Credit Ratings

Published on February 2017 | Categories: Documents | Downloads: 39 | Comments: 0 | Views: 383
of 58
Download PDF   Embed   Report

Comments

Content

Credit Ratings across Asset Classes: A ≡ A?
Jess Cornaggia, Kimberly J. Cornaggia, and John E. Hund*

August 31, 2011

Abstract

Contrary to assertions by the Big 3 credit raters, we demonstrate that credit ratings are not comparable across asset classes. Default frequencies, ratings transition matrices, hazard rate models, and ratings adjustment regressions all indicate that differences exist across asset classes both in the levels of credit ratings and the distributions of their changes. Relative to traditional corporate bond ratings, municipal and sovereign bonds have been rated more harshly and structured products have been rated more generously. These findings exist to varying degrees throughout our entire 30-year sample period. Consistent with a conflict of interest in an issuerpays compensation structure, ratings standards are inversely correlated with revenue generation among the asset classes. Our results are less consistent with the more benign explanation that ratings inflation is a result of issuer opacity. These results contribute to the debate surrounding regulatory reliance on credit ratings and the current SEC proposal to standardize credit ratings across asset class.

JEL classification: G14, G24, G28, G32 Keywords: Credit Ratings, NRSRO, Municipal Bonds, Sovereign Bonds, CDOs, Capital Markets Regulation

*

J. Cornaggia ([email protected]) is at the Kelley School of Business at Indiana University, K. Cornaggia ([email protected]) is at the Kogod School of Business at American University, and Hund ([email protected]) is at the Jones Graduate School of Business at Rice University. The authors are grateful to John Griffin for helpful discussion, as well as audience members at Indiana University. The authors thank Toby Kearn for research assistance. J. Cornaggia is grateful for financial support from the Kelley School of Business Research Database Committee. Any errors belong to the authors.

Electronic copy available at: http://ssrn.com/abstract=1909091

"We have always had one scale, a consistent scale that we have tried to adopt across all our asset classes." -- Deven Sharma, President, Standard & Poor‘s, July 27, 20111 One aspect of the Dodd-Frank Wall Street Reform and Consumer Protection Act of 2010 (DoddFrank) requires federal regulators to diminish their reliance on credit ratings issued by Nationally Recognized Statistical Ratings Organizations (NRSROs). As the Securities and Exchange Commission (SEC) considers alternative benchmarks, existing literature advocates credit risk measures that are timely rather than ‗stable‘, cardinal rather than ordinal, quantitative rather than qualitative, and funded by investors and other end users rather than issuers.2 However, there is a dearth of studies considering the importance of measuring absolute risk comparably across asset classes.3 Regulations drawing hard lines at ‗investment grade‘ heretofore draw no distinction between Baa3 rated municipal bonds and Baa3 rated collateralized debt obligations (CDOs). Why should they have? Moody‘s Investor Service (Moody‘s), Standard & Poor‘s (S&P), and Fitch Ratings (Fitch), collectively referred to as the Big 3 credit rating agencies (CRAs), report that their ratings are comparable across asset classes; see Appendix A.2. In this paper, we document otherwise. For each asset class, we report default frequencies by initial rating, we construct transition matrices, we investigate the distribution of times to upgrade and downgrade via hazard rate models, and we estimate rating change regression models. The evidence overwhelmingly suggests that while ratings of structured products were significantly more generous (optimistic) than those assigned to corporate bonds, those assigned to municipals and sovereign bonds were significantly less generous (more pessimistic).

1

Testimony before the U.S. House of Representatives, Committee on Financial Services, Oversight and Investigations Subcommittee, 2129 Rayburn Office Building, Washington DC, July 27, 2011.
2 3

See Partnoy (1999), Griffin and Tang (2010), Xia (2010) and Cornaggia and Cornaggia (2011).

Morse and Deely (1983) document state and regional biases in municipal bond ratings and conclude that northern states have an unfair access to regulated bank capital relative to southern states. However, this cross-sectional difference may reflect variation in state-mandated accounting requirements; Ingram and Copeland (1982). These studies consider only municipalities, without comparison to structured products, corporate, or sovereign issuers. Page 1

Electronic copy available at: http://ssrn.com/abstract=1909091

Moreover, we document a similar pattern within the broad category of structured finance. Tranches of collateralized debt obligations (CDOs) and residential mortgage backed securities (RMBS) were rated most generously at issuance, whereas public finance (PF) tranches were rated least generously of all structured products. Assertions by the Big 3 notwithstanding, evidence that rating performance varies by asset class may not surprise the informed reader. Previous literature addresses inflated credit ratings – particularly among structured finance products – and their contribution to the recent financial crisis; i.e. Coval, Jurek, and Stafford (2009). Moody‘s (2002, 2007) analysts explicitly discuss the municipal rating dichotomy, although they assert that sovereign and structured issues are intentionally assessed according to the same scale as corporate issuers. To our knowledge, we are the first to document comprehensively the differences in the applied rating standards across all asset classes. This is the primary contribution of this paper and we hope that it will aid regulators as they reconsider appropriate risk metrics for establishing bank capital requirements and prudent investments by pension funds and insurance companies. Because these regulatory restrictions ultimately effect capital allocation, our results should be of interest to a host of market participants. As a secondary contribution, we hope to shed some light on the reason for the difference in rating standards. One explanation for the inflated ratings of structured products is that structured ratings are ‗noisy‘ due to the opacity of the underlying assets. The intuition is appealing: synthetic CDOs are opaque relative to corporations with audited financial statements; this opacity results in greater dispersion among credit assessment by competing CRAs; this dispersion results in a greater opportunity for ratings shopping by the issuers of the CDOs. 4 In the model of Skreta and Veldkamp (2009), ratings inflation results from issuer opacity even if all

4

Sangiorgi, Sokobin, and Spatt (2009) provide a detailed discussion of rating shopping. Becker and Milbourn (2010) and Bongaerts, Cremers, and Goetzman (2010) provide evidence that is consistent with the notion that increased ratings shopping is an unintended consequence of increased competition in the credit ratings industry. Page 2

CRAs endeavor to produce accurate ratings. Further, Mathis, McAndrews, and Rochert (2009) suggest that CRA concern over reputation capital is diminishing in issuer opacity. Issuer opacity is a more compelling explanation for ratings inflation among synthetic CDOs backed by credit default swaps than for traditional RMBS or ABS. Indeed, one could argue that a pool of mortgages or credit card receivables should be less opaque than corporate issuers with synthetic leases and other exotic off-balance-sheet liabilities.5 However, we find even these more transparent structured products exhibit significant ratings inflation relative to corporate bonds. Moreover, to the extent that issuer opacity is a compelling explanation for ratings inflation, it should also apply to municipalities and sovereign issuers. Dispersion in the qualitative credit assessment of the government of Indonesia should, like complex structured products, be greater than the dispersion in CRA assessment of corporations‘ audited financial statements.6 Indeed, Ingram, Brooks, Copeland (1983, page 997) conclude that ―financial accounting information about municipalities is generally less reliable, less comparable crosssectionally, and less timely than information about corporations‖. Yet we find the ratings among municipal and sovereign bonds do not reflect the ratings inflation observed in the structured products. To the contrary, they appear deflated at issuance relative to corporate bonds. The evidence appears more consistent with a conflict of interest in the issuer-pays compensation structure. We find ratings optimism (leniency or inflation) increases in the revenue generation by asset class. Revenues generated from structured finance products are significantly higher than those generated from corporate issuers which are, in turn, higher than those generated from sovereign issuers and municipalities. We lack sufficient data to document

5

See Cornaggia, Franzen, and Simin (2011) regarding the impact of increased off-balance-sheet financing on conventional financial risk metrics including credit ratings.
6

Moody‘s (2008) Sovereign Analytics Report reviews extensive case studies of sovereign crises, discusses deposit freezes and debt moratoria, transfer and convertibility risk, country debt ceilings, the distinction between sovereign and country risk, changing legal provisions in various countries, correlation of sovereign defaults and banking crises, and Moody‘s qualitative approach to sovereign bond ratings. Page 3

variation in profit margin by asset class, which would be more compelling than correlation with revenues (i.e., deal volume). Still, we note that unlike corporations and municipalities that typically issue debt at lower ratings, structured products would not be marketable without sufficient Aaa tranches. Given the strong correlation of rating standards with revenue generation and the apparent lack of successful ratings shopping among complex sovereign issuers, our results are more consistent with a conflict of interest than the more benign interpretation of CRA best efforts to rate opaque issuers. Our results are important and timely given the current SEC mandate to consider the feasibility and desirability of standardizing credit ratings. Our results do not necessarily imply that ratings should be (or even could be) standardized across asset classes. 7 However, our results imply that reliance on ratings without standardization likely results in over-allocation of regulated funds in higher risk CDOs with under-allocation to less risky municipal bonds. Our results also contribute to the ongoing debate surrounding the value of municipal bond insurance to taxpayers. The argument against paying insurance premiums is twofold. First, if the municipal issues constitute a ‗moral obligation‘ on the part of higher levels of government, there is negligible risk of default. Second, if bond insurers can – in the event of a municipal default – refuse to insure all future issues throughout the state, thereby implicitly threatening the state‘s credit rating, state governments make every effort to cover their municipalities facing potential default. So, why do local governments pay insurance premiums when filing claims is cost-prohibitive? The argument in favor of bond insurance is simple: insurance allows local governments to ―buy‖ the Aaa rating of the bond insurer and therefore lowers the interest rate paid by taxpayers.8 We contribute to this debate by analyzing municipal bond performance vis-àvis bonds issued by corporations, sovereign nations, and structured finance products. In our

7

The CRAs argue that it is not feasible and some market participants suggest that it is not desirable; www.sec.gov/comments/4-622/4-622.shtml.
8

Indeed, the downgrades in 2008 (Ambac from Aaa to Aa3 and MBIA from Aaa to A2) had a significant negative impact on municipal bond insurance; McDonald and Richard (2008). Page 4

sample of bonds issued between 1980 – 2010, corporate bonds with Aaa ratings at issuance defaulted more frequently than municipal bonds issued with a single-A rating (six notches lower).9 Moreover, municipal bonds were significantly more likely to be upgraded over this 30 year period. If historical performance is indicative of future performance (as ratings models typically assume), then single-A-rated municipal bonds (without insurance) should not face higher interest rates than their Aaa-rated insurers.10 The paper proceeds as follows. Section II provides institutional details, Section III describes the sample, Section IV discusses the empirical results pertaining to credit rating performance across asset classes, and Section V concludes. II. Institutional Detail A. Dichotomous municipal rating scales In contrast to Moody‘s public assertion of comparable ratings (Appendix A.2), their analysts report a lower default rate among all municipals when compared to Aaa-rated corporate issues; Moody‘s (2002).11 This 2002 report explains that their municipal bond rating scale is distinct from the corporate bond rating scale, though the report attests that the latter scale is applicable to non-U.S. sovereign issuers and all structured products. By 2007, this latter scale ―used to rate all bonds outside of the U.S. public finance market‖ is referred to as the Global Rating Scale; Moody‘s (2007). The duality is attributed in part to the tax-exempt nature of the municipal bond to U.S. investors and in part to a finer gradation in the more stringent municipal rating scale. Unlike the global rating scale which measures ―expected loss‖ among corporate and other non-municipal issuers, the municipal rating scale reflects the probability that the

9

The default frequency of municipal bonds issued in the range Aaa - A3 was 0.3% over the period 1980-2010; the corresponding frequency for corporate bonds (structured products, sovereign debt) was 1.3% (50%, 0%).
10

Peng (2002) suggests that investors are increasingly relying on the underlying ratings of the insured municipal bonds rather than relying on the rating of the bond insurer.
11

