Credit Ratings and the Cost of Issuing Seasoned Equity† Garrett A

Credit Ratings and the Cost of Issuing Seasoned Equity†

Garrett A. McBrayer*

Abstract

This study examines the effects of issuer credit ratings on the underwriting costs incurred in seasoned equity offerings (SEOs). The evidence from a panel of common stock SEOs from 1990-2014 shows that when firms issue seasoned equity, those with issuer credit ratings pay significantly reduced investment banking fees. Recognizing the potential endogeneity, I confirm these results by conducting a matched sample analysis of firms who obtain new, long-term issuer credit ratings to an unrated control group. Controlling for factors identified in prior literature as determinants of SEO issuance and of SEO fees, I find that firms who obtain a new credit rating are more likely to issue seasoned equity and pay reduced investment banking fees in doing so. The findings suggest that firms can reduce their costs to issuing equity by actively seeking to mitigate issues of value uncertainty prior to a seasoned equity offering, i.e., obtaining a credit rating prior to initiating a SEO.

JEL Classification: D82, G14, G24

Keywords: credit ratings, seasoned equity offerings, information asymmetry

Draft Version: May 2017 v5.5

†I am grateful for the invaluable comments and suggestions of Josh Filzen, Wayne Lee, Alexey Malakhov, Tim Yeager, Pu Liu, as well as to the seminar participants at the University of Arkansas and Boise State University.

*Assistant Professor of Finance, College of Business and Economics, Boise State University, Boise, ID 83725. [email protected].

Credit Ratings and the Costs of Issuing Seasoned Equity

1. Introduction

Information asymmetries pertaining to the valuation of a firm’s assets have direct effects on the risks inherent in investing in the firm’s financial claims. In secondary equity markets, the adverse selection costs associated with an investment in a firm’s equity are a positive function of information asymmetry (Benston and Hagerman, 1974; Kyle, 1985; Glosten and Harris, 1988; Stoll, 1989; Huang and Stoll, 1997; Lin, Sanger, and Booth, 1995; among others). Investors account for the potential losses faced when trading with an information-motivated trader. In primary markets, the observed underpricing of initial public offerings (IPOs) has been attributed, in part, to information asymmetry. Rock (1986) argues that the observed underpricing results from a winner’s curse problem where information heterogeneity induces informed investors to bid on only the best IPOs leaving less-informed investors with only the overpriced issues. Habib and Ljungqvist (2001) show that issuing firms take costly action to reduce information asymmetry prior to IPO issuance, e.g., hiring a reputable, more expensive underwriter for their certification ability. Mechanisms which ameliorate the adverse selection costs of equity issuance serve to improve the efficiency of primary equity markets. In this paper I explore one such mechanism, specifically the presence of an issuer credit rating.

Credit rating agencies play critical roles in alleviating the information asymmetries that exist between borrowers and lenders and in apportioning risks in financial markets. In debt markets, firms who obtain credit ratings from Standard and Poor’s and/or Moody’s have increased leverage

(Faulkender and Petersen, 2006), are able raise more funds in syndicated debt financing (Sufi, 2009), and suffer less from underinvestment due to capital constraints (Harford and Uysal, 2014). What is it, then, that makes credit ratings a mechanism for a reduction in information asymmetry? Boot,

Milbourn, and Schmeits (2006) argue that credit ratings serve as a coordinating mechanism. The

1 threat of adverse rating changes motivates firms to take corrective actions. Thus, in the model of

Boot et al. (2006), this threat leads to homogeneity in investor beliefs.

The literature on the role of credit ratings in debt markets is well developed. As for the role of credit ratings in equity issuances, An and Chan (2008) find that credit rated firms exhibit reduced underpricing at IPO relative to unrated issuers. When they examine credit rating levels, i.e., the rating obtained by the firm, they do not find an association between the level obtained and IPO underpricing. An and Chan (2008) conclude that the credit rating itself conveys useful information in reducing value uncertainty of the issuing firm, i.e., credit ratings serve to reduce information asymmetry. One methodological challenge recognized in An and Chan (2008) is the potential problem of endogeneity. For example, a firm would choose to become rated when the benefits to doing so, such as reduced equity issue costs, outweigh the costs of the rating. As such, the decision to become credit rated is endogenous to the decision to issue equity.

In this study, I examine the SEO underwriting costs for credit rated firms relative to their unrated contemporaries. Using a sample of U.S. common share SEOs from 1990-2014, I document two main results. First, I advance the literature exploring the investment banking fees associated with issuing seasoned equity. I find that the fees paid by credit rated firms are significantly reduced relative to their unrated contemporaries. The presence of a credit rating mitigates the issues of information asymmetry thus lessening the value uncertainty regarding the issuing firm’s assets resulting in a reduction of the fees charged by the underwriting investment bank. This finding contributes to the results of An and Chan (2008) as it provides evidence that credit ratings reduce value uncertainty for firms who are, relatively speaking, better known to financial markets. The reduction in SEO underwriting fees for rated firms suggests that the benefits to rated firms when issuing equity extend beyond the IPO.

Secondly, I address the potential endogeneity problem noted in An and Chan (2008) by examining the differential effects that becoming credit rated imparts on the SEO activity and underwriting costs of issuing firms. The decision to become credit rated and the decision to issue seasoned equity may be endogenous to each other when firm characteristics affecting SEO behavior also affect the decision to be rated. To alleviate this concern, I employ a propensity-score, matched- sample approach wherein I identify previously unrated firms who obtain a credit rating and compare their SEO activity and characteristics to an unrated, propensity-score matched control group.

Using a unique dataset from Bloomberg Data Services, I identify 738 instances of credit rating initiations (i.e, instances where a previously unrated firm obtains a Standard and Poor’s long-term issuer credit rating) over the period 1990-2014. Credit rating initiation firms are matched to an unrated, propensity matched control group following the methodology of Faulkender and Petersen

(2006). I then examine the SEO activity of the matched-sample exploring both the likelihood that a firm issue an SEO and their costs to doing so. Controlling for the factors identified in prior literature as determinants of SEO issuance, I find a 12.47 percentage point increase in the likelihood of SEO issuance for credit rating initiation firms. Additionally, I document a reduction in the investment bank fees paid by newly rated firms who choose to issue seasoned equity. Obtaining a credit rating prior to an SEO issue is associated with an economically significant reduction of between 4.88% and 8.12% in the gross fees charged by the underwriting investment bank.

The results suggest that the adverse selection costs faced by a firm for issuing seasoned equity are not wholly exogenous to the firm. Improving the information environment in which the firm’s secondary market equity trades mitigates the costs associated with add-on equity issues.

Specifically, obtaining a credit rating prior to the issuance of seasoned equity reduces value uncertainties and information asymmetries thus leading to a reduction in underwriting fees. The findings provide support to prior literature documenting the economic importance of credit ratings.

The identification, certification, and validation that occur with an existing, or new, credit rating seems to affect the information environment with which the rated firm’s equity trades. Credit ratings serve to mitigate problems of uncertainty regarding asset valuations enabling less-costly placements of seasoned equity issues.

The remaining sections of this paper are organized as follows. Section 2 reviews the related literature and develops the concept. Section 3 discusses the sample used in the panel analysis and provides evidence on the relation between credit ratings and SEO costs. The results of the examination of the association between the credit rating initiations and firm SEO characteristics are provided in Section 4. Section 5 concludes.

2. Related Literature and Concept Development

The purpose of this study is to examine the impacts of credit ratings on the costs associated with SEO issuance. More precisely, I test the relation between credit ratings and the underwriting fees charged by the associated investment bank(s). As such, this paper relates two strands of literature. The first examines the effects of credit ratings on information asymmetry. And the second, the primary market costs associated with the issuance of equity.

