Expanded Auditor’s Report Disclosures and Loan Contracting

University of Groningen

Expanded Auditor’s Report Disclosures and Loan Contracting

Porumb, Vlad; Zengin-Karaibrahimoglu, Yasemin; Lobo, Gerald J. ; Hooghiemstra, Reggy; Waard, de, Dick

Published in:

Contemporary Accounting Research DOI:

10.1111/1911-3846.12697

IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from it. Please check the document version below.

Document Version

Version created as part of publication process; publisher's layout; not normally made publicly available

Publication date: 2021

Link to publication in University of Groningen/UMCG research database

Citation for published version (APA):

Porumb, V., Zengin-Karaibrahimoglu, Y., Lobo, G. J., Hooghiemstra, R., & Waard, de, D. (2021). Expanded Auditor’s Report Disclosures and Loan Contracting. Contemporary Accounting Research.

https://doi.org/10.1111/1911-3846.12697

Copyright

Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).

Take-down policy

If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.

Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum.

Expanded Auditor’s Report Disclosures and Loan Contracting

C.T. Bauer College of Business,

University of Houston Department of Accountancy & Taxation

gjlobo@uh.edu Reggy Hooghiemstra University of Groningen Department of Accounting r.b.h.hooghiemstra@rug.nl Dick de Waard University of Groningen Department of Accountancy d.a.de.waard@rug.nl

* Accepted by Miguel Minutti-Meza. We are grateful to Daniel Bens, John Donovan, Aloke (Al) Ghosh, Zhongwei Huang, Miguel Minutti-Meza, Haresh Sapra, Joe Schroeder, Aleksandra Zimmerman, seminar participants at the University of Groningen, KU Leuven, Vrije Universiteit Amsterdam, University of Miami, University of Memphis, University of Houston, and University of Zurich, and participants at the 2017 Dutch Accounting Research Conference (Tilburg University), 2018 AAA Auditing Section Midyear Meeting (Portland), 2018 European Accounting Association Congress (Milan), 2018 International Symposium on Audit Research (Maastricht), and 2018 PCAOB/JAR Conference on Auditing and Capital Markets (Washington, DC) for many helpful comments. The paper previously circulated under the title “Is More Always Better? Disclosures in the Expanded Audit Report and Their Impact on Loan Contracting.”

†Corresponding author.

This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process which may lead to differences between this version and the Version of Record. Please cite this article as doi: 10.1111/1911-3846.12697

Accepted

Expanded Auditor’s Report Disclosures and Loan Contracting

ABSTRACT

Starting in October 2013, auditors of premium-listed firms in the United Kingdom are mandated to prepare an expanded auditor’s report that provides details on audit procedures, risks of material misstatement (RMMs), and materiality thresholds. This regulatory change is important to study, because it aims to increase the informational value of the traditional, highly standardized, pass-or-fail auditor’s report. We examine whether the disclosures in the expanded auditor’s report provide information that is relevant for adopting firms’ loan contracting terms in the post-adoption period. Our results indicate that the introduction of the expanded auditor’s report is associated with reduced loan spread and longer maturity for loan facilities of adopting firms relative to non-adopting UK firms. When we focus on adopting firms in the post-adoption period, we find that the number of “unique RMMs” mentioned in the auditor’s report, but not in the audit committee report, are positively associated with loan spread but are not associated either with loan maturity or the number of lenders in the loan syndicate. Additional tests show that the benefits, in terms of a reduced spread, of having a lower number of “unique RMMs” accrue mostly to adopters with a poor information environment. Taken together, our results provide preliminary evidence that the expanded auditor’s report disclosures contain relevant information for loan contracting in the United Kingdom. This study highlights the unique role of the expanded auditor’s report in providing information relevant to lenders and supports standard setters’ efforts to enrich its informational content.

Keywords: expanded auditor’s report, audit risk disclosures, risks of material misstatement (RMMs), loan contracting terms

JEL classification: G20, G21, G32, K22, M42

Accepted

1. Introduction

To enhance the information content of the auditor’s report, regulators and audit standard setters worldwide took steps to require enhanced disclosures beyond the traditional “pass-or-fail” model (IAASB 2015a; PCAOB 2017). Starting in October 2013, the United Kingdom’s Financial Reporting Council (FRC) issued International Standard on Auditing (ISA) 700, making it the first to require auditors of listed firms to disclose such an “expanded” auditor’s report. Specifically, ISA 700 mandates that auditors of UK premium-listed firms on the London Stock Exchange (LSE) Main Market disclose significant risks of material misstatement (RMMs), how the audit procedures addressed these risks, and the materiality thresholds used in the auditor’s report.1

Several studies assess the information content of the expanded auditor’s report from an equity market perspective in the United Kingdom (e.g., Gutierrez et al. 2018; Reid et al. 2019; Lennox et al. 2021) and find mixed results. Reid et al. (2019) document higher earnings response coefficients for premium-listed firms in the post-adoption period relative to the pre-adoption period, which suggests that the expanded auditor’s report has information content for equity market participants. Gutierrez et al. (2018) and Lennox et al. (2021), on the other hand, do not find that equity market participants react to the expanded disclosures in the auditor’s report. We attribute the lack of clear empirical evidence to the characteristics of the UK setting, which is not conducive to finding results from an equity perspective. Specifically, the pre-earnings announcements of UK premium-listed firms are very detailed, as they consist of complete financial statements as well as the audit opinion (Gutierrez et al. 2018). The comprehensive nature of these disclosures suggests that equity investors are likely to

1 _{The adoption of the expanded auditor’s report in the United Kingdom is particularly important, because the FRC}

announced it in June 2013, only four months before its mandatory adoption. Therefore, relative to other settings, where the adoption of the new report can be foreseen and preempted by preparers and auditors through pre-adoption voluntary disclosures, the United Kingdom provides the only instance where the regulatory change is unexpected.

Accepted

incorporate this information in investment decisions at the time of the pre-earnings announcements, consequently putting less emphasis on the annual report information. By extension, this may explain why equity investors do not seem to use information from the expanded auditor’s report disclosures.2

In this study, we examine the informativeness of the disclosures in the expanded auditor’s report for the private debt market. This is potentially important because providers of debt and equity not only draw different payoffs from their investments, but also have different risk preferences (Hasan et al. 2014; Chiu et al. 2018). Moreover, relative to other types of investors, who rely mainly on publicly available information, private lenders assign more importance to the auditor’s report (i) in decision making and (ii) when estimating a firm’s future viability (Asare and Wright 2012). We focus on the information content of the expanded auditor’s report on one segment of the private debt market, syndicated loans, which are loans for which a borrowing firm obtains a loan from a group of lenders. Focusing on the information content of the expanded auditor’s report on the private debt market is important because research on the role of auditing in debt markets is relatively limited (DeFond and Zhang 2014; Baylis et al. 2017). Additionally, private debt represents a major source of capital for firms worldwide - companies raise more capital from private lending than from equity markets and public debt combined (Sufi 2007; Ferreira and Matos 2012).

