Credit Ratings and the Cost of Debt: The Sovereign Ceiling Channel

Transcription

1 Credit Ratings and the Cost of Debt: The Sovereign Ceiling Channel Felipe Restrepo Carroll School of Management Boston College This Draft: October 30, 2013 Please do not distribute Abstract A sovereign s credit rating generally represents the highest attainable rating by most issuers domiciled within their respective country. In this paper I show that the sovereign ceiling represents a meaningful institutional friction, and that through this channel credit ratings have an important effect on borrowing costs in the private sector. Specifically, I estimate the differential effect of contractions and relaxations in the sovereign ceiling on the bond spreads of firms that are exactly at the sovereign bound, relative to other firms that are near but not at the bound. I find that following a sovereign downgrade, the spreads of bound firms increase significantly more relative to non-bound firms. I also show that firms that are bound tend to be rated more unfavorably and their default rates tend to be lower relative to non-bound firms. Keywords: Sovereign Ceiling; Corporate Ratings; Cost of Debt; Event Study, Fuzzy Regression Discontinuity Design. JEL Classifications: G15, G23, G32. I would like to especially thank Phil Strahan for his advice and support. I would also like to thank Clifford Holderness, Darren Kisgen, Jun Qian (QJ), Jérôme Taillard, and seminar participants at Boston College for helpful comments and suggestions. 1

2 1 Introduction Credit rating agencies (CRAs) play a central role in providing information about the ability and willingness of issuers, including governments and firms, to repay their debts. A crucial link between sovereign and corporate credit ratings is the sovereign ceiling, a policy still strongly implemented by CRAs whereby a government s foreign-currency rating generally represent the highest attainable rating by corporate issuers within that country. 1 CFO Magazine (2013) summarized the key implication of the sovereign ceiling as follows: If a company is a better credit risk than its home country, it might still have trouble getting a credit rating agency to recognize that fact. In this paper I show that the sovereign ceiling represents a meaningful institutional friction, and that through this channel credit ratings have an important effect on borrowing costs in the private sector. Specifically, I estimate the differential effect of contractions and relaxations in the sovereign ceiling on the bond spreads of firms that are exactly at the sovereign bound, relative to other firms that are near but not at the bound. The main hypothesis I test in this study is whether the sovereign ceiling policy by CRAs represents a meaningful friction that affects the cost of debt of firms issuing USDdenominated bonds in international markets. I evaluate two main implications of this hypothesis. First, if sovereign ratings represent a meaningful constraint for firms that are exactly bound by the sovereign ceiling, then being bound by the ceiling should be associated with credit ratings that are more pessimistic relative to firms that are not bound. Second, if the sovereign ceiling is an important channel through which credit ratings affect a firm s cost of debt, then contractions and relaxations in the ceiling (i.e. sovereign downgrades and upgrades respectively) should lead to more pronounced changes in the yield spreads of those firms that are exactly at the sovereign bound, relative to non-bound firms. The main chal- 1 Rating agencies assign different types of ratings depending on the maturity (short term or long term) and currency denomination (local currency or foreign currency) of an issuance. The focus of this study is on foreign-currency long-term ratings, where CRAs use a sovereign s rating as a strong upper bound on the ratings of bonds issued by firms domiciled within each country. 2

3 lenge when tracing the effect of the sovereign ceiling on corporate outcomes is the inherent endogeneity between a sovereign s credit quality and the creditworthiness of firms in that country. I explicitly address this concern in my empirical design by focusing on the differential effect stemming from contractions and relaxations in the sovereign ceiling on firms that are at the sovereign bound, relative to other firms in the same country that are near but not at the bound. To evaluate these testable implications of the sovereign ceiling channel hypothesis, I exploit two important empirical regularities associated with the sovereign ceiling. First, I document that the distribution of corporate ratings across sovereign rating levels is systematically concentrated exactly at each country s sovereign rating. As Borensztein et al. (2013) point out, this implies that even though CRAs have moved away from fully enforcing the sovereign ceiling over the last two decades, sovereign ratings still represent a strong upper bound for the ratings of corporate borrowers. Second, I show that within the month of a sovereign rating change, the probability of a corporate issuer obtaining a rating adjustment in the same direction and magnitude as its corresponding sovereign jumps exactly at the sovereign rating bound. Specifically, conditional on the event of a sovereign rating change, firms that are at the bound have a probability of approximately 60% of obtaining the same rating adjustment as the sovereign within a month, compared to less than 10% for firms that are within three notches of the sovereign rating. This jump in the probability of obtaining a corporate rating change allows me to use a firm s bound status as an instrument to estimate the effect of a rating adjustment directly related to the sovereign ceiling channel. Since sovereign rating changes provide no additional firm-level private information, any differential effect between bound and non-bound firms should result from a jump in the probability of a rating change for those firms that are at the sovereign ceiling bound. My identification strategy assumes that changes in fundamentals are the same for treatment (firms bound by the ceiling) and control firms (firms near but not bound by the ceiling). I focus on a large sample of bond and firm level data between 1999 and 2012 for 51 countries that had at least 3

