
of common and preferred stock to lagged total assets. Panel A presents net debt issuance estimates
and Panel B presents net equity issuance estimates. Treated firms have a credit rating equal to or
above the sovereign rating in the year before a sovereign downgrade. Control firms are matched
firms using the Abadie and Imbens matching estimator. The covariates are country, year, size, in-
vestment, Tobin’s Q, cash flow, cash, leverage, foreign sales, government ownership, and exposure
to government spending (pretreatment values). The sample consists of 73 treated and control ob-
servations. Robust standard errors clustered by country-event are reported in parentheses. ***,**,
and * indicate significance at the 1%, 5%, and 10% level, respectively.
Panel A: Net Debt Issuance
Treated Firms 7.52* 2.41** −5.11

(4.08) (0.89) (4.54)
Control Firms 3.35 1.07 −2.28
(2.45) (1.08) (2.82)
(2.52) (1.15)
(3.19)
(2.71)
Panel B: Net Equity Issuance
Treated Firms −0.20 −0.75 −0.55

(0.65) (0.44) (0.56)
Control Firms 1.92 −0.49 −2.41*
(1.40) (0.49) (1.37)
Difference −2.12** −0.26
(0.86) (0.21)
Difference in Differences 1.86**
(0.87)
Matching Estimator 2.56**
(1.15)
raise debt following a sovereign downgrade, and that this shock leads firms to
replace debt with equity.
We also examine the relative change in leverage and cash holdings for treat-
ment and control groups. The predictions for leverage and cash are less clear-
cut. The negative shock to treated firms’ ability to issue debt should cause them
to have lower leverage. However, the sharp reduction in investment should de-
crease asset growth and push toward higher leverage. Cash holdings are also
affected by multiple forces. On the one hand, treated firms could use some of
their cash to help weather the negative financial shock associated with the
Figure 5. Debt and equity issuance around a sovereign downgrade. This figure shows the
evolution of net debt and net equity issuance of treatment and control groups around a sovereign
downgrade (which occurs between year −1 and year 0). Net debt issuance is the ratio of debt
issuance to lagged total assets. Net equity issuance is the ratio of sales of common and preferred
stock minus repurchases of common and preferred stock to lagged total assets. Treated firms have
a credit rating equal to or above the sovereign rating in the year before a sovereign downgrade.
Control firms are matched firms using the Abadie and Imbens matching estimator. The covariates
are country, year, industry (two-digit SIC), size, investment, Tobin’s Q, cash flow, cash, leverage,
foreign sales, government ownership, and exposure to government spending (pretreatment values).
downgrade. On the other hand, they may have incentives to save more cash
to regain their predowngrade credit rating or for precautionary reasons. Table
IA.IX in the Internet Appendix presents the results. We do not find significant
differential effects in leverage and cash holdings.
Overall, the evidence suggests that rating downgrades induce a contraction

in the supply of debt capital and an increase in its cost, which leads to lower
investment rates and less use of debt.
D. Linear Regression Model