The relevance of the information in the auditor’s report for private debt market participants is the subject of considerable academic debate. One stream of literature advocates that the auditor’s report assists lenders to more accurately assess a borrower’s risks because it provides credible information on the quality of the accounting numbers (Blackwell et al. 1998; Christensen et al. 2014) and transmits auditors’ private information about the creditworthiness

2 _{Another potential explanation for the lack of consistent findings of this stream of research embedded in the UK}

setting is that annual report filing dates from databases (e.g., Worldscope, Bloomberg) or other open sources are largely unreliable (Gutierrez et al. 2018; Lennox et al. 2021).

Accepted

of borrowers (Chen et al. 2016). In contrast, another stream of literature highlights that due to lenders’ access to private information, they are less likely to use information incorporated in the auditor’s report in loan contracting (e.g., Cheng et al. 2019; Carrizosa and Ryan 2017; Baylis et al. 2017). This may be of particular importance in the context of syndicated loans because they are associated with at least two types of ex ante information asymmetries (e.g., Ball et al. 2008a; Ivashina 2009; Bharath et al. 2011; Fang et al. 2016), one between the borrowing firm and the lenders and another within the syndicate of lenders. Given that the expanded auditor’s report has a wider scope and includes additional disclosures, such as information regarding materiality levels and RMMs, relative to the traditional pass-or-fail report, it could arguably cater to the informational needs of lenders and have implications for loan contracting. This motivates our first research question, which is: Is the adoption of the expanded auditor’s report associated with changes in loan contracting terms?

Our second research question relates to cross-sectional differences in the information content of the expanded auditor’s report for loan contracting terms of adopting firms in the post-adoption period. More specifically, we examine how banks adjust their lending terms within the sample of adopting firms, conditional on disclosures in the expanded auditor’s report. The literature on debt contracting posits that the debt market is characterized by imperfect information (Leland and Pyle 1977) and strong adverse selection problems (Jaffee and Russel 1976; Stiglitz and Weiss 1981), which may be particularly the case for the private debt market, in general, and the syndicated loans market, in particular. From the pool of loan applicants, lenders attempt to distinguish “good” or high-quality borrowers (i.e., those that are more likely to repay their loans) from “poor” or low-quality borrowers (i.e., those that are less likely to repay their loans). This screening process is difficult because lenders have an information disadvantage, and borrowers have incentives to supply positive information about their firms when requesting debt financing (Leland and Pyle 1977). To overcome this information

Accepted

disadvantage, lenders rely on information from external parties to perform an initial screening of borrowers and determine how much effort is required in their due diligence process and subsequent monitoring (Best and Zhang 1993). Since lenders face an asymmetric payoff structure on loans, their risk assessments are particularly focused on downside risk (Bae et al. 2013; Hasan et al. 2014; Florou and Kosi 2015; Chiu et al. 2018), which drives their preference for negative information (Ball et al. 2008 a, b). In the context of the expanded auditor’s report, RMMs represent disclosures on downside risks that provide a credible signal on a borrower’s risk of financial misstatement (Christensen et al. 2014).3 _{Consequently, if RMMs are indicative}

of risks, lenders are likely to incorporate this information into loan contracting terms. We, therefore, formulate our second research question as follows: Are RMM disclosures in the expanded auditor’s report associated with loan contracting terms?

We perform two sets of tests to answer our research questions. We first assess whether the loan contracting terms differ following the switch from the traditional, pass-or-fail auditor’s report to the expanded auditor’s report. Based on a sample of UK firms listed on the LSE Main Market with syndicated loans in the two-year window around the introduction of the expanded auditor’s report, we find that relative to UK non-adopters, loan facilities of adopters have lower spreads and longer maturities in the post-adoption period. These differences are largely attributable to the significant pre-post change in spread and maturity for the adopting firms.4

Specifically, we find that a typical borrowing firm experiences a decrease in the average all-in-drawn spread of approximately 38 basis points after the adoption of the expanded auditor’s report. The results are robust to using both shorter and longer windows around the adoption

3 _{We focus on RMMs in the second part of our analyses because they represent a common feature of all standard}

setters’ policy proposals for enhancing the information content of the current auditor’s report (Lennox et al. 2021). According to the joint description issued by the PCAOB, the IAASB, and the FRC, RMMs convey information about sensitive practices that are likely to be associated with decreases in firm value.

4 _{In line with Gutierrez et al. (2018) and Reid et al. (2019), we use firms listed on the LSE Alternative Investment}

Market (AIM) segment to construct our control sample (non-adopters), because these firms are not required to adopt the expanded auditor’s report.

Accepted

date and a placebo test, as well as to using alternative difference-in-differences (DID) designs, where we use comparable non-adopting UK firms and US firms as control samples. However, in none of the model specifications do we find a significant association between the adoption of the expanded auditor’s report and the number of lenders. Overall, results suggest that the introduction of the expanded auditor’s report had an impact on the contractual, but not on the structural terms of syndicated loans.

To answer our second research question, we focus solely on the adopting firm sample and assess the relationship between loan contracting terms and the number of RMMs disclosed in the expanded auditor’s report. To better gauge these relationships, we focus on unique RMMs— that is, those RMMs that are not also disclosed in the audit committee’s report.5 _{The results}

indicate a positive relation between unique RMMs and loan spread. Specifically, a decrease of one unique RMM is associated with an average decrease in spread of approximately 20 basis points. We do not find a reliable relationship between either loan maturity or number of lenders and unique RMMs. In supplemental analyses, we obtain similar results when we use unexpected (or abnormal) unique RMMs when we consider unique and relevant RMMs (i.e., unique RMMs that are more likely to be relevant for lenders),6 _{when we control for the quality of a firm’s}

corporate governance, and when we employ a seemingly unrelated regression approach to take into account that loan contracting terms may be jointly determined. We further show that these relationships hold primarily for borrowers with a poorer information environment.

Our study makes several contributions. First, it contributes to the growing body of research on the information content of the expanded auditor’s report (e.g., Bédard et al. 2019; Gutierrez et al. 2018; Liao et al. 2019; Reid et al. 2019; Zeng et al. 2020; Minutti-Meza, 2020; Lennox et

5 _{See Appendix 1 for examples of unique and non-unique RMMs.}

6 _{We consider lenders to be particularly concerned about RMMs dealing with fraud, management override of}

internal controls, asset valuation, going concern, credit risk, goodwill impairment, and IT failures.

Accepted

al. 2021).7 _{Whereas these studies focus on the information content for equity market}

participants of adopting the expanded auditor’s report in the United Kingdom (Gutierrez et al. 2018; Reid et al. 2019; Lennox et al. 2021), France (Bédard et al. 2019), Mainland China (Zeng et al. 2020), and Hong Kong (Liao et al. 2019), we assess its importance from a private debt market perspective. To the best of our knowledge, ours is one of the first studies to empirically assess the information content or usefulness of the expanded auditor’s report for the private debt market. This is potentially important given that the findings of prior studies on equity market consequences may not be generalizable to the debt market because both the payoffs and the risk preferences of debt and equity providers differ (Hasan et al. 2014; Chiu et al. 2018). By assessing the importance of the expanded auditor’s report for the private debt market, we also respond to the call for research on the role of auditing in debt markets (DeFond and Zhang 2014; Baylis et al. 2017). We are aware of two studies that rely on an experimental approach to investigate whether the information in the expanded auditor’s report influences the decision on whether to grant a loan (Ruhnke et al. 2018; Boolaky and Quick 2016). Both studies use German banks’ executive board members but find mixed results; Ruhnke et al. (2018) find that information disclosed in the expanded auditor’s report has implications for credit lending conditions, whereas Boolaky and Quick (2016) do not. Unlike these two studies, which focus only on the likelihood of granting a (private) loan, our study assesses the information content of the expanded auditor’s report for the contractual and structural terms of syndicated loans.