4 one sovereign rating change between this period. The sample includes not only developing countries (29) but also developed economies (22). I first find that the sovereign ceiling policy is associated with ratings that tend to be more pessimistic for firms that are bound by the sovereign ceiling, relative to firms not bound by it. Specifically, I use as a benchmark financial statements data on firms in AAA countries (where the sovereign ceiling does not represent a constraint) to estimate the predicted ratings of firms in non-aaa countries, and find that bound firms tend to be rated more pessimistically relative to non-bound firms. Similarly, I find that the probability of transitioning into default during a 5-year window is lower for firms that are bound compared to non-bound firms, conditional on their rating. I also show that investors, to some extent, recognize that bound firms are on average of better credit quality when compared to non-bound firms, and thus their yield spreads tend to be lower for a given corporate rating. Furthermore, even though the market appears to partially incorporate the fact that bound firms tend to be more pessimistically rated, my main results indicate that the sovereign ceiling still has a sizable impact on the cost of debt of firms bound by it. I find that in the month following a sovereign rating downgrade, the spread of bound firms increases by approximately 54 bp more relative to other firms in the same country that are near but below the sovereign ceiling. This differential effect is even more pronounced as the pre and post event window widens, while remaining statistically significant. Although sovereign upgrades tend to be associated with a decrease in the spreads of bound relative to nonbound firms, the magnitude and statistical significance of their effect is lower. Finally, I also find evidence suggesting that through the sovereign ceiling channel firms investment policies are affected following a sovereign rating downgrade. In particular, in the year following a sovereign downgrade, capital expenditures scaled by total assets decline by 3.2% more for bound firms relative to non-bound firms. I perform a falsification test of these results by examining whether the effect on both corporate spreads and investment are also present one year before each actual sovereign rating change event in my sample, and find no evidence 4

5 supporting the alternative explanation that pre-event trends are driving the results. Overall, the results from this paper are consistent with the hypothesis that the sovereign ceiling policy is a meaningful institutional friction for firms that issue USD-denominated debt in international markets, and that through this channel sovereign ratings have an important effect on borrowing costs in the private sector. The findings of this paper also indicate that since corporate debt costs can increase when public finances deteriorate, policies that affect a sovereign s creditworthiness have a potentially important impact on the cost of external financing in the private sector through the sovereign rating channel. 2 Literature Review and Relative Contribution Previous research documents that credit ratings affect a firm s cost of capital (e.g. Kliger and Sarig (2000); Jorion et al. (2005); Kisgen and Strahan (2010)), and that sovereign ratings are an important determinant of corporate ratings (e.g. Borensztein et al. (2013); Williams et al. (2012)). I extend the existing literature by identifying an unexplored dimension, the sovereign ceiling channel, through which credit ratings affect a firm s cost of debt. Furthermore, I contribute to prior research examining the link between sovereign and corporate credit risk by implementing an identification strategy that allows me to estimate the impact of the sovereign ceiling on corporate bond spreads, while explicitly controlling for the endogenous component of the country-level information contained in sovereign rating changes. 2.1 The Sovereign Ceiling and Credit Ratings This paper is related to prior research that examines the link between sovereign and corporate ratings. Borensztein et al. (2013) show that, even though CRAs have moved away from a policy of never rating a firm above the sovereign ceiling, sovereign ratings still have a 5

6 significant effect on corporate ratings after controlling for country specific and firm-level characteristics. Williams et al. (2012) analyze the effect of sovereign rating changes on banks ratings in a sample emerging markets countries between 1999 and 2009, and find that sovereign rating upgrades (downgrades) are strongly associated with bank rating upgrades (downgrades). The evidence from these studies indicates that the sovereign ceiling still represents a strong bound for the credit ratings of firms that issue USD-denominated bonds in international markets. However, these studies stop short of tracing any causal link between the sovereign ceiling policy and a firm s cost of debt or corporate policies, which is the main goal of my study. Relative to these Researchers have also examined the relationship between sovereign and corporate credit risk using bond spreads data. Durbin and Ng (2005) compare the yield spreads of foreign currency denominated corporate bonds to those of bonds issued by their respective home government. Although Durbin and Ng are limited by a small sample size, they show that 11 corporate bonds out of 108 in their sample have yields that are below their comparable government spreads. They find that these bonds tend to be issued by firms that are export driven or that have a close relationship with either a foreign firm or the home government. Similarly, Lee et al. (2013) use credit default swaps (CDS) data on a larges sample of 2,364 companies in 54 countries to examine the characteristics under which firms obtain lower CDS spreads relative to their respective sovereigns. They find that firms exposed to better property rights institutions and firms listed on stock exchanges with stricter disclosure requirements are more frequently able to obtain CDS spreads below their sovereign counterparts. These studies show that firms that manage to pierce the sovereign ceiling are typically export driven, are exposed to better property rights institutions through their foreign assets positions, and have their stocks listed on exchanges with stricter disclosure requirements. 2 However, these papers do not examine whether the sovereign ceiling still 2 The findings from these papers are overall consistent with the key features of the relaxation of the sovereign ceiling by CRAs. For instance, Fitch (2004) explained that:... exceptional banks and corporates can be rated above the sovereign ceiling if their stand-alone credit fundamentals imply that they are more creditworthy than the sovereign and they are shielded from the risk of exchange controls by substantial 6