Reported five- and ten-year cumulative default rates for all Moody‘s-rated municipal bond issuers were 0.0233% and 0.0420%, respectively, over the 1970-2000 period. These compare to 0.1237% and 0.6750% for Aaa rated corporates over the same period. Page 5

municipality will need support from higher government. Historically, state governments cover bond payments for distressed municipalities resulting in trivial ―expected loss‖ and thus Aaa rated municipalities – if they had been rated according to the same scale as corporate issuers. In 2010, following the release of the senate proposal to mandate such change, Moody‘s announced that it would move away from the dichotomous municipal rating scale. 12 Moody‘s reported that General Obligation (GO) bond ratings would rise by two notches on average, with a range of zero to three notches, most among those rated below Aa3 on the municipal scale. B. Implications for regulatory compliance and capital allocation Financial regulators have long relied on NRSRO ratings to limit the financial risk borne by banks, money market and pension funds, and insurance companies.13 Because fixed income markets are dominated by these regulated institutional investors, such regulations have important implications for capital allocation in the U.S. economy.14 Dodd-Frank calls for the removal of NRSRO ratings in SEC rules and other federal regulations; see Appendix A.1. However, state (i.e., insurance) and international (i.e., Basel) regulators are not subject to Dodd-Frank mandates. Efficient capital allocation requires reliable risk metrics that reflect cardinal (absolute) credit risk – at least contemporaneously and preferably consistently through time. Prior literature explores the consequence of regulatory reliance on ratings that are ordinal (relative), based on qualitative analysis, paid for by issuers, and intentionally slow to update.15 Another strand of literature considers variation in credit standards for corporate bonds over time; Blume, Lim, and MacKinlay (1998) and Jorion, Si, and Zhang (2009). We consider here the potential

12 13

www.bondbuyer.com/issues/119_300/moodys-moving-to-global-muni-scale-1009615-1.html Regulations either prohibit speculative-grade assets or impose large capital requirements to support them. See Cantor and Packer (1997) and Cornaggia and Cornaggia (2011) regarding the pervasive reliance on credit ratings.
14

For example, Ellul, Jotikasthira, and Lundblad (2010) document fire sales of downgraded bonds induced by regulatory constraints imposed on insurance companies.
15

See Partnoy (1999), Griffin and Tang (2010), Xia (2010) and Cornaggia and Cornaggia (2011). Kraft (2010) provides evidence of CRA catering to client issuers, which further supports reliance on independent credit analysis. Page 6

misallocation of capital resulting from regulatory reliance on ratings with standards that vary by asset class. The investment- versus speculative-grade threshold (drawn at Baa3) is the most prominent, but some regulations draw the line at Aaa. For example, money market funds have historically been required to hold Aaa rated commercial paper, and banks face lower reserve requirements for loans to Aaa rated borrowers. However, to date none of the ratings-based regulations differentiates thresholds according to asset class. Given variation in absolute credit risk, regulated institutional investors can to some extent circumvent regulatory constraints (e.g., by overinvesting in CDOs and underinvesting in municipal bonds). We assume that large institutional investors rely on more sophisticated risk metrics for bond valuation; see Cornaggia and Cornaggia (2011). However, retail investors relying on credit ratings for information – and assuming uniform standards across asset classes – may unknowingly misallocate their capital.16 III. Data We employ Moody‘s Default and Recovery Database (DRD) and Moody‘s Structured Finance Default Risk Service Database (SFD).17 The DRD includes complete Moody‘s credit ratings histories for debt obligations issued by corporations (industrials and transportation companies), financial institutions (U.S. banks, U.S. bank holding companies, securities companies, and insurance companies), sovereign nations, and local and regional governments. The SFD includes Moody‘s ratings histories for structured finance products including Asset Backed Securities (ABS are backed by various receivables including credit cards, auto loans, student loans, equipment leases, etc.), Collateralized Debt Obligations (CDOs), Commercial

16

Although retail participation in structured finance markets has historically been indirect through regulated institutional investors, the retail market for structured products has grown in recent years; www.finra.org/Newsroom/NewsReleases/2011/P123744
17

Ideally, we would like to examine the difference across asset classes for each of the NRSROs. However, previous work indicates that ratings by the Big 3 are highly correlated suggesting that results for S&P or Fitch would likely be similar to those reported here for Moody‘s, see Bongaerts, Cremers, and Goetzman (2010). Thus, the costs associated with purchasing complete ratings history from each CRA likely outweigh the potential benefit of any cross-CRA analysis. Page 7

Mortgage Backed Securities (CMBS), structured Public Finance (PF) deals, and Residential Mortgage Backed Securities (RMBS). We examine Moody‘s credit ratings from both databases that fall along a 21-point alphanumeric scale. The scale ranges from most creditworthy to least creditworthy: Aaa, Aa1, Aa2 Aa3, A1, A2, A3, Baa1, Baa2, Baa3, Ba1, Ba2, Ba3, B1, B2, B3, Caa1, Caa2, Caa3, Ca, and C. We map alphanumeric ratings Aaa, …, C to numeric ratings 21, …, 1 such that ratings are increasing in credit quality and decreasing in credit risk. Obligations with credit ratings equal to Baa3 or higher (12 through 21) are ―investment grade‖ and obligations with credit ratings equal to Ba1 or lower (1 – 11) are ―speculative grade.‖ Our analysis focuses on credit ratings issued by Moody‘s between 1980 and 2010. A. Sample description We describe the sample, by asset class, in Table I. The median face value of corporate bonds ($150M) is more than twice the size of the median municipal issue ($64M). Sovereign issues are the largest by a wide margin; $822M at the median. Financial issues and structured tranches are considerably smaller; $30M and $18M at the median, respectively. However, as noted above, the broad ―structured finance‖ category includes securities backed by a variety of underlying assets. The ABS are the largest sub category ($36M median) followed by CDOs ($30M), CMBS ($24M), RMBS ($13M). Public finance (PF) tranches are the smallest ($6M median). Another noteworthy difference is in the duration of the structured products averaging 24.1 years with a 30 year median. The other asset classes have means and medians closer to 5 to 7 years. Initial ratings vary by asset class, as do frequencies of ratings changes. The median corporate issue is initially rated Baa2, which is investment grade. Further, these bonds are more likely to be downgraded (30 percent probability over the life of the bond) than upgraded (20 percent probability over the life of the bond). Conversely, municipal bonds are three times more likely to be upgraded than downgraded (30 percent versus 10 percent), and are more likely to be upgraded than corporate bonds even though the median is issued at a much higher rating (Aa2)
Page 8

than the median corporate bond. Sovereign issues are quite similar to municipals in this regard; three times as likely to be upgraded than downgraded even though the median bond has a higher rating at issuance (Aa3) than the median corporate bond. Conversely, structured tranches behave more like corporates, but with an even greater frequency of downgrades and an even lower frequency of upgrades. The median structured tranche was issued with an Aa2 rating over our sample period. The notable counter examples within the broad structured finance category are the PF tranches which behave more like the municipals and sovereigns: more likely upgraded than downgraded even from higher initial ratings (Aaa at the median). Finally, the financial issues are broadly similar to the corporate bonds along these dimensions. Frequency of default also varies by asset class. Six percent of our benchmark corporates default over the sample period. The handful of municipals that defaults is a rounding error; 1% of sovereigns default; 3% of financials default; and 14% of structured tranches default. This percentage varies considerably by the underlying asset types: 20% of ABS, 29% of CDOs, 15% RMBS and only 4% of CMBS. No tranches of the PF deals default over our sample period. We provide greater detail on defaults by asset class over time in Figures 4 and 5 below. The correlation matrix in Table II suggests multicolinearity among the descriptive variables and asset classes. Table I indicates that sovereign issues have the highest face values; indeed there is little intersection between (sovereign, non-sovereign) in this dimension. Likewise, there is little intersection between (structured, non-structured) in issue maturity. As expected, structured products are strongly positively correlated with initial rating, downgrades and defaults. Although they are also significantly positively correlated with initial rating, municipals are negatively correlated with downgrades and defaults. This effect leads to the somewhat counterintuitive negative correlation between initial rating and downgrades. Ratings migration matrices presented in Table IV below help explain this finding, which is a result of upward momentum of municipals, sovereigns and PF tranches that are issued with higher ratings than the average corporate bond.
Page 9

We plot annual issuance volume by asset class in Figure 1. The overwhelming importance of structured finance products post-2000 is clear. Figure 2 provides greater detail of initial ratings for each asset class and how these evolved over time. In each panel, the proportions are cumulative with the issues rated Aaa appearing at the top, Aa second from top, and so on. The A, Baa, Ba, and B levels are most prominent among corporate issues in Panel A. The Aaa and Aa slivers along the top are just discernable in greyscale, although combined they comprise as much as 28% (in 1991) of original issues. The Caa range spikes in 1999 (6.8%) and grows again from 2001 to reach 6.7% of original issues in 2004. The Ca and C issues are discernable in 1999 (1.1% of new issues). Panel B is less complicated. Municipal bonds are generally rated Aaa, Aa or A at issuance (although the A level is difficult to discern in greyscale). The diminishing proportion of Aaa beginning in 2000 is remarkable given the other changes beginning in this year, including the shift in focus toward structured products. Sovereign issues in our sample (Panel C) were generally investment grade (Baa or higher) prior to the wave of sovereign crises beginning in 1995, a year in which 29.6% of new sovereign issues were awarded speculative ratings at issuance.18 The percentage of new issues in speculative territory peaked at 49.9% in 1999 and fell back to 28% in 2002 and remains in the 18-35% range thereafter, reflecting continued issues of downgraded countries and issues by an increasing pool of emerging markets. These initial ratings were frequently upgraded after issuance, a phenomenon documented in Table IV below. Financial issuers in Panel D include U.S. banks, U.S. bank holding companies, insurance companies, and securities firms. The initial ratings of these issues are overwhelmingly in the Aa and A categories. Our sample contains 298 new issues by financial firms in 1990 of which 80.9%

18

See Bartram, Brown, and Hund, (2007) for details on the Mexican crisis (December 1994), Asian crisis (July 1997), Russian crisis (August 1998), LTCM crisis (September 1998), the Brazilian crisis (January 1999) and the disruption to the financial system following the appalling attacks on the U.S. in September 2001. Page 10

was rated A or higher. By 1995, this percentage is 66.8% of 709 new issues; by the year 2000, 96.5% of 1565. At the peak in 2007, 96.5% of 3399 new issues are rated A or higher. Panel E contains all structured issues (tranches) and provides detail regarding the change in initial ratings of these products over time. Proportionally, the greatest increase is observed in the Baa tranches, which are the lowest investment grade ratings. A common criteria for regulatory compliance is that investors may only hold investment grade-rated securities. Virtually nonexistent prior to 1990, the Baa tranches represent 10.2% of structured issues in 1999, grow to 15% in 2000 and peak at 26% in 2006 before declining back to 5 and 7% in 2009 and 2010. Conversely, 95% to 100% of tranches in the earliest years of the sample received Aaa ratings. This proportion declined to 52% in 1995, ranging from 50% to 65% thereafter. We break down the broad ‗structured finance‘ category in Panels E.1 through E.5. These individual figures display similar qualitative patterns to the pooled figure in Panel E. The one clear departure from typical is Panel E.4 containing Public Finance issues which are consistently rated Aaa or Aa. Similar to the trend observed in municipal bonds – but to less extent here – the proportion of Aaa is declining over time. B. Moody’s revenue by asset class In Figure 3, we plot annual revenue by asset class as reported in Moody‘s 10-K filings.19 By 2005, the earliest year for which we have revenues by asset class, revenues generated by rating structured products ($709M) are 2.5 times revenues generated by rating corporate bonds ($277M). By 2008, the difference became smaller ($405M and $307M, respectively). By 2009 revenues from corporate bonds once again surpassed those of structured products. Taken together, Figures 1, 2, and 3 suggest that Moody‘s over-reliance on structured products primarily occurred between 2000 and 2007. Before this time, corporate and structured markets appear comparable; following the collapse of the structured finance market in 2008 (Figure 2), corporate
19