2.1 Credit Ratings and Information Asymmetry

Prior literature suggests that credit ratings convey information beyond that which is incorporated in the observed prices of financial claims. Credit ratings and credit rating changes inform markets as to the economic prospects of the rated firm (Holthausen and Leftwich, 1986;

Ederington and Goh, 1998; Dichev and Piotroski, 2001). Norden and Weber (2004) show that credit default swap (CDS) and equity markets respond to the news of rating downgrades or reviews for downgrade as the news reveals private information to markets. Information asymmetry is reduced when the assessment and monitoring expertise of credit rating is engaged (Boot et al., 2006; Odders-

White and Ready, 2006; He, Wang, and Wei, 2010). Odders-White and Ready (2006) document

4 improved secondary market liquidity for higher credit quality firms. The authors argue that improvements in secondary market liquidity result from reductions in asymmetric information. Boot et al. (2006) provide a concise justification for credit ratings as a mechanism to reduce information asymmetry. The authors suggest that the threat of adverse rating changes motivates firms to take corrective actions thus leading to homogeneity in investor beliefs.

He et al. (2010) find that the reduction of information asymmetries is not simply due to the presence of a credit rating. Looking at a sample of credit rating changes (upgrades and downgrades), the authors show that measures of information asymmetry in secondary equity markets improve

(erode) as a firm experiences a credit rating upgrade (downgrade). Further, the effect is a function of the magnitude of the change. He et al. (2010) conclude that the composition of informed and uninformed traders changes with a credit rating change. Overall, this research suggests that credit rating agencies disseminate private information to public markets.

2.2 Information Asymmetry and the Costs of Issuing Equity

The adverse selection costs of asymmetric information in IPOs has been well developed.

Rock (1986) argues IPOs are underpriced, to some extent, to attract uniformed investors thus avoiding the problems of a winner’s curse. Firms and underwriters leave some money “on-the-table” to ensure the continued participation of uniformed investors. Beatty and Ritter (1986) suggest that the underpricing is an artifact of the ex-ante uncertainty regarding the issuing firm, avoiding any inferred intentions of the firm/underwriter. Treating an investment in the firm’s equity at IPO as a call option with a strike price of the issue price, Beatty and Ritter (1986) argue that the value of the investment increases with the level of uncertainty regarding the value of the firm’s assets. The greater the uncertainty, the lower the price investors are willing to bear thus leading to IPO underpricing. Armitage, Dionysiou, and Gonzalez (2014) study the role that inelastic demand, or illiquidity, is a plays in SEO underpricing. The provide evidence that inelastic demand, or illiquidity,

5 is a primary determinant of SEO underpricing. However, Armitage et al. (2014) acknowledge the link between illiquidity and information asymmetry and conclude that asymmetry could be a contributing factor to inelastic demand.

Underpricing that occurs as a result of uncertainty imposes costs on the firm as it requires firms to issue additional shares for a given level of capital need thus diluting the positions of current shareholders. Habib and Ljungqvist (2001) model the behavior of issuers and conclude that firms will take costly action to avoid uncertainty surrounding equity issues, specifically IPOs. One such method suggested by the authors is the hiring of a more reputable, costly underwriter as a means of leveraging the underwriter’s reputation.

An and Chan (2008) connect the credit rating and IPO literatures. The authors examine a sample of U.S. common stock IPOs over the period 1986-2004 and find that underpricing is reduced for issuing firms who obtain a credit rating prior to their IPO. Interestingly, the authors find that the level of the rating does not affect the uncertainty surrounding the IPO; the effect is driven by the simple presence of a rating. An and Chan (2008) conclude that the credit rating provides

“value certainty” in the IPO process. In an unpublished working paper, Liu and Malatesta (2006) study the relation for SEOs. The authors find that underpricing is less pronounced and the abnormal returns are less-negative for credit rated firms who issue seasoned equity. The authors conclude that the presence of a credit rating reduces the adverse selection costs associated with add- on equity issues.

The underpricing which arises as a result of the issues of adverse selection are not the only costs to the issuing firm. More directly, the firm faces the underwriting costs required by the underwriting investment bank. Butler et al. (2005) examine these costs directly in the context of

SEOs. The authors document an inverse association between underwriter fees and the secondary market equity liquidity of the issuing firm. The findings of Butler et al. (2005) are very intuitive, i.e.,

6 issuing firms with less-liquid equity prior to the SEO pay increased investment bank fees due to the additional difficulty faced by the bank in placing the issue. The relation between information asymmetry and liquidity documented in Odders-White and Ready (2006) manifest in the liquidity premium found in Butler et al. (2005). Ginglinger, Matsoukis, and Riva (2013) advance the work of

Butler et al. (2005) by showing that liquidity is an important determinant of SEO flotation choice.

Liquidity concerns contribute both to the flotation choice and to the cost of issuance.

In this study I extend the work of Butler et al. (2005), An and Chan (2008), and Armitage et al. (2014) by examining the extent to which the issuing firm can take active steps to reduce the adverse effects of asymmetric information prior to the issuance of seasoned equity. Specifically, I explore the economic impacts that being credit rated imparts on the underwriting costs of issuing seasoned equity.

3. Credit Ratings and SEO Characteristics: Panel Analysis

3.1 Sample Construction and Descriptive Statistics

This study uses seasoned equity offering data from the Securities Data Company’s (SDC)

Global New Issues database. First, I collect the full sample of U.S. common stock offerings from

January 1, 1990 through December 30th, 2014, excluding initial public offerings, unit offerings, rights offerings, mutual conversions, and issues by closed-end funds.1 This results in a sample of 9,194 offers. Following prior literature, to be included in the final sample, an issue must: 1) include at least some primary shares; 2) be issued by a firm listed on NYSE, NASDAQ, or AMEX; 3) have an offer price of at least $3.00 and less than $400.00; 4) be issued by a firm whose has at least 6 months of prior trading data available on CRSP and who is tracked by Compustat; and, 5) not be originated by a firm in the finance or utility industries. The results of the sample identification and restrictions yields

1 The sample period ends in 2014 to avoid right-censoring the data in the post-rating initiation analyses. Offerings types are identified using SDC’s classifications and CRSP share codes.

7 a final sample of 4,731 seasoned equity issues by 2,765 firms. Descriptive statistics on the SEOs and firms in the sample are provided in Table 1.

[Insert Table 1 here]

The left-half of Table 1 lists descriptive statistics of the SEOs in the sample by year, while the right-half list results by Fama and French (1997) 17-industry classifications. Incidence of SEO offers tend to be somewhat pro-cyclical. The results by year suggest that SEO issues tend to cluster in years of economic expansion. The gross fees charged by investment banks have declined nearly monotonically throughout the sample period likely as a result of technological advances and increased competition in the underwriting space. SEOs tend to cluster by industry as well as year.

Firms in the Machinery, Oil, Retail, Transportation, and Consumer Goods industries account for roughly 51% of all SEO issues over the period. Gross investment banking fees as a proportion of the proceeds raised in the issue range from an average of 3.75% in the Transportation industry to

5.46% in Consumer Goods.

Table 2 provides summary statistics for the sample of SEOs. The left-half of Table 2 lists the results for the entire sample and the right-half by whether or not the issuing firm has a long-term, issuer credit rating from S&P at the time of the issue. For each variable, Table 2 reports distribution statistics as well as difference in means (medians) by the presence of a credit rating. Statistical results on the differences in means (medians) are from t-tests (k-sample tests). Variable definitions are provided in Appendix A.

The average SEO over the sample period raised $142.7 million in proceeds. The average issue proceeds of the SEO relative to market capitalization of the issuing firm (i.e., Issue Size) at issue is 19.21%. The gross investment bank fees (in dollar terms) as a percent of SEO proceeds incurred over the sample are 4.92%. Multiple bookrunners are used in 28.09% of the issues in the

8 sample and are from underwriters who underwrite, on average, 3.08% of the entire SEO market in a given year.