7 affects corporate spreads. That is, even if a corporate that is constrained by the sovereign rating ceiling has a lower spread than its corresponding sovereign, the fact that the firm is bound by the ceiling could still be hindering that issuer from obtaining an even lower cost of debt. I explicitly test this hypothesis in my main empirical analysis. 2.2 Credit Ratings, Capital Structure and the Cost of Capital Existing literature also provides strong evidence that corporate credit ratings affect firms financial policies. Kisgen (2006) finds that firms near a rating change issue less debt relative to equity than firms not near a rating adjustment. Kisgen (2009) shows that following a rating downgrade, a firm is more likely to reduce leverage. Sufi (2009) shows that the introduction of syndicated bank loan ratings in 1995 by S&P and Moody s resulted in an increase in the use of debt by firms, and also in an increase in cash acquisitions and investment in working capital. This paper also relates to papers examining how ratings affect a firm s cost of capital. Kliger and Sarig (2000) use the refinement of Moody s ratings in April, 1986, and find that corporate rating changes contain additional information that help explain bond spreads. By focusing on this specific event, Kliger and Sarig (2000) are able to examine the effect of rating changes that exclusively reflect rating information and not fundamental changes in the issuer s risk. Tang (2009) also exploit Moody s 1982 rating refinement to identify the impact of information asymmetry on real outcomes: firms with refinement upgrades experience an additional decrease in their cost of debt compared with firms with downgrades. Tang also show that this lower cost of debt ultimately leads firms to issue more debt, to invest more and to hold less cash. Jorion et al. (2005) find that following the implementation of Regulation Fair Disclosure (Reg FD) in October, 2000, rating downgrades are related to larger declines in stock prices relative to the pre-reg FD period. They also find that rating upgrades, foreign exchange revenues, off-shore assets or more highly rated foreign partners/parents willing to provide financial support. 7

8 which are found to be generally insignificant in previous studies, are associated with greater increases in stock prices. Finally, recent literature also shows that credit ratings are important because securities rules, regulations and institutional investors investment policies often depend on them. Kisgen and Strahan (2010) examine the introduction in the U.S. of DBRS, a fourth credit rating agency, and show that investments rules and regulations related to bond ratings also have an impact on corporate borrowing costs. Bongaerts et al. (2012) find that additional credit ratings matter mainly for regulatory purposes, and they do not appear to add significant additional information related to credit quality. 3 The Sovereign Ceiling and Corporate Ratings: 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 CRAs -Standard and Poor s, Moody s and Fitch- assign different types or ratings depending on the maturity (short term or long term) and currency denomination of an issuance (foreign currency or local currency). This study focuses on the foreign-currency, long-term borrowing space, where CRAs use a sovereign s rating as a strong upper bound on the credit ratings of firms that operate within each country. Even though the sovereign ceiling has typically represented a more important constraint for firms in developing countries where sovereign ratings tend to be lower, the relationship between the credit risk of a sovereign and private sector borrowers has received increased attention following the recent European sovereign debt crisis, where several developed countries including Greece, Italy, Ireland, France, Portugal and Spain experienced sovereign rating downgrades. Until 1997, rating agencies strictly followed the policy of not granting a private company a 8

9 rating higher than the sovereign rating. In April of that year, S&P first relaxed its sovereign ceiling rule in three dollarized economies: Argentina, Panama, and Uruguay. 3 Fitch and Moody s followed suit in 1998 and 2001 respectively. Although rating agencies have moved away from strictly enforcing the sovereign ceiling over the last two decades, corporate ratings that pierce the ceiling are still not common. For instance, in 2012, S&P reported, that only 114 corporate and local government FC LT ratings in 20 countries exceeded the ratings on their corresponding sovereign (S&P s Rating Services, 2012). 4 The limited number of firms above the sovereign ceiling coupled with the fact that prior research shows that firms that pierce the ceiling are systematically different than firms at or below the ceiling, are the central reasons I do not use these firms as part of the control group in my empirical analysis. The key credit rating implication of the sovereign ceiling is shown in figure 1, where I plot the distribution of corporate credit ratings by each sovereign rating level. This figure show that although there are corporate issues that manage to pierce the ceiling, almost systematically the largest concentration of ratings is located exactly at the sovereign rating. Evidence from figure 1 also suggests that, as expected, the sovereign ceiling policy represents less of a rating constraint for firms in countries with the highest sovereign ratings (i.e. there is a lower concentration of corporates at the ceiling in countries with AA-, AA, and AA+ ratings). Why do CRAs use sovereign rating as a strong upper bound when rating corporate issues? The central argument by rating agencies is that a sovereign default may result in the government imposing exchange controls and other restrictive measures that limit a firm s access to the foreign currency necessary to service their financial obligations. Moody s explains that most non-structured locally-domiciled issuers are rated at or below the level of the sovereign because they operate in the same economic and financial environment and are therefore vulnerable to the same broad credit pressures as the sovereign (Moody s Investor 3 For instance, in the case of Argentina S&P increased the credit rating of 14 firms above the sovereign rating of BB. 4 The focus of this study is solely on corporate ratings and not on local or state governments issues. 9