Moody‘s changed its asset classification for revenue reporting in 2007. The 10-K filed in February 2008 provides retroactive revenues revisions by segment back to 2005. Moody‘s did not report revenue by asset class prior to 2005. Thus, the time series between 2005 and 2010, using the retroactive data from 2008, is the largest span available. Page 11

issuers are again the primary source of revenue generation (Figure 3). Municipalities and sovereign issuers always appear to be a far less relevant source of revenue for Moody‘s. C. Defaults by asset class through time We provide more detail regarding default frequency over time in Figure 4. Prior to the year 2000, the graph depicts low default frequency in general, with corporates higher than financial services and a trivial incidence among municipals. Corporate defaults correspond generally with NBER business cycles.20 In the year 2001, we observe a spike in the sovereign default frequency (approaching 3%) followed by a minor uptick (1%) among municipals in 2002. The most recent financial crisis is apparent in the tranches of structured finance products, although corporate issues and financial institutions also reach in-sample peaks. IV Ratings Performance and Comparability across Asset Classes In this section, we compare various rating performance metrics across asset classes. In order to ease interpretation, we employ bonds issued by corporations (industrials and transportation firms) as our benchmark. Although we find that they perform similarly in general, we consider financial institutions separately from corporates because Moody‘s reports them as a separate category (see Figure 3). We analyze structured finance deals at the tranche level because it is possible for tranches to perform differently; i.e., Moody‘s could downgrade a B1-rated tranche without downgrading the Aaa-rated tranches in the same deal. However, we recognize that the tranches of any particular deal are not entirely independent of one another and we thus cluster standard errors at the deal level in all regression models. As noted in Section III above, the broad category of structured finance products contains distinct classes of underlying assets. These deal types may vary in terms of deal complexity and the various issuer types also contribute differently to CRA revenue. We thus break structured products into various types in most analyses.
20

NBER reports an 8 month contraction July 1990-March 1991, an 8 month contraction March-November 2001, and an 18 month contraction December 2007 – June 2009. Find complete cycle data here: www.nber.org/cycles.html Page 12

A. Default percentages by asset class and initial credit ratings We document the frequency of default by initial ratings across asset classes in Table III. That is, we report the % of bonds issued Aaa (or Aa, etc.) that later default within our sample period – separately for each asset class.21 Moody‘s reports that its ratings are intended to be ordinal in nature; Moody‘s (2002b). If properly ordinal, we should find the default frequency strictly decreasing in credit rating. This appears to be the case only for the corporate issues. However, the 8% (10%) default rate among municipals issued Baa (sovereigns issued B) is somewhat misleading given the very small number of issues with these ratings. More concerning is the 4.9% default rate among A-rated financial bonds and the default rate among investment grade tranches of the structured products, including Aaa. Table III clearly indicates a material difference in the absolute credit risk implied by any given rating across asset classes. For example, consider the default frequencies in the A range: corporations 1.83%, municipals 0.49%, sovereigns 0.00%, financials 4.92%, and structured tranches 27.21%. Perhaps most importantly to taxpayers, we note that municipal bonds rated Ba have a lower default frequency than the Baa rated corporate issues. This is important because the line between investment grade and speculative bonds lies between Baa and Ba. A host of financial regulations (noted in Section II above) deny Ba rated municipals access to the regulated capital of financial institutions. Without access to the largest investors in fixed income, taxpayers faced higher interest rates due to higher liquidity premiums associated with thinner markets for speculative grade debt. The decomposed structured products (according to their underlying assets) better indicate the problematic issue types. The pervasive defaults of investment grade tranches are primarily among ABS, RMBS and CDOs. The defaulting CMBS tranches were largely issued as speculative grade. There are no defaults among any of the Public Finance tranches. B. Transition matrices
21

A missing default frequency implies that there were no issues in that particular rating bin. A default frequency of 0.00% implies that there were issues but none defaulted. Page 13

In order to better understand the path from investment grade to default, and to better appreciate the variation in ratings migration by asset class, we report ratings transition matrices in Table IV. Transition probability matrices sometimes report ratings migration frequencies as percentages of the initial rating total. We choose instead to report ratings migration using the number of issues in each ratings bin five years after the original date of issue.22 Reporting the number of issues rather than percentages allows the reader to better visualize the relative mass across asset classes and across ratings bins within each class. (For example, there are three ABS issues rated Ca at issuance in our sample. Reporting one migration to Caa and one migration to C is more informative than reporting 33.3% in each of Caa, Ca, and C.) The sum column conveys the relative likelihood of each initial rating and allows for comparison across asset classes. We also summarize the percentage of upgrades and downgrades in the rightmost columns, by initial rating. We report migration of ratings among corporate issuers first (in Panel A) as they serve as the benchmark for comparison. First, we note the disparity between upgrade (9.5%) and downgrade (27.5%) frequency which is consistent with a bias at the time of issuance in favor of issuing corporations from whom Moody‘s receives its compensation. This bias is consistent with the conclusions that Moody‘s caters to issuing firms (Kraft (2010)) and that Moody‘s favors Type I classification errors (Cornaggia and Cornaggia (2011)). We also note the higher downgrade frequency among the higher ratings and higher upgrade frequency among the lower ratings. This apparently ―contrarian‖ rating migration is not surprising as both ends of the rating distribution (Aaa and C) can only change in one direction. The migration of municipal bond ratings in Panel B is remarkably different from that of corporate bonds. First, we note that although they have a higher percentage of initial ratings in the upper echelon (35% Aaa and 33% Aa) they are far less likely to downgrade (9.3%) than corporates (27.5%) and more likely to upgrade (11.7%) than corporates (9.5%). Moreover,
22

Reported totals are somewhat incomplete as the transition matrix imposes a five year survivorship bias. Page 14

unlike the more ―contrarian‖ ratings changes among corporates (higher downgrade frequency among the higher ratings and higher upgrade frequency among the lower ratings) the municipal bonds exhibit ratings changes better characterized as ―momentum‖ (higher downgrade frequency among the lower ratings and higher upgrade frequency among the higher ratings). More than one in ten municipal bonds initially rated Aa is subsequently upgraded to Aaa. It is important to note that our sample precedes Moody‘s advertised recalibration following the introduction of DoddFrank legislation. The upgrades we document are on the original, more granular, municipal scale. The migration of sovereign issuers is Panel C is more closely resembles that of municipal bonds than corporates with a 17% (17.8%) frequency of downgrades (upgrades) and a relatively high incidence of upgrades among the upper echelons (24% of Aa and 27% of A are upgraded). Financial issues in Panel D behave more similarly to corporates (27% downgrade with a ―contrarian‖ migration pattern), although the frequency of upgrades is higher (20% versus 9.5%). The ratings migration found in Panel E suggests that the structured finance products enjoyed the most inflated initial ratings of the broad asset classes. We break these down into subcategories in panels E.1 through E.5. The grade inflation appears most severe among the CDOs (51.4% downgraded versus 3.2% upgraded) followed by RMBS (37.2% downgraded versus 6.7% upgraded) and ABS (30% downgraded versus 3.6% upgraded). The inflation among CMBS at issuance (26.6% downgraded versus 14.3% upgraded) appears less egregious than the aforementioned structured product types, and the Public Finance tranches appear more evenly split (10.8% downgraded versus 9.6% upgraded). C. Preliminary work on mobility indices A drawback of the transition matrices in the previous section is they combine the performance of credit ratings over the entire sample period (1980-2010). This analysis clearly demonstrates that ratings of asset classes have behaved differently over the last three decades, but it does not allow us to specify the periods of time in which the individual asset classes experienced the greatest and least amounts of transition, nor whether these transitions were
Page 15

significantly different from other asset classes. This section extends the analysis in the previous section by calculating annual mobility indices – scalars that summarize the amount of transition exhibited by the ratings of each asset class each year. Specifically, we calculate Euclidean distance metrics (Equation 1) and Singular Value Distance metrics (Equation 2) as proposed by Jafry and Schuermann (2004). M
∑ ucl dean ∑ √ (̃ ̃ ) ∑ |̃|

, where: ̃

-

(1)

M

, where: ̃

-

(2)

̃ is a mobility matrix. We calculate mobility matrices by subtracting P, the raw transition matrix which contains the probability elements Pi,j, the conditional probability of having rating j in the next time period given a current rating of i, from the identity matrix. We calculate each of these metrics for each asset class and each type of structured product each year of the sample. We construct the raw transition matrices based on transitions over five year periods. For example, the metrics for the year 1985 would be based on how bonds existing in the sample in 1980 had transitioned from 1980 to 1985. with the matrix, ̃ ̃. Jafry and Schuermann (2004) provide a thorough discussion of the relative merits of the Euclidean and Singular Value Distance metrics. For our purposes, these mobility indices allow us to distill the transitionary properties of the asset classes‘ credit ratings into simple metrics. Further, we develop the distributional properties of these metrics for each asset class through the resampling technique of bootstrapping. This approach, although imperfect because the underlying data do not represent a random sample from a given population, allows us to construct standard errors, and thus reasonably evaluate the statistical significance of asset classes‘ mobility indices. Our preliminary untabulated results indicate the relative probabilities of ratings transitions across asset classes are broadly similar through time. Specifically, corporate (̃ ̃), i = 1 to N is the set of eigenvalues associated

Page 16

bonds are more likely to transition than municipal bonds or sovereign bonds throughout the entire sample. Bonds issued by financial companies are also more likely to transition than municipal bonds or sovereign bonds. Not surprisingly, tranches of structured products transition to a much greater extent than other asset classes in the final years of the sample. However, perhaps surprisingly, tranches of structured products are also more likely to transition than municipal bonds or sovereign bonds through the 1990s, and are roughly equally likely to transition in comparison to corporate bonds and financial bonds. Finally, within the broad structured finance category, RMBS and CDO tranches appear far more likely to transition over the most recent years of the sample, and PF tranches are the least likely to transition. D. Hazard rate models This section formally investigates the differing probabilities of ratings changes across asset classes by comparing hazard rates for downgrades and upgrades. Specifically, we denote the instantaneous downgrade (or upgrade) rate for bond j as hj(t) and estimate:
h j t   h0 t exp X 

(3)

for a vector of covariates X. This approach is a single-failure Cox proportional hazard model
 with ―failure‖ denoting a downgrade (upgrade), the unit of observation being the time until a

downgrade (upgrade) for each rating change, and allowing observations to exit or censor upon upgrade (downgrade), maturity, default, or the end of the sample period. For the vector of covariates X representing dummy variables corresponding to membership in various asset classes, the coefficient β represents the proportional shift in the instantaneous baseline downgrade/upgrade intensity, which we set to correspond to corporate bonds. For example, β i = 2 would indicate asset class i has a downgrade rate which is twice that of corporate bonds; β i = .5 means that asset class i has a downgrade rate half that of corporate bonds.23 Coefficients for all asset classes statistically insignificant from 1 imply strict ratings comparability in the sense that the distributions of ratings changes are indistinguishable.
23

This specification implies that significance testing is versus the null of  i = 1. Page 17

Table V presents the results of Cox regressions on ratings changes over the full sample period for all ratings changes and for those ratings changes broken down into those originating from investment and speculative grades. It is apparent that ratings change intensities are dramatically different across asset classes. In fact, all coefficients in the table are significantly different from the baseline at the 1% level. More striking is the pattern across the full sample of low relative downgrade intensities and high relative upgrade intensities for municipal and sovereign bonds and exactly the opposite pattern for structured products, with ABS and CDO exhibiting especially high downgrade intensities. This result is consistent with the notion that initial ratings for municipal and sovereign bonds are too low and ratings for structured products, especially CDOs, are too high. Additional insight into ratings volatility comes from estimating relative upgrade and downgrade intensities separately; upgrade and downgrade intensities both greater than the baseline corporate class point to highly volatile yet unbiased ratings changes, as is the case for the financial asset class and speculative grade municipals. Such a pattern might emerge if these bonds were more opaque than corporates; it is much harder to reconcile the pattern of higher downgrade and lower upgrade intensities exhibited by all the other asset classes with theories of asymmetric information. A potential drawback of using all ratings changes as observations is that the estimation could be skewed by differences in ratings momentum across asset classes.24 In Table VI we limit the sample to the first rating change after issuance to more directly measure the potential implied bias in the initial rating. Municipal and sovereign bonds have nearly ¼ the relative downgrade intensity of CDOs and almost 10 times the upgrade intensity, strongly suggesting that initial ratings were biased in favor of at least the CDO class of structured products. Other structured products fare much better, although ABS and RMBS securities have somewhat skewed upgrade and downgrade relative intensities. Financial bonds continue to have much higher upgrade and
24