[Insert Table 2 here]

Following Butler et al. (2005), I construct a liquidity index which captures the secondary market liquidity of the issuing firm’s secondary market equity. The liquidity index is an average of four ranked measures of equity liquidity in the pre-SEO period: the inverse of Amihud (2002) illiquidity, the inverse of the ask-bid spread, volume, and turnover, and ranges from 0 to 1 with 0.5 being the expected mean by construction. The average liquidity index for the issuing firms in the sample is

0.501. The average issuing firm has a market capitalization of $885 million and share price of $25.63.

The average daily standard deviation of equity returns for issuing firms over the sample period is

3.49%. Finally, a majority of the offerings are from firms listed on Nasdaq. Issues by firms listed on

Nasdaq, NYSE, and AMEX exchanges account for 64.57%, 32.56%, and 2.87% of all issues in the sample, respectively.

SEO characteristics show distinct differences when offerings are contrasted by the presence of an issuer credit rating at issue (the right-half of Table 2). Interestingly, rated firms are responsible for larger issues as measured by Offer Proceeds, but smaller issues as measured by Issue Size. The investment banking fees as a percentage of offer proceeds paid by rated firms are 116 basis points less than those paid unrated issuers at the mean and 125 basis points less at the median. Issues by rated firms are more than twice as likely to use multiple bookrunners and tend to use underwriters who control a greater portion of the total SEO market. As it relates to firm characteristics, rated issuers are larger, have higher share prices, are more likely to trade on the NYSE, and have equity which trades with greater liquidity and at a lower volatility, on average.

3.2 Methodology

OLS estimates of the effects of being credit rated on the costs of issuing seasoned equity are only unbiased if the decision to become credit rated is independent of the decision, or costs, of issuing seasoned equity. However, the decision of a firm to obtain a credit rating is related, at least in part, to the benefits incurred by the firm for being rated. For example, a firm would choose to become rated when the benefits to doing so, such as reduced SEO issue costs, outweigh the costs of the rating. In such a case the two are endogenously determined creating a potential sample construction issue if used for analysis.

To account for the endogenous selection, I follow An and Chan (2008) and estimate a

Heckman treatment effect model. In the first stage I use a probit estimation to model the likelihood that a firm is credit rated at the time of its issue. Specifically, I model the firm’s decision to obtain a credit rating by:

∗ 푅푎푡푒푑푖 = 휸 ∗ 풁푖 + 휂푖

∗ 푅푎푡푒푑푖 = 1 if 푅푎푡푒푑푖 > 0

∗ 푅푎푡푒푑푖 = 0 if 푅푎푡푒푑푖 < 0

∗ where 푅푎푡푒푑푖 is a latent variable. 푍푖 is a set of observable variables affecting the firm’s choice of being credit rated. 휸 is a set of coefficients, and 휂푖 is an error term. I use the vector of coefficient estimates, denoted 휸̂, to construct the 휆̂ or inverse Mills ratio, as follows:

̂ 휙(휸̂∗풁푖) 휆푖 = if 푅푎푡푒푑푖 = 1 Φ(휸̂∗풁푖)

̂ −휙(휸̂∗풁푖) 휆푖 = if 푅푎푡푒푑푖 = 0 1−Φ(휸̂∗풁푖) where 휙 and Φ denote the density and cumulative distribution functions of the standard normal distribution, respectively. The inverse Mills ration, 휆̂, is then added in the second-stage, OLS regression testing to address the endogenous selection.

Table 3 provides the results from the first-stage, probit estimation modeling the decision to become credit rated. The probit specification follows Faulkender and Petersen (2006) and includes, as regressors, measures of firm size [Ln(Mkt. Assets)], firm age [Ln(1+Age)], profitability [Profit], tangibility of firm assets [Tangible Assets], firm growth opportunities [Market-to-Book], firm investments in brand name and intellectual capital [Advertising/Sales], volatility of a firm assets

[(Asset Return)], two indicators which take a value of 1 if the firm trades on the NYSE or is a member of the S&P500 [NYSE and SP500] respectively, the natural log of the proportion of firms within the issuing firm’s Fama and French 17-industry which are rated [Ln(1+Pr(Rating))], an indicator variable for non-linearities in a firm’s age which takes a value of 1 if the firm has less than three years of Compustat coverage [Young], and an indicator which takes a value of 1 if the firm’s market value of assets times the median debt ratio (0.183) is less than the minimum bond size required for the Barclay’s US Aggregate Bond Index (formerly Lehman Brother’s Bond Index)

[Barclay’s]. Formal variable definitions are provided in appendix A. In addition to the aforementioned regressors, the model controls for year and Fama-French (1997) 17-industry fixed effects with robust standard errors clustered at the industry level.

[Insert Table 3 here]

Overall the results are consistent with those of Faulkender and Petersen (2006) with the exception being the coefficient estimate on SP500. Faulkender and Petersen (2006) find that inclusion in the S&P 500 is positively related to the likelihood of a firm being credit rated. For the firms in the SEO sample, this association is negative. The remaining coefficient estimates are generally consistent. Again, the primary purpose of this specification is to construct the inverse Mills ratio used in the second phase of the Heckman tests.

3.3 Effects of Credit Ratings on SEO Costs

The results of testing on the dynamics of being credit rated at issue and SEO issue costs are presented in this section. The dependent variable in all three specifications the natural log of the ratio of the gross fees (in dollar terms) charged by the underwriting investment bank scaled by the offer proceeds. Following Butler et al. (2005), I include as regressors measures of issue and issuer characteristics [i.e., Ln(Proceeds), Ln(Mkt. Cap.), Ln(Share Price), Ln(Equity Vol.), Und.

Reputation, Mult. Book Runner, AMEX, and Nasdaq] and a measure of secondary market liquidity preceding the SEO issue [i.e., Ln(Liquidity Index)]. Regressions results are from an ordinary-least-squares framework controlling for year and Fama-French (1997) 17-industry fixed effects with robust standard errors clustered at the industry level.

[Insert Table 4 here]

Column (1) of Table 4 corroborates the findings of Butler et al. (2005), i.e., smaller, less well- known firms with more illiquid and volatile secondary market equity pay higher investment banking fees. Additionally, issues with multiple bookrunners and issues underwritten by more reputable underwriters are associated with higher fees. The second column of Table 4 presents the results of the previous specification including Rated, an indicator variable which takes a value of 1 if the issuing firm is credit rated at the time of the issue. The coefficient estimate on Rated is negative and statistically significant indicating that SEOs by firms with a long-term S&P issuer rating carry lower investment banking fees. Economically speaking, the coefficient estimate on Rated suggests that the presence of a credit rating reduces at the time of SEO issue reduces the investment bank fees by about 3.15%.

The third column of Table 4 provides results controlling for the potential issue of endogenous selection. The presence of a long-term issuer credit rating at the time of SEO issue is associated with a reduction in SEO issue costs after accounting for the selection bias. The

12 coefficient on Rated moves closer to zero when including the Inverse Mills ratio, but is still negative and statistically significant.

4. Credit Ratings and SEO Characteristics: Matched Sample

4.1 Sample Construction and Descriptive Statistics

The Heckman selection results produce unbiased estimates of the relation between the presence of an issuer credit rating and SEO issue costs if the models are appropriately specified. If, for example, the specifications include omitted regressors, then inferences gained from the approach are suspect. To account for this potential issue, I employ an event study approach where the SEO behavior and characteristics of firms are examined subsequent to the initiation of a new long-term issuer rating. The SEO characteristics of the newly rated firms are then contrasted to an unrated, propensity score matched control group. The results of these tests are presented in this section.

The sample of credit rating initiations come from Bloomberg Data Services. Firms are identified as having obtained a new issuer rating if their prior Standard and Poor’s long-term issuer rating, as identified by Bloomberg Data Services, is either missing or contains a value of "NR" which identifies a firm as being not-rated. This results in a sample of 2,269 credit rating initiations. To be included in the final sample of credit rating initiations I impose two additional restrictions: 1) the new rating must not be a “personal opinion” or “credit watch” rating; and, 2) the rating initiation must be by a firm with at least 6 months of prior trading data available on CRSP and who is tracked by Compustat.

The sample identification procedure yields a final sample of 738 credit rating initiations. Statistics on the distribution of credit rating initiations by year, Fama-French (1997) 17-industry, and credit rating obtained are provided in Table 5.