10 Service (2012)). Similar arguments are also made by S&P and Fitch (S&P s Rating Services, 2012: FitchRatings, 2012). However, the extent to which the sovereign ceiling policy is implemented, although similar, is not identical across CRAs. S&P is the least likely to assign a corporate rating above the sovereign, followed by Fitch and Moody s respectively. As depicted in figure 2, the percentage of firms with a rating at or below the ceiling in countries with a sovereign rating below AAA is 90.7% for S&P, 79.3% for Moody s, and 85.2% for Fitch. 5 There are two key factors CRAs use when rating foreign-currency corporate issues: 1) the issuer s inherent likelihood of repayment (which is the same as local ratings), and 2) the issuer profile after taking into account the risk of exchange controls being imposed by the government that would hinder the ability of non-sovereign issuers to convert local currency into foreign currency to meet their financial obligations. Thus, firms that pierce the ceiling are particularly strong corporates whose exposure to the risk of not been able to meet their foreign currency obligations in the case of a sovereign default is clearly very limited. Firms with foreign assets, high export earnings and foreign parents tend to have a higher probability of being rated above their corresponding sovereign. 6 In general, CRAs only grant an issuer a rating above the sovereign if it is able to demonstrate a strong resilience and low default dependence with the sovereign, as well as a degree of insulation from the domestic economic and financial disruptions that are typically associated with sovereign defaults. Note however that there is not a clear reason why the creditworthiness of firms that are exactly at the bound should be affected more by a change in a sovereign s credit quality than other firms also near the sovereign ceiling but not exactly at it, which is a key component of my identification strategy. 5 Moody s for instance explains that when rating corporate debt in foreign currency, they are willing to assign a rating that is up two to notches above the sovereign (Moody s Investor Service, 2012). 6 The sovereign ceiling can also be pierced if bonds are offered with sufficient collateral. For example, The Economist (2006) reported that in 2005 the Emirates National Securitization Corporation (ENSeC) issued notes for $350m, using as collateral property leases in The Palm, a property development on islands shaped like a palm tree, as well cash collateral. The notes received a rare AAA rating by both S&P and Moody s. 10

11 4 Data and Summary Statistics Since the goal of this paper is to examine how, through the sovereign ceiling channel, credit ratings affect the cost of debt of firms, I start my sample construction by collecting data on long-term foreign-currency sovereign ratings directly from S&P, Moody s and Fitch. Similarly, I obtain from Bloomberg corporate ratings for USD-denominated bonds by non-us firms, and I match each issue with its corresponding sovereign data. I obtain a bond s S&P, Moody s and Fitch ratings, as well as its end of the month yield where available. I also collect issue specific information (issuance and maturity dates, amount issued, coupon payment and frequency and collateral type), firm-level information (e.g. issuer ticker, industry classification) and financial statements data where available. I convert both sovereign and corporate ratings to a numerical scale ranging from 0 to 21, where 21 represents a AAA rating (see table A.1 in the Appendix for the numerical conversion). Since bond pricing data is available from Bloomberg starting in 1999, I construct my matched sample of sovereign ratings, corporate ratings and corporate bond yields from 1999 to The fact that I use USD denominated bonds implies that spreads above US treasury yields represent default risk, rather than currency risk (Domowitz et al. (1998); Durbin and Ng (2005)). Thus, I calculate corporate yield spreads by subtracting the equivalent maturity US Treasury yield for each issue. 7 I eliminate a small number of observations with negative spreads, I require that yields for consecutive months are not equal, and I winsorize at the 1% level to reduce the influence of outliers. Since my empirical strategy exploits sovereign rating changes for identification, I exclude from my sample countries and firms where sovereign ratings for all three CRAs were unchanged between 1999 and Table 1, panel A shows the number of bonds (CUSIPs), firms and countries each year in my sample. Panel B in table 1 displays the composition of firms by industry, using the Dow Jones s Industry Classification Benchmark (ICB). The final matched sample consists 7 I obtain constant maturity U.S. treasury rates data from the Federal Reserve Economic Data (FRED) website: 11