Lando and Skodeberg (2002) document significant ratings momentum using a method similar to our Cox regressions. Page 18

downgrade relative intensities, again potentially consistent with higher issuer opacity in this asset class. Overall, the results strongly reject the idea that ratings changes are comparable across asset classes, and also support the notion that initial ratings on municipal and sovereign classes were too low, and those on CDO securities too high. Differential ratings changes across asset classes are obviously dominated by changes during the financial crisis period of 2007-2010 where large numbers of downgrades clearly affect the downgrade intensity rate. Table VII confirms this effect by presenting results for ratings changes by time period. Nearly half of the ratings changes occur after 2006, reflecting both the massive waves of issuance in the prior period and rush to downgrade during the financial crisis. Downgrade intensities for structured products change dramatically, first declining for all types in the period after Moody‘s goes public in the year 2000, and then spiking far higher than the baseline corporate intensity during the crisis. Municipal and sovereign bonds continue to have consistently lower downgrade intensities than corporate bonds and higher upgrade intensities, with the exception of municipals during the financial crisis.25 Interestingly, the high fee CDOs seem to be over-rated even in the pre-crisis period with their downgrade intensity only approaching that of corporates during the ―boom‖ period of 2000-2006. Results from the Cox proportional regression framework overwhelmingly reject the hypothesis of ratings comparability, and are highly suggestive of bias in initial ratings for municipal, sovereign, and structured products relative to the corporate benchmark. These biases correspond directly to the fees associated with rating each asset class, with municipal and sovereign bonds persistently underrated and structured products (especially CDOs) persistently overrated. In addition, asset classes differ not only in levels of ratings and default behavior, but also in their distributions of ratings changes and their volatility. Credit risk models (such as CreditMetrics) that use ratings transition matrices as an input yet ignore asset classes will not

25

The upgrade probabilities would be dramatically higher if our sample included the re-calibration of Moody‘s bond scale for municipals and the concomitant upgrade of virtually all municipal bonds. Page 19

only under- or over-estimate ratings volatility, they will also have biased distributions of rating change intensities. E. Rating change regressions Table IV demonstrates migration from initial rating by asset class and summarizes upgrades and downgrades in a binary sense. We further explore the extent of these ratings changes by asset class using discrete (but not binary) ratings changes as the dependent variable in the regression models found in Table V. For ease of interpretation, corporate issues are employed as the benchmark class. The dependent variable in Panel A is a discrete indication of the ratings changes for each issue (corporates, municipals, sovereign bonds, bonds issued by financial institutions, or structured product tranches) from issuance until maturity, default, or until the end of available data. Numeric ratings are increasing in credit quality; i.e. Aaa = 21 and C = 1 such that positive coefficients correspond to upgrades and negative coefficients correspond to downgrades. Results from the full sample indicate that, relative to corporate issues, municipal and sovereign bonds are significantly more likely to be upgraded and structured finance products are significantly more likely to be downgraded. This is consistent with a more generous rating of structured products at issuance and a more stringent rating of the municipal and sovereign issues. While the results generally hold through time for the municipal and sovereign issuers, there is a marked change in the year 2000 for the structured products. In fact, the negative coefficient in the full sample is driven by bonds issued after Moody‘s went public in the year 2000. Prior thereto, structured products were more likely upgraded. In the full sample, ratings of financial institutions appear similar to those of the corporate (industrial) issues but again with variation over time. Given the financial crisis of the era, it is not surprising that bonds issued by financial institutions after 2006 were more likely downgraded. The independent variable in Panel B is a binary variable indicating a downgrade from investment to speculative grade. Crossing this threshold has important consequences for regulated institutional investors; Ellul, et al. (2010). Comparing the number of observations to
Page 20

those in Panel A indicates the proportion of bonds (tranches) issued in the speculative territory. The results broadly tell the same story as those in Panel A. Only structured products were more likely downgraded into speculative territory than were corporates, and these results are again driven by bonds issued after the year 2000. The dependent variable in Table VI is the same as in Panel A of Table V; discrete but not binary. In Table VI, we disentangle structured products based on their underlying assets. Corporations are again employed as the benchmark class. This table suggests that the results in Table V Panel A are driven by CDOs, RMBS, and to a lesser extent ABS, which are all more likely downgraded than the corporates in the overall sample. Conversely, ratings of Public Finance tranches are more likely upgraded and ratings changes among CMBS are not significantly different from the corporate issues. Table VII reports results from one model containing each of the various structured finance types. Corporations remain the benchmark class. The results are consistent with Tables V and VI. Relative to corporates, municipal and sovereign issues are consistently more likely to be upgraded. In general, structured finance products are more likely upgraded prior to the year 2000 and more likely to be downgraded thereafter. These downgrades are less significant among the CMBS and PF deals than among the CDOs, RMBS and ABS. Overall, the regression results reported in Tables V, VI, and VII are consistent with a generally more stringent rating of the municipal and sovereign issues at issuance and a more generous rating of structured products, those issued after the year 2000 in particular. F. Cumulative distributions of default prediction ability and accuracy ratios Perhaps the most common metrics of ratings performance are empirical cumulative distributions of default prediction and accuracy ratios (see Cantor and Mann (2003)). Figure 5 displays cumulative distributions of default and the corresponding accuracy ratios for the five main asset classes in our sample (Panel A), as well as individual structured product types (Panel B). For each asset class and type of structured product we count the number of bonds with a
Page 21

given credit rating as of January 1 of any year of the sample and the number of those bonds that default over the following year. For each credit rating, we divide the full sample count of defaulted bonds by the full sample count of bonds. This approach calculates a default percentage associated with each credit. Panels A and B plot the cumulative distribution of these percentages for each asset class and type of structured product, moving from the lowest credit rating to the highest. The solid black line in both panels represents the cumulative distribution of ratings that have no predictive content. In other words, if Moody‘s randomly assigned credit ratings, then we would expect equal percentages of defaults among the ratings, and the dashed line representing a uniform cumulative distribution function would emerge. In Panel A, the cumulative distribution for municipal bonds lies higher and further to the left than the other four asset classes. Closest to the cumulative distribution for municipal bonds is that of corporate bonds. The cumulative distribution for tranches of structured products lies closest to the ―randomly assigned‖ cumulative distribution. This comparison indicates that more of the lowest-rated municipal bonds default than the lowest-rated bonds of other asset classes, and less of the lowest-rated tranches of structured products default than the lowest-rated bonds of other asset classes. In other words, Moody‘s credit ratings for municipal bonds outperform Moody‘s credit ratings of other asset classes in terms of ordinal performance, with ratings of corporate bonds performing second best, and ratings of tranches of structured products performing the worst. We compute accuracy ratios in order to formally express the difference between these cumulative distributions. Accuracy ratios measure the area between the cumulative distribution and the dashed line. The larger the ratio, the more accurate the ratings are in an ordinal sense. To be concrete, we calculate the accuracy ratios as follows:
u ber of ssues t at default over t e ne t year

Accuracy ratio = ∑

[∑

[

u ber of ssues



u ber of ssues t at default over t e ne t year u ber of ssues

- ]]

(4)

Page 22

N = the number of credit rating classifications (we combine ratings of Caa1, Caa2, Caa3, Ca, and C since so few bonds take on these ratings) and i, j, and k are numerical translations of issues‘ credit ratings. The accuracy ratios for the five asset classes are as follows: municipal bonds = 0.44, corporate bonds = 0.40, sovereign bonds = 0.36, financial bonds = 0.30, and tranches of structured products = 0.16. The accuracy ratios of the individual structured products are as follows: CMBS = 0.33, ABS = 0.17, RMBS = 0.14, and CDO = 0.09. Importantly, we cannot construct a cumulative distribution of default prediction ability or calculate an accuracy ratio for PF tranches because none defaulted in our sample. Taken together, the accuracy ratios provide additional evidence that credit ratings across asset classes behave differently. The credit ratings of municipal bonds perform best in terms of ordinal performance, with those of corporate bonds performing second best. The credit ratings of structured products perform worst in an ordinal sense, with those of CDOs exhibiting the worst performance of the asset class. V Conclusion We examine the differential performance of credit ratings across asset classes. We find that rating standards are inversely correlated with revenue generation. Relative to traditional corporate issues, structured finance products generate higher rating agency revenues and receive significantly higher (more optimistic) ratings. Conversely, municipalities and sovereign issuers generate the lowest rating agency revenues and receive significantly lower (more stringent) ratings relative to their corporate counterparts. Like structured products, municipal and sovereign issuers are more complex than corporations with audited financial statements, and yet they do not exhibit successful ratings shopping. Thus, our results are more consistent with a conflict of interest in the issuer-paid rating agencies than with the more benign explanation that inflated ratings reflect difficulty in rating opaque issuers. This interpretation is further strengthened by the marked change in the performance of both municipal bonds and structured products issued after Moody‘s went public in the year 2000.
Page 23

Our results contribute to the debate surrounding regulatory reliance on credit ratings that do not reflect absolute credit risk and the associated misallocation of capital. Failure of regulators to distinguish the credit risk associated with single-A rated Asset Backed Securities (ABS exhibited 31.5% default frequency) compared to similarly-rated corporate issues (1.83% default) and municipals (0.49% default) allowed banks, money market and pension funds, and insurance companies to circumvent regulatory safeguards. Equally importantly, failure to distinguish among assets classes at the highest speculative grade (Ba) denies relatively safe municipal issuers access to regulated capital; municipals exhibit 2.26% default frequency at this initial rating level compared to 6.58% of corporates and 67.1% of ABS. Without access to the largest investors in fixed income, taxpayers faced higher interest rates due to higher liquidity premiums associated with thinner markets for speculative grade debt. Meanwhile, corporate issues with higher absolute risk (3.79% default) were certified investment grade (Baa). These results contribute to the debate surrounding municipal bond insurance. If the expected losses among municipalities are continually lower than those of their bond insurers, the value of insurance to taxpayers remains questionable.

Page 24

References Bartram, Sohnke, Greg Brown, and John E. Hund, 2007 Estimating systemic risk in the international financial system, Journal of Financial Economics 86, 835-869. Becker, Bo and Todd Milbourn, 2010, Reputation and competition: evidence from the credit rating industry, Journal of Financial Economics 101(3), 493-514. Blume, Marshall E., Felix Lim, and A. Craig MacKinlay, 1998, The declining credit quality of U.S. corporate debt: Myth or reality?, Journal of Finance 53(4), 1389-1413. Bongaerts, Dion, Martijn Cremers, and William Goetzman, 2010, Tiebreaker: certification in multiple credit ratings, Journal of Finance, forthcoming. Cantor, Richard, and Frank Packer, 1997, Differences of opinion and selection bias in the credit rating industry, Journal of Banking & Finance 21(10), 1395-1417. Cornaggia, Jess, and Kimberly J. Cornaggia, 2011, Does the bond market want informative credit ratings?, Indiana University working paper. Cornaggia, Kimberly, Laurel Franzen, and Timothy Simin, 2011, Manipulating the balance sheet? Implications of off-balance-sheet lease financing, Penn State University working paper. Coval, Joshua, Jakub Jurek, and Erik Stafford, 2009, The economics of structured finance, Journal of Economic Perspectives 23(1), 3–25. Ellul, Andrew, Chotibhak Jotikasthira, and Christian Lundblad, 2010, Regulatory pressure and fire sales in the corporate bond markets, Journal of Financial Economics, forthcoming. Griffin, John M. and Dragon Tang, 2010, Did subjectivity play a role in CDO credit ratings? University of Texas working paper. Ingram, Robert W., Leroy D. Brooks, and Ronald M. Copeland, 1983, The information content of municipal bond rating changes: A note, Journal of Finance 38(3), 997-1003. Ingram, Robert W. and Ronald M. Copeland, 1982, Municipal market measures and reporting practices: An extension, Journal of Accounting Research 20(2), 766-772. Jafry, Yusuf, and Til Schuermann, 2004, Measurement, estimation and comparison of credit migration matrices, Journal of Banking and Finance 28(11), 2603-2639. Jorion, Philippe, Charles Shi, and Sanjian Zhang, 2009, Tightening credit standards: the role of accounting quality, Review of Accounting Studies 14, 123-160.