While the nonparametric matching approach is well suited for our test strat-
egy, we also implement reduced-form linear regressions to examine whether
firms’ investment rates drop for firms that are bound by the sovereign ceiling
following a sovereign downgrade.
First, we run a pooled OLS regression using the sample of all firms in the 1990
to 2013 period. The dependent variable is the annual change in the investment
rate (Investment) in year t. The main explanatory variables are a dummy
variable that takes a value of one if a firm has a rating equal to or above the
sovereign rating in year t − 1 (Bound), a dummy variable that takes a value of
one if a firm’s country rating is downgraded in year t (Sovereign Downgrade),
and the interaction term Bound × Sovereign Downgrade. The coefficient on the
interaction term captures the difference in the reaction of the investment rates
of treated firms and nontreated firms following a sovereign downgrade.
Table IX reports the results. Column (1) shows that treated firms reduce
their investment by 9.6 percentage points more than other firms as indicated
by the interaction term’s coefficient. The group difference estimate is signifi-
cant at the 5% level. In column (2), we estimate the investment regressions
including the same covariates (coefficients not shown) as in Table II. While
these controls predict changes in investment in their own right, their inclusion
does not materially alter the coefficient on the interaction term. The estimated
group-mean difference increases slightly to 10.2 percentage points. Columns
(3) to (5) present additional results using combinations of year, industry (two-
digit SIC), country, and firm fixed effects. In particular, column (5) presents
estimates using country × industry × year fixed effects, so that the effect is
driven only by variation within a given country-industry-year.
Overall, the magnitudes of the group difference estimates in Table IX are
similar at about 10 percentage points and statistically significant in all speci-
fications. Furthermore, the difference between the two groups of firms outside
the sovereign downgrade period becomes statistically insignificant when we
include country fixed effects, as indicated by the coefficient on Bound. The lin-
ear model regression estimates are consistent with those reported under the
matching estimator approach in Table IV.
E. Robustness
We perform several robustness checks of our primary findings on the effect
of rating downgrades on corporate investment. The results of these tests are
reported in the Internet Appendix.
We first account for the presence of serial correlation in the data and consider
a wider event window. Following Bertrand, Duflo, and Mullainathan (2004),
Table IX
Linear Regression: Difference-in-Differences in Investment around a
Sovereign Downgrade
This table presents difference-in-differences linear regression estimators for investment rate
around a sovereign downgrade. The dependent variable is the annual change in investment rate
(Investment) in year t. Investment rate is the ratio of annual capital expenditure to lagged net
property, plant, and equipment. Bound is a dummy variable that takes the value of one if a firm
has a credit rating equal to or above the sovereign rating in year t − 1, and Sovereign Downgrade
is a dummy variable that takes the value of one if a firm’s country rating is downgraded in year
t. The control variables (coefficients not shown) are firm size, Tobin’s Q, cash flow, cash holdings,
leverage, foreign sales, a dummy that takes the value of one if the firm has a credit rating, gov-
ernment ownership, and exposure to government spending. The regressions also include year,
industry (two-digit SIC), country, and firm fixed effects. The sample consists of WRDS-Factset
Fundamentals Annual Fiscal (North America and International) nonfinancial firms in the 1990
to 2013 period. Robust standard errors clustered by country-event are reported in parentheses.
***,**, and * indicate significance at the 1%, 5%, and 10% level, respectively.
(1) (2) (3) (4) (5)
Bound × Sovereign Downgrade −9.62** −10.18** −9.44* −11.42*** −12.50**