12 of rating history data for 51 countries, 566 firms and 1,935 distinct issues. Table 2 provides corporate yield spreads summary statistics by rating category, and shows that conditional on a firm s corporate rating, yield spreads tend to be on average 90 bp lower if the firm is bound by the sovereign ceiling. Table 2 also shows that this difference is generally more pronounced for lower ratings (e.g. the bound vs. below bound difference is -1.7% for B+ firms) and less important for higher ratings (e.g. for A+ firms the difference is +0.2%, although it is statistically no different to zero). 8 Table 4 summarizes the coverage of the data as well as the number of sovereign rating changes by country. In table 3 I report several bond and firm level characteristics for bound and non-bound firms, and compare whether their averages are statistically different from each other. I find that bonds issued by bound firms do not have a statistically different maturity when issued, although they do appear to be issued with a higher face value (the average issue for a bound firm has a face value of 436million, vs.385 million for non-bound firms). In terms of firm-level characteristics, bound firms on average have a leverage ratio of 60%, which is slightly higher than the 56% ratio for non-bound firms. A firm s EBITDA over Assets, Capex over Assets, and Total Debt Issuance over Assets are not statistically different from each other depending on a firm s bound status. 5 Methodology and Results The fundamental hypothesis of this paper is that the sovereign ceiling represents a meaningful institutional friction that affects the cost of debt of firms that issue USD-denominated bonds in international markets. I evaluate two main empirical implications of this hypothesis. First, if sovereign ratings do in fact represent a meaningful constraint for firms that are at the sovereign ceiling, then being bound should be systematically associated to credit ratings that are more pessimistic relative to firms that are not bound. Second if the sovereign 8 I perform a more formal test of the bound vs. below bound effect on bond spreads in the following section to also account, for instance, for time variation in spreads. 12

13 ceiling is an important channel through which credit ratings affect firms cost of debt, then contractions and relaxations in the ceiling should lead to more pronounced changes in the yield spreads of those firms that are at the sovereign bound, relative to non-bound firms. The main challenge when tracing the effect of sovereign ceiling contractions or relaxations on corporate outcomes is the inherent endogeneity between a sovereign s credit quality and the creditworthiness of firms in that country. I explicitly address this concern in my empirical strategy by examining the differential effect stemming from sovereign rating changes on firms that are bound by the sovereign ceiling, relative to other firms in the same country that are near but not bound by it. I do this by exploiting two important empirical regularities associated with the sovereign ceiling. First, as discussed earlier and depicted in figure 1, the distribution of corporate ratings across sovereign rating levels is systematically concentrated exactly at each country s sovereign rating. This implies that even though CRAs have moved away from fully enforcing the sovereign ceiling over the last two decades, sovereign ratings still represent a meaningful upper bound for corporate borrowers issuing USD-denominate bonds in international markets. Second, as figure 3 shows, the probability of a corporate issuer obtaining a rating adjustment in the same direction and magnitude as its sovereign within the month of a sovereign rating change is also discontinuous exactly at the sovereign rating bound (where a firm s distance-from-sovereign, the difference between a firm s rating and its corresponding sovereign, is equal to zero). More precisely, the middle panel in figure 3 shows that conditional on the event of a sovereign rating change, firms that are at the bound have a probability of approximately 60.3% of obtaining the same rating adjustment as the sovereign within a month, compared to 9.9%, 5.0% and 2.5% for firms that are respectively one, two and three notches below the sovereign rating. The left and right panels in figure 3 also show that this disparity in the response of corporate of ratings is not observed either the month before or the month after the sovereign change. As a result, the key identifying assumption in my empirical strategy is that sovereign rating changes do not provide additional firm-specific information, and thus the differential 13

14 effect on corporate yields between bound and non-bound firms in the event of a contraction or relaxation of the sovereign ceiling should be stemming from an increased probability of obtaining a corporate rating change in the same direction as the sovereign for those firms that are exactly at the ceiling bound. Consequently, if the sovereign ceiling constrains firms from potentially obtaining higher ratings, and if this is not fully incorporated in market prices, then following an upgrade in a country s sovereign rating, corporate borrowing costs should decrease more for those firms that are bound by the sovereign ceiling, relative to firms in the same country that are not bound by the constraint. Conversely, if a contraction in the sovereign ceiling results in firms at the bound obtaining a lower rating, then yield spreads should increase more for bound firms relative to non-bound firms. 5.1 Does being bound by the sovereign rating lead to a pessimistic rating? I first test whether firms that are bound by the sovereign ceiling (i.e. that their rating is equal to the sovereign rating) have a more pessimistic rating relative to firms that are not bound by it. If the sovereign ceiling represents a meaningful friction and not just an unbiased and accurate assessment of a firm s creditworthiness, then this rating practice should be systematically associated with ratings that are more pessimistic for bound firms relative to other firms with the same actual ratings but that are not bound by the sovereign ceiling. Thus, I first examine whether the sovereign ceiling policy is consistent with CRAs providing an unbiased assessment on the creditworthiness of borrowers by comparing the differential effect of being bound on a firm s predicted rating, as well as on its probability of transitioning into default. 14