Second, our study complements the recent stream of research that examines the auditor’s role in private debt contracting (e.g., Menon and Williams 2016; Chen et al. 2016; Francis et al. 2017; Robin et al. 2017). These studies underscore the unique role of auditors and the

7 _{Papers relying on experimental research design also show mixed results. While Carver and Trinkle (2017) are}

unable to document that the new disclosures are useful, other studies document that investors do find the disclosures in the expanded auditor’s report useful for their decision-making (e.g., Christensen et al. 2014; Kachelmeier et al. 2020; Moroney et al. 2020; Rapley et al. 2020) or reduce auditor liability judgments (Brasel et al. 2016).

Accepted

auditor’s report in providing information to the private debt market. For instance, while Chen et al. (2016) examine the effects of modified audit opinions on the cost of debt and covenants, Francis et al. (2017) investigate the association between auditor changes and the cost of debt. We add to this emerging stream of research by documenting the relationship between the new disclosures in the expanded auditor’s report and debt contracting terms.

Third, the results of our study are relevant to policy makers because they provide evidence on the relevance of the adoption of the expanded auditor’s report to a specific group of investors - providers of private debt. Indeed, a major goal of the expanded auditor’s report is to allow investors to better incorporate the uncertainties inherent in financial statements when making their investment decisions (Kelton and Montague 2018). In this regard, our private debt market evidence provides some support for standard setters’ efforts to enrich the informational content of the current auditor’s report (EU law 2014; IAASB 2015a; PCAOB 2017).

2. Institutional background, related literature, and research questions

The structure of the UK auditor’s report8

Responding to calls to increase the informational value of the auditor’s report, regulators and standard setters (e.g., IAASB, PCAOB, and FRC) proposed the use of an expanded auditor’s report to replace the highly standardized, pass-or-fail auditor’s report. According to these recently implemented standards, the auditor’s report should include information regarding the auditor’s responsibilities, the procedures applied during the audit, and the materiality levels used. One of the most significant changes is the disclosure in the auditor’s report of Key Audit Matters (KAMs; IAASB 2015a), Critical Audit Matters (CAMs; PCAOB 2013, 2017), and the Risks of Material Misstatements (RMMs, FRC 2013). Comparable to KAMs and CAMs, RMMs can be defined as “[t]hose matters that, in the auditor’s professional judgment, were of

8 _{For more detailed information about the history of the expanded auditor’s report, see Minutti-Meza (2020). This}

paper also provides an excellent discussion of the implementation in various jurisdictions, including the United Kingdom and the United States.

Accepted

most significance in the audit of the financial statements of the current period” (ISA 701, paragraph 8).9 _{Furthermore, according to ISA 701 (paragraph 13), auditors have to explain not}

only why the matter was considered to be one of the most significant but also how the risks were addressed in the audit.

As part of the IAASB’s consultations prior to issuing its draft standards, several interested parties emphasized the importance of the expanded auditor’s report from a debt market perspective. For instance, the Basel Committee on Banking Supervision (BCBS) stated that “in the credit decision process, improvements to the auditor’s report may help banks improve their overall credit risk management” (BCBS 2012, 1). Moreover, Standard & Poor’s (S&P’s) highlighted that “the expanded disclosures…will create a better understanding of the financial statements and provide relevant information on issues identified in the audit that are meaningful to credit analysts,” and “will better enable users to understand and compare the financial condition, results, and cash flows among peer companies” (S&P 2012, 2).

In June 2013, the FRC in the UK issued International Standard on Auditing (UK and Ireland) 700, “The Independent Auditor’s Report on Financial Statements.”10 _{This standard}

requires that auditors of premium-listed companies on the LSE present an expanded auditor’s report for fiscal years ending on or after October 1, 2013. Besides including a discussion of the RMMs, ISA 700 requires an explanatory paragraph of how the auditor applied the concept of materiality and a discussion of the audit scope (FRC 2013). The two reviews (2015, 2016) conducted by the FRC after the implementation suggest that investors value the enhanced disclosures. For instance, the 2016 review highlights not only that disclosures about risk, scope,

9 _{As indicated by Gutierrez et al. (2018, 1548) “[a]n important new requirement in both the IAASB and PCAOB}

standards is to communicate information about key audit matters (KAMs in the IAASB proposal) or critical audit matters (CAMs in the PCAOB proposal). Similar to the UK requirements regarding the assessment and disclosure of risks of material misstatement identified by the auditor [RMMs, added by the authors], the purpose of both KAMs and CAMs is to provide additional information about the auditor’s work.” Similar to the United Kingdom, EU-Regulation No 537/2014 prescribes that the auditor’s report shall provide, among other disclosures, a description of the most significant risks of material misstatement.

10 _{ISA (UK and Ireland) 700, “The Independent Auditor’s Report on Financial Statements” should be read together}

with ISA (UK and Ireland) 701 “Communicating Key Audit Matters in the Independent Auditor’s Report”.

Accepted

and materiality are perceived as clear, concise, and transparent, but also that “the language used in auditors’ reports move away from generic risk descriptions and language generally, in favor of more granular descriptions” (FRC 2016a, 4).

In the US, the PCAOB issued a new auditing standard on the auditor’s report on June 1, 2017, which was approved by the SEC in October 2017. According to the standard, the auditor’s report should include CAMs as well as information on audit firm independence and auditor tenure. The standard requires the expanded auditor’s report to be fully implemented by large, accelerated filers for fiscal years ending on or after June 30, 2019, and for fiscal years ending on or after December 15, 2020, for other companies.

Syndicated loans and the relevance of publicly available information

In this study, we focus on the implications of the adoption of the expanded auditor’s report for one important part of the private debt market, syndicated loans. Syndicated loans, in which a borrowing firm obtains a loan from a group of lenders, represent a major source of finance for listed firms worldwide (e.g., Ivashina 2009; Amiram et al. 2017). These loans involve two types of lending parties, namely the “lead arranger” (or lead lender) and the participating lenders. In a syndicated lending set-up, the lead lender receives a fee from the loan syndicate to perform due diligence on the borrowing firm and to decide on the initial contract terms (Sufi 2007). According to Best and Zhang (1993), lead lenders use third-party information to perform an initial screening of borrowers and, depending on this assessment, determine how much effort the due diligence process is likely to require. If the information obtained is generated by reliable sources, and if it indicates a positive borrower outlook, lending parties significantly reduce their subsequent monitoring efforts.