(4.42) (5.58) (5.42) (3.72) (6.43)
Bound 4.69*** 1.85** −0.76 −1.66 0.08
(0.78) (0.81) (0.82) (1.21) (1.66)
Sovereign Downgrade 2.23 3.46* 1.50 0.44
(1.49) (2.04) (1.97) (2.55)
Controls No Yes Yes Yes Yes
Year FE No Yes Yes Yes No
Industry FE No Yes Yes No No
Country FE No No Yes No No
Firm FE No No No Yes No
Country × Industry × Year FE No No No No Yes
Observations 424,903 348,593 348,593 348,593 348,593
R2 0.000 0.016 0.017 0.135 0.090
we collapse the investment and debt issuance data into periods before and
after the downgrade: window (−2,−1) and window (0,+2). Tables IA.X and
IA.XI in the Internet Appendix report the results. The differential change
in investment produced by the matching ATT is −6.0 percentage points and
statistically significant. The ATT of net debt issuance is also significant at −3.9
percentage points.
To check the sensitivity of our findings to outliers, we first obtain estimates
by deleting from our sample one firm at a time. In all cases we obtain similar
estimates of the matching ATT of investment, at about −9 percentage points.
We next exclude from the treatment group 18 observations corresponding to
Japanese firms. We again obtain similar estimates of the ATT of investment,
at −14.5 percentage points and statistically significant.
We also exclude from the treatment group firms with a rating above the
sovereign, as these firms may be systematically different from firms with
a rating equal to the sovereign. Table IA.XII in the Internet Appendix re-
ports difference-in-differences estimates for investment using this alternative
treatment group. Not surprisingly, the effect is even stronger than in Table IV.
The differential change in investment rate produced by the matching estimator
of the ATT is −10.7 percentage points and is statistically significant.
In our next robustness test we also compare the effects for firms that actually
get downgraded following a sovereign downgrade with those that do not; 36
treated firms out of 73 are actually downgraded in the year of the sovereign
downgrade. We find that the reduction in investment is driven by the firms
that are actually downgraded. The average change in investment rate for firms
that are downgraded is −19.3 percentage points, while for firms that are not
downgraded it is 1.4 percentage points. This result provides additional evidence
that the reduction in investment is in fact associated with rating downgrades.
We also address the concern that pretreatment average Q is higher for firms
in the treatment versus the control group. We construct a control group in which
we use only Q and firm size (and categorical variables) as matching covariates.
Under this alternative, the differential effect on investment is similar at −11.1
percentage points and is again statistically significant.
We further check whether treatment and control groups differ in terms of
their exposure to foreign currency debt and bank debt using Capital IQ data.
Figure IA.2 in the Internet Appendix shows the evolution of foreign currency
debt (as a percentage of total debt) in the three quarters before and after the
sovereign downgrade. We find that the treatment and control groups have
similar exposure to foreign currency debt before the downgrade. Following
the sovereign downgrade, firms seem to increase their exposure to foreign
currency debt, but the increase is more pronounced for the control group. We
also examine the evolution of bank debt (as a percentage of total debt) before
and after the sovereign downgrade. We find that the debt structure of the
treatment and control groups contains similar levels of bank debt at 12% and
10% of assets, respectively. Thus, there is no indication that treated firms are
more affected because they rely more on bank debt and lending relationships
(and consequently less on public debt markets) than control firms.
Another concern that we address is that the relative decrease in the supply
of capital to treated firms may be due to a financial repression story. Becker
and Ivashina (2014b) show that, during the recent eurozone sovereign debt cri-
sis, governments use the domestic financial sector (e.g., banks, pension funds,
insurance companies) to absorb government debt, which monopolizes investor
demand for highly rated corporate securities. Such financial repression may
affect our treated firms more than control firms because government debt is a
closer substitute to debt securities issued by firms with high ratings. To exam-
ine whether financial repression may explain our results, we limit the control
group to firms within three notches of the sovereign rating. If our differential
effects are due to financial repression, then they should disappear or at least be
reduced in these alternative control groups in which corporate debt is a closer
substitute to government debt. We find that the results are similar when we
consider this alternative control group: the matching ATT in the investment
tests is about −10.3 percentage points and statistically significant. Thus, we
do not find support for a financial repression explanation for our results.
IV. Effect on Cost of Debt

The evidence indicates that the ratings of bound firms are more affected
by sovereign downgrades than the ratings of nonbound firms. While rating
changes resulting from the sovereign ceiling downgrade do not reveal any new
information, the cost of debt may be affected due to rating-based regulatory
frictions. An increase in the cost of debt would be consistent with a contraction
in debt capital supply and a negative impact on firm investment and financial
policy following a sovereign downgrade.
To evaluate the impact on the cost of debt, we rely on corporate bond yields.
We collect data on end-of-month yield to maturity of bond issues for treated
and matched control firms from 1990 to 2013 from Bloomberg. We use yields for
local currency bond issues because they are available for the majority of firms
in the treatment and control groups, while yields for U.S. dollar-denominated
bonds are limited to a small fraction of our sample. We drop floating-rate
notes, insured bonds, and bonds with option-like features (i.e., convertible,
callable, and puttable bonds). We also require that bond issues have at least
one observation in the predowngrade period and one in the postdowngrade
period. The final sample of bond yields includes 20 treated firms (342 issues)
and 11 control firms (134 issues).
We estimate regressions to examine whether bond yields increase for bound
firms versus nonbound firms following a sovereign downgrade. The depen-
dent variable is the change in yield around the sovereign downgrade, that is,
the yield on firm i’s bond j measured t months after each sovereign down-
grade minus the yield on firm i’s bond j measured s months prior to the event
(Y ieldi, j,t−s ). Because we perform these tests at the bond issue level, we focus
on the change in yield around sovereign downgrades to control for issue-specific
effects. We perform event studies with different values of t around the time of
the sovereign downgrade to capture the response of bond markets and to ac-
count for the possibility that rating changes can be anticipated.
The coefficient on Bound captures the differential effect on the yield of bound
firms relative to nonbound firms, as the dependent variable is the change in
yield around the sovereign downgrade. We control for coupon rate, issue amount
(in U.S. dollars), maturity, and ratio of issue amount to total amount issued
(outstanding for each firm). The regressions also include country-event fixed
effects, which correspond to estimating the differential impact of the sovereign
downgrade on the bond yield of bound firms versus nonbound firms for the
same country-event. Standard errors are clustered by country-event.22
Table X reports the results for event windows starting three months prior
to a sovereign rating change. When looking at the yield three months after
the event, the average yield for bound firms increases by 34 basis points more
than for nonbound firms. As the event window widens to six months after a
sovereign downgrade, the differential effect increases to 61 basis points.
22 We include country-event fixed effects because the sample of yields is unbalanced with a
different number of treated and control observations for each event. We are not able to include
industry-event fixed effects due to the small sample size.
Table X
Difference-in-Differences in Corporate Bond Yields around a
Sovereign Downgrade
This table presents difference-in-differences matching estimators for corporate bond yield around
a sovereign downgrade. The dependent variable is the change in corporate bond yield around a
sovereign downgrade for different event windows (in months). Bound is a dummy variable that
takes a value of one if a firm has a credit rating equal to or above the sovereign rating before
a sovereign downgrade. Bond issue controls (coefficients not shown) include coupon rate, issue
amount (in U.S. dollars), maturity, and ratio of issue amount to total amount issued. Treated
firms have a credit rating equal to or above the sovereign rating in the year before a sovereign
downgrade. Control firms are matched firms using the Abadie and Imbens matching estimator. The
covariates are country, year, size, investment, Tobin’s Q, cash flow, cash, leverage, foreign sales,
government ownership, and exposure to government spending (pretreatment values). The sample
consists of treated and control firms for which yields on local currency bond issues are available.
A bond issue is required to have at least one observation in the predowngrade period and one
in the postdowngrade period. Robust standard errors clustered by country-event are reported in
parentheses. ***,**, and * indicate significance at the 1%, 5%, and 10% level, respectively.
(−3,+1) (−3,+3) (−3,+4) (−3,+5) (−3,+6)
Bound −0.067 0.340** 0.385** 0.521*** 0.608***