15 5.1.1 Rating Analysis: The effect of bound status on ratings I explore whether bound firms tend to be pessimistically rated using a two-step procedure. First, I use as a benchmark annual financial data on rated firms that issue USD-denominated debt in AAA countries (where this friction does not matter) to predict the corporate ratings of firms in non-aaa countries, where the sovereign ceiling rule potentially matters. 9 Using this sample of firms in countries with a AAA sovereign rating I estimate a regression using a set of explanatory variables used in previous studies predicting credit ratings (see Kisgen, 2006 and Horrigan, 1966 for a similar implementation). The dependent variable is a firm s credit rating, which takes a value of 1 for a rating of D, up to a value of 21 to AAA (see table A.1 in appendix for the numerical conversion). I estimate the following regression for firms in AAA countries: Rtg i,t =β 1 (NI/A i,t ) + β 2 (D/T otcap i,t ) + β 3 Ln(A i,t ) + β 4 Square(NI/A i,t ) + β 5 Square(D/T otcap i,t ) + β 6 Square[Ln(A i,t )] + T imef E t + SectorF E i + ɛ i,t (1) For firm i and year t. NI/A is net income over assets, D/T otcap is the ratio of debt to total capitalization and Ln(A) is the log of total assets. I include year fixed effects to control for time specific shocks common to all firms, as well as sector fixed effects. I estimate the model above for firms in AAA countries and then I use the estimated coefficients to calculate the predicted credit ratings for the sample of firms in non-aaa countries (which I denote as Rtg), where the sovereign rating ceiling potentially represent a meaningful institutional friction. 10 In the second step, after having calculated the predicted rating for the sample of non- AAA countries, I compare, for each actual corporate rating level, whether predicted rating are systematically higher for firms that are bound relative to other firms with the same actual rating but that are not bound by the sovereign ceiling. Thus, I estimate the following 9 I do not include data for U.S. firms, as the ratings for USD-denominated debt are in that case localcurrency, and not foreign-currency as they are elsewhere in my data. 10 Estimating the model in equation 1 for the sample of firms in AAA countries results in an adjusted R 2 of Table A.2 in the appendix shows the estimated coefficients obtained from this regression. 15

16 regression: Rtg i,t = β 0 + β 1 RtgF E i,t + β 2 (RtgF E i,t Bound i,t ) + Sov.RatF E i,t + T imef E t + ɛ i,t (2) Where β 2 is a vector of coefficients that captures the differential effect, for each rating level (denoted by Rtg.F E), of being bound by the sovereign ceiling on a firm s predicted rating. If firms that are bound are rated fairly relative to firms that are not at the ceiling bound, then predicted ratings should not systematically differ based on whether firms are below or at the sovereign bound. On the other hand, if firms that are bound tend to be pessimistically rated, then their predicted rating should be higher relative to other firms with the same actual corporate rating but that are not bound by the sovereign rating. ). I focus only on firms that have a corporate rating below the sovereign rating. As highlighted earlier, previous research and CRAs themselves indicate that firms that are above the ceiling have typically strong ties to more financially developed countries and other specific characteristics that make them fundamentally different than other firms that do not pierce the ceiling. Table 5 shows that when comparing the predicted rating for bound versus below bound firms by estimating equation 2, bound firms tend to have a predicted rating that is above the predicted rating of the latter. For example, a bound firm with a B+ rating has a predicted rating that is 1.1 notches higher than a firm that is also rated B+ but that is not bounded by the sovereign ceiling. The difference between the predicted ratings of bound vs. below bound firms is positive and statistically significant in 12 of the 14 actual rating levels evaluated. The difference between the predicted rating of bound versus below bound firms is only not positive in two of the highest rating levels (AA- and AA) where, as indicated before, the sovereign ceiling rule represents a less meaningful restriction I also perform this comparison for bound firms relative to above-bound issuers in unreported tests and find similar results. Note however that due to the limited number of observations for firms above the sovereign ceiling, the power of the tests is significantly smaller. 16

17 5.1.2 Default analysis: The effect of bound status on transitioning into default An alternative test to evaluate whether a firm s bound status leads to a systematically pessimistic rating is to examine whether bound firms are associated with a lower default rate than non-bound firms, for the same actual rating. Specifically, I asses the unbiasedness of credit ratings in predicting that a firms transitions into default, depending on whether the firm is bound by the sovereign rating or not. The power of any default based test is constrained by the fact that actual defaults do not occur frequently. Out of the 566 firms in my full sample of non-aaa countries, only 15 had at any time a default rating ( D ). Thus, to perform this particular test I extend this sample to include firms with a rating of CCC+ and below, which more precisely proxy for being close to default. These firms are characterized by S&P as having significant speculative characteristics, currently vulnerable. The number of close-to-default firms in the sample is 64. I estimate a logit regression where the dependent variable is a dummy variable that indicates whether a firm had a rating of CCC+ or below during the last five years. I examine if a firm s bound status affects its probability of being close to default, after controlling for its credit rating by estimating the following model: Close-to-default i,j,[t,t+t ] =β 0 + β 1 RtgF E i,j,t + β 2 (RtgF E i,j,t Bound i,j,t ) + SovRatF E i,t + T imef E t + ɛ i,j,t (3) where the coefficients vector β 2 measures the interaction between each corporate rating and Bound, a dummy variable that takes a value of one for bounded firms and zero otherwise. β 2 captures, for each corporate rating, whether being bound by the sovereign rating is associated with a lower default. I include time (year) fixed effects to account for variations in default rates through the business cycle. Similarly, I include sovereign rating fixed to control for differences in default rates that vary dependent on the overall level of creditworthiness 17