[Insert Table 5 here]

Credit rating initiations exhibit clustering by year, industry, and rating obtained. Rating initiations in the years 1996 through 2000 comprise 39.2% of the total initiations in the sample.

Similarly, credit rating initiations in the Machinery, Oil, Retail, Transportation, and Consumer

Goods industries make up 37.4% of the initiations sample. Finally, the newly obtained rating tends to be a speculative grade rating, i.e., 68.6% of the credit rating initiations are BB+ and below.

4.2 Methodology

To evaluate the effects of credit rating initiations on SEO issue activity and costs, I follow the methodology of Faulkender and Petersen (2006) and Li and Zhao (2006). Specifically, I construct a propensity-score matched sample of firms who are similar to the credit rating initiation firms in the rating initiations sample, but who elect not to become credit rated. This is accomplished by using the probit model of Faulkender and Petersen (2006) where the presence of a credit rating is model as a function of firm size [Ln(Mkt. Assets)], firm age [Ln(1+Age)], profitability [Profit], tangibility of firm assets [Tangible Assets], firm growth opportunities [Market-to-Book], firm investments in brand name and intellectual capital [Advertising/Sales], volatility of a firm assets

To accomplish the propensity score matching, I first estimate the probit model across the universe of firm-years for firms who are covered in both CRSP and Compustat to obtain coefficient estimates. The probit estimation includes fixed effects for year and industry using Fama and French

17-industry classifications and computes robust standard errors clustered by industry. Propensity scores are calculated for each firm using the coefficient estimates from the probit estimation and the

14 firm’s values for the regressors in a given year. Matching the credit rating initiation firms to an unrated control group is then achieved by selecting the unrated firm whose propensity score from the year preceding a credit rating initiation event most closely matches that of newly rated firm in absolute difference. Matched firms are used only once in a given year yielding a sample of 1,476 firms, half of which being newly rated.2 Validation of the matching procedure is provided in Table 6.

[Insert Table 6 here]

Table 6 provides the descriptive statistics for the event and control firms in the matched sample and tests for differences in means (medians). Statistical significance on differences in means and medians is computed using t-tests for mean estimates and k-sample tests for median estimates.

At the means, credit rating initiation firms tend to be larger, younger, have less tangible assets, and have greater growth potential (i.e., higher market-to-book values) than the unrated firms in the sample. The results are largely consistent at median values with the exception that asset volatility of credit rating initiation firms is greater than that of the firms in the control sample. The results from the probit estimation used to compute propensity scores can be found in Appendix B.

The matched sample is then merged with the SEO sample described in section 3.1 to identify the SEOs issued by the firms in the credit rating initiation matched sample over the period

1990-2014. Specifically, I identify the first instance of an SEO issue to occur within three years of the credit rating initiation event for both event (newly rated firms) and control (unrated) firms.3 This matching yields a sample of 246 SEOs issued between the event and control subsamples.

4.3 Effects of Credit Ratings on SEO Incidence and Costs

2 In unreported results, I impose the restriction that a firm can enter the control sample only once over the sample period. The results are qualitatively unchanged regardless of this restriction. 3 In unreported results, the time to SEO from the credit rating initiation date was varied from as little as one year to as much as five years. The results were qualitatively unchanged.

Table 7 provides descriptive statistics on the SEOs issued by firms in the matched sample.

The variable SEO Flag is an indicator which takes a value of 1 if a firm issues an SEO within three years of entering the matched sample and 0 otherwise. The left-half of Table 7 presents summary statistics for the entire matched sample. Firms in the matched sample issue an SEO within three years of entering the sample 16.67% of the time, i.e., SEO Flag equals 0.1667. The average issue by firms in the matched sample generates $194.9 million in proceeds, accounts for 36.77% of issuing firm’s current market capitalization, incurs gross investment banking fees as a percent of total proceeds of 4.47%, uses more than one bookrunner 38.6% percent of the time, and is underwritten by investment banks who underwrite 3.47% of the total SEO market in a given year on average.

[Insert Table 7 here]

The right-half of Table 7 displays descriptive statistics and comparisons of means (medians) by the rated and unrated subsamples. Credit rating initiation firms are more than three and a half times as likely to issue an SEO following the credit rating initiation relative to their unrated counterparts. The incidence of issue is 26.02% for newly rated firms versus 7.32% for control firms.

Issues tend to be larger for rating initiation firms in dollar terms, but smaller as a percentage of current market capitalization. On average, the gross investment bank fees as a percent of issue proceeds paid by credit rating initiation firms are 33 basis points lower than those paid by unrated firms at the mean, and 50 basis points lower at the median.

The univariate results suggest that credit rating initiation firms are more likely to issue and do so at a reduced cost. However, the results in Table 7 indicates differences in the issue and financial characteristics of the firms in these two subsamples, e.g., newly rated firms are larger and have greater secondary market liquidity. To account for these differences, I examine the SEO activity and costs across the subsamples in a multivariate framework.

Table 8 provides the results of probit estimations where the likelihood to issue an SEO post- credit rating initiation is modeled as a function of the demand-side determinants of obtaining a credit rating (i.e, the variables identified in Faulkender and Petersen (2006)), Share Price, and three variables identified in prior literature as determinants of SEO issuance, i.e, Leverage, Prior

Return, and Ln(Liquidity Index). Leverage is the change in leverage from the quarter immediately preceding the credit rating initiation date to the quarter immediately preceding the SEO issue date and serves as a control for a firm’s desire to rebalance their capital structure (Fama and

French, 2002; Leary and Roberts, 2006). Prior Return is the equity return for a given firm in the six months preceding the SEO issuance stopping ten days before the issue date and controls for the likelihood that a manager issues seasoned equity when she believes the market currently overvalues the firm’s equity (Lee, 1997; Baker and Wurgler, 2002). Ln(Liquidity Index) controls for managerial attempts at timing their issue when secondary market liquidity is favorable (Lin and Wu,

2013). The probit regressions include controls for year and Fama-French (1997) 17-industry fixed effects and compute robust standard errors clustered at the industry level.

[Insert Table 8 here]

Column (1) of Table 8 provides the results of a probit estimation where the dependent variable, SEO Flag, takes a value of 1 if the firm issues an SEO within three-years of the credit rating initiation date and 0 otherwise. The primary variable of interest in the probit specifications is

New Rating which takes a value of 1 if the issuing firm is from the credit rating initiation subsample and 0 otherwise. The coefficient estimate on New Rating in Column (1) is positive and statistically significant indicating that firms in the credit rating sample are more likely to issue seasoned equity. The estimate translates to a 12.47 percentage point increase in the likelihood of an

SEO issue when evaluated at the mean values of the other covariates. Columns (2) and (3) present similar results with the exception that Column (2) requires the SEO to occur within the two-year

17 period (730 days) following the rating initiation and Column (3) within one-year (365 days). The coefficient estimates on New Rating are both positive and statistically significant despite the additional time restrictions. Economically speaking, the coefficient estimates on New Rating in

Columns (2) and (3) translate to an 11.00 percentage point and 7.55 percentage point increase in the likelihood of SEO issuance for newly rated firms, respectively. Across all three specifications the likelihood that a firm issue seasoned equity following a credit rating initiation is greater for newly rated firms than it is for unrated firms in the control group.

I repeat the examination of the relation between credit ratings and the investment banking fees associated with issuing seasoned equity. The regression specifications follow those of Butler et al. (2005), and include, as regressors, measures of issue and issuer characteristics [i.e., Ln(Proceeds),

Ln(Mkt. Cap.), Ln(Share Price), Ln(Equity Vol.), Und. Reputation, Mult. Book Runner,

AMEX, and Nasdaq], and a measure of secondary market liquidity preceding the SEO issue [i.e.,

Ln(Liquidity Index)]. Regressions results are from an ordinary-least-squares framework controlling for year and Fama-French (1997) 17-industry fixed effects with robust standard errors clustered at the industry level. The primary variable of interest is New Rating. The results are presented in

Table 9.

[Insert Table 9 here]

The first column of Table 9 presents the results of the specification used in Butler et al.