Syndicated loans come with at least two types of ex ante information asymmetries (e.g., Ball et al. 2008a; Ivashina 2009; Bharath et al. 2011; Fang et al. 2016). The first type relates mainly to information asymmetry between the borrowing firm and the lead lender. Despite the

Accepted

lead lender’s direct access to the borrowing firm and notwithstanding the possibility that the two parties may have established a relationship via previous interactions, the lead lender always faces the risk, inherent in any agency setting (Jensen and Meckling 1976; Christensen et al. 2016), that the borrowing firm provides information opportunistically in an attempt to secure more favorable loan contracting terms. The second type of asymmetry relates to the adverse selection problem the participating lenders face. As the participating lenders lack direct access to the borrowing firm prior to signing the loan contract, they do not know whether the lead lender is fully forthcoming about the (lack of) quality of the loan. Furthermore, while participating lenders rely on the lead lender to perform costly due diligence on the borrowing firm, they cannot actually observe it, thus creating the potential that the lead lender shirks from conducting a thorough due diligence (e.g., Ball et al. 2008a).11

The literature on debt contracting (e.g., Armstrong et al. 2010; Christensen et al. 2016) suggests that publicly available accounting information plays a crucial role in reducing information asymmetry between contracting parties involved in a syndicated loan in at least two ways. First, the lead lender screens and processes accounting information about the borrowing firm’s performance, future cash flows, and risks to assess the probability of periodic interest payment and timely repayment of the loan, and hence the risk of default, as well as to estimate the market value of the underlying collateral (Chen et al. 2016). Second, accounting information helps in reducing adverse selection problems the “less-informed” participating lenders face by enabling them to evaluate the quality of the loan themselves and, hence, to assess whether the lead lender is fully forthcoming about the (lack of) quality of the loan, and to thus mitigate their concerns about the lead lender shirking on the due diligence.

11 _{We acknowledge that the literature mentions instances when syndicated loan participants have access to private}

information about firms or have private access to auditors through the lead lender (Carrizosa and Ryan 2017; Baylis et al. 2017). Nonetheless, the access to borrower information is costly and syndicate participants are likely to scrutinize borrowers only when their perceived riskiness is high or when they have serious concerns regarding the lead lender’s ability or incentives to perform due diligence on the borrower (Bharath et al. 2008).

Accepted

More recently, scholars have demonstrated the usefulness of information communicated in the auditor’s report for debt contracting purposes (e.g., Minnis 2011; Chen et al. 2016). Although our knowledge of the role of auditing in debt markets remains limited to date (Baylis et al. 2017), it does seem that information presented in the auditor’s report is useful in the private debt contracting process because it provides lenders with credible information on the quality of the accounting numbers (e.g., Christensen et al. 2014), helps to reduce uncertainty about future cash flows, and incorporates the auditor’s private information about the borrower’s creditworthiness (DeFond and Zhang 2014; Chen et al. 2016). Moreover, the auditor’s report alleviates the concerns of participating lenders about being misled and/or exploited by the lead lender, given that they do not have direct access to the borrowing firm (e.g., Ivashina 2009; Kim and Song 2011). While these arguments also apply to the “pass-or-fail model” auditor’s report (Asare and Wright 2012), the additional disclosures mandated in the expanded auditor’s report are likely to further reduce the information asymmetry between the parties involved in syndicated loans, affect their risk assessments, and, ultimately, the contractual terms at origination date.

Research questions

Current empirical research analyzing the UK setting finds mixed evidence regarding the information content of disclosures in the expanded auditor’s report for equity market participants (Gutierrez et al. 2018; Reid et al. 2019; Lennox et al. 2021). So a key question is, why, given that prior research does not seem to find information content of the expanded auditor’s report for equity investors, would the report be likely to inform private debt market participants?

One possibility is that decisions about the terms of a syndicated loan at origination can be characterized as “episodic” because borrowing firms secure funds from this source of the private debt market on an irregular basis. Moreover, it is likely that the group of lenders

Accepted

involved in a given firm’s current syndicated loan is different from the group involved in a syndicated loan secured on a prior occasion. This means that lenders are unlikely to systematically follow a firm during a period in which they do not have a contractual relationship with that firm. When a borrowing firm requires financing, both the lead lender and, at a later stage, the participating lenders search for and process firm-specific accounting information to decide on the terms of a syndicated loan. The relatively short period in which these syndicated loans are set-up, negotiated, and contracted (Rhodes and Campbell 2004) puts considerable pressure on lead lenders to use relevant and credible information when they have to assess the borrowing firm’s creditworthiness. In support of this view, some suggest that the auditor’s report assists lenders in this process to more accurately assess a borrower’s risks because it provides credible information on the quality of the accounting numbers (Christensen et al. 2014) and at the same time transmits the auditor’s private information about the creditworthiness of the borrower (Chen et al. 2016).

Further, research suggests that, relative to equity investors, private lenders are likely to rely more on the work of external auditors. Specifically, relative to other types of investors, which mainly rely on public information, private lenders attach more importance to the auditor’s report both (i) in decision making and (ii) when estimating a firm’s future viability (Asare and Wright 2012). If the expanded auditor’s report disclosures contain relevant information, private lenders would be more likely to consider these disclosures when shaping loan contracting terms.

Another reason is that the due diligence process gives rise to an agency conflict between the lead lender and the other participating syndicate members. To mitigate the information asymmetry between parties, the lead lender has incentives to use observable external information to justify its decisions regarding lending terms to the other members of the loan syndicate (Call et al. 2018).12 _{Given that the auditor’s report represents an important}

12 _{Although non-lead lenders in a syndicate rely on the lead lender to perform due diligence on the borrower, they}

also access (or use) publicly available information to assess the borrower’s credit riskiness before agreeing to the

Accepted

information source for lenders (Booth 1992), participating syndicate members will, therefore, consider it as a credible source that could justify lead lenders’ choice of loan contracting terms.

Although the new disclosures in the expanded auditor’s report have the potential to be relevant when the contractual terms are set at the origination date, their impact could also be negligible due to reasons related to lenders’ access to alternative information channels (e.g., Bushman et al. 2010). The extensive access to borrowers’ private information could provide lenders with information on sensitive firm practices, which could render the information in the expanded auditor’s report uninformative. Apart from the direct access to the management of the borrowing firm, in certain instances, lenders have been documented to obtain borrower-specific information directly from the external auditor. For example, according to a recent US-based study by Cheng et al. (2019), in certain instances, lenders have the ability to access borrower-specific information through special clauses added to loan contracts. Specifically, these authors argue that contractual clauses allow the lender to request: “(1) any direct communication and discussion between the lender and auditor on borrower issues, (2) provisions providing lender with copies of written communications between auditors and borrowers…” (Cheng et al. 2019, 1).

Overall, given the multiple information channels available to banks when searching for borrower-specific information, whether the expanded auditor’s report has implications for loan contracting is an empirical question.We, therefore, formulate our first research question as:

RESEARCH QUESTION 1 (RQ1): Is the adoption of the expanded auditor’s report associated

with changes in loan contracting terms?