(0.093) (0.156) (0.157) (0.149) (0.094)
Country-Event FE Yes Yes Yes Yes Yes
Controls Yes Yes Yes Yes Yes
Observations 533 520 523 515 445
R2 0.714 0.527 0.537 0.550 0.544
Figure 6 shows the monthly evolution of bond yields for bound and nonbound
firms in the 10 months before and after the sovereign downgrade. While the
yields of nonbound firms stay fairly constant over time, the yields of bound
firms increase steadily over time after the sovereign downgrade.23
We conduct a placebo test using periods of large increases in sovereign bond
yields without sovereign downgrades. This test addresses the concern that
treated firms may have greater exposure to a deterioration in government fiscal
position as they are more likely to receive government support. We identify
placebo events by taking the month with the maximum increase in sovereign
bond yields in a given country and year. For each country, we exclude periods
in which the country is downgraded in the six months before or after the
month of the event. We estimate the same model as in Table X. Table IA.XIII
in the Internet Appendix reports the results. We find that the coefficients
are statistically insignificant in this falsification test, which indicates that
differences between bound and nonbound firms do not appear during periods
of deterioration in government fiscal position without a sovereign downgrade.
23 It takes approximately two months after a sovereign downgrade for the bond yields of bound
firms to increase. If the effect of ratings is indeed induced by ratings-based regulatory constraints,
it takes time for investors to adjust their holdings of bond securities. Kisgen and Strahan (2010)
find that the regulatory effect on the cost of debt is important, but that it takes a few months to be
observed in a sample of U.S. firms.
Figure 6. Corporate bond yields around a sovereign downgrade. This figure shows es-
timates of linear regressions of corporate bond yield of treatment and control groups around a
sovereign downgrade (between month −1 and month 0). The dependent variable is the corporate
bond yield. The explanatory variables are event-time dummies (month relative to the sovereign
downgrade) for the treatment and control groups, and bond-event fixed effects (a dummy variable
for each bond-sovereign downgrade pair). The coefficients can be interpreted as the change in
bond yields around sovereign downgrades. Treated firms have a credit rating equal to or above
the sovereign rating in the year before a sovereign downgrade. Control firms are matched firms
using the Abadie and Imbens matching estimator. The covariates are country, year, size, invest-
ment, Tobin’s Q, cash flow, cash, leverage, foreign sales, government ownership, and exposure to
government spending (pretreatment values). The sample consists of treated and control firms for
which yields on local currency bond issues are available. A bond issue is required to have at least
one observation in the predowngrade period and one in the postdowngrade period.
V. Mechanism: Why Do Rating Downgrades Matter?