(2005) without the addition of New Rating. SEOs with greater offer proceeds and those by larger issuers exhibit reduced investment banking fees consistent with the results of Butler et al. (2005).

Column (2) repeats the analysis including New Rating as a covariate. For the SEOs in the event- study sample, the initiation of a credit rating is associated with a reduction in the investment banking fees associated with the issuance of seasoned equity. The investment bank fees paid by firms who choose to become credit rated prior to the issuance of seasoned equity are statistically lower than

18 those paid by unrated issuers. The coefficient on New Rating in Column (2) indicates that obtaining a credit rating prior to an SEO issue is associated with an economically significant reduction of roughly 4.88% in the gross fees charged by the underwriting investment bank. The results in Column (3) include the inverse Mills ratio from the probit regression presented in Table 8.

The effect becomes more pronounced controlling for the sample selection bias. The coefficient estimate on New Rating indicates an 8.12% reduction in the investment bank fees charged for an

SEO issue.

5. Conclusion

In this study, I contribute to the results of Butler et al. (2005), An and Chan (2013), and

Armitage et al. (2014) by examining the SEO underwriting fees of credit rated firms. Using a sample of U.S. common share SEOs from 1990-2014, I document two main findings. First, I advance the literature exploring the determinants of SEO investment banking fees by including the presence of a credit rating. The investment bank fees incurred by credit rated firms are significantly lower than those of unrated firms. Being credit rated prior to SEO issuance reduces the value uncertainty and information asymmetry regarding the asset valuation of the issuing firm resulting in a reduction in underwriter fees of 3.15% on average. This evidence contributes to the results of An and Chan

(2008) as it suggests that the benefits to rated firms when issuing equity extend beyond the IPO.

In addition to documenting a reduction in fees, I address the endogeneity problems noted by prior studies by examining the differential effects that become credit rated imparts on the SEO costs incurred by issuing firms. Using a propensity score matched sample controls for the determinants of firms to be credit rated at SEO issue thus better disentangling the endogeneity. The matched sample approach confirms the panel results and shows that the reduction in the fees charged for SEO holds after addressing the potentially confounding effects. Obtaining a new credit rating prior to an SEO

19 issue is associated with an economically significant reduction of between 4.88% and 8.12% in the gross fees charged by the underwriting investment bank.

The results suggest that reducing information asymmetries prior to SEO issuance mitigates the adverse selection costs associated with add-on equity issues. Obtaining a credit rating prior to the issuance of seasoned equity mitigates information symmetries lessening the adverse impact of private information in SEO placement. In addition to the hiring of a more reputable underwriter suggested in Habib and Ljungqvist (2001), managers can obtain an issuer credit rating as a means to resolve uncertainty. This study supports prior literature documenting the economic importance of credit ratings. The engagement of a credit rating agency serves to mitigate the problems of information asymmetry facilitating the less-costly placement of seasoned equity issues.

References:

Amihud, Y., 2002, Illiquidity and Stock Returns: Cross-Section and Time-Series Effects, Journal of Financial Markets 5: 31-56.

Armitage, S., Dionysiou, D., Gonzalez, A., 2014, Are the Discounts in Seasoned Equity Offerings Due to Inelastic Demand?, Journal of Business Finance & Accounting 41: 743-772.

An, H., Chan, K., 2008, Credit Ratings and IPO Pricing, Journal of Corporate Finance 14: 584-595.

Baker, M., Wurgler, J., 2002, Market Timing and Capital Structure, Journal of Finance 57: 1-30.

Beatty, R., Ritter, J., 1986, Investment Banking, Reputation, and the Underpricing of Initial Public Offerings, Journal of Financial Economics 15: 212-232.

Benston, G., Hagerman, R., 1974, Determinant of Bid-Asked Spreads in the Over-the-Counter Market, Journal of Financial Economics 1: 353-364.

Boot, A., Milbourn, T., Schmeits, A., 2006, Credit Ratings as Coordination Mechanisms, Review of Financial Studies 19: 81-118.

Butler, A., Grullon, G., Weston, J., 2005, Stock Market Liquidity and the Cost of Issuing Equity, Journal of Financial and Quantitative Analysis 40: 331-348.

Dichev, L., Piotroski, J., 2001, The Long-Run Stock Returns Following Bond Ratings Changes, Journal of Finance 56: 173-203.

Ederington, L., Goh, J., 1998. Bond Rating Agencies and Stock Analysts: Who Knows What When?, Journal of Financial and Quantitative Analysis 33: 569–585.

Fama, E., French, K., 1997, Industry Costs of Equity, Journal of Financial Economics 43: 153-193.

Fama, E., French, K., 2002, Testing Tradeoff and Pecking Order Predictions about Dividends and Debt, Review of Financial Studies 15: 1-33.

Faulkender, M., Petersen, M., 2006, Does the Source of Capital Structure Affect Capital Structure?, Review of Financial Studies 19: 45-79.

Ginglinger, E., Matsoukis, L., Riva, F., 2013, Seasoned Equity Offerings: Stock Market Liquidity and the Rights Offer Paradox, Journal of Business Finance & Accounting 40: 215-238.

Glosten, L., Harris, L., 1988, Estimating the Components of the Bid-Ask Spread, Journal of Financial Economics 21: 123-142.

Habib, M., Ljungqvist, A., 2001, Underpricing and Entrepreneurial Wealth Losses in IPOs: Theory and Evidence, Review of Financial Studies 14: 433-458.

Harford., J., Uysal, V., 2014, Bond Market Access and Investment, Journal of Financial Economics 112: 147-163.

He., Y., Wang, J., Wei, K., 2010, Do Bond Rating Changes Affect the Information Asymmetry of Stock Trading, Journal of Empirical Finance 18: 130-116.

Holthausen, R., Leftwich, R., 1986, The Effect of Bond Rating Changes on Common Stock Prices, Journal of Financial Economics 17: 57-89.

Huang, R., Stoll, H., 1997, The Components of the Bid-Ask Spread: A General Approach, Review of Financial Studies 10: 995-1034.

Kyle, A., 1985, Continuous Auctions and Insider Trading, Econometrica 53: 1315-1335.

Leary, M., Roberts, M., 2006, Do Firms Rebalance their Capital Structures?, Journal of Finance 60: 2575- 2619.

Lee, I., 1997, Do Firms Knowingly Sell Overvalued Equity?, Journal of Finance 52: 1439-1466.

Li, X., Zhao, X., 2006, Propensity Score Matching and Abnormal Performance After Seasoned Equity Offerings, Journal of Empirical Finance 13: 351-370.

Lin, J., Wu, Y., 2013, SEO Timing and Liquidity Risk, Journal of Corporate Finance 19: 95-118.

Lin, J., Sanger, G., Booth, G., 1995, Trade Size and Components of the Bid-Ask Spread, Review of Financial Studies 8: 1153-1184.

Liu, Y., Malatesta, P., 2006, Credit Ratings and the Pricing of Seasoned Equity Offerings, Working paper, University of Washington.

Norden, L., Weber, M., 2004. Informational efficiency of credit default swap and stock markets: The impact of credit rating announcements. Journal of Banking & Finance 28 (11): 2813-2843.

Odders-White, E., Ready, M., 2006, Credit Ratings and Stock Liquidity, Review of Financial Studies 19: 119-157.

Rock, K., 1986, Why New Issues are Underpriced, Journal of Financial Economics 15: 187-212.

Stoll, H., 1989, Inferring the Components of the Bid-Ask Spread: Theory and Empirical Tests, Journal of Finance 44: 115-134.

Sufi, A., 2009, The Real Effects of Debt Certification: Evidence from the Introduction of Bank Loan Ratings, Review of Financial Studies 22: 1659-1691.