Following the introduction of the expanded auditor’s report, research has emphasized the importance of auditors’ disclosure of financial reporting risks (i.e., RMMs) (e.g., Gutierrez et

lending terms of the syndicate. The lead lender therefore needs to base the proposed terms of lending on public information about the borrower that is easily observable by the other syndicate participants (Call et al. 2018).

Accepted

al. 2018; Reid et al. 2019; Lennox et al. 2021). As observed by Lennox et al. (2021, 8) “The auditor can issue an unqualified opinion together with RMMs if the auditor believes the audited financial statements are fairly stated. In other words, the auditor can issue an unqualified opinion as long as the auditor believes the financial reporting risks were mitigated during the course of the audit.”13 _{For various reasons outlined below, RMMs are likely to be relevant for}

lenders.

Given that the providers of debt face an asymmetric payoff structure on loans, their risk assessments are particularly focused on downside risk (Bae et al. 2013; Hasan et al. 2014; Florou and Kosi 2015; Chiu et al. 2018), which likely drives their preference for negative information (Ball et al. 2008 a, b). In a recent report, the PCAOB indicates that the CAMs (which are similar to RMMs) disclosed in the expanded auditor’s report should provide information regarding “especially challenging, subjective, or complex aspects of the audit as they relate to the relevant financial statement accounts and disclosures” (PCAOB 2016, 2) and that they should represent “areas of high financial statement and audit risk; unusual transactions; and other significant changes in the financial statements” (PCAOB 2016, 2). As this type of disclosure is indicative of borrower riskiness and likely corresponds to lenders’ informational needs, we expect that it will be reflected in the contract terms of syndicated loans. Moreover, these disclosures likely provide a “roadmap to help users better navigate complex financial reports” (IAASB 2011, para. 36) and may assist users in prioritizing the most significant financial reporting risks (CFA Institute 2013). These characteristics make RMMs highly salient and thus likely to draw lenders’ attention. We, therefore, investigate whether the informativeness of the expanded auditor’s report adoption is contingent upon the number of disclosed RMMs. More specifically, we focus on unique RMMs (i.e., RMMs mentioned in the auditor’s report, but not in the audit committee report), as they are likely to capture new

13 _{Please note that the disclosure of RMMs, despite differences in terminology used by standard setters and}

regulators, is the main commonality in the new standards on more extensive auditor reporting across jurisdictions.

Accepted

information that is not disclosed by the audit committee or management in the annual reports. The purpose of this analysis is to investigate whether or not the number of unique RMMs is associated with loan contracting terms. As with RQ1, we acknowledge that in the case of RMMs, similar to the overall disclosures in the expanded auditor’s report, the effect could be insignificant if banks obtain relevant risk-related information via private information channels. Moreover, to the extent that RMM disclosures are relevant for loan contracting, it is not obvious as to what direction the effect might be. On the one hand, since lenders are likely to use the number of RMMs as a pre-screening device to reduce their adverse selection problem, they will allocate fewer resources for the ex ante screening of firms perceived as less risky (Strahan 1999). In turn, this will lead to less use of resources when lenders subsequently perform due diligence and improve the loan contracting terms of firms perceived to be less risky in the post-adoption period. On the other hand, we acknowledge that the potential effect of RMM disclosures on loan contracting terms could arise from increasing lenders’ confidence in the numbers disclosed in the financial statements. Specifically, the disclosures could provide lenders comfort, as they would be likely to perceive that the RMMs have been addressed by the external auditors.

Accordingly, we formulate the following research question:

RESEARCH QUESTION 2: Are RMM disclosures in the expanded auditor’s report associated

with loan contracting terms?

3. Sample and research design

Sample

Auditors of premium-listed firms on the LSE are required to disclose an expanded auditor’s report for fiscal years ending on or after October 1, 2013. Therefore, our treatment sample consists of mandatory expanded auditor’s report adopters in the United Kingdom. We start our sample selection procedure with available syndicated loan data for all non-financial UK firms

Accepted

on the Thomson Reuters LPC’s DealScan database between 2010 and 2016.14 _{We use the}

DealScan-Compustat link from Chava and Roberts (2008) to identify borrowers’ GVKEYs in DealScan, and then use Compustat Global to match GVKEYs with ISIN codes. This process allows us to successfully match 35.34% of the loans (1,285 loans for 299 firms). To increase our sample size, we use the websites of LSE and Bloomberg and manually identify ISIN codes for 756 firms in the United Kingdom with loan information in DealScan. We further search the company name in DealScan and manually match the borrower ID and ISIN codes to successfully identify an additional 14.71% of the loans (535 loans from 155 firms).15 _After

identifying borrowers’ ISIN codes, we determine which ones pertain to the LSE premium-listed firms. Further, following Kim, Song, and Zhang (2011), Kim, Tsui, and Cheong (2011), and Florou and Kosi (2015), we use various company identifiers (i.e., GVKEY, ISIN, CUSIP, ticker, and company name) to match loan facilities with borrower-specific financial data from Worldscope. We restrict the sample to firms with financial reporting dates within two years before and two years after the expanded auditor’s report mandate.16 _{In line with Brown (2016),}

we use the most recent annual auditor’s report data publicly available before the loan contract is signed. We collect borrower-specific financial data from Worldscope. To construct our control sample, we consider all non-financial firms with available syndicated loan data that are listed on the AIM market segment of the LSE. These firms were not mandated to adopt the expanded auditor’s report in 2013 and share a common institutional setting with our treatment firms. The disadvantage of using this control sample is that firms listed on the AIM segment are generally younger, growing, and smaller than the firms with a premium listing (Gutierrez et al. 2018; Reid et al. 2019; Florou et al. 2020).

14 _{We exclude financial firms (4-digit SIC codes 6000–6999) and loans with sole lenders.}

15 _{We detected differences between the DealScan-Compustat link from Chava and Roberts (2008) and the}

hand-collected ISIN codes. After confirming the validity of our collection process, we used the hand-hand-collected ISIN codes in our data selection process.

16_{To decrease the effect of other potential confounding events around the adoption, following DeFond et al. (2015)}

and Gutierrez et al. (2018) we use a (-2 to +2) year window.

Accepted

After eliminating observations with missing borrower and loan-specific controls, we have a final sample of 561 facility-year observations from 204 adopting firms (henceforth, Adopters) and 174 facility-year observations from 54 non-adopting control firms (henceforth, Non-Adopters), which we use for addressing RQ1.17 _{Further, to understand whether our sample}

represents the full universe of premium-listed firms in the LSE, we test whether characteristics (Firm size, ROA, Leverage, and MTB) of the premium-listed firms included in our sample differ from those of other premium-listed firms. The latter firms also had to adopt the expanded auditor’s report but were not included in our sample because they were not covered by the DealScan database. Our results (see Appendix 2) indicate that the two groups of firms do not significantly differ from each other, except for firm size.