Page 25

Kraft, Pepa, 2010, Do rating agencies cater? Evidence from rating-based contracts, NYU working paper. Lando, David, and Torben M. Skødeberg, 2002, Analyzing rating transitions and rating drift with continuous observations, Journal of Banking and Finance 26(2), 423-444 Mathis, Jerome, James McAndrews, and Jean-Charles Rochet, 2009, Rating the raters: Are reputation concerns powerful enough to discipline rating agencies?, Journal of Monetary Economics 56, 657-674. McDonald, Michael and Christine Richard, 2008, Insurance drops for municipal debt, undermines MBIA (Update 3), Bloomberg News, March 13. Moody‘s (2002) Special Comment: Moody‘s US municipal bond rating scale, November. Moody‘s (2002b) Special Comment: Understanding Moody‘s corporate bond ratings and rating process, May. Moody‘s (2006) Special Comment: Analyzing the tradeoff between ratings accuracy and stability, September. Moody‘s (2007) Rating Methodology: The U.S. Municipal Bond Rating Scale: Mapping to the Global Rating Scale and assigning Global Ratings to municipal obligations, March. Moody‘s (2008) Sovereign Analytics: Sovereign defaults and interference: Perspectives on government risks, August. Morse, Dale, and Cathy Deely, 1983, Regional differences in municipal bond ratings, Financial Analysts Journal 39 (6), 54-59. Partnoy, Frank, 1999, The Siskel and Ebert of financial markets: two thumbs down for the credit rating agencies, Washington University Law Quarterly 77, 619-712. Peng, Jun, 2002, Do investors look beyond insured triple-A rating? An analysis of Standard & Poor‘s underlying ratings, Public Budgeting and Finance 22, 115-131. Sangiorgi, Francesco, Jonathan Sokobin, and Chester Spatt, 2009, Credit-rating shopping, selection and the equilibrium structure of ratings, working paper. Skreta, Vasiliki, and Laura Veldkamp, 2009, Ratings shopping and asset complexity: A theory of ratings inflation, Journal of Monetary Economics 56, 678-695. Xia, Han, 2010, The Issuer-Pay Rating Model and Rating Inflation: Evidence from Corporate Credit Ratings, University of North Carolina working paper.

Page 26

Appendix A.1. Relevant Sections of Dodd-Frank Wall Street Reform and Consumer Protection Act SEC. 939A. REVIEW OF RELIANCE ON RATINGS. (a) AGENCY REVIEW.—Not later than 1 year after the date of the enactment of this subtitle, each Federal agency shall, to the extent applicable, review— (1) any regulation issued by such agency that requires the use of an assessment of the creditworthiness of a security or money market instrument; and (2) any references to or requirements in such regulations regarding credit ratings. (b) MODIFICATIONS REQUIRED.—Each such agency shall modify any such regulations identified by the review conducted under subsection (a) to remove any reference to or requirement of reliance on credit ratings and to substitute in such regulations such standard of credit-worthiness as each respective agency shall determine as appropriate for such regulations. In making such determination, such agencies shall seek to establish, to the extent feasible, uniform standards of credit-worthiness for use by each such agency, taking into account the entities regulated by each such agency and the purposes for which such entities would rely on such standards of credit-worthiness. (c) REPORT.—Upon conclusion of the review required under subsection (a), each Federal agency shall transmit a report to Congress containing a description of any modification of any regulation such agency made pursuant to subsection (b).

Page 27

A.2. CRA assertion of comparable rating standards across asset classes26 "Standard & Poor‘s strives to make its rating symbols reflect a broadly comparable view of creditworthiness wherever they appear. Standard & Poor‘s believes that maximizing comparability makes Standard & Poor‘s‘ ratings more useful to investors. Thus, when Standard & Poor‘s assigns a given rating symbol to multiple issuers, it intends to connote roughly the same opinion of creditworthiness, irrespective of whether the issuers are a Canadian mining company, a Japanese financial institution, an Illinois school district, a British mortgage-backed security, or a sovereign nation." Deven Sharma, President, Standard & Poor‘s, February 7, 2011

"To meet needs over time, credit ratings have developed important attributes including insightful, robust and independent analysis, symbols that succinctly communicate opinions, and broad coverage across markets, industries and asset classes. These attributes have enabled credit ratings to serve as a point of reference and common language of credit that is used by financial market professionals worldwide to compare risk across jurisdictions, industries and asset classes, thereby facilitating the efficient flow of capital worldwide." Farisa Zarin, Managing Director, Moody‘s Investors Service, February 18, 2011.

"Fitch‘s first and primary goal is that over the longer term, default rates will be broadly similar for like-rated securities across all asset classes. As a secondary goal, Fitch aspires to greater comparability of ratings transition/volatility across asset classes, especially at the highest end of the rating scale." John S. Olert, Chief Credit Officer, Fitch Ratings, March 7, 2011. (Reference is taken from appended ―Ratings Comparability‖ Special Report dated June 21, 2010, and coauthored by Mr. Olert.)

26

Each of these quotes are taken from comment letters found in response to the SEC‘s proposed Credit Rating Standardization: www.sec.gov/comments/4-622/4-622.shtml Page 28

45000 40000 35000 30000 25000 20000 15000 10000 5000
1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010

Structured Financial Sovereign Municipality Corporate

0

Figure 1 Number of issues by asset class through time
This figure displays the number of issues rated by Moody‘s Investors Service every year from 1980 to 2010 partitioned by asset class. The asset classes include tranches of structured products, bonds issued by financial companies (U.S. banks, U.S. bank holding companies, securities companies, and insurance companies), bonds issued by sovereign nations, bonds issued by municipalities, and bonds issued by corporations (industrials and transportation companies). The data come from Moody‘s Default and Recovery Database, and Moody‘s Structured Finance Default Risk Service Database.

Page 29

200

300

400

500

600

700

800

100
0

1000

2000

3000

4000

5000

6000

7000

0

Panel A. Corporate issues

Panel B. Municipal issues
C B C

1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010

B

A

A

Ca

Ba

Ca

Ba

Aa

Aa

Caa

Baa

Caa

Baa

Aaa

Aaa

Page 30

1000 200 400 600 800

1500

2000

2500

3000

4000

1000

1200

3500

500 0

0

Panel C. Sovereign issues

Panel D. Financial issues
C B C A Ca Ba Ca Aa Caa Baa Aaa

1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010

1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010

B

A

Ba

Aa

Caa

Baa

Aaa

Page 31

5000

15000

20000

25000

30000

35000

40000

10000

0

10000

12000

2000

4000

6000

8000

0

Panel E. Structured issues

Panel E.1. Structured issues – Asset Backed Securities
C B C B A Ca Ba Ca Ba Aa Caa Baa Caa Baa Aaa

1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010

1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010

A

Aa

Aaa

Page 32

1000

1500

2000

2500

3000

500 2000 4000 6000 8000 0

10000

12000

Panel E.3. Structured issues – Commercial Mortgage Backed Securities Panel E.2. Structured issues – Collateralized Debt Obligations
1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010

0 C B A Ca Ba

Aa

Caa

Baa

Aaa

C

B

A

Ca

Ba

Aa

Caa

Baa

Aaa

Page 33

6000 5000 4000 3000 2000 1000 0

Aaa
Aa A Baa Ba B Caa Ca C
1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010

Panel E.4. Structured issues – Public Finance
12000
10000 8000 6000 4000 2000 0 Aaa Aa A Baa Ba B Caa Ca C
1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010

Panel E.5. Structured issues – Residential Mortgage Backed Securities Figure 2 Number of issues by initial credit rating through time
Panels A through E display the number of new issues rated by Moody‘s Investors Service every year from 1980 to 2010 for each asset class and initial credit rating. The asset classes include tranches of structured products, bonds issued by financial companies (U.S. banks, U.S. bank holding companies, securities companies, and insurance companies), bonds issued by sovereign nations, bonds issued by municipalities, and bonds issued by corporations (industrials and transportation companies). Panels E.1. through E.5. display the tranches of structured products decomposed into their respective product types: Asset Backed Securities, Collateralized Debt Obligations, Commercial Mortgage Backed Securities, Public Finance, and Residential Mortgage Backed Securities. The data come from Moody‘s Default and Recovery Database, and Moody‘s Structured Finance Default Risk Service Database. The rating scale in this figure is a simplified version of Moody‘s traditional 21-point scale. For example, we lump initial credit ratings of Aa1, Aa2, and Aa3 into one bin, Aa.

Page 34

1000 900 800 700 Corporate Finance Structured Finance Financial Institutions Public, Project, and Infastrcuture Finance

$ Millions

600 500 400

300
200 100 0 2005 2006 2007 2008 2009 2010

Figure 3 Moody’s revenue by asset class through time
This figure displays revenue generated by Moody‘s Investors Service from 2005 to 2010 decomposed by asset class. We collect this information from Moody‘s 10-k filings.

Page 35

12% 10% 8% 6% 4% 2% 0% 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010

Corporate Municipal Sovereign Financial Structured

Figure 4 Percent of outstanding issues that default by asset class through time
This figure displays the percentage of outstanding issues of each asset class that default within calendar years 1980 to 2010. The asset classes include tranches of structured products, bonds issued by financial companies (U.S. banks, U.S. bank holding companies, securities companies, and insurance companies), bonds issued by sovereign nations, bonds issued by municipalities, and bonds issued by corporations (industrials and transportation companies). The data come from Moody‘s Default and Recovery Database, and Moody‘s Structured Finance Default Risk Service Database.

Page 36

1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 C - Caa1 B3 B2 B1 Ba3 Ba2 Ba1 Baa3 Baa2 Baa1 A3 A2 A1 Aa3 Aa2 Aa1 Aaa

Cumulative distribution

Corporate Municipal Sovereign Financial Structured

Randomly assigned ratings

Moody's credit rating

Panel A. Cumulative distributions of default prediction ability by asset class
1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0
C - Caa1 B3 B2 B1 Ba3 Ba2 Ba1 Baa3 Baa2 Baa1 A3 A2 A1 Aa3 Aa2 Aa1 Aaa

Cumulative distribution

ABS

CDO
CMBS RMBS Randomly assigned ratings

Moody's credit rating

Panel B. Cumulative distributions of default prediction ability for structured issues decomposed by product type Figure 5 Cumulative distributions of default prediction ability
Panel A of this figure plots empirical cumulative distributions of default prediction ability for each asset class (corporate bonds; bonds issued by local and regional governments; sovereign bonds; bonds issued by U.S. banks, U.S. bank holding companies, insurance companies, and securities firms; and tranches of structured products). For each asset class we count the number of bonds with a given credit rating as of January 1 of any year of the sample and the number of those bonds that default over the following year. For each credit rating classification, we then divide the full sample count of defaulted bonds by the full sample count of bonds. The figure plots the cumulative sum of these values, moving from the lowest credit rating to the highest. Panel B plots the same for different types of structured products (Asset Backed Securities, Collateralized Debt Obligations, Commercial Mortgage Backed Securities, and Residential Mortgage Backed Securities).

Page 37

Table I Summary Statistics
Panels A through E display summary statistics for debt issues by asset class. The asset classes include bonds issued by corporations (industrials and transportation companies), bonds issued by municipalities, bonds issued by sovereign nations, bonds issued by financial companies (U.S. banks, U.S. bank holding companies, securities companies, and insurance companies), and tranches of structured products. Panels E.1. through E.5. decompose the issues in Panel E by deal type: Asset Backed Securities, Commercial Mortgage Backed Securities, Collateralized Debt Obligations, Public Finance, or Residential Mortgage Backed Securities. Face represents the face value of debt obligations measured in millions of dollars. Maturity represents the number of years between when the debt obligation was issued and when it matures, assuming it does not default. Coupon represents the coupon rate expressed as a percentage. Initial rating is a numerical translation of an obligation‘s first Moody‘s credit rating. The highest credit rating, Aaa, equals 21 and the lowest credit rating, C, equals 1. Downgrade (Upgrade) is a dummy variable taking a value of one if Moody‘s downgrades (upgrades) the issue between the date of issuance and the earlier of the issue‘s maturity date, default date, or the end of the sample, and zero otherwise. Rating change represents the difference between the numerical translation of an issue‘s credit rating when the issue matures, defaults, or the sample ends and the initial rating. Default is a dummy variable taking a value of one if the issue defaults, and zero if it matures or has not defaulted by the end of the sample period. The data come from Moody‘s Default and Recovery Database, and Moody‘s Structured Finance Default Risk Service Database.

Face Maturity Coupon Initial rating Downgrade Upgrade Rating change Default

N 57,931 48,183 35,723 67,809 67,809 67,809 62,020 67,809

Panel A. Corporate issues Mean SD 25% 422.9 1,121.3 51 9.0 8.2 5 7.0 3.2 5.1 12.4 4.5 8 0.3 0.5 0 0.2 0.4 0 -0.9 2.5 -1 0.06 0.24 0 Panel B. Municipal issues Mean SD 25% 286.6 940.4 9 9.5 8.0 4 5.6 2.7 4 18.7 2.9 18 0.1 0.3 0 0.3 0.5 0 0.2 1.7 0 0.00 0.06 0

Median 150 7 6.9 13 0 0 0 0

75% 359 10 9.2 16 1 0 0 0

Face Maturity Coupon Initial rating Downgrade Upgrade Rating change Default

N 6,108 5,869 4,997 6,410 6,410 6,410 6,350 6,410

Median 64 7 5.3 19 0 0 0 0

75% 200 10 7 21 0 1 1 0

Page 38

Face Maturity Coupon Initial rating Downgrade Upgrade Rating change Default

N 10,759 10,697 9,467 11,168 11,168 11,168 11,144 11,168

Panel C. Sovereign issues Mean SD 25% 3,278.9 4,431.9 150 9.2 7.8 4 5.7 3.8 3 16.7 4.5 14 0.1 0.3 0 0.3 0.4 0 0.2 1.6 0 0.01 0.12 0 Panel D. Financial issues Mean SD 25% 370.1 1,385.7 5 7.2 7.6 2 5.0 3.1 3.2 16.8 2.4 16 0.4 0.5 0 0.2 0.4 0 -1.0 3.3 -2 0.03 0.16 0 Panel E. Structured products N Mean SD

Median 822 7 5.6 18 0 0 0 0

75% 5,148 11 8.3 21 0 1 1 0

Face Maturity Coupon Initial rating Downgrade Upgrade Rating change Default

N 30,596 28,523 16,812 33,861 33,861 33,861 33,507 33,861

Median 30 5 5.5 17 0 0 0 0

75% 197.5 10 6.8 18 1 0 0 0

25% 1 0.0 16 0.0

Median 2 50.0 101 92.8

75% 5 100.0 498 100.0

Deal characteristics N tranches % N tranches rated Aaa Face % Face rated Aaa Tranche characteristics Face Maturity Coupon Initial rating Downgrade Upgrade Rating change Default

43,069 43,069 43,069 43,069

4.5 54.0 859.2 65.5

6.4 42.7 3,667.9 43.7

194,978 193,980 0 194,978 194,978 194,978 185,340 194,978

189.8 24.1 -18.7 0.4 0.1 -3.5 0.14

1,007.8 11.1 -3.4 0.5 0.3 5.8 0.35

5 14 -17 0 0 -6 0

18 29 -21 0 0 0 0

65 30 -21 1 0 0 0

Page 39

Panel E.1. Structured products – Asset Backed Securities N Mean SD 25% Median Deal characteristics N tranches % N tranches rated Aaa Face % Face rated Aaa Tranche characteristics Face Maturity Coupon Initial rating Downgrade Upgrade Rating change Default 10,762 10,762 10,762 10,762 5.5 57.3 1,501.1 77.4 7.2 36.2 5,004.7 34.1 1 31.3 188 79.7 3 50.0 482 91.7

75% 7 100.0 967 100.0

59,254 58,521 0 59,254 59,254 59,254 58,888 59,254

272.6 21.7 -18.2 0.4 0.0 -3.2 0.20

1,138.7 11.2 -3.6 0.5 0.2 5.2 0.40

12 9 -16 0 0 -7 0

36 29 -21 0 0 0 0

132 30 -21 1 0 0 0

Panel E.2. Structured products – Collateralized Debt Obligations N Mean SD 25% Median Deal characteristics N tranches 5,494 3.6 3.6 1 2 % N tranches rated Aaa 5,494 35.7 35.7 0.0 28.6 Face 5,494 1,027.6 3,302.3 50 278 % Face rated Aaa 5,494 53.9 41.3 0.0 73.5 Tranche characteristics Face Maturity Coupon Initial rating Downgrade Upgrade Rating change Default

75% 6 50.0 521 90.0

19,862 19,787 0 19,862 19,862 19,862 19,810 19,862

284.2 19.2 -17.2 0.6 0.0 -5.5 0.29

1,336.3 14.5 -3.9 0.5 0.2 6.6 0.45

13 9 -13 0 0 -11 0

30 13 -19 1 0 -3 0

92 34 -21 1 0 0 1

Page 40

Panel E.3. Structured products – Commercial Mortgage Backed Securities N Mean SD 25% Median Deal characteristics N tranches 1,649 9.2 8.0 3 7 % N tranches rated Aaa 1,649 37.2 29.4 20.0 33.3 Face 1,649 3,830.8 8,919.7 284 702 % Face rated Aaa 1,649 65.1 32.9 55.3 76.8 Tranche characteristics Face Maturity Coupon Initial rating Downgrade Upgrade Rating change Default

75% 14 48.0 1,700 88.7

15,189 15,172 0 15,189 15,189 15,189 15,178 15,189

415.9 26.6 -16.1 0.3 0.1 -1.1 0.04

1,645.6 12.5 -4.7 0.5 0.4 3.8 0.19

7 13 -12 0 0 -2 0

24 32 -17 0 0 0 0

102 36 -21 1 0 0 0

Panel E.4. Structured products – Public Finance N Mean SD 25% Deal characteristics N tranches % N tranches rated Aaa Face % Face rated Aaa Tranche characteristics Face Maturity Coupon Initial rating Downgrade Upgrade Rating change Default 17,633 17,633 17,633 17,633 1.9 54.6 33.3 54.4 2.7 49.7 100.5 49.7 1 0.0 8 0.0

Median 1 100.0 14 100.0

75% 2 100.0 29 100.0

34,162 34,121 0 34,162 34,162 34,162 25,034 34,162

17.2 19.7 -20.0 0.2 0.3 -0.7 0

58.4 9.9 -1.5 0.4 0.5 1.9 0

0 12 -19 0 0 -1 0

6 20 -21 0 0 0 0

17 29 -21 0 1 0 0

Page 41

Panel E.5. Structured products – Residential Mortgage Backed Securities N Mean SD 25% Median Deal characteristics N tranches 7,531 8.8 9.0 2 6 % N tranches rated Aaa 7,531 54.6 49.7 0.0 100.0 Face 7,531 1,102.2 3,400.4 165 380 % Face rated Aaa 7,531 83.4 32.9 91.9 97.0 Tranche characteristics Face Maturity Coupon Initial rating Downgrade Upgrade Rating change Default

75% 13 100.0 789 100.0

66,511 66,379 0 66,511 66,511 66,511 66,430 66,511

124.8 29.4 -19.5 0.4 0.0 -4.7 0.15

793.0 7.0 -3.0 0.5 0.2 6.6 0.35

2 29 -20 0 0 -10 0

13 30 -21 0 0 0 0

51 30 -21 1 0 0 0

Page 42

Table II Correlation Matrix
This table displays correlation coefficients for issue characteristics and dummy variables representing asset class. Face represents the face value of debt issues measured in millions of dollars. Maturity represents the number of years between the date of issuance and when the debt it matures, assuming it does not default. Coupon represents the coupon rate expressed as a percentage. Initial rating is a numerical translation of an issue‘s first Moody‘s credit rating. The highest credit rating, Aaa, equals 21 and the lowest credit rating, C, equals 1. Downgrade (Upgrade) is a dummy variable taking a value of one if Moody‘s downgrades (upgrades) the issue between the date of issuance and the earlier of the issue‘s maturity date, default date, or the end of the sample, and zero otherwise. Rating change represents the difference between the numerical translation of an issue‘s credit rating when the issue matures, defaults, or the sample ends and the initial rating. Default is a dummy variable taking a value of one if the issue defaults, and zero if it matures or has not defaulted by the end of the sample period. Corporate is a dummy variable taking a value of one if an industrial or transportation firm issued the bond, and zero otherwise. Municipal is a dummy variable taking a value of one if a municipality issued the bond, and zero otherwise. Sovereign is a dummy variable taking a value of one if a sovereign nation issued the bond, and zero otherwise. Financial is a dummy variable taking a value of one if a U.S. bank, U.S. bank holding company, securities company, or insurance company issued the bond, and zero otherwise. Structured is a dummy variable taking a value of one if the bond is a tranche of a structured product, and zero otherwise. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively. The data come from Moody‘s Default and Recovery Database, and Moody‘s Structured Finance Default Risk Service Database.

Maturity Coupon Initial rating Downgrade Upgrade Rating change Default Corporate Municipal Sovereign Financial Structured

Face -0.01*** -0.01*** -0.01*** 0.00** 0.00** 0.00** 0.00 0.00* 0.00 0.04*** 0.00 -0.01***

Maturity 0.17*** 0.17*** 0.23*** -0.13*** -0.37*** 0.26*** -0.32*** -0.11*** -0.16*** -0.28*** 0.53***

Coupon

Initial rating

Downgrade

Upgrade

Rating change

Default

-0.39*** 0.02*** 0.08*** -0.07*** 0.19*** 0.25*** -0.05*** -0.06*** -0.21*** --

-0.06*** -0.20*** -0.09*** -0.18*** -0.55*** 0.04*** -0.02*** -0.03*** 0.48***

-0.30*** -0.72*** 0.38*** -0.01*** -0.06*** -0.08*** 0.04*** 0.03***

0.31*** -0.09*** 0.11*** 0.06*** 0.06*** 0.08*** -0.18***

-0.54*** 0.15*** 0.07*** 0.10*** 0.09*** -0.24***

-0.07*** -0.04*** -0.05*** -0.08*** 0.14***

Page 43

Table III Default Percentages by Asset Class and Initial Credit Rating
The table displays default percentages for issues by asset class and initial Moody‘s credit rating. The asset classes include bonds issued by corporations (industrials and transportation companies), bonds issued by municipalities, bonds issued by sovereign nations, bonds issued by financial companies (U.S. banks, U.S. bank holding companies, securities companies, and insurance companies), and tranches of structured products. The types of structured products include Asset Backed Securities, Collateralized Debt Obligations, Commercial Mortgage Backed Securities, Public Finance, and Residential Mortgage Backed Securities. The rating scale in this table is a simplified version of Moody‘s traditional 21-point scale. For example, we lump initial credit ratings of Aa1, Aa2, and Aa3 into one bin, Aa. The data come from Moody‘s Default and Recovery Database, and Moody‘s Structured Finance Default Risk Service Database.

Initial rating Aaa Aa A Baa Ba B Caa Ca C

Corporate 0.47% 0.61% 1.83% 3.79% 6.58% 13.95% 17.45% 32.74% 10.00%

Municipal 0.10% 0.06% 0.49% 8.18% 2.26% 0.00% 0.00% 0.00%

Sovereign 0.00% 0.00% 0.00% 2.20% 5.14% 9.99% 6.17% 0.00%

Financial 0.27% 0.10% 4.92% 1.69% 5.04% 9.70% 27.27% 0.00% 0.00%

Structured 3.11% 17.15% 27.21% 39.72% 37.92% 25.97% 14.05% 3.77% 0.00%

Structured decomposed by deal type ABS CDO CMBS PF RMBS 2.05% 22.35% 0.11% 0.00% 3.73% 29.31% 28.94% 0.34% 0.00% 33.51% 31.49% 32.30% 2.79% 0.00% 45.73% 51.83% 36.06% 3.17% 0.00% 48.68% 67.11% 32.97% 8.58% 0.00% 38.64% 28.22% 40.52% 22.10% 35.58% 2.53% 8.33% 51.35% 28.57% 0.00% 33.33% 0.00% 8.00% 0.00% 0.00% 0.00%

Page 44

Table IV Transition Matrices
This table displays five-year transition matrices for issues by asset class. The asset classes include bonds issued by corporations (industrials and transportation companies), bonds issued by municipalities, bonds issued by sovereign nations, bonds issued by financial companies (U.S. banks, U.S. bank holding companies, securities companies, and insurance companies), and tranches of structured products. The types of structured products include Asset Backed Securities, Collateralized Debt Obligations, Commercial Mortgage Backed Securities, Public Finance, and Residential Mortgage Backed Securities. The rating scale in this table is a simplified version of Moody‘s traditional 21-point scale. For example, we lump initial credit ratings of Aa1, Aa2, and Aa3 into one bin, Aa. The vertical access represents the issues‘ initial credit ratings and the horizontal access represents the issues‘ credit ratings five years later. The data come from Moody‘s Default and Recovery Database, and Moody‘s Structured Finance Default Risk Service Database.

Page 45

Initial rating

Aaa Aa A Baa Ba B Caa Ca C Sum

Aaa 1,400 181 44 3 6

Aa 460 2,389 482 33 2 2 2

A 99 1039 6,230 633 56 23 4 1 1

Panel A. Corporate issues Rating after five years Baa Ba B Caa 40 5 5 1 127 24 10 9 1,911 482 103 37 3,918 819 332 124 614 2,303 822 265 116 642 3,294 999 7 8 120 368 2 4 10 5 2 3

Ca 2 11 67 84 527 100 74

C 2 1 25 47 93 36 4 14

Sum 2,010 3,783 9,301 5,954 4,193 5,702 645 100 20 31,708

% Down 30.3% 32.1% 27.4% 23.0% 29.0% 28.4% 21.1% 4.0% 27.5%

% Up 4.8% 5.7% 11.2% 16.0% 13.8% 21.9% 22.0% 30.0% 9.5%

Initial rating

Aaa Aa A Baa Ba B Caa Ca C Sum

Aaa 762 84

Aa 59 625 152 2

A 1 57 357 19

Panel B. Municipal issues Rating after five years Baa Ba B Caa 3 7 26 2 5 56 30 3 3 8 19 5 1 6 14 2 1 1

Ca

C

1 12 3

Sum 825 773 542 113 34 34 2 3 0 2,326

% Down 7.6% 8.3% 6.1% 31.9% 20.6% 41.2% 0.0% 0.0% 9.3%

% Up 10.9% 28.0% 18.6% 23.5% 17.6% 50.0% 0.0% 11.7%

Page 46

Initial rating

Aaa Aa A Baa Ba B Caa Ca C Sum

Aaa 952 255 3 1 1

Aa 247 697 186 1 1 1

A 22 127 480 126 12

Panel C. Sovereign issues Rating after five years Baa Ba B Caa 3 3 1 1 2 2 1 15 5 1 2 230 47 113 1 70 304 53 6 17 64 155 16 1 26 5 1 3 2

Ca 1 1

C

31 35 3

Sum 1,229 1,086 692 518 478 289 32 9 0 4,333

% Down 22.5% 12.3% 3.3% 31.1% 18.8% 17.6% 0.0% 0.0% 17.0%

% Up 23.5% 27.3% 24.5% 17.6% 28.7% 84.4% 66.7% 17.8%

Initial rating

Aaa Aa A Baa Ba B Caa Ca C Sum

Aaa 99 15 2 13 1

Aa 125 2,393 1,650 172 9 2 2 1

A 21 1,416 3,306 448 31 13 1 2

Panel D. Financial issues Rating after five years Baa Ba B Caa 4 1 111 22 11 1 557 376 26 7 386 66 29 14 51 62 79 4 2 10 72 2 1 1 4 1

Ca 2 3 168 20 5 3

C 1 12 107 59 20 4

60

Sum 251 3,983 6,034 1,355 277 110 6 6 64 12,086

% Down 60.6% 39.5% 17.8% 24.8% 44.4% 10.0% 0.0% 0.0% 27.1%

% Up 0.4% 27.4% 46.7% 33.2% 24.5% 33.3% 50.0% 20.1%

Page 47

Initial rating

Aaa Aa A Baa Ba B Caa Ca C Sum

Aaa 29,792 1,822 353 114 13 3 3 1 5

Aa 2,904 6,964 1,327 195 29 21 14 13 22

A 1,486 1,093 4,736 559 80 12 8 3 14

Panel E. Structured issues Rating after five years Baa Ba B Caa 940 740 1,433 2,856 647 328 284 363 816 606 320 362 3,472 756 585 636 167 1,233 192 241 10 62 659 152 3 3 2 47 1 1 1 1

Ca 570 266 269 465 180 75 7 25

C 695 852 833 888 341 179 9 3 39

Sum 41,416 12,619 9,622 7,670 2,476 1,173 96 48 81 75,201

% Down 28.1% 30.4% 33.3% 43.4% 38.5% 34.6% 16.7% 6.3% 31.1%

% Up 14.4% 17.5% 11.3% 11.7% 9.2% 34.4% 41.7% 51.9% 6.5%

Initial rating

Aaa Aa A Baa Ba B Caa Ca C Sum

Aaa 8,215 213 75 36 1

Aa 370 1,510 256 26 3

Panel E.1. Structured issues – Asset Backed Securities Rating after five years A Baa Ba B Caa Ca C 236 233 125 173 120 54 5 273 204 125 120 160 77 248 2,228 496 289 132 159 105 517 95 1,346 382 289 243 178 528 4 14 154 33 42 39 112 1 3 32 12 5 17 4 5 1 1 1 1

Sum 9,531 2,930 4,257 3,123 402 70 9 3 1 20,326

% Down 13.8% 41.2% 39.9% 51.9% 56.2% 48.6% 55.6% 33.3% 30.0%

% Up 7.3% 7.8% 5.0% 5.5% 5.7% 0.0% 33.3% 0.0% 3.6%

Page 48

Initial rating

Aaa Aa A Baa Ba B Caa Ca C Sum

Aaa 1,147 54 7 4 1

Aa 339 579 71 11

Panel E.2. Structured issues – Collateralized Debt Obligations Rating after five years A Baa Ba B Caa Ca C 152 102 105 84 121 135 64 203 123 67 40 89 57 92 472 137 150 63 58 104 110 50 576 161 163 178 133 141 3 14 299 52 127 74 53 35 10 11 25 1 1 1 1 1 2

Sum 2,249 1,304 1,172 1,417 623 81 3 1 3 6,853

% Down 49.0% 51.5% 53.1% 54.8% 49.1% 56.8% 66.7% 0.0% 51.4%

% Up 4.1% 6.7% 4.6% 2.9% 0.0% 0.0% 100.0% 33.3% 3.2%

Initial rating

Aaa Aa A Baa Ba B Caa Ca C Sum

Aaa 1,068 195 92 38 4

Panel E.3. Structured issues – Commercial Mortgage Backed Securities Rating after five years Aa A Baa Ba B Caa Ca C 64 46 26 11 3 16 7 8 339 45 71 35 23 17 4 9 183 408 50 56 56 41 10 20 43 171 590 79 60 95 35 81 4 7 36 437 73 44 18 112 1 1 9 373 109 21 105 1 16 6 2 1

Sum 1,249 738 916 1,192 735 619 25 1 0 5,475

% Down 14.5% 27.6% 25.4% 29.4% 33.6% 38.0% 32.0% 0.0% 26.6%

% Up 26.4% 30.0% 21.1% 6.9% 1.8% 4.0% 0.0% 14.3%

Page 49

Initial rating

Aaa Aa A Baa Ba B Caa Ca C Sum

Aaa 4,403 179 50 10 5 3 1 5

Aa 492 2,252 440 51 27 14 15 16 14

A 153 261 735 38 15 7 9 2 14

Panel E.4. Structured issues – Public Finance Rating after five years Baa Ba B Caa Ca 31 3 1 52 9 13 137 23 1 1 3 47 1 33 19 1 1 12 2

C

29

Sum 5,083 2,753 1,238 261 98 54 46 33 64 9,630

% Down 13.4% 11.7% 1.1% 9.6% 1.0% 0.0% 0.0% 0.0% 10.8%

% Up 6.5% 39.6% 37.9% 51.0% 38.9% 58.7% 63.6% 54.7% 9.6%

Initial rating

Aaa Aa A Baa Ba B Caa Ca C Sum

Aaa 12,274 1,017 121 30 3

Panel E.5. Structured issues – Residential Mortgage Backed Securities Rating after five years Aa A Baa Ba B Caa Ca C 1,075 876 538 495 1,162 2,594 367 617 1,658 250 188 93 99 95 127 503 236 634 94 91 67 102 52 183 61 202 654 69 68 110 111 133 3 47 99 279 22 26 49 59 2 1 7 49 151 15 36 33 1 1 1 1 1 2 1

Sum 19,998 4,030 1,580 1,438 587 294 4 3 1 27,935

% Down 38.6% 33.6% 37.3% 34.1% 26.6% 28.6% 25.0% 66.7% 37.2%

% Up 25.2% 22.6% 20.4% 25.9% 20.1% 50.0% 0.0% 0.0% 6.7%

Page 50

Table V Cox Proportional Hazard Regressions on Credit Rating Adjustments
This table presents regression results from Cox proportional hazards regressions to estimate the relative downgrade and upgrade intensities of bonds by asset class. The asset classes include bonds issued by corporations (industrials and transportation companies), bonds issued by municipalities, bonds issued by sovereign nations, bonds issued by financial companies (U.S. banks, U.S. bank holding companies, securities companies, and insurance companies), and tranches of structured products. The types of structured products include Asset Backed Securities, Collateralized Debt Obligations, Commercial Mortgage Backed Securities, Public Finance, and Residential Mortgage Backed Securities. The coefficients represent the hazard rate of each asset class relative to the baseline hazard of the corporate asset class, and the unit of observation is a rating change. The data come from Moody‘s Default and Recovery Database, and Moody‘s Structured Finance Default Risk Service Database.

Corporate Municipal Sovereign Financial ABS CDO PF CMBS RMBS

Downgrade Full Sample Inv. Grade Spec. Grade 1 1 1 0.446
(0.010)

Upgrade Full Sample Inv. Grade Spec. Grade 1 1 1 1.460
(0.026)

0.399
(0.010)

1.618
(0.074)

1.848
(0.035)

1.434
(0.095)

0.472
(0.008)

0.375
(0.008)

0.868
(0.022)

1.516
(0.021)

1.733
(0.029)

1.563
(0.040)

1.554
(0.010)

1.591
(0.012)

3.299
(0.045)

1.620
(0.015)

2.008
(0.022)

2.085
(0.049)

1.251
(0.007)

0.925
(0.007)

3.307
(0.030)

0.237
(0.004)

0.290
(0.005)

0.246
(0.009)

1.889
(0.013)

1.629
(0.015)

2.784
(0.030)

0.605
(0.010)

0.628
(0.013)

0.724
(0.019)

1.067
(0.009)

1.186
(0.011)

2.778
(0.247)

0.837
(0.011)

1.087
(0.016)

0.402
(0.134)

1.071
(0.011)

0.824
(0.012)

1.624
(0.023)

0.985
(0.015)

1.595
(0.027)

0.265
(0.011)

1.084
(0.006)

0.765
(0.006)

2.973
(0.027)

0.195
(0.003)

0.238
(0.004)

0.203
(0.008)

N

629,222

449,096

180,126

629,222

449,096

180,126

Page 51

Table VI Cox Proportional Hazard Regressions on First Credit Rating Adjustments
This table presents regression results from Cox proportional hazards regressions to estimate the relative downgrade and upgrade intensities of bonds by asset class. The asset classes include bonds issued by corporations (industrials and transportation companies), bonds issued by municipalities, bonds issued by sovereign nations, bonds issued by financial companies (U.S. banks, U.S. bank holding companies, securities companies, and insurance companies), and tranches of structured products. The types of structured products include Asset Backed Securities, Collateralized Debt Obligations, Commercial Mortgage Backed Securities, Public Finance, and Residential Mortgage Backed Securities. The coefficients represent the hazard rate of each asset class relative to the baseline hazard of the corporate asset class, and the unit of observation is the first rating change after issuance. The data come from Moody‘s Default and Recovery Database, and Moody‘s Structured Finance Default Risk Service Database.

Corporate Municipal Sovereign Financial ABS CDO PF CMBS RMBS

Downgrade Full Sample Inv. Grade Spec. Grade 1 1 1 0.337
(0.012)

Upgrade Full Sample Inv. Grade Spec. Grade 1 1 1 1.691
(0.044)

0.349
(0.013)

0.803
(0.112)

2.139
(0.060)

1.750
(0.212)

0.395
(0.010)

0.331
(0.010)

0.848
(0.038)

1.468
(0.032)

1.621
(0.043)

1.848
(0.076)

1.177
(0.014)

1.301
(0.017)

0.534
(0.042)

1.759
(0.027)

2.290
(0.042)

1.121
(0.079)

0.733
(0.007)

0.752
(0.009)

1.830
(0.051)

0.173
(0.004)

0.221
(0.006)

0.149
(0.017)

1.431
(0.017)

1.550
(0.021)

1.264
(0.040)

0.189
(0.008)

0.247
(0.011)

0.114
(0.014)

0.826
(0.011)

0.895
(0.012)

15.437
(10.927)

0.692
(0.014)

0.893
(0.020)

0.000
(0.000)

0.664
(0.019)

0.653
(0.022)

0.808
(0.045)

0.938
(0.034)

1.165
(0.048)

0.713
(0.055)

0.608
(0.006)

0.648
(0.007)

0.793
(0.028)

0.207
(0.004)

0.241
(0.006)

0.529
(0.029)

N

283,668

243,470

40,198

283,668

243,470

40,198

Page 52

Table VII Cox Proportional Hazard Regressions on Credit Rating Adjustments by Time Period
This table presents regression results from Cox proportional hazards regressions to estimate the relative downgrade and upgrade intensities of bonds by asset class over different time periods. The asset classes include bonds issued by corporations (industrials and transportation companies), bonds issued by municipalities, bonds issued by sovereign nations, bonds issued by financial companies (U.S. banks, U.S. bank holding companies, securities companies, and insurance companies), and tranches of structured products. The types of structured products include Asset Backed Securities, Collateralized Debt Obligations, Commercial Mortgage Backed Securities, Public Finance, and Residential Mortgage Backed Securities. The time periods represent the date of the rating change and exclude rating changes that don‘t lie in the specified interval. The coefficients represent the hazard rate of each asset class relative to the baseline hazard of the corporate asset class, and the unit of observation is a rating change. The data come from Moody‘s Default and Recovery Database, and Moody‘s Structured Finance Default Risk Service Database.

Corporate Municipal Sovereign Financial ABS CDO PF CMBS RMBS

Downgrade pre-2000 2000-2006 2007-2010 1 1 1 0.877
(0.028)

Upgrade pre-2000 2000-2006 2007-2010 1 1 1 0.926
(0.037)

0.179
(0.008)

0.567
(0.023)

2.206
(0.047)

0.242
(0.024)

0.605
(0.017)

0.509
(0.012)

0.245
(0.010)

1.201
(0.031)

1.240
(0.030)

2.580
(0.065)

1.038
(0.016)

0.753
(0.009)

2.474
(0.027)

2.402
(0.036)

1.182
(0.019)

1.730
(0.033)

0.327
(0.014)

0.264
(0.004)

1.710
(0.017)

0.268
(0.015)

0.227
(0.006)

0.277
(0.007)

1.533
(0.146)

0.915
(0.018)

2.004
(0.022)

0.808
(0.141)

0.593
(0.022)

0.716
(0.016)

0.782
(0.064)

0.187
(0.006)

1.294
(0.015)

1.865
(0.135)

1.097
(0.025)

0.869
(0.018)

0.756
(0.059)

0.255
(0.008)

1.361
(0.018)

1.416
(0.105)

1.838
(0.038)

0.692
(0.019)

0.163
(0.006)

0.013
(0.001)

1.602
(0.016)

0.488
(0.013)

0.315
(0.008)

0.070
(0.003)

N

111,415

195,306

322,390

111,415

195,306

322,390

Page 53

Table VIII Credit Rating Adjustment Regressions
This table displays regression results from measures of credit rating adjustment regressed on asset class dummy variables. The asset classes include bonds issued by corporations (industrials and transportation companies), bonds issued by municipalities, bonds issued by sovereign nations, bonds issued by financial companies (U.S. banks, U.S. bank holding companies, securities companies, and insurance companies), and tranches of structured products. Panel A displays results from OLS regressions. The dependent variable in Panel A is Rating change, the difference between the numerical translation of an issue‘s credit rating when the issue matures, defaults, or the sample ends and the initial rating. The highest credit rating, Aaa, equals 21 and the lowest credit rating, C, equals 1. Panel B displays results from probit regressions. The dependent variable in Panel B is a dummy variable taking a value of one if the issue‘s first credit rating was investment grade but Moody‘s downgraded the issue to speculative grade by the earlier of the issue‘s maturity date, default date, or the end of the sample, and zero otherwise All of the issues in Panel B had initial credit ratings of investment grade. We cluster the standard errors at the issuer level. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively. The data come from Moody‘s Default and Recovery Database, and Moody‘s Structured Finance Default Risk Service Database.

Panel A. Dependent variable is Rating change Full sample Municipal Sovereign Financial Structured Constant Year FE? Adj R2 N 1.291
(0.229)***

Issued 1980 - 1989 -0.549
(0.745)

Issued 1990 - 1999 1.421
(0.206)***

Issued 2000 - 2006 1.273
(0.147)***

Issued 2007 - 2010 0.335
(0.078)***

0.825
(0.162)***

1.265
(0.325)***

1.180
(0.227)***

0.790
(0.191)***

0.862
(0.208)***

0.215
(0.487)

-0.544
(0.230)**

1.117
(0.267)***

-0.700
(0.557)

-1.283
(0.757)*

-1.364
(0.318)***

1.279
(0.146)***

0.887
(0.139)***

-3.183
(0.777)***

-4.960
(2.304)**

-1.668
(0.359)***

-1.246
(0.146)***

-1.115
(0.132)***

-0.573
(0.049)***

-0.525
(0.050)***

Yes 0.26 300,038

No 0.06 13,146

No 0.03 73,891

No 0.07 157,140

No 0.13 55,861

Page 54

Panel B. Dependent variable is a dummy variable taking a value of one if Moody‘s downgrades the issue to speculative grade Full sample Municipal Sovereign Financial Structured Constant Psuedo R2 N -0.975
(0.194)***

Issued 1980 - 1989 0.139
(0.338)

Issued 1990 - 1999 -1.496
(0.178)***

Issued 2000 - 2006 -1.297
(0.155)***

Issued 2007 - 2010

-0.644
(0.267)**

-1.792
(0.326)***

-0.820
(0.228)***

-0.258
(0.418)

-0.096
(0.282)

-0.294
(0.116)**

-0.520
(0.217)**

0.225
(0.379)

0.535
(0.421)

0.614
(0.249)**

-1.254
(0.074)***

-0.760
(0.124)***

0.843
(0.235)***

1.696
(0.515)***

-1.331
(0.064)***

-1.226
(0.074)***

-1.144
(0.081)***

-1.479
(0.074)***

-2.082
(0.090)***

0.05 274,612

0.09 11,268

0.07 63,635

0.04 146,210

0.09 50,804

Page 55

Table IX Full Sample Credit Rating Adjustment Regressions with Individual Structured Products
This table displays regression results from OLS regressions with Rating change regressed on asset class dummy variables. Rating change is the difference between the numerical translation of an issue‘s credit rating when the issue matures, defaults, or the sample ends and the initial rating. The highest credit rating, Aaa, equals 21 and the lowest credit rating, C, equals 1. The asset classes include bonds issued by corporations (industrials and transportation companies), bonds issued by municipalities, bonds issued by sovereign nations, bonds issued by financial companies (U.S. banks, U.S. bank holding companies, securities companies, and insurance companies), and tranches of structured products. The types of structured products include Asset Backed Securities, Collateralized Debt Obligations, Commercial Mortgage Backed Securities, Public Finance, and Residential Mortgage Backed Securities. We cluster the standard errors at the issuer level. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively. The data come from Moody‘s Default and Recovery Database, and Moody‘s Structured Finance Default Risk Service Database.

Municipal Sovereign Financial ABS CDO CMBS PF RMBS Constant Year FE? Adj R2 N

1.128
(0.161)***

1.071
(0.134)***

1.066
(0.129)***

1.049
(0.125)***

1.300
(0.247)***

0.861
(0.137)***

0.908
(0.121)***

0.920
(0.121)***

0.928
(0.119)***

0.819
(0.171)***

-0.019
(0.414)

-0.101
(0.374)

-0.118
(0.371)

-0.148
(0.362)

0.219
(0.505)

-1.829
(0.188)***

-4.218
(0.444)***

-0.115
(0.112)

0.178
(0.087)**

-2.703
(0.417)***

-1.044
(0.145)***

-0.927
(0.080)***

-0.878
(0.067)***

-0.865
(0.061)***

-1.365
(0.306)***

Yes 0.19 172,326

Yes 0.22 132,918

Yes 0.05 128,271

Yes 0.05 138,339

Yes 0.34 179,462

Page 56

Table X Credit Rating Adjustment Regressions with Combined Structured Products by Time Period
This table displays regression results from OLS regressions with Rating change regressed on asset class dummy variables. Rating change is the difference between the numerical translation of an issue‘s credit rating when the issue matures, defaults, or the sample ends and the initial rating. The highest credit rating, Aaa, equals 21 and the lowest credit rating, C, equals 1. The asset classes include bonds issued by corporations (industrials and transportation companies), bonds issued by municipalities, bonds issued by sovereign nations, bonds issued by financial companies (U.S. banks, U.S. bank holding companies, securities companies, and insurance companies), and tranches of structured products. The types of structured products include Asset Backed Securities, Collateralized Debt Obligations, Commercial Mortgage Backed Securities, Public Finance, and Residential Mortgage Backed Securities. We cluster the standard errors at the issuer level. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively. The data come from Moody‘s Default and Recovery Database, and Moody‘s Structured Finance Default Risk Service Database.

Full sample Municipal Sovereign Financial ABS CDO CMBS PF RMBS Constant Year FE? Adj R2 N 1.325
(0.239)***

Issued 1980 - 1989 -0.548
(0.745)

Issued 1990 - 1999 1.419
(0.206)***

Issued 2000 - 2006 1.273
(0.147)***

Issued 2007 - 2010 0.290
(0.078)***

0.824
(0.169)***

1.265
(0.325)***

1.178
(0.227)***

0.790
(0.191)***

0.817
(0.208)***

0.242
(0.490)

-0.544
(0.230)**

1.114
(0.266)***

-0.700
(0.557)

-1.329
(0.758)*

-1.245
(0.262)***

1.163
(0.147)***

0.382
(0.140)***

-2.908
(0.403)***

-4.580
(1.487)***

-2.629
(0.404)***

1.247
(0.145)***

-0.728
(0.416)*

-4.701
(0.458)***

-6.244
(1.302)***

1.078
(0.468)**

-0.812
(0.145)***

2.547
(0.184)***

-0.235
(0.148)

-2.906
(0.950)***

1.052
(0.470)**

0.812
(0.145)***

1.106
(0.132)***

-0.084
(0.049)*

-0.630
(0.050)***

-2.683
(0.376)***

1.324
(0.145)***

1.220
(0.133)***

-4.743
(0.682)***

-9.377
(2.042)***

-1.651
(0.367)***

-1.247
(0.145)***

-1.113
(0.132)***

-0.572
(0.049)***

-0.479
(0.050)***

Yes 0.31 300,038

No 0.06 13,146

No 0.06 73,891

No 0.15 157,140

No 0.33 55,861

Page 57

Sponsor Documents

Or use your account on DocShare.tips

Hide

Forgot your password?

Or register your new account on DocShare.tips

Hide

Lost your password? Please enter your email address. You will receive a link to create a new password.

Back to log-in

Close