In this section, we provide evidence that rating-based regulatory frictions
help explain the effects of downgrades on investment and financial policy. Un-
der Basel II, ratings affect the capital requirements applied to banks and in-
surance companies when they have claims on specific sovereigns or firms. The
rating bins on sovereign claims and their corresponding risk weights are as
follows: AAA to AA− (0%), A+ to A− (20%), BBB+ to BBB− (50%), BB+ to
B− (100%), and below B− (150%). We split the sample into two subsamples:
(1) when the sovereign downgrade implies a change in the rating bin, and (2)
when the sovereign downgrade does not imply a change in the rating bin. Our
prediction is that a sovereign downgrade will have stronger effects when the
sovereign changes rating bin as a result of the downgrade. In this case, the
sovereign downgrade should have a particularly large effect on the supply of
capital to bound firms, if they are downgraded together with the sovereign.24
24 Consistent with our previous analysis, we use sovereign downgrades and sovereign rating
bins in this test. We obtain similar estimates of the ATT of investment, at −20.4 percentage points
and statistically significant, when we use actual corporate downgrades and rating bins.
Table XI
Difference-in-Differences in Investment around a Sovereign
Downgrade: Rating-Based Regulation
property, plant, and equipment. Panel A presents results for a subsample of firms located in coun-
tries whose sovereign rating migrates to a new rating bin, defined by Basel II capital requirement
rules, as a consequence of a downgrade. Panel B presents results for a subsample of firms located in
countries whose sovereign rating does not migrate to a new rating bin as a consequence of a down-
grade. Treated firms have a credit rating equal to or above the sovereign rating in the year before
a sovereign downgrade. Control firms are matched firms using the Abadie and Imbens matching
estimator. The covariates are country, year, size, investment, Tobin’s Q, cash flow, cash, leverage,
foreign sales, government ownership, and exposure to government spending (pretreatment val-
ues). The sample consists of 23 treated and control observations in Panel A, and 50 treated and
control observations in Panel B. Robust standard errors clustered by country-event are reported
in parentheses. ***,**, and * indicate significance at the 1%, 5%, and 10% level, respectively.
Panel A: Sovereign Downgrade to a New Rating Bin
Treated Firms 33.35 19.29*** −14.06

(19.54) (4.11) (20.48)
Control Firms 29.14* 25.66** −3.47
(15.01) (9.78) (18.96)
Difference 4.21 −6.37
(5.71) (7.52)
(8.80)
Matching Estimator −25.29
(21.22)
Panel B: Sovereign Downgrade within the Same Rating Bin
Treated Firms 22.38*** 16.57*** −5.81