To address RQ2, we focus on the post-adoption period, starting on October 1, 2013, and ending on December 31, 2016. This filter results in a restricted sample of 301 facility-year observations from 167 firms. For each firm, we hand-collect the information on the characteristics and content of the expanded auditor’s report. Specifically, we obtain the number of RMMs, the number of words in the material risk statement section of the auditor’s report, and the materiality threshold. Table 1 defines the variables used in our analyses, and Table 2 summarizes our sample selection procedure for Adopters.

Empirical models

We build on previous literature to analyze the relationship between the expanded auditor’s report disclosures and loan contracting terms (Bharath et al. 2008; Kim, Tsui, and Cheong 2011; Florou and Kosi 2015). Specifically, in line with Melnik and Plaut (1986), who state that syndicated lending results in packages of various terms that need to be assessed collectively, and similar to several recent studies (e.g., Bharath et al. 2008, 2011; Kim, Tsui, and Cheong

17 _{In line with previous studies, we conduct a loan level analysis. As loans have different features, we consider}

each loan in our sample as a separate observation.

Accepted

2011; Florou and Kosi 2015), we analyze the contractual (interest rate spread and loan maturity) and structural (number of lenders) terms of syndicated loans. We focus on interest rate spread because banks use it to protect themselves from ex ante information risk (Diamond and Verrecchia 1991). As an outcome of the borrower screening process, it is positively associated with perceived risk. We focus on loan maturity because banks use it to control the frequency of the ex post loan renegotiation/renewal process. Therefore, it has a direct effect on the monitoring ability of lenders; shorter loan maturities improve the ability of lenders to ex post monitor borrower credit risk through more frequent loan renegotiations (Graham et al. 2008). The number of participating lenders in a syndicate is negatively associated with information uncertainty (Dennis and Mullineaux 2000; Qian and Strahan 2007; Sufi 2007), because a smaller number of lenders reduces potential free-rider problems in both ex ante information gathering and ex post monitoring.18

To examine RQ1 (i.e., whether the adoption of the expanded auditor’s report is associated with changes in private loan contracting terms), we use the following models:

Loan featuret = δ0 + δ1Post + δ2Borrower-specific controls(lagged)+ δ3Loan-specific

controlsz + (Industry and Country of syndication dummies) + ε

(1)

Loan featuret = δ0 + δ1Post + δ2Adopt + δ3Post×Adopt

+ δ4Borrower-specific controls(lagged) + δ5Loan-specific controlst

+ (Industry and Country of syndication dummies) + ε

(2)

Loan featuret represents one of the following three loan contracting features: (i) interest rate

spread, measured as the all-in-drawn spread (Spread), (ii) loan maturity (Maturity), measured as the natural logarithm of the length of the loan in months, and (iii) the number of lenders participating in the syndication of the loan (NLenders).19 _{Post is an indicator variable that}

18 _{Regarding the ex post implications of the loan syndicate size, the literature also discusses the costs of loan}

renegotiations or recovery from default.

19 _{An additional measure that is closely related to adverse selection problems between lenders and syndicate}

participants is the lead arranger’s share of the loan (Ball et al. 2008a). Although highly relevant for our study, this

Accepted

equals one for all firms with a fiscal year-end on or after October 1, 2013, and zero otherwise. The benchmark (control) sample in the DID model (equation (2)), which includes non-financial and non-premium-listed UK firms, controls for the impact of other potential borrower- and institutional-specific confounding factors. Adopt is an indicator variable that equals one for non-financial UK premium-listed firms (Adopters), and zero for non-non-financial and non-premium-listed UK firms (Non-Adopters).

In all the empirical models, we follow prior studies (e.g., Bharath et al., 2008; Kim, Song, and Zhang 2011; Kim, Tsui, and Cheong 2011) and control for loan-specific characteristics and lagged borrower-specific characteristics. We use ROA, to control for firm performance; Size, to control for firm size; MTB, to control for growth opportunities;Leverage to control for potential agency cost and financial risk; LoanAmt, Maturity, NLenders, Tloan, Relationship, and Foreign_Cur to control for the joint determination of loan contracting terms.20 _{We include}

industry and country of syndication fixed effects in all models to address the potential effect of time-invariant unobserved heterogeneity.

In equation (1), we are interested in the coefficient of Post, which indicates the average difference between the change in loan contracting terms from pre- to post-adoption of the expanded auditor’s report. In equation (2), we are interested in the coefficient of Post×Adopt, which indicates the difference between the change in loan contracting terms from pre- to post-adoption of the expanded auditor’s report for the Adopters and the Non-Adopters—that is, the difference in differences.

To examine RQ2, whether RMM disclosures from the expanded auditor’s report are associated with loan contracting terms, we use the following cross-sectional model only for the Adopters in the post-adoption period:

field is not well-populated in DealScan, thus preventing its use in our analysis. Future research could make use of larger datasets to harvest the saliency of this measure.

20 _{We include Maturity and NLenders in the model as independent variables when they are not used as the}

dependent variable.

Accepted

Loan feature = δ0 + δ1UniqueRMMs(lagged) + δ2Borrower-specific Controls(lagged)

+ δ3Loan-specific Controls + δ4AuditReport_features(lagged)

+ (Year, Industry, and Country of syndication dummies) + ε

(3)

UniqueRMMs represents the total number of unique RMMs disclosed in the expanded auditor’s report. To construct this variable, we manually collect information about the risks disclosed in the audit committee’s report and compare them to the RMMs identified in the expanded auditor’s report. This allows us to isolate the effect of the RMMs included only in the expanded auditor’s report that are non-overlapping with the risks disclosed in the audit committee’s report and to control for potential confounding effects. For ease of interpretation of our results, we multiply the number of unique RMMs by (-1), so that a less negative value indicates a lower number of unique RMMs. Furthermore, in addition to the previously defined borrower-specific and loan-specific characteristics, we also control for characteristics of the expanded auditor’s report in equation (3)—that is, Materiality and Average Words_RMM. We also include year fixed effects, in addition to industry and country of syndication fixed effects, in equation (3).

Of interest in equation (3) is the coefficient of UniqueRMMs. Per RQ2, because firms with a low number of unique RMMs could benefit from more favorable loan contracting terms, we expect the coefficient of UniqueRMMs to be negative (positive) when we use Spread and NLenders (Maturity) as the dependent variable. To mitigate the potential undue influence of extreme values, we winsorize all continuous variables at the 1% and 99% levels. We cluster standard errors by auditor and firm to correct for unobserved within-auditor and within-firm correlations.

4. Results

Summary statistics and univariate analysis

Figure 1, panel A, presents a box plot of the distribution of Spread for Adopters and Non-Adopters in the pre-adoption and post-adoption periods. Each box indicates the 25th _{and 75}th

Accepted

percentile range, together with the line (50th _{percentile), representing the median. The square}

dot in the boxplot indicates the mean value of Spread for each group. The vertical lines at the top and bottom of each boxplot are the whiskers that show +/- 1.5 times the 25th _{and 75}th

percentile values. As depicted in Figure 1, panel A, in the pre-adoption period, the mean Spread is higher for Adopters (µ = 232.55) than for Non-Adopters (µ = 182.77), whereas in the post-adoption period, the mean Spread decreases substantially for Adopters (µ = 181.29) and increases slightly for Non-Adopters (µ = 192.02).