18 of a sovereign. The results from estimating the default model in equation 3 are shown in tables 6 and 7, which respectively report the estimated coefficients and the marginal effects from the estimated logit model. Note that there are no estimated coefficients for firms with ratings above A, as there are no firms with these higher ratings that transition into close-to-default in my sample. Consistent with the finding from the previous subsection that bound firms tend to be rated more pessimistically, the predicted probabilities in table 7 indicate that bound firms tend to have a lower probability of transitioning into default, for a given rating, than non-bound firms. For instance, the probability of a non-bound firm with a B+ rating transitioning into close-to-default in a 5-year window is 7.1%. This is significantly higher than the 4.3% probability of a firm also with a B+ rating but bound by the sovereign ceiling of transitioning into close-to-default. As it was the case in the predicted ratings test, the difference between bound and non-bound firms is more pronounced for firms in lower rating levels Are bound firms perceived as more creditworthy by the market? The evidence from the previous two tests indicates that bound firms tend to be more unfavorably rated by CRAs and tend to have lower probabilities of transitioning into default than non-bound firms with the same rating. Thus, the next natural question is whether investors in the corporate bond market follow the ratings, or do they see past them and recognize bound firms relatively higher credit quality. To test whether the market actually recognizes bound firms potentially higher credit quality, I estimate the following regression: Spread i,j,t =β 0 + β 1 RtgF E i,j,t + β 2 (RtgF E i,j,t Bound i,j,t ) + Sov.RatF E i,t + T imef E t + ɛ i,j,t (4) where the dependent variable is a bond s yield spread, and the coefficients vector β 2 18

19 measures the interaction between each corporate rating and Bound, a dummy variable that takes a value of one for bound firms and zero otherwise. The coefficients vector β 2 captures, for each corporate rating, whether being bound by the sovereign rating or not results in differences in corporate spreads for a given rating level, after including sovereign rating fixed effects and time (month) fixed effects. Table 8 shows the estimation results from model 4. Overall, the average spread of bound firms, after being demeaned by sovereign rating and time effects, tends to be lower for each rating category when compared to non-bound firms. For instance, a B+ rated firm bound by the sovereign ceiling, the average, time-demeaned spread is 7.13%, 66 basis points higher than the average spread of 6.46% for a non-bound B+ rated firm. This suggest that investors do incorporate, to an extent, the better quality and lower probability of default of bound firms relative to non-bound firms given the same corporate rating. 5.2 The Effect of the Sovereign Ceiling Channel on Corporate Yields The results from the empirical tests in the previous subsection show that bound firms tend to be rated more pessimistically, have lower probabilities of transitioning into default, and are often priced at lower yield spreads by the market than non-bound firms. However, the fact that investors price the debt of bound firms at lower yields does not in itself mean that the sovereign ceiling policy has no impact on firms affected by it. More precisely, it is possible that relaxations or contractions in the sovereign ceiling result in even a more pronounced reduction or increase in the cost of debt of firms bound by the sovereign rating. To test this, in this subsection I first implement reduced form regressions to examine whether firms borrowing costs increase (decrease) more for those firms that are bound by the ceiling following a contraction or a relaxation in the sovereign ceiling. I then use a fuzzy regression discontinuity design (RDD) to instrument a one-notch corporate rating change directly related to 19

20 the sovereign ceiling channel, using as an instrument the interaction between a firm s bound status and a sovereign rating change Reduced Form Regressions I examine the change in corporate spreads around sovereign rating downgrades and upgrades for firms that are bound by the sovereign ceiling, relative to firms that are near but not bound by it. That is, I use a firm s bound status as an instrument to estimate the effect of a contraction or a relaxation in the sovereign ceiling on corporate spreads. I estimate the following reduced form pooled regression for sovereign downgrades and upgrades: Spread i,j,[t s] =β 1 Sov.Down i,j + β 2 Sov.Up i,j + β 3 (Sov.Down i,j Bound i,j ) + β 4 (Sov.Up i,j Bound i,j ) + ɛ i,j (5) where the dependent variable is the change in spread around sovereign rating changes: the spread on a firm s bond t months after each sovereign event minus its spread s months prior to the event. I focus on the differential effect on bound firms relative to firms that are near but not bound by the sovereign ceiling. The main coefficient of interests are β 3 and β 4, the interaction terms between sovereign downgrades and upgrades respectively, and the bound status identifier. I focus only on firms that have a corporate rating below the sovereign rating, as firms that pierce the ceiling tend to have certain specific characteristics that make them fundamentally different. I then face a trade-off between using as controls firms that are not bound by the sovereign rating but that are not too far away from it, and having enough firms as controls. Thus, I constrain non-bound firms to be three rating notches or less below the sovereign. I also limit my analysis to firms that have a rating of B- or higher. 12 Because rating changes can be anticipated, I perform event studies with different values of t around the time of the sovereign rating announcements to capture the response of financial markets. The indicator variables Sov.Down takes a value of one the 12 Only 0.1% of the observations in my sample have a corporate rating in the CCC, CC or D categories. 20

