The influence of credit ratings on auditor’s going concern decision

(1)

The influence of credit ratings on auditor’s going concern decision.

Harmen Zegers

Universiteit van Amsterdam

Abstract

Purpose – Recent attention is given by legislators to auditors reporting on an entity’s ability to continue

as a going concern. I argue in this paper that Credit Rating Agencies are more specialized in this subject and investigate if auditors use the information imbedded in credit ratings in their going concern assessment.

Design/methodology/approach – A logistic regression model was employed in a unique sample

comprising of 14,754 observations.

Findings – Weak evidence is found that auditors use credit ratings in their going concern assessment.

However based upon this study it cannot be out ruled whether this is due to limitations of the model or auditors and Credit Rating Agencies assessing the same information equal.

Practical implications – Auditors seem to use credit ratings in their assessment of entities ability to

continue as a going concern. This gives rise to question for legislators if they should not formalize reporting requirements for credit rating agencies where it comes to going concern instead of auditors.

Originality/value – One of the first studies to consider the use of ‘external’ experts by auditors in

making their going concern assessment.

Keywords Going Concern Opinion, Corporate governance, United States of America, Auditing, Credit

Rating

Name: Harmen Zegers Studentnr: 10428402

Study: Accountancy & Control Supervisor: Peter Kroos, PhD

(2)

1 Introduction

1.1 Background

In the aftermath of the financial crisis much attention was given to the audited financial statements of the companies that failed to survive the economical heat. Why didn’t these financial statements show the investors the financial precarious situation of the company? More important, why did they receive unqualified opinions from their auditors with no disclosure on the possibility of going concern issues? Sikka (2009) shows in his study a non-complete list of 28 banks that were under severe financial stress at the end of 2008, but still received an unqualified clean audit opinion. Major examples are Fannie Mae, Freddie Mac which were placed into conservatorship of the Federal Housing Finance Agency on the 7th of September 2014, Bear Stearns that went into financial distress and was acquired by JP Morgan Chase for $2 dollar a share (NY Times 2008) 90% under the previous trading price and Lehman Brothers which filed for Chapter 11 on the 15th_{of September 2008 (SEC).}

As response investigation committees were formed, like the Financial Crisis Inquiry Commission in the US and the Sharman Enquiry in the UK. Both committees identified that the reporting of auditors is currently not sufficient were it comes to going concern (GC) issues. The Sharman enquiry for instance proposed to require that auditors do not report on GC on an exception basis, but that always a paragraph in the auditor’s report should be included, reviewing the management’s GC assessment. The recent attention of government institutions and governing bodies increases the pressure on the auditors to extensively review the ability of the audited entity to continue as a going concern. However this doesn’t take away that the GC assessment by nature is out of the comfort zone of the auditor. The auditor his regular business is providing backward assurance. The GC assessment is forward looking. I argue that other professions are more capable of assessing an entity’s ability to continue as a GC. Professionals like security analysts and especially credit rating agencies are assessing on a continuous basis if an entity is capable to meet its future obligations.

1.2 Research question

The auditing standard for GC states the following: SAS 59; ‘’The auditor has a responsibility to evaluate

whether there is substantial doubt about the entity's ability to continue as a going concern for a reasonable period of time, not to exceed one year beyond the date of the financial statements being audited’’. Apart from the specific reference to the auditor, little else in this sentence specifically triggers

to think that an auditor would be the ultimate professional to assess the GC issue. The regular business of an auditor is to provide backward looking assurance on the preceding year. This can be either via an audit, review or specific agreed upon procedures. The common factor remains that the viewpoint of an auditor remains backward looking. SAS 59 requires the auditors to do something which is out of the regular business they perform and likely outside of their comfort zone. The auditor is required to assess if the entity will be able to meets its future obligations and as such will be able to continue as a going concern. If the auditor doesn’t expect that an entity will be able to continue as a GC, financial

statements should be prepared at liquidation basis and more importantly, the doubt of the auditor needs to be communicated in the auditor’s opinion. By communicating this to the financial statement users, the downward risk of investors of investing in the audited company is limited as the investors are provided additional information regarding the ability of the entity to continue as a GC.

Other professions like the already mentioned security analysts and credit rating agencies perform similar services to investors and are arguably more specialized in this type of business, as this is for them day to day routine. Security analysts try to predict future price trends for company stock or other financial products. Credit rating agencies (CRA) try to predict if companies are able to meet their future debt obligations. The service performed by the CRA is similar to what SAS 59 requires from auditors.

(4)

Furthermore both auditor and the CRA have the same risk focus. Neither of the 2 is reporting on the ability of an entity to make abnormal profits in the future, their only focus is to limit the downside risk of a company failing to meet its future obligations and due to that not able to continue as a GC.

As the CRA is more specialized and has the same risk focus as the auditor, I expect that auditors will use the information imbedded in the credit ratings issued by CRA’s in their assessment if an entity is GC, to gain assurance on these for themselves non-routine procedures. This leads to my research question:

‘’Do auditors make use information embedded in credit ratings in their going concern assessments?”