Similarly, Figure 1, panels B and C, depict the distributions of Maturity and NLenders, respectively, for Adopters and Non-Adopters in the pre-adoption and post-adoption periods. As depicted in Figure 1, panel B, for Non-Adopters, the mean value of Maturity is relatively stable in the pre-adoption (µ = 52.65) and post-adoption periods (µ = 51.29). In contrast, the mean Maturity for Adopters increases slightly from 47.77 months in the pre-adoption period to 52.32 months in the post-adoption period. Figure 1, panel C, illustrates that, for Non-Adopters, the mean NLenders changes from 8.32 to 9.88, and for Adopters it changes from 7.24 to 8.37—that is, both groups increase NLenders after the adoption.

In Table 3, panel A, we use the pre- and post-adoption periods combined to present descriptive statistics for the sample used to test RQ1 (pre-post analysis). On average, in the full sample, the means of our main dependent variables, Spread, Maturity, and NLenders are 201.69 basis points, 50.61 months, and 8.19 lenders, respectively. Univariate tests show that Adopters and Non-Adopters are not significantly different regarding Spread, Maturity, Relationship, ROA, and MTB.

To provide a more detailed depiction of our sample, in Table 3, panel B, we present descriptive statistics for Adopters and Non-Adopters in the pre-adoption and post-adoption periods. Univariate tests show that, for Adopters, Spread (Loan Maturity and NLenders) is (are) significantly lower (higher) (p<0.01) in the post-adoption period than in the pre-adoption

Accepted

period. In contrast, for the non-adopting UK control sample, we do not find significant differences in the mean values of these variables between the pre-adoption and the post-adoption periods.

In Table 3, panel C, we report the characteristics of the sample that we use to test RQ2 (RMM analysis). On average, there are 4 RMMs disclosed in the expanded auditor’s report, which is comparable to the average RMMs reported in contemporaneous studies (e.g., Reid et al. 2019; Lennox et al. 2021). Firms in our sample disclose approximately 1 unique RMM in the auditor’s report, implying an average of 3 (overlapping or non-unique) RMMs that are also included in the audit committee’s report. Table 4 shows Pearson correlations between the variables. Except for Size and Materiality, the correlations are under 0.5, suggesting a reduced potential for multicollinearity in our sample.21

Results of the pre-post analysis (RQ1)

Table 5, panel A presents estimation results for equation (1), which assesses the change in loan features for the adopting firms between the pre-adoption and the post-adoption period. We find that the coefficient of Post is negative and significant (p<0.01) in model (1), positive and significant (p<0.05) in model (2), and positive and insignificant in model (3). In economic terms, this translates into a decrease in the average all-in-drawn spread of approximately 38 basis points (the coefficient of Post (-0.310) × the standard deviation of Spread (123.604), (Table 3, panel A, Adopters)). Given the average loan amount of $752 million in the post-adoption period (see Table 3, panel B, Adopt=1, Post=1), this implies that a typical borrower has to pay about $2.8 million less in interest per year after the adoption of the expanded auditor’s report(=$752 million × 38.32 basis points).22

21 _{The high correlations between Size and NLenders (0.58) and Size and LoanAmt (0.69) are not unexpected. Larger}

firms likely require bigger loans, which are typically provided by larger syndicates. Similarly, the correlations of 0.58, 0.69, and 0.89 between, respectively, Materiality and NLenders, LoanAmt, and Size simply suggest that auditors set higher materiality thresholds for larger firms, which are likely to obtain bigger loans.

22 _{Using a similar computation to that for Spread, we first multiply the coefficient of Post (0.224) by the sample}

standard deviation of Maturity (0.458) (Table 3, panel A, Adopters). The result (0.103) is in natural logs. The

Accepted

In Table 5, panel B, we show the estimation results for equation (2), which assesses the difference between the changes in loan features from the pre-adoption to the post-adoption period for Adopters versus Non-Adopters. We find that the coefficient of the interaction term, Post×Adopt, is negative and significant (p<0.01) in model (1), positive and significant (p<0.01) in model (2), and negative and insignificant in model (3).Economically, the results indicate that relative to Non-Adopters, the loan spread of Adopters decreases by about 51 basis points in the post-adoption period (the coefficient of Post×Adopt (-0.406) × the standard deviation of Spread (125.895), (Table 3, panel A, Full sample)). The results also show that δ1, the coefficient on

Post, is not reliably different from zero for all three dependent variables. This result indicates that the pre-post changes in Spread, Maturity, and NLenders for Non-Adopters are not significantly different from zero. By contrast, (δ1+δ3), which represents the corresponding

changes for Adopters, is significantly negative (positive) for Spread (Maturity) and implies a pre-post change of 46 basis points. Together, these results show that the difference-in-differences results are driven by the pre-post changes for Adopters and not by the corresponding changes for Non-Adopters. The results in panel B of Table 5 indicate that, relative to non-adopting UK firms, the UK non-adopting firms obtain loans with lower spreads and longer maturities after the adoption of the expanded auditor’s report.

RMM analysis in the post-period (RQ2)

In Table 6, we present the test results of investigating RQ2. Model 1 presents the estimation results of equation (3). The coefficient of our main variable of interest, Unique_RMMs, is negative and significant (p<0.01), indicating lower loan spreads for firms with fewer unique RMMs disclosed in the auditor’s report. This relationship is also economically significant. It indicates that a decrease of one unique RMM is associated with an average decrease in spread

coefficient estimate in percentage is, therefore, equal to: 100 × ((exp(0.103))-1) = 10.78 percent. Given a post-adoption average maturity of 52.32 months (Table 3, panel B, Adopters-Post), this translates into an increase in maturity of 5.6 months.

Accepted

of approximately 20 basis points. This reduction in loan spread implies a decrease in yearly interest payments of about $1.5 million. 23 _{Estimation results for models 2 and 3 show}

insignificant coefficients for Unique_RMMs, implying that unique RMMs are not reliably related to maturity or to the number of lenders. Taken together, the findings from Table 6 suggest that disclosures of unique RMMs have a detectable effect only on loan spread, which appears to incorporate information presented in the expanded auditor’s report.

Additional analyses24

Role of the information environment

So far, we have presented evidence that supports the idea that the new disclosures in the auditor’s report are helpful in screening borrowers and shaping loan contracting terms. We next examine whether the benefits from the new disclosures, in terms of more favorable debt contracting terms, differ systematically across firms. One factor that is likely to result in systematic cross-sectional differences in benefits is the quality of the information environment. Indeed, the FRC argued that the benefits of the expanded auditor’s report would be “particularly important for audited entities where there are fewer sources of other information, including smaller companies” (FRC 2016a, 4). Accordingly, and in line with Gutierrez et al. (2018) and Lennox et al. (2021), we examine the role of the richness of the information environment as a conditioning variable.