Table 1: Distribution of Seasoned Equity Offerings The sample consists of all seasoned equity offerings recorded in the Securities Data Company's (SDC) Global New Issues Database from January 1, 1990 through December 31, 2014 that satisfy the following criteria: the offering is not an initial public offering, unit offering, rights offering, mutual conversion, or an offering issued by a closed-end fund. Additionally, an issue must include at least some primary shares, be issued by a firm listed on NYSE, NASDAQ, or AMEX, have an offer price of at least $3.00 and less than $400.00, be issued by a firm whose has at least 6 months of prior trading data available on CRSP and who is tracked by Compustat; and not be originated by a firm in the finance or utility industries. Variable definitions are provided in Appendix A.

SEOs by Year SEOs by Fama-French 17 Industry Rated Unrated Rated Unrated Year # of SEOs Firms Firms Gross Fees % Industry # of SEOs Firms Firms Gross Fees % 1990 36 3 33 5.263 Food 91 30 61 4.873 1991 136 9 127 5.545 Mining 54 35 19 4.300 1992 122 16 106 5.470 Oil 479 277 202 4.321 1993 287 65 222 5.226 Clths 52 9 43 4.934 1994 186 44 142 5.077 Durbl 73 12 61 5.285 1995 282 40 242 5.219 Chem 54 21 33 4.566 1996 318 60 258 5.197 Cnsum 297 21 276 5.466 1997 271 51 220 5.232 Cnstr 119 51 68 4.978 1998 194 44 150 4.905 Steel 67 33 34 4.748 1999 229 69 160 4.897 Fabpr 26 7 19 5.153 2000 217 50 167 4.933 Machn 630 115 515 5.069 2001 184 70 114 4.900 Cars 54 22 32 4.690 2002 168 94 74 4.675 Trans 304 231 73 3.758 2003 217 80 137 4.893 Rtail 335 89 246 4.784 2004 182 70 112 4.729 Other 2096 379 1717 5.141 2005 148 45 103 4.941 2006 168 52 116 4.817 2007 190 66 124 4.554 2008 112 38 74 4.633 2009 259 92 167 4.664 2010 202 86 116 4.637 2011 195 53 142 4.740 2012 216 76 140 4.499 2013 150 41 109 4.454 2014 62 18 44 4.628 Total 4,731 1,332 3,399 4.919 Total 4,731 1,332 3,399 4.919

Table 2: Descriptive Statistics of SEOs Table reports descriptive statistics on SEO and firm characteristics for the seasoned equity offerings (SEO) firms in the sample. The sample consists of all seasoned equity offerings recorded in the Securities Data Company's (SDC) Global New Issues Database from January 1, 1990 through December 31, 2014 that satisfy the following criteria: the offering is not an initial public offering, unit offering, rights offering, mutual conversion, or an offering issued by a closed-end fund. Additionally, an issue must include at least some primary shares, be issued by a firm listed on NYSE, NASDAQ, or AMEX, have an offer price of at least $3.00 and less than $400.00, be issued by a firm whose has at least 6 months of prior trading data available on CRSP and who is tracked by Compustat; and not be originated by a firm in the finance or utility industries. Statistical significance on differences in means and medians is computed using t-tests for mean estimates and k-sample tests for median estimates. Variable definitions are provided in Appendix A. a, b, and c indicate statistical significance at the 10%, 5%, and 1% levels, respectively.

Issuer Rated Firms Unrated Firms Difference N Mean Median Std. Dev. N Mean Median N Mean Median Mean Median

SEO Characteristics Offer Proceeds (million $) 4,731 142.6703 75.3730 236.9281 1,332 272.6145 175.2840 3,399 91.7477 58.0000 180.8668 c 117.2840 c Issue Size (%) 4,731 0.1921 0.1600 0.1528 1,332 0.1504 0.1146 3,399 0.2084 0.1765 -0.0579 c -0.0619 c Gross Fee (%) 4,731 4.9188 5.0000 1.1023 1,332 4.0822 4.0000 3,399 5.2466 5.2520 -1.1645 c -1.2520 c 24 c c

Mult. Bookrunner 4,731 0.2809 0.0000 0.4495 1,332 0.4805 0.0000 3,399 0.2027 0.0000 0.2778 0.0000 Underwriter Reputation 4,731 0.0308 0.0080 0.0444 1,332 0.0370 0.0105 3,399 0.0283 0.0074 0.0087 c 0.0031 c Firm Characteristics Liquidity Index 4,731 0.5012 0.5126 0.2287 1,332 0.5951 0.6168 3,399 0.4644 0.4594 0.1307 c 0.1574 c Ln(Market Assets) 4,731 6.7855 6.7118 1.4386 1,332 8.1504 8.0484 3,399 6.2513 6.2137 1.8991 c 1.8347 c Share Price 4,731 25.6269 21.8700 16.9049 1,332 31.3373 28.5845 3,399 23.3891 19.7733 7.9481 c 8.8112 c σ(Equity Return) 4,731 0.0349 0.0319 0.0179 1,332 0.0265 0.0224 3,399 0.0382 0.0354 -0.0117 c -0.0130 c Amex 4,731 0.0287 0.0000 0.1671 1,332 0.0098 0.0000 3,399 0.0362 0.0000 -0.0264 c 0.0000 c Nasdaq 4,731 0.6457 1.0000 0.4783 1,332 0.2695 0.0000 3,399 0.7932 1.0000 -0.5237 c -1.0000 c

Table 3: Propensity to be Rated at SEO Issue This tables reports the results of a probit regression where the likelihood that an SEO firm has an S&P long-term credit rating at the time of the SEO issue is modeled as a function of the characteristics of the firm, following Faulkender and Petersen (2006). Variable definitions are provided in Appendix A. The regression specification includes fixed effects for year and industry using Fama and French (1997) 17-industry classifications and computes robust standard errors clustered by industry. t-statistics are reported in the parentheses below the coefficients. a, b, and c indicate statistical significance at the 10%, 5%, and 1% levels, respectively.

Dependent Variable = Rated (1 if yes) SP500 -0.362 b (-1.999) NYSE 0.236 c (2.715) Ln[1 + Pr(Rating)] 1.746 c (% of other firms in industry) (3.599) Young 0.100 (0.925) Barclay's -0.107 (-0.355) Ln(Mkt. Assets) 0.783 c (14.599) Ln(1 + Age) 0.271 c (4.600) Profit 0.044 (0.862) Tangible Assets 0.465 b (2.359) Market-to-Book (Assets) -0.325 c (-12.628) Advertising/Sales 2.165 (0.685) σ(Asset Return) 0.042 (0.131) Constant -7.205 c (-17.579) Observations 4,731 Pseudo R2 0.498

Table 4: Credit Ratings and SEO Issue Fees This table reports the results of ordinary-least-squares testing on the relation between the investment bank fees charged at SEO offerings and the presence of an S&P long-term issuer credit rating controlling for firm characteristics following Butler et al. (2005). The inverse Mills ratio in Column (3) is computed using the estimates generated from the probit specification presented in Table 3. Variable definitions are provided in Appendix A. All specifications include fixed effects for year and industry using Fama and French 17-industry classifications and compute robust standard errors clustered by industry. t-statistics are reported in the parentheses below the coefficients. a, b, and c indicate statistical significance at the 10%, 5%, and 1% levels, respectively.

Dependent Variable = Ln(Gross Fees) (1) (2) (3) Ln(Liquidity Index) -0.0118 b -0.0112 b -0.0125 c (-2.514) (-2.371) (-2.594) Ln(Proceeds) -0.0322 c -0.0309 c -0.0275 c (-6.149) (-5.907) (-5.155) Ln(Mkt. Cap.) -0.0892 c -0.0864 c -0.0894 c (-18.796) (-18.153) (-18.269) Ln(Share Price) 0.0030 0.0003 0.0033 (0.530) (0.058) (0.564) Ln(Equity Vol.) 0.1035 c 0.1003 c 0.0945 c (13.656) (13.160) (12.207) Und. Reputation 0.1090 b 0.1220 b 0.1690 c (1.968) (2.212) (2.989) Mult. Book Runner 0.0626 c 0.0646 c 0.0650 c (7.230) (7.436) (7.387) AMEX 0.0798 c 0.0726 c 0.0652 c (6.342) (5.736) (5.106) Nasdaq 0.0357 c 0.0289 c 0.0203 c (5.191) (4.081) (2.786) Rated -0.0310 c -0.0228 c (-4.175) (-3.050) Inverse Mills Ratio -0.1297 c (-6.441) Constant 3.1669 c 3.1286 c 3.1432 c (60.918) (59.564) (58.867) Observations 4,731 4,731 4,731 Adj. R2 0.597 0.599 0.600

Table 5: Distribution of Credit Rating Initiations This table provides summary statistics on the distribution of credit rating initiations by year, industry, and rating class. The sample consists of new long-term issuer credit rating initiations by Standard and Poor's over the time period January 1st, 1990 through December 31st, 2014. Firms are identified as having obtained a new issuer rating if their prior rating, as identified by Bloomberg Data Services, is either missing, blank, or contains a value of "NR" which identifies a firm as being not-rated. Firms are classified in to 17 industries following the classification methodology of Fama and French (1997).