(4.82) (1.66) (4.85)
Control Firms 15.67*** 13.15*** −2.52
(1.64) (1.46) (2.49)
(4.27) (2.08)
(3.30)
(1.41)
Table XI reports the estimates. We rerun the analysis for investment in Ta-
ble IV separately for the two subsamples described above. Panel A of Table XI
reports the results when there is a change in rating bin, and Panel B reports
the results when there is no change in rating bin. The results indicate that the
differential reduction in investment of treated firms relative to control firms
after a sovereign downgrade is driven mostly by those observations in which
there is a change in rating bin. The differential reduction in investment is
much smaller when there is no change in rating bin. Panel A shows that the
difference-in-difference estimate is −10.6 percentage points, while in Panel B
the estimate is −3.3 percentage points. The ATT estimates are −25.3 percent-
age points in Panel A and 2.0 percentage points in Panel B. The estimates in
the subsample in which the downgrade implies a change in rating bin (Panel
A) are economically important but statistically insignificant, likely because of
the small size of the sample (23 treated firms). Figure IA.3 in the Internet
Appendix shows the evolution of investment in the two years before and after
the sovereign downgrade for the two subsamples. The differential effects that
we measure in Table XI occur only in the year of the sovereign downgrade.
We also perform this test for net debt issuances. Table IA.XIV and Figure
IA.4 in the Internet Appendix report the results. We find that the differential
reduction in net debt issuance of treated firms relative to control firms following
a sovereign downgrade is driven by the subsample in which the downgrade
implies a change in rating bin. The estimates are not statistically significant,
as in Table XI.
This evidence suggests that changes in capital requirements are one of the
mechanisms through which sovereign downgrades affect corporate debt mar-
kets and in turn firm investment and financial policy. This evidence is subject
to two important limitations. First, we split the sample according to the initial
sovereign rating to identify the regulation channel. Given that our original
sample is small, tests based on sample partitions generate large standard er-
rors and limit our ability to precisely identify the magnitude of effects that are
due to ratings. Second, there may be other mechanisms through which ratings
may affect the supply of capital to bound firms. A likely channel is the effect of
ratings on contracts such as debt and supply contracts. We attempt to analyze
instances of debt covenant violations around sovereign downgrades by collect-
ing and reading firms’ annual reports (including 10-K forms for U.S. companies
and 20-F forms for international firms cross-listed on U.S. exchanges). While
we find evidence that debt covenants do in fact depend on ratings, we are un-
able to find direct evidence that violations (as far as disclosed by firms) are due
to ratings triggers. As a result, it is difficult to measure the contract channel
empirically.
VI. Conclusion
We show that sovereign debt impairments can affect financial markets and
real economic activity through a credit ratings channel. In particular, we find
that firms with ratings at the sovereign bound reduce investment and reliance
on credit markets more than firms with ratings below the bound following a
sovereign downgrade. The bond yields of bound firms also increase significantly
more than the yields of firms whose rating is below the bound. This is consistent
with an increase in firms’ costs of borrowing and a reduction in the supply of
debt capital that is caused by the effect of sovereign downgrades on corporate
ratings, rather than a reduction in the demand for debt capital.
We develop a new strategy to identify the effects of sovereign debt impair-

ments and rating downgrades on firm investment and financial policy. Our
results show that sovereign downgrades have effects on firm policies that
are unlikely to be related to variation in unobservable firm characteristics or
macroeconomic conditions. Our results also uncover unintended consequences
for real economic activity of the sovereign ceiling policies that the rating agen-
cies typically follow. While rating agencies have been gradually moving away
from a policy of never rating a private borrower above the sovereign, our results
suggest that sovereign ceiling policies still apply. The ceiling rule pushes down
ratings and may be responsible for significant effects on firm investment and
financial policy in the aftermath of a sovereign downgrade.
The results also have implications for public debt management. They show
that sovereign downgrades matter for real economic activity, over and above the
deterioration in macroeconomic fundamentals. Governments should be aware
of the potential adverse effects of sovereign downgrades on the corporate sector,
and should factor these negative externalities into public debt management
decisions.
Initial submission: September 15, 2014; Accepted: December 20, 2015