21 month of a sovereign downgrade. Similarly, the dummy variable Sov.U p takes a value of one the month of a sovereign upgrade. The dummy variable Bound takes a value of one if a corporate issue has the same rating as the corresponding sovereign prior to a sovereign rating change, and zero if a firm is three notches or less below the sovereign bound. Since firms can have two or more bonds at any given point, I weight observations based on the number of CUSIPs observed for each firm at each point in time, and I cluster standard errors by each country-sovereign event. If the sovereign ceiling represents a meaningful constraint for firms, I expect the interaction coefficients β 3, which measures the differential effect of sovereign downgrades on the change in spreads of bound firms, to be positive, indicating a higher increase in yields for bound firms. Similarly, I expect β 4, which measures the added effect of sovereign upgrades on bound firms, to be negative. Table 9, panel A reports the results from estimating the model in eq. 5 for event windows starting three months prior to a sovereign rating change. Results from panel A indicates that following a sovereign rating downgrade, and when comparing the change in spread one month after the event vs. the spread three months prior to the sovereign rating adjustment, the average spread for firms that are bound by the sovereign ceiling increases by 103 bp more than for firms that are not bound by the sovereign ceiling. As the event windows widens to three months after a sovereign downgrade the differential effect increases to 117 bp, while remaining statistically significant. Results for the upgrade interaction coefficient are statistically significant although economically weaker: the spread for bound firms decreases by 27 bp more following a sovereign upgrade. The fact the differential effect of sovereign rating changes is stronger for downgrades than for upgrades is consistent with previous studies who also find and asymmetry in the response to each type of event (e.g. Brooks et al., 2004; Gande and Parsley, 2005; Ferreira and Gama, 2007) Since I focus on the change in spreads around sovereign events for bound firms and firms that are not bound but still close to the bound, the empirical design I set up in equation 5 should not require the use for other baseline covariates; that is, their inclusion should not 21

22 affect the consistency of the parameters of interest β 3 and β 4. Nevertheless, to further pin down the identification of the effect of changes in the sovereign rating ceiling on spreads I extend the specification in model 5 by including country-event specific fixed effects (i.e. fixed effects for each sovereign rating downgrade or upgrade for each country). This approach, which reduces the likelihood that the coefficient estimates β 3 and β 4 will be biased by a correlation between country-time specific effects and spread changes, has a high cost in terms of lost degrees of freedom. Including country-event fixed effects equates to estimating the differential impact of the sovereign ceiling on the cost of debt of firms that are bound by the sovereign ceiling, relative to firms in the same country that are not bound by the constraint. I estimate the following model: Spread i,j,[t s] = β 1 (Sov.Down i,j Bound i,j ) + β 2 (Sov.Up i,j Bound i,j ) + CountryEventF E + ɛ i,j (6) where the main coefficients of interest are β 1 and β 2 (note that including country-event specific fixed effects absorbs both the constant term and the Sov.Down term included in the pooled regression). Panel B in table 9 shows the estimation results of this specification. Although including country-event fixed effects dampens the overall magnitudes compared to the pooled OLS results in panel A, the coefficients for the interaction term Sov.Down i,j,t Bound i,j,t remain statistically and economically significant. For example, the spread for firms that are bound by the sovereign ceiling increases by 54 bp more than for firms that are not bound by the sovereign ceiling using a one month after the event. When using a three-month pre vs. one month post window, the effect of a sovereign rating upgrade on firms bound by the ceiling is a decrease of 12 bp relative to non-bound firms. However, results for sovereign upgrades yield no statistical significance as the event window widens once country-event fixed effects are introduced. 22

23 5.2.2 Fuzzy Regression Discontinuity Design (RDD) In this subsection I implement a fuzzy RDD to instrument a one-notch corporate rating change directly related to the sovereign ceiling channel, using as an instrument the interaction between a firm s bound status with a sovereign rating downgrade (or upgrade). More precisely, I implement a fuzzy RDD where the effect evaluated is the impact on a firm s yield spread resulting from a corporate credit rating change directly related to a sovereign ceiling contraction or relaxation. As it has been noted, CRAs do not strictly follow the sovereign ceiling policy and thus a sovereign rating change does not lead to a 100% probability in bound firms obtaining the same rating change, as it would be required for a sharp RDD. However, as previously shown in figure 3, the probability of treatment jumps sharply from less than 10% for firms that are three notches or less below the sovereign ceiling to just above 60 % for firms that are exactly at the bound. The reduced form estimated using equations 5 and 6 and the fuzzy RD are closely related to each other. However, since the probability of a firm obtaining the same rating change as the sovereign is less than one at the sovereign bound, the jump in the relationship between δspread and Bound can only be interpreted as the average treatment effect of a corporate rating change stemming from the sovereign ceiling channel if Bound does not affect the changes in spreads outside of its influence through treatment receipt. As Lee and Lemieux (2010) point out, the treatment effect can still be obtained by instrumenting the treatment dummy (obtaining a corporate downgrade or upgrade due to the sovereign ceiling) with a firm bound s status. Thus, I implement the following fuzzy RD design described by the following equation system: Corp.Down i,j = α 1 (Bound i,j Sov.Down i,j ) + CountryEventF E + ɛ i,j Corp.Up i,j = γ 2 (Bound i,j Sov.Up i,j ) + CountryEventF E + ɛ i,j (7a) (7b) 23