To proxy for differences in richness of information environment, we create a composite measure similar to Duchin et al. (2010) and Krishnaswami and Subramaniam (1999). This measure combines the (i) number of analysts that follow the firm; (ii) dispersion of analysts’ forecasts (i.e., thestandard deviation of analysts’ forecasts divided by share price at fiscal

year-23 _{The 20 basis points spread reduction is computed as follows: -0.288 (the coefficient on UniqueRMMs in Table}

6, model 1) multiplied by 110.946 (the standard deviation of Spread, see Table 3, panel C) and divided by 1.586 (the standard deviation of UniqueRMMs, see Table 3, panel C). The $1.5 million reduction in yearly interest payments is computed as follows: (=$755 million (the average loan amount in the RMM sample, see Table 3, panel C) multiplied by 20.15 basis points [-0.002015 × $755 million].

24 _{See Appendix 2 for details of the robustness tests and additional tests reported in online appendices.}

Accepted

end); and (iii) analysts’ forecast error calculated as the absolute difference between the mean analyst forecast and the actual earnings per share divided by share price at fiscal year-end. For each of these measures, we obtain the firm’s percentile ranking in the sample using the reverse ranking for the number of analysts. We then scale the index to range from zero to one, such that higher values indicate a poorer information environment. Lastly, based on a median split of this composite measure, we create two subsamples comprising firms with high and low levels of information asymmetry and re-estimate both the pre-post DID analysis and the post-adoption analysis that focuses on the effects of the number of unique RMMs.

In Table 7, panels A and B, we present the results of estimating equation (2) for firms with a rich and a poor information environment, respectively. We find that the coefficient of the interaction term, Post×Adopt, for the rich information environment subsample is negative but not significantly different from zero in any of the models in panel A. By contrast, the coefficient of Post×Adopt for the poor information environment subsample is significantly negative (p<0.01) in model 1, significantly positive (p<0.05) in model 2, and insignificant in model 3 of panel B. These results indicate a significant pre-post decrease (increase) in Spread (Maturity) for Adopters relative to Non-Adopters for the low information environment subsample but not for the high information environment subsample. Furthermore, our results also indicate that NLenders does not show a significant pre-post change for Adopters relative to Non-Adopters for either the low or the high information environment subsamples. However, in all estimations, the coefficient of Post×Adopt is not significantly different between the poor information subsample and the rich information subsample. Overall, our results suggest that the loan contracting benefits (in terms of spread, maturity, and number of lenders) of the expanded auditor’s report adoption do not differ between firms with poor and rich information environments.

Accepted

Table 8, panels A and B show the estimation results of equation (3) for firms with a rich and a poor information environment, respectively. Although the coefficient of Unique_RMMs when Spread is the dependent variable is insignificant in model (1) of panel A, it is negative and significant in model 1 of panel B (p<0.05). When the dependent variable is Maturity, we observe that while the coefficient of Unique_RMM in model 2 of panel A is positive and marginally significant (p<0.10), it is insignificant in model 2 of panel B. Finally, in model 3 of panels A and B, when the dependent variable is NLenders, the coefficient of Unique_RMM is insignificant. The coefficients of Unique_RMM are significantly different (p< 0.01) between the two groups in the Spread model (model 1 of panels A and B). However, the coefficients are not significantly different between the two groups in the Maturity model (model 2 of panels A and B) and in the NLenders model (model 3 of panels A and B). This evidence suggests that having fewer unique RMMs decreases spread only for adopting firms with a poor information environment. Our findings also suggest that, regardless of the information environment, the number of unique RMMs does not significantly influence the maturity and the number of lenders of adopters.

Robustness analyses

Controlling for corporate governance

Agency and information risks are considered by lenders when designing loan contracts (Bhojraj and Sengupta 2003; Rajan and Winton 1995). Since one objective of corporate governance is to ensure a firm's resources are used with all stakeholders' interests in mind, lenders may also consider the mitigating effects of borrowers’ corporate governance structure on the agency and information risks. If so, lenders are likely to provide more favorable loan contracting terms to borrowers with stronger corporate governance. Consistent with this reasoning, Francis et al. (2012) find that borrowers receive better price and non-price loan contracting terms, and more lenders participate in the loan syndication, when borrowers’ boards of directors and audit

Accepted

committees are more independent. Similarly, Li et al. (2014) document that debt contracts have fewer restrictions when the board is larger, board members have more expertise, and the borrower has more activist shareholders. Additionally, their results indicate that borrowers with block holders face more restrictions due to the greater concerns about expropriation risk associated with block holders.

To ensure that our results are not sensitive to differences in borrowers’ corporate governance, we repeat our main tests after controlling for borrowers’ corporate governance structure. Specifically, we include a composite score, taken from the corporate governance pillar of Thomson Reuters’ ASSET4 database, in our estimations. Our results, discussed in Appendix 2, are similar to our main findings in Tables 5 and 6, suggesting that lenders consider the information in RMMs independently of borrowers’ corporate governance quality.

Implications for debt covenants

In our main analyses, we focus on three dimensions of loan contracting: spread, maturity, and number of lenders. Another important non-price dimension of syndicated loans is covenants. We focus on the intensity (or number) of debt covenants included in loan contracts because lenders generally impose a larger number of covenants on the borrower in an attempt to limit their risk exposure arising from information asymmetry (e.g., Nikolaev 2010; Kim, Song, and Zhang 2011; Kim, Tsui, and Cheong 2011).

To investigate whether the new disclosures in the auditor’s report are associated with a decrease in debt covenant intensity and whether debt covenant intensity in the post-adoption period varies with the number of RMMs, we repeat our analyses using the logarithm of (1 + number of covenants) as the dependent variable. Due to the lack of covenant information in DealScan, the sample sizes for these analyses are considerably smaller. Our results, discussed in Appendix 2, indicate that although we do not find a statistically significant decrease in debt

Accepted

covenant intensity in the post-adoption period compared to the pre-adoption period, the number of covenants decreases with the number of RMMs in the post-adoption period.

Decomposing RMMs into expected and unexpected RMMs

To alleviate the concern that because lenders may foresee some of the risks mentioned in the expanded auditor’s report and therefore these risks may not be new information, we follow an approach similar to Lennox et al. (2021) and decompose total unique RMMs into expected and unexpected unique RMMs by estimating a model that predicts the number of unique RMMs in the auditor’s report. As explanatory variables for the expected number of RMMs, we use ROA, Size, MTB, number of audit committee risks, number of geographic segments, number of business segments, number of recorded subsidiaries, loss indicator, cross-listed indicator, industry indicators, auditor indicators, and year indicators. The fitted values and error term from this regression are proxies for expected and unexpected unique RMMs, respectively. We then re-estimate equation (3) using both Expected_UniqueRMMs and Unexpected_UniqueRMMs as independent variables instead of UniqueRMMs.

We find that while the coefficient of Expected_UniqueRMMs is insignificant, the coefficient of Unexpected_UniqueRMMs is significantly associated with Spread. We find insignificant associations for the Maturity and NLender models (see Appendix 2 for details). Taken together, these results suggest that the RMM disclosures are relevant to lenders only when they are not expected.

We further refine our approach by focusing on the unique RMMs that are also likely to be particularly relevant from a debt-market perspective. We create Unique_RelevantRMMs, which are the unique RMMs related to asset valuation, fraud, management override of internal

Accepted