Editors: Bruno Biais, Michael R. Roberts, and Kenneth J. Singleton
REFERENCES
Abadie, Alberto, David Drukker, Jane Herr, and Guido Imbens, 2004, Implementing matching
estimators for average treatment effects in Stata, Stata Journal 4, 290–311.
Abadie, Alberto, and Guido Imbens, 2011, Bias-corrected matching estimators for average treat-
ment effects, Journal of Business & Economic Statistics 29, 1–11.
Adelino, Manuel, and Miguel Ferreira, 2016, Bank ratings and lending supply: Evidence from
sovereign downgrades, Review of Financial Studies 29, 1709–1746.
Adrian, Tobias, Paolo Colla, and Hyun Shin, 2013, Which financial frictions? Parsing the evidence
from the financial crisis of 2007–9, Working paper, Federal Reserve Bank of New York.
Almeida, Heitor, Murillo Campello, Bruno Laranjeira, and Scott Weisbenner, 2012, Corporate debt
maturity and the real effects of the 2007 credit crisis, Critical Finance Review 1, 3–58.
Augustin, Patrick, Hamid Boustanifar, Johannes Breckenfelder, and Jan Schnitzler, 2014,
Sovereign credit risk and corporate borrowing costs, Working paper, Stockholm School of
Economics.
Baghai, Ramin, Henri Servaes, and Ane Tamayo, 2014, Have rating agencies become more conser-
vative? Implications for capital structure and debt pricing, Journal of Finance 69, 1961–2005.
Becker, Bo, and Victoria Ivashina, 2014a, Cyclicality of credit supply: Firm level evidence, Journal
of Monetary Economics 62, 76–93.
Becker, Bo, and Victoria Ivashina, 2014b, Financial repression in the European sovereign debt
crisis, Working paper, Harvard Business School.
Bedendo, Mascia, and Paolo Colla, 2015, Sovereign and corporate credit risk: Spillover effects in
the eurozone, Journal of Corporate Finance 33, 34–52.
Belo, Frederico, Vito Gala, and Jun Li, 2013, Government spending, political cycles, and the cross
section of stock returns, Journal of Financial Economics 107, 305–324.
Bertrand, Marianne, Esther Duflo, and Sendhil Mullainathan, 2004, How much should we trust
differences-in-differences estimates?, Quarterly Journal of Economics 119, 249–275.
Borensztein, Eduardo, Kevin Cowan, and Patricio Valenzuela, 2013, Sovereign ceilings lite? The
impact of sovereign ratings on corporate ratings, Journal of Banking & Finance 37, 4014–4024.
Brooks, Robert, Robert William Faff, David Hillier, and Joseph Hillier, 2004, The national market
impact of sovereign rating changes, Journal of Banking & Finance 28, 233–250.
Carvalho, Daniel, Miguel Ferreira, and Pedro Matos, 2015, Lending relationships and the effect of
bank distress: Evidence from the 2007-2008 financial crisis, Journal of Financial and Quan-
titative Analysis 50, 1165–1197.
Chava, Sudheer, and Amiyatosh Purnanandam, 2011, The effect of banking crisis on bank-
dependent borrowers, Journal of Financial Economics 99, 116–135.
Chernenko, Sergey, and Adi Sunderam, 2012, The real consequences of market segmentation,
Review of Financial Studies 25, 2041–2069.
Chodorow-Reich, Gabriel, 2014, The employment effects of credit market disruptions: Firm-level
evidence from the 2008-09 financial crisis, Quarterly Journal of Economics 129, 1–59.
Durbin, Erik, and David Ng, 2005, The sovereign ceiling and emerging market corporate bond
spreads, Journal of International Money and Finance 24, 631–649.
Gande, Amar, and David Parsley, 2005, News spillovers in the sovereign debt market, Journal of
Financial Economics 75, 691–734.
Graham, John, Mark Leary, and Michael Roberts, 2014, How does government borrowing affect
corporate financing and investment? Working paper, Duke University.
Harford, Jarrad, and Vahap Uysal, 2014, Bond market access and investment, Journal of Financial
Economics 112, 147–163.
Kaminsky, Graciela, and Sergio Schmukler, 2002, Emerging market instability: Do sovereign
ratings affect country risk and stock returns?, World Bank Economic Review 16, 171–195.
Kisgen, Darren, 2006, Credit ratings and capital structure, Journal of Finance 61, 1035–1072.
Kisgen, Darren, 2007, The influence of credit ratings on corporate capital structure decisions,
Journal of Applied Corporate Finance 19, 65–73.
Kisgen, Darren, 2009, Do firms target credit ratings or leverage levels?, Journal of Financial and
Quantitative Analysis 44, 1323–1344.
Kisgen, Darren, and Philip Strahan, 2010, Do regulations based on credit ratings affect a firm’s
cost of capital?, Review of Financial Studies 23, 4324–4347.
Lemmon, Michael, and Michael Roberts, 2010, The response of corporate financing and investment
to changes in the supply of credit, Journal of Financial and Quantitative Analysis 45, 555–587.
Manso, Gustavo, 2013, Feedback effects of credit ratings, Journal of Financial Economics 109,
535–548.
Manso, Gustavo, Bruno Strulovici, and Alexei Tchistyi, 2010, Performance-sensitive debt, Review
of Financial Studies 23, 1819–1854.
Reinhart, Carmen, and Kenneth Rogoff, 2009, This Time Is Different: Eight Centuries of Financial
Folly (Princeton University Press, Princeton, NJ).
Reisen, Helmut, and Julia von Maltzan, 1999, Boom and bust and sovereign ratings, International
Finance 2, 273–293.
Standard & Poor’s Rating Services, 2012, Corporate and government ratings that exceed the
sovereign rating, October.
Standard & Poor’s Rating Services, 2013, Ratings above the sovereign-corporate and government
ratings: Methodology and assumptions, November.
Sufi, Amir, 2009, The real effects of debt certification: Evidence from the introduction of bank loan
ratings, Review of Financial Studies 22, 1659–1691.
Tang, Tony, 2009, Information asymmetry and firms’ credit market access: Evidence from Moody’s
credit rating format refinement, Journal of Financial Economics 93, 325–351.
Supporting Information
Additional Supporting Information may be found in the online version of this
article at the publisher’s website:
Appendix S1: Internet Appendix.