1.3 Contribution

Prior studies always have focused on the individual assessment made by the auditor on whether or not an entity is able to continue as a GC. Financial ratios like leverage and recurring losses have been used as predictors for GC opinions. Other studies have included market variables, material weaknesses or audit fees. However to my knowledge no previous study has focused if auditors use external information in their assessment. With external information I mean, information produced by another expert or profession like a credit rating agency. My study can contribute to regulators in their evaluation on how to implement reporting to the market on GC. If the auditor makes use of credit ratings in his assessment, this could indicate that not the auditor, but the CRA is more suitable to be the required reporting

profession on GC.

1.4 Structure of the paper

The paper proceeds as follows. Reporting on GC by the auditor is regulated by accounting standards as issued by the FASB as well as by auditing standards as issued by the AICPA/PCAOB, I will start describing this legal environment in paragraph 1 & 2 of chapter 2. In paragraph 3 & 4 I will provide a literature review on prior research on auditors GC opinion decision and the market’s reaction to these opinion. Regulatory background and prior literature on Credit Rating agencies will be described in paragraph 5 & 6. In paragraph 7 I will present my hypothesis.

Chapter 3 & 4 I present the logistic model used & the results from this model. The final chapter is a summary and conclusion of the paper.

(5)

2 Regulatory Framework

2.1 Management’s responsibility (FASB)

Currently the going concern (GC) assumption is not specifically described in US GAAP. The GC

assumption is only an inherent assumption. This is described by the FASB in the 2013 exposure draft as following ‘Under U.S. generally accepted accounting principles (GAAP), financial statements are

prepared under the inherent presumption that the reporting entity will be able to continue as a going concern; that is, the entity will continue to operate such that it will be able to realize its assets and meet its obligations in the ordinary course of business (the going concern presumption)’

This implies that management has no responsibility to disclose any information regarding the ability of an entity to continue as GC. Only when an entity’s liquidation is imminent, management is required to prepare the financial statements under liquidation basis (Subtopic 205-30, Presentation of Financial statement –liquidation basis of accounting).

Currently, AU Section 341 (SAS 59), The Auditor’s Consideration of an Entity’s Ability to Continue as a

Going Concern, of the AICPA Codification of Statements on Auditing Standards contains the guidance

about the going concern assessment.

In this respect US Gaap is different from IFRS where the GC assumption is included in the conceptual framework. In IFRS the going concern principle is the underlying assumption, described in paragraph 4.1 of the conceptual framework. ‘’The financial statements are normally prepared on the assumption that

an entity is a going concern and will continue in operation for the foreseeable future. Hence, it is

assumed that the entity has neither the intention nor the need to liquidate or curtail materially the scale of its operations; if such an intention or need exists, the financial statements may have to be prepared on a different basis and, if so, the basis used is disclosed.’’

Being the one and only underlying assumption of the IFRS framework, the going concern assumption is of big importance. Besides that the GC assumption is included in the conceptual framework,

management responsibility regarding GC disclosure is described in IAS 1, paragraph 25: “When

preparing financial statements, management shall make an assessment of an entity’s ability to continue as a going concern. An entity shall prepare financial statements on a going concern basis unless

management either intends to liquidate the entity or to cease trading, or has no realistic alternative but to do so.’’ If material uncertainties exist that may cast significant doubt on the upon the entity’s ability

to continue as a going-concern this has to be disclosed by management in in the financial report. Interpreting this, management of the entity has to make a prediction of the future based upon current information. For how long is described in paragraph 26 as follows: ‘’In assessing whether the going concern assumption is appropriate, management takes into account all available information about the future, which is at least, but is not limited to, twelve months from the end of the reporting period.’’

To fill the gap and provide guidance on the preparation of financial statements as a going concern and on management’s responsibility to evaluate a reporting entity’s ability to continue as a going concern the FASB started in 2008 working on an exposure draft. Furthermore the Exposure draft also would require certain disclosures when either the financial statement is not prepared on a going concern basis or when there is substantial doubt as to an entity’s ability to continue as a going concern.

The Board decided to carry forward the going concern guidance from AU Section 341, subject to several modifications to align the guidance with International Financial Reporting Standards (IFRSs). One of the

(6)

modifications is to use the time horizon from IAS 1, all available information about the future, which is

at least, but is not limited to, twelve months from the end of the reporting period. Opposing the fixed

time limit in AU Section 341, ‘not to exceed one year beyond the date of the financial statements being

audited’. The Board decided to use the time horizon in IAS 1 because it avoids the inherent problems

that a bright-line time horizon (ED 2008) would create for events or conditions occurring just beyond the one-year time horizon that are significant and most likely would have to be disclosed. Final standard is expected in to be finalized in the second quarter of 2014 and will be effective as from December 15, 2015 (FASB project update page). