0 Street earnings definitions of credit rating agencies ...

4 downloads 141 Views 607KB Size Report
difference between the street earnings of credit rating agencies and equity analysts are ... conservatism incentives of credit rating agencies than those of equity ...
Street earnings definitions of credit rating agencies and equity analysts

George Batta Claremont McKenna College [email protected] Volkan Muslu The University of Texas at Dallas [email protected]

September 2010

Abstract We compare street earnings issued by credit rating agencies with those issued by equity analysts. We find that the street earnings of credit rating agencies are lower in level, higher in sensitivity to contemporaneous negative news, and less useful in predicting future year‘s earnings. The difference between the street earnings of credit rating agencies and equity analysts are greater when the underlying company stocks are more volatile, the company‘s bonds are rated as speculative, and outstanding levels of company debt are high. Market participants perceive the street earnings differences between rating agencies and equity analysts as informative, as corporate bond spreads are higher when the difference is greater. Our evidence indicates greater conservatism incentives of credit rating agencies than those of equity analysts, at the expense of diminished earnings predictability.

JEL Classification: G24; G17. Key Words: Analysts; Debt ratings; Earnings forecasts; Street earnings; Conservatism.

We acknowledge Moody‘s and Thomson Financial Services Inc. for providing earnings per share forecast data as part of a broad academic program to encourage earnings expectation research. 

0

1. INTRODUCTION In an effort to portray companies‘ earnings potentials, information intermediaries define street earnings by selectively adjusting for non-recurring and other components of companyreported GAAP earnings. Considerable research has documented that the street earnings definitions of equity analysts explain future earnings better than GAAP earnings do, pointing to equity analysts‘ ability in processing earnings information—notwithstanding their opportunistic incentives to promote company stocks (Gu and Chen, 2004; Bhattacharya et al., 2003; Bradshaw and Sloan, 2002; Johnson and Schwartz, 2005).1 While the credit rating agencies are mostly known for their ratings about the default risk of corporate debt, the agencies also present assessments of recurring earnings potentials of their client companies. In these reports, which are primarily prepared for institutional investors, the agencies disclose how they work out the ‗as reported‘ quarterly financials to compute the ‗as adjusted‘ or ‗street‘ financials. The street earnings definitions of credit rating agencies serve as key inputs into agencies‘ own ratings, and studies have shown that the adjusted financials better explain variation in credit spreads than reported financials (Kraft, 2009; Batta, Ganguly, and Rosett, 2010). Despite the significant role of credit rating agencies in debt markets, the literature is silent on the agencies‘ street earnings definitions and how these definitions compare with those of equity analysts. In this paper, we attempt to fill this gap. Massive accounting and governance scandals in the recent decade turned the spotlight on the companies‘ financial reporting practices, as well as on the information intermediaries, notably investment banks, brokerage houses, and rating agencies. Sell-side equity analysts are known to issue optimistic stock recommendations because of conflicts of interest, i.e., their

1

Doyle, Lundholm, and Soliman (2003) and Baik, Farber, and Petroni (2010) show that managers and analysts opportunistically defining street earnings, especially when analysts have greater incentives to do so.

1

incentives to generate business or curry favour with company management (Dugar and Nathan, 1995; Lin and McNichols, 1998; Ertimur, Zhang, and Muslu, 2010). Likewise, many observers have suggested that rating agencies positively bias their ratings on corporate debt or structured products like mortgage-backed securities and CDOs (Lynch, 2009; Riddiough and Zhu, 2010), either owing to conflicts of interest or because of putative investor demand for a high standard of justification for ratings downgrades (Altman and Rijken, 2006). Contrasting their incentives to generate optimistic ratings, the credit rating agencies (CRAs, henceforth) also have incentives to be conservative for the very reason that the default of a positively rated corporate debt will hamper the CRA‘s credibility. Such reputational incentives come not only from market forces but also from regulators. The Securities and Exchange Commission (SEC) registers some CRAs as ‗Nationally Recognized Statistical Rating Organizations (NRSRO)‘ based on factors including past integrity and performance. Such status is vital in the industry, because the users of ratings data such as the institutional investors, regulators, and bond issuers themselves demand that bond ratings come from an NRSRO (Beaver, Shakespeare, and Soliman, 2006). The conservatism incentives for CRAs are likely to be higher than those for equity analysts. This is because CRAs focus on the credit risk of their client companies, and stand to lose more credibility and business if their optimism proves inaccurate. In other words, in addition to their investment advisory role, which they share with equity analysts, CRAs also assume a certification role about corporate debt—unlike equity analysts. The certification role results in their greater asymmetric loss function compared to that of equity analysts. We assess how these differing incentives for optimism and conservatism manifest themselves in differences in the information output of equity and CRA analysts. To do so, we

2

assess one dimension of their respective outputs that can be readily compared, namely, their street earnings definitions. We predict that incentives for conservatism should dominate for CRA analysts, which would produce street earnings definitions that (a) are lower on average, (b) are lower under conditions of high uncertainty about firm outcomes, and (c) incorporate negative economic news in a more timely manner, relative to equity analyst street earnings. We test our predictions using a sample of street earnings information from the quarterly reports of Moody‘s Inc., one of the two largest CRAs globally. Our sample spans years 20042008 and covers all industrial U.S. public companies that had outstanding Moody‘s ratings in May 2008. We also obtain I/B/E/S-reported Actual earnings, which represent street earnings definitions of equity analysts. Our empirical analyses are based on comparisons of the two street earnings samples at the company and fiscal quarter level. In our first set of tests, we show that the street earnings definitions of Moody‘s analysts are, on average, 20% lower than I/B/E/S earnings definitions. At the same time, Moody‘s earnings definitions more strongly incorporate bad news, proxied by negative contemporaneous stock returns. In our second set of tests, we cross-sectionally check our ‗conservatism‘ interpretation. We show that the gap between the street earnings definitions of Moody‘s and equity analysts is greater when the underlying company bonds are rated as speculative, company stocks are more volatile, and outstanding levels of company debt are relatively high in the sample. In our final set of tests, we assess the valuation consequences of Moody‘s greater street earnings conservatism. We show that corporate bond credit spreads increase with the gap between the street earnings definitions of rating agencies and equity analysts, suggesting that investors find the gap informative about the credit risk of the underlying companies. The empirical results are overall consistent with our predictions that credit rating agency analysts are

3

more conservative in defining street earnings than the equity analysts; this gap is stronger when the downside risk for the agencies is more apparent. Our findings offer several insights on the economics of financial intermediation. As information intermediaries, rating agencies and equity analysts attempt to reduce the information asymmetry between investors and companies. The literature devotes particular attention to examining the outputs of equity analysts in the form of stock recommendations and earnings forecasts (Ramnath, Rock, and Shane, 2008). Researchers have also examined the effect of rating changes and announcements of ratings reviews and watchlists on security and credit derivative valuation (Hand, Holthausen, and Leftwich, 1992; Dichev and Piotroski, 2001; Hull, Predscu, and White, 2004), yet have devoted less attention to the other, more regular research output of credit rating agencies. This study thus helps to highlight an additional means by which credit rating agencies help reduce information asymmetries, in their role as information intermediaries who are more critical of the companies‘ going concern positions than equity analysts. We also contribute to the research on the role of accounting conservatism in debt contracting.

Positive accounting theory suggests that accounting conservatism, defined as

asymmetric verification standards for losses versus gains, enhances efficiency in the debt contracting process (Watts and Zimmerman, 1986; Watts, 2003). Beatty, Weber, and Yu (2008) find that private debt contracts modify accounting numbers for greater conservatism, and the evidence of rating analysts‘ adjustments in this paper can be seen as another version of conservative contract modifications. An alternative conservatism argument predicts that CRAs faced with asymmetric loss functions would generate systematically lower \ratings, rather than lower their street earnings

4

definitions. The rationale of this argument is that analysts want to use the best possible inputs into their ratings model. Therefore street earnings definitions should not show any bias, even if the ultimate ratings are shaded downwards. Our empirical evidence is inconsistent with this alternative story. Instead, our results are consistent with a scenario in which street earnings definitions of CRAs serve as readily observable indicators of conservatism. That is, agencies can point to these inputs as evidence of their caution in the event of an increase in the credit risk of a highly-rated company. It might be more challenging to do so by pointing to the rating itself, which is the product of many inputs, both observable (like street earnings) and unobservable. This interpretation complements Kraft (2009), who documents that rating agencies make ―soft adjustments‖ to their rating models, to incorporate more qualitative factors, and which on average produce published ratings that are more optimistic than ratings that would be predicted by ―harder‖ inputs. Our results are also consistent with a scenario in which the CRA‘s are optimistic in ratings levels—owing either to the aforementioned high standard of justification for downgrades or to efforts to please management—but reveal their conservative outlook in their inputs into the ratings. Less visible, yet observable, street earnings definitions will convey the agencies‘ more reserved views of the firm‘s credit quality, at least to sophisticated parties. This argument is analogous to findings that equity analysts strategically bias their stock recommendations upwards, but keep their earnings forecasts less biased for consumption by more sophisticated investors (Malmendier and Shantikumar, 2009; Ertimur, Zhang, and Muslu, 2010). We also note a competing story for our evidence on conservative street earnings definitions of credit rating agencies is that agency analysts may have privileged access to information given that CRA analysts are not constrained by Regulation FD, and street earnings

5

of CRA analysts on average reflect some of this privileged information. This explanation is unlikely, as our analyses show that CRA street earnings definitions are inferior to equity analysts‘ street earnings definitions in predicting one-year-ahead company earnings and cash flows. We organize our paper as follows. The next section describes the related literature and develops our hypotheses. Section 3 describes our sample selection, and Section 4 provides empirical tests. Section 5 concludes.

2. RELATED LITERATURE AND HYPOTHESES DEVELOPMENT 2.1. Conservatism in street earnings definitions Corporate debt securities constitute an important capital market. The total U.S. debt market capitalization amounts to $6.9 trillion in 2009, versus total U.S. equity market capitalization of $11.7 trillion.2 The credit rating agencies serve two critical functions in debt markets. First, they rank the default risk of debt securities, helping investors to assess risk and make informed investment decisions (investment advisory role). Second, the agencies certify the securities as investment versus non-investment-grade, and, by doing so, help in regulatory oversight, portfolio governance, and private credit arrangements (certification role). As a result of accounting and corporate governance scandals in the recent decade, many observers have suggested that, rating agencies positively bias their ratings (Lynch, 2009). The turmoil in equity and credit markets in 2008 have also turned the spotlight on rating agencies‘ incentives to curry favour with the rated companies who pay them, although the focus has been on overly optimistic ratings on structured products like asset-backed securities and insurance products (Riddiough and Zhu, 2010). The agencies‘ incentives to issue optimistic ratings are 2

Source: Securities Industry and Financial Markets Association (SIFMA) and Wilshire Associates.

6

similar to equity analysts‘ incentives to issue optimistic stock recommendations (Dugar and Nathan, 1995; Lin and McNichols, 1998), especially since issuers pay to have their securities rated by the agencies. Additionally, ratings may be positively biased because agencies have a stated goal of trying to ensure that declines in credit quality will be permanent before agencies lower a rating. Given that institutional investors can hold only very limited amounts of speculative grade debt, downgrades require costly portfolio rebalancing, so agencies wish to avoid debt downgrades that are soon followed by upgrades. Such demand for less volatile rankings may also come from financial regulators seeking to avoid procyclical capital requirements for banks whose capital requirements are tied to asset risk, which in turn is tied to ratings (Loffler, 2004). This phenomenon known as ―through the cycle ratings‖ cause the ratings to be sticky and result in agencies slowly incorporating (especially negative) information into their ratings (Altman and Rijken, 2006). Countering these tendencies towards the production of optimistic ratings and research are agencies‘ incentives to produce more conservative research. Positive accounting theory predicts that different constituents demand conservative financial reporting for different reasons (Watts, 2003). Regulators are conservative due to political considerations, as they are held responsible for failing to prevent or mitigate large investor losses. Institutional investors are conservative due to their fiduciary duties to make prudent investments (Del Guercio, 1996; Gompers and Metric, 2001). Consistent with prior literature (Watts, 2003; Holthausen and Leftwich, 1986), we then expect that rating agencies have incentives to be conservative because of their certification role for both regulators and institutional investors. Specifically, the possible default of a positively rated client company or product hampers the credibility of the CRA in question.

7

Furthermore, the Securities and Exchange Commission (SEC) gives some CRAs ‗Nationally Recognized Statistical Rating Organizations (NRSRO)‘ status based on factors including past integrity and performance. 3 Such status is vital in the industry, because the users of ratings data such as the institutional investors, regulators, and corporate bond issuers themselves demand that bond ratings come from a rating agency that is a NRSRO (Beaver, Shakespeare, and Soliman, 2006). We predict that this certification role in particular will prompt agency analysts to produce more conservative research than equity analysts. Sell-side equity analysts also have incentives to be conservative, because they advise institutional investors for their investments (Hugon and Muslu, 2010). However, unlike bond rating analysts, equity analysts do not certify the downside risk of the companies they follow. Furthermore, equity analysts‘ incentives to be conservative are constrained due to following reasons that do not necessarily follow for rating analysts: 1) the desire to win investment-banking business, 2) incentives to generate trading commissions, and 3) appease investors long in the stock (Lin and McNichols, 1998). As a result, we expect that rating analysts have greater asymmetric loss functions, prompting them to have more pessimistic views about company prospects and to expect negative outcomes. Our expectation of greater conservatism of rating analysts over equity analysts is consistent with Beaver, Shakespeare, and Soliman (2006), who find that NRSRO rating analysts are more conservative than non-NRSRO rating analysts due to NRSRO analysts‘ greater certification role.

3

The SEC gives some CRAs ―Nationally Recognized Statistical Rating Organizations (NRSRO)‖ status based on the rating agencies‘ market position and past performance The Credit Rating Agency Reform Act of 2006 provides clear guidelines for NRSRO qualifications. The CRAs that are currently registered as NRSRO‘s for corporate debt issues are LACE Financial Corporation, A.M. Best Company, DBRS, Egan-Jones Company, Japan Credit Rating Agency, Rating and Investment Information, Fitch, Moody‘s Investor Service, and Standard and Poor‘s. The ratings industry is more concentrated than the brokerage industry in equity markets. Moody‘s and Standard and Poor‘s are the largest and oldest rating agencies, and they are the only two agencies rating almost all corporate bond issues. Fitch is the third largest, rating about half of the bond issues (Bongaerts, Cremers, and Goetzmann, 2009).

8

We also expect that the greater conservatism of rating agencies will primarily manifest itself in the agency-prepared inputs into the ratings. While there exist many inputs into agency ratings, many are either unobservable or are difficult to benchmark in order to assess relative optimism or pessimism. In contrast, street earnings definitions are both observable and can be benchmarked against GAAP earnings or equity analyst street earnings definitions in the event of a sudden decline in credit quality. Additionally, similar to how equity analysts strategically bias their recommendations upwards due to misaligned incentives, but keep their forecasts less biased for consumption by more sophisticated investors (Malmendier and Shantikumar, 2009; Ertimur, Zhang, and Muslu, 2010), rating inputs like street earnings can be used to convey the agencies‘ finer views of the firm‘s credit quality at least to these sophisticated parties; ratings themselves may be shaded upwards in order to please company management or to achieve the ratings stability objective described above. The above discussion serves as a basis for our first hypothesis. Hypothesis 1a: Street earnings definitions of credit rating analysts are lower than those of equity analysts. Hypothesis 1b: Compared to those of equity analysts, street earnings definitions of credit rating analysts are timelier with respect to negative news. 2.2. Uncertainty and conservatism in street earnings definitions One corollary of Hypothesis 1 is that, facing an asymmetric loss function, rating agency analysts will be more conservative where there is more uncertainty about the credit risk outcomes. Facing significant reputation loss for failing to predict a credit risk downturn, agency analysts choose to err on the side of greater conservatism. This uncertainty about credit risk comes in two forms. First is a general uncertainty over firm value and earnings realizations. High stock return volatility and equity analyst forecast dispersion proxy for this type of

9

uncertainty. Second is the uncertainty on company‘s prospects as going concern. Low bond ratings and high indebtedness are proxies for this type of uncertainty. The above discussion serves as a basis for our second hypothesis.

Hypothesis 2: The gap between the street earnings definitions of credit rating and equity analysts increases with greater uncertainty.

2.3. Consequences of conservatism in street earnings definitions We also investigate whether investors factor in differential conservatism in the street earnings definitions of credit rating and equity analysts. Given that rating analysts specifically focus on creditworthiness of bond issues, the incremental conservatism of rating analysts should indicate reduced debt-paying ability of the companies in question. If the street earnings definitions of credit rating analysts are informative and convey agencies‘ more nuanced views of credit quality, then investors should react by requiring differential rate of returns to the differences in the street earnings definitions of the two sets of analysts. The above discussion serves as a basis for our third hypothesis. H3: Credit spreads increase with the gap between the street earnings definitions of credit rating and equity analysts.

2.4. Predictability of future earnings in street earnings definitions Rating agency analysts possess an informational advantage over equity analysts, because the Regulation Fair Disclosure (Reg FD), which is enacted in 2000, does not extend strictures on privileged communication of managerial information to rating agencies. Jorion, Liu, and Shi (2005) confirm that credit rating changes generate greater stock returns after Reg FD, suggesting privileged information offered by rating analysts. Given that rating analysts have access to privileged information, they may be able to use that information to better identify nonrecurring 10

items and, thus, create street earnings definitions that better predict future earnings and cash flows. On the other hand, our previous hypotheses predict that rating analysts choose to be conservative in their street earnings definitions, regardless of their informational advantage. Such conservative bias and focus on the credit risk should reduce rating analysts‘ ability to predict future company earnings and cash flows, because rating analysts‘ street earnings definitions will be weighed down by more transitory loss items, or else fail to reflect more persistent gain items. The above discussion, with opposite predictions, serves as a basis for our fourth hypothesis.

H4: Street earnings definitions of credit rating analysts better predict future company earnings and cash flows than street earnings definitions of equity analysts.

3. SAMPLE SELECTION AND RESEARCH DESIGN 3.1. Sample selection Our empirical tests are based upon comparisons of company-reported GAAP earnings, Moody‘s street earnings, and equity analysts‘ street earnings. We obtained information on quarterly reported and adjusted financial numbers from Moody‘s Financial Metrics, a division of Moody‘s Inc. Moody‘s Financial Metrics provides ‗as reported‘ and ‗as adjusted‘ financial information on nearly all Moody‘s-rated industrial entities, and provides their original worksheets with detailed company-level information on all Moody‘s-rated clients.4 For all industrial U.S. companies that had Moody‘s ratings outstanding in May 2008, we obtain quarterly financial data from the first quarter of 2004 to the first quarter of 2008. Our initial dataset comprises 9,315 firm-quarters from 1,590 firms.

4

According to Moody‘s representatives, in rare cases, a rated company may not be included in the database if the analysis sheets contain material non-public information that cannot be disaggregated from the larger set of data contained in the analysis worksheets.

11

Moody‘s adjusts GAAP earnings for a number of different items. The prominent of the adjustments include adjustments for unusual and non-recurring income statement items; those incorporating changes in the fair value of pension plan assets, imputed interest expense on the projected pension benefit obligation, and the pension service cost; expensing interest capitalized during the period; recognition of the fair value of stock-based compensation, at least prior to the implementation of SFAS No. 123(R); and the recharacterization of preferred dividends as interest expense for hybrid debt securities.5 Financial Metrics reports the after-tax net income effect of all these adjustments in the ‗Unusual & Non-recurring Items-Adjustment, After-tax Adjustment‘ line item. To develop a measure of Moody‘s street earnings, we add this line item to the ‗Reported Net Profit After-tax Before Unusual Items‘6, which is equivalent to earnings before extraordinary items and discontinued operations (ibq) in the Compustat Fundamentals Quarterly file. We label the resulting street earnings figure as Moody‘sTOT. Because equity analysts generally adjust for only unusual or non-recurring income statement items, we also compute a Moody‘s street earnings specification that only includes these adjustments to net income. We label this alternative street earnings definition as Moody‘sIS. For the I/B/E/S street earnings definition, we obtain the I/B/E/S ‗Actual Unadjusted fiscal quarter‘ figures. The I/B/E/S Actuals (hereafter, IBES) represent each company‘s reported

5

Moody‘s also adjusts for the effect of capitalizing operating leases obligations; however, their implementation only involves an above-the-line shift, lowering operating income and increasing interest expense, rather than a net income effect. Moody‘s also adjusts income for ―non-standard‖ adjustments based on public information, which are ad hoc or new categories of adjustments that do not fit into the standard categorization scheme. Examples include reclassifying minority interest expense as interest expense on General Mills, Inc.‘s financials for the year ended May 31, 2007 or recharacterizing cost of cost of goods sold as depreciation and amortization expense for MagnaChip Semiconductor, L.L.C.‘s financials for the year ending December 31, 2006. In most cases, non-standard adjustments do not result in a net income effect, although they do affect operating profits and interest expense. Finally, Moody‘s imputes interest expense on deemed financing from securitizing assets, which again has no bottom-line effect. 6 This amount excludes any quarterly discontinued and extraordinary items.

12

results adjusted for non-recurring items, discontinued operations, and extraordinary items as defined by the majority of equity analysts following a firm. To convert these estimates to dollar value amounts, we multiply by either the number of diluted earnings per share from Compustat for the relevant fiscal quarter or number of common shares for basic EPS.7 To measure company-reported GAAP numbers, we use the ‗Reported Net Profit After-tax Before Unusual Items‘ figure in the Financial Metrics database (hereafter, GAAP). We then deflate Moody‘sTOT, Moody‘sIS, IBES, GAAP earnings figures by the average number of basic shares outstanding (cshprq in Compustat Fundamentals Quarterly) multiplied by the price per share as of the first day of the fiscal quarter from CRSP. The final dataset encompasses deflated street earnings definitions of 3,860 firm-quarters, representing 841 firms. The final dataset represents more than 50% data attrition from the initial dataset from Financial Metrics due to data requirements about equity analysts and company-reported earnings.

3.2. Research Design Hypothesis 1a We examine whether the mean difference between Moody‘sTOT (or Moody‘sIS) and IBES, is negative using the following model: Moody’sk, it - IBESit = α1 + εit where k=TOT or IS, and with standard errors clustered at both the firm and time period level (Cameron, Gelbach, and Miller 2010). To test Hypothesis 1a, we assess whether α1