12-The Real Effects of Credit Ratings - Cleaned

Uploaded by

Copyright:

Available Formats

12-The Real Effects of Credit Ratings - Cleaned

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

12-The Real Effects of Credit Ratings - Cleaned

Uploaded by

Copyright:

Available Formats

THE JOURNAL OF FINANCE • VOL. LXXII, NO.

The Real Effects of Credit Ratings: The Sovereign

SOVEREIGN DEBT IMPAIRMENTS have become a significant problem for developed

identification strategy exploits the variation in corporate ratings that is due

long-term contracts. Because of these effects, firms appear to react to rating

sample according to variables such as the likelihood of a change in capital requirements.

I. Methodology and Data

A. Sovereign Downgrades and Ceilings: Institutional Background

to represent a strong upper bound and an important determinant of ratings

Figure 2. Frequency of corporate downgrades around a sovereign downgrade by dis-

Figure 2 illustrates the key empirical regularity that we use to identify

11 We also include industry (Fama-French 12-industry classification) as a covariate in an alter-

native control group.

D. Sample and Variable Construction

Argentina 2001 BB− SD 4

of shares outstanding (using Factset ownership data), and Government Expo-

Gala, and Li (2013).

diversification, the high quality of our generation fleet, our resilient

The effect of the sovereign downgrade on EDP’s investment and financial

Although EDP signed a credit line of 2 billion euros in November 3, 2010,

Median Mean Kolmogorov-

Panel A: Matched Sample

Size 5.25 9.58 9.64 1.00 5.20 9.01 8.87 0.50

(0.11) (1.21) (0.86)

Median Mean Kolmogorov-

Panel B: Matched Sample with Industry Match (Fama-French 12-industry classification)

Size 5.25 9.79 9.75 0.91 5.20 9.26 9.08 0.90

(0.24) (2.97) (2.00)

After the matching procedure, there are no statistically significant differ-

II. Effect on Corporate Ratings

variables) as matching covariates. In this alternative sample, there is no difference in Q across

Year before Downgrade Year of Downgrade Difference

Panel A: Matched Sample

Treated Firms 16.11*** 14.75*** −1.36***

Panel B: Matched Sample with Industry Match (Fama-French 12-industry classification)

Treated Firms 16.95*** 16.13*** −0.83

III. Effect on Investment and Financial Policy

26.6% to 17.7% of capital, a reduction of 8.9 percentage points. For control

Year before Downgrade Year of Downgrade Difference

Panel A: Matched Sample

Treated Firms 26.62*** 17.70*** −8.92

Panel B: Matched Sample with Industry Match (Fama-French 12-industry classification)

Treated Firms 23.49** 16.15*** −7.33

ATT is equal to −8.9 percentage points, which corresponds to 33% of the

Panel A: Recession without a Sovereign Downgrade

explicit or implicit subsidies.20 Additionally, treated firms might be subject

Year before Downgrade Year of Downgrade Difference

Panel A: Corporate Yield

(1) (2) (3)

Bound 0.504 2.161 0.472

Panel B: Stock Return

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

Bound 0.002 0.000 0.004 0.005*

to or above the sovereign rating in year t − 1 (Bound), the 10-year constant-

effect of sovereign downgrades on the profitability of treated firms in the year

Year before Downgrade Year of Downgrade Difference

Panel A: Net Debt Issuance

Treated Firms 7.52* 2.41** −5.11

Panel B: Net Equity Issuance

Treated Firms 16.11* 14.75* −1.36***

Treated Firms 16.95* 16.13* −0.83

Treated Firms 26.62* 17.70* −8.92

Treated Firms 23.49 16.15* −7.33

Bound × Sovereign Downgrade −9.62 −10.18 −9.44* −11.42* −12.50

Bound −0.067 0.340 0.385 0.521* 0.608*

Treated Firms 22.38* 16.57* −5.81