No. of Fama-French No. of Issuer Credit No. of Year Observations 17-Industry Observations Rating Observations 1990 17 FOOD 23 AAA 1 1991 18 MINING 12 AA+ - 1992 21 OIL 62 AA 6 1993 16 CLTHS 12 AA- 9 1994 21 DURBL 10 A+ 13 1995 28 CHEM 20 A 24 1996 52 CNSUM 24 A- 37 1997 73 CNSTR 22 BBB+ 33 1998 72 STEEL 16 BBB 53 1999 52 FABPR 4 BBB- 56 2000 40 MACHN 94 BB+ 38 2001 26 CARS 16 BB 68 2002 35 TRANS 40 BB- 136 2003 22 RTAIL 56 B+ 129 2004 27 OTHER 327 B 95 2005 29 B- 30 2006 20 CCC+ 10 2007 20 CCC - 2008 15 CCC- - 2009 13 CC - 2010 23 C - 2011 28 2012 22 2013 22 2014 26 Total 738

Table 6: Validation of Control Sample This table reports summary statistics on the financial characteristics of the newly rated and matched control firms in the sample for the fiscal year end prior to the date of the rating initiation. Control firms are unrated firms matched based on the closest absolute-difference in their propensity score to that of an initiation firm following the methodology of Faulkender and Petersen (2006). Propensity scores are calculated from the probit estimation results provided in Appendix B. Statistical significance on differences in means and medians is computed using t-tests for mean estimates and k-sample tests for median estimates. Variable definitions are provided in appendix A. a, b, and c indicate statistical significance at the 10%, 5%, and 1% levels, respectively.

Matched New Issuer Rated Firms Unrated Firms Difference Variable N Mean Median Mean Median Mean Median SP500 738 0.148 0.0000 0.070 0.0000 0.0772 c 0.0000 c NYSE 738 0.580 1.0000 0.631 1.0000 -0.0515 b 0.0000 b Ln[1 + Pr(Rating)] 738 0.185 0.1457 0.182 0.1508 0.0029 -0.0051 Young 738 0.077 0.0000 0.047 0.0000 0.0298 b 0.0000 b Barclay's 738 0.008 0.0000 0.023 0.0000 -0.0149 b 0.0000 b Ln(Mkt. Assets) 738 7.536 7.4728 7.323 7.2129 0.2131 c 0.2599 c Ln(1 + Age) 738 2.584 2.5649 2.648 2.6391 -0.0642 a -0.0741 Profit 738 0.138 0.1532 0.114 0.1416 0.0244 0.0116 a Tangible Assets 738 0.321 0.2498 0.378 0.3268 -0.0569 c -0.0770 c Market-to-Book (Assets) 738 1.872 1.4498 1.615 1.3430 0.2570 c 0.1068 b Advertising/Sales 738 0.007 0.0000 0.006 0.0000 0.0007 0.0000 b σ(Asset Return) 738 0.095 0.0686 0.089 0.0560 0.0051 0.0125 c

Table 7: SEO Descriptive Statistics for Matched Sample This table presents descriptive statistics on seasoned equity offerings for newly rated and matched control firms following the initiation of a new long-term issuer credit rating. The sample covers the time period January 1st, 1990 through December 31st, 2014. SEO Flag is an indicator which takes a value of one if the firm issues a seasoned equity offer within three years following the new rating initiation date and zero otherwise. Variable definitions are provided in appendix A. Statistical significance on differences in means and medians is computed using t-tests for mean estimates and k-sample tests for median estimates. a, b, and c indicate statistical significance at the 10%, 5%, and 1% levels, respectively.

New Issuer Matched Rated Firms Unrated Firms Difference N Mean Median Std. Dev. N Mean Median N Mean Median Mean Median SEO Flag 1,476 0.1667 0.0000 0.3728 738 0.2602 0.0000 738 0.0732 0.0000 0.1870 c 0.0000 c

SEO Characteristics Offer Proceeds (million $) 246 194.8656 132.5000 232.8292 192 201.1677 136.9690 54 172.4580 98.8125 28.7097 38.1565 Issue Size (%) 246 0.3677 0.1836 0.7608 192 0.3205 0.1790 54 0.5353 0.2660 -0.2147 a -0.0870 a Gross Fee (%) 246 4.4685 4.5000 0.9250 192 4.3960 4.4985 54 4.7263 4.9935 -0.3303 b -0.4950 a Mult. Bookrunner 246 0.3862 0.0000 0.4879 192 0.3958 0.0000 54 0.3519 0.0000 0.0440 0.0000 29 Underwriter Reputation 246 0.0347 0.0097 0.0461 192 0.0361 0.0100 54 0.0301 0.0073 0.0060 0.0027 Firm Characteristics Liquidity Index 246 0.4327 0.4343 0.1868 192 0.4476 0.4514 54 0.3798 0.3938 0.0678 b 0.0576 a Ln(Market Assets) 246 7.2456 7.2171 1.0148 192 7.3114 7.2361 54 7.0116 6.9716 0.2998 a 0.2645 Share Price 246 22.1635 18.8050 14.8769 192 22.3845 19.3225 54 21.3777 16.6250 1.0068 2.6975 σ(Equity Return) 246 0.0289 0.0268 0.0142 192 0.0285 0.0265 54 0.0307 0.0275 -0.0022 -0.0010 Amex 246 0.0163 0.0000 0.1267 192 0.0156 0.0000 54 0.0185 0.0000 -0.0029 0.0000 Nasdaq 246 0.3577 0.0000 0.4803 192 0.3646 0.0000 54 0.3333 0.0000 0.0313 0.0000

Table 8: Propensity to Issue Seasoned Equity This tables reports the results from probit regressions where the likelihood that a firm issues seasoned equity following a credit rating initiation is modeled as a function of the financial characteristics of the firm and an indicator, New Rating, which takes a value of one if the observation is for a newly-rated firm and zero otherwise. ∆Leverage is the change in firm leverage from the quarter immediately preceding the credit rating initiation to the quarter immediately preceding the SEO offering. Prior Return is the equity return for a given firm in the six months preceding the SEO issuance stopping 10 days before the SEO issue date. Remaining variable definitions are provided in appendix A. All specifications compute robust standard errors clustered by Fama-French (1997) 17-industry. t-statistics are reported in the parentheses below the coefficients. a, b, and c indicate statistical significance at the 10%, 5%, and 1% levels, respectively

Three-Year Two-Year One-Year Dependent Variable = SEO Issue (1 if yes) (1) (2) (3) New Rating 0.8907 c 0.9761 c 1.1435 c (9.750) (10.041) (8.175) ∆Leverage 1.9569 c 0.9942 b -0.4965 (4.746) (2.180) (-0.828) Prior Return 0.3348 c 0.3121 c 0.2772 b (3.880) (4.368) (2.410) Ln (Liq. Index) -0.2556 -0.4830 -0.4647 (-0.690) (-1.396) (-1.229) Ln (Mkt. Assets) -0.0691 a -0.0234 0.0514 (-1.847) (-0.467) (0.645) Ln (1 + Age) -0.3440 c -0.4065 c -0.4517 c (-4.133) (-4.731) (-3.593) Profit -0.4305 c -0.4753 c -0.5802 c (-4.110) (-3.970) (-3.692) Tangible Assets -0.3051 b -0.3972 b -0.2057 (-2.378) (-2.095) (-1.418) Market-to-Book (Assets) -0.0785 c -0.0650 b -0.0656 a (-2.582) (-2.248) (-1.722) Advertising/Sales 0.0445 -4.6066 -17.4532 a (0.013) (-1.048) (-1.851) σ(Asset Return) -1.3575 b -1.8890 c -1.4920 c (-2.331) (-4.168) (-2.618) S&P500 -0.4358 b -0.5111 a -0.4591 (-2.138) (-1.732) (-1.631) NYSE 0.3003 b 0.3038 b 0.3101 a (2.494) (2.290) (1.675) Ln [1 + Pr(Rating)] 0.3355 0.0437 1.1628 (% of other firms in industry) (0.262) (0.038) (0.829) Young -0.5641 -0.5417 a -0.3718 (-1.620) (-1.764) (-1.073) Barclay's -0.2915 (-0.409) Constant -0.7828 b -0.8304 a -1.6710 c (-2.077) (-1.920) (-2.592) Observations 1,476 1,476 1,476 2 Pseudo R 0.214 0.220 0.235

Table 9: Credit Rating Initiations and SEO Investment Bank Fees This table reports the results of ordinary-least-squares testing on the relation between the investment bank fees charged at SEO offerings and the presence of a credit rating controlling for the financial characteristics of the issuing firm. New Rating is an indicator variable which takes a value of 1 if the observation if for a firm which obtained an S&P long-term issuer credit rating preceding the SEO offering and 0 otherwise. The inverse Mills ratio in Column (3) is computed using the estimates generated from the probit specification presented in Table 8. Variable definitions are provided in appendix A. All specifications include fixed effects for year and industry using Fama and French (1997) 17-industry classifications and compute robust standard errors clustered by industry. t-statistics are reported in the parentheses below the coefficients. a, b, and c indicate statistical significance at the 10%, 5%, and 1% levels, respectively.

Dependent Variable = Ln(Gross Fees) (1) (2) (3) Ln(Liquidity Index) 0.0066 0.0201 0.0071 (0.186) (0.671) (0.207) Ln(Proceeds) -0.0745 c -0.0750 c -0.0709 c (-2.998) (-3.116) (-3.075) Ln(Mkt. Cap.) -0.0561 c -0.0598 c -0.0577 c (-4.430) (-5.813) (-6.149) Ln(Share Price) 0.0032 0.0097 0.0273 (0.310) (0.744) (1.519) Ln(Equity Vol.) 0.0081 0.0057 0.0562 (0.198) (0.133) (1.109) Und. Reputation 0.1012 0.1133 0.0227 (0.370) (0.474) (0.078) Mult. Book Runner 0.0652 0.0677 a 0.0617 (1.621) (1.674) (1.490) AMEX -0.0288 -0.0283 -0.0251 (-0.425) (-0.392) (-0.320) Nasdaq 0.0011 0.0046 -0.0101 (0.032) (0.145) (-0.358) New Rating -0.0476 b -0.0781 b (-2.415) (-2.445) Inverse Mills Ratio -0.0285 (-0.853) Constant 2.2436 c 2.2922 c 2.4843 c (11.091) (12.105) (14.944) Observations 246 246 246 Adj. R2 0.560 0.566 0.570

Appendix A: Variable Definitions Variable Definition New Rating An indicator variable which takes a value of one if the firm obtains an S&P long-term issuer rating over the sample period, and zero otherwise. Panel A: Faulkender and Peterson Variables Rated An indicator variable which takes a value of one if the firm has an S&P long-term issuer rating, and zero otherwise.

SP500 An indicator variable which takes a value of one if the firm is a constituent of the S&P 500 index, and zero otherwise.

NYSE An indicator variable which takes a value of one if the firm's equity trades on the NYSE, and zero otherwise.

Ln[1 + Pr(Rating)] The natural log of one plus the percentage of firms in the same three-digit SIC industry with a long-term issuer credit rating.

Young An indicator variable which takes a value of one if the firm's age is three years or less, and zero otherwise.

Barclay's An indicator variable which takes a value of one if the firm's market value of assets times the median debt ratio is less than the minimum bond size required to be included in the Barclay's US Aggregate Bond Index (≤2002: $150 mil, 2003: $200mil, ≥2004: $250 mil).

Ln(Mkt. Assets) The natural log of the sum of the market value of the firm equity plus the book value of debt.

Ln(1+Firm Age) The natural log of one plus the number of years the firm has existed in Compustat.

Profit EBITDA scaled by total revenue.

Tangible Assets Net property, plant, and equipment scaled by book total assets.

Market-to-Book (Assets) The market value of firm assets (described above) scaled by its book value of assets.

Advertising/Sales Advertising expenses scaled by total revenue.

σ(Asset Return) The standard deviation of the firm's daily equity returns over a six-month period ending 10-days before the fiscal year-end date multiplied by the firm's equity-to-asset ratio.

Panel B: SEO Characteristics SEO Flag An indicator variable which takes a value of one if the firm issues seasoned equity in a defined interval, and zero otherwise.

Offer Proceeds The total dollar amount of the SEO offering in millions.

Issue Size SEO offer proceeds scaled by the market capitalization of the issuing firm 20-trading days prior to the date of the issue.

Gross Fee The total fees paid to the SEO underwriter scaled by the offer proceeds.

Mult. Bookrunner An indicator variable which takes a value of one if the SEO has more than one bookrunner, and zero otherwise.

Underwriter Reputation The market share of the lead underwriter in the year of the SEO.

Ln(Mkt. Cap.) The natural log of the market capitalization of the firm 10-days before the SEO date.

Share Price The price at which the firm's equity trades 10-days before the SEO date.

σ(Equity Return) The standard deviation of the firm's daily returns over a six-month period ending 10- days before the date of the SEO.

AMEX An indicator variable which takes a value of one if the firm's equity trades on the AMEX, and zero otherwise.

Nasdaq An indicator variable which takes a value of one if the firm's equity trades on the Nasdaq, and zero otherwise.

Liquidity Index Each observation is ranked from least liquid to most liquid across Amihud, Volume, Ask-Bid Spread, and Turnover. The average rank is computed for each observation and this value is scaled by the total number of observations. The liquidity Index ranges from zero to one.

∆Leverage The difference in leverage from the quarter-ending immediately preceding the SEO date less the leverage from the quarter-ending immediately preceding the initiation date where leverage is the total long-term debt plus the current portion of long-term debt scaled by book assets.

Prior Return The firm's equity return over a six-month period ending 10-days before the date of the SEO.

Appendix B: Construction of the Matched Sample This table provides the results from a probit estimation following Faulkender and Petersen (2006) where the likelihood that a firm has an S&P long-term issuer credit rating is modeled as a function of the firm's financial characteristics. The sample used is the universe of firm's who exist in both CRSP and Compustat over the time period 1990-2014. The specification includes fixed effects for year and industry using Fama and French (1997) 17-industry classifications and computes robust standard errors clustered by industry. t-statistics are reported in the parentheses below the coefficients. Variable definitions are provided in appendix A. a, b, and c indicate statistical significance at the 10%, 5%, and 1% levels, respectively.

Dependent Variable = Rated (1 if yes) SP500 0.626 c (36.065) NYSE 0.446 c (42.246) Ln[1 + Pr(Rating)] 0.034 (% of other firms in industry) (0.612) Young -0.207 c (-8.400) Barclay's -0.573 c (-19.930) Ln(Mkt. Assets) 0.455 c (108.287) Ln(1 + Age) 0.222 c (27.476) Profit -0.153 c (-12.794) Tangible Assets 0.794 c (40.056) Market-to-Book (Assets) -0.229 c (-46.038) Advertising/Sales 1.102 c (3.188) σ(Asset Return) 0.170 c (5.495) Constant -4.553 c (-120.149) Observations 157,271 Pseudo R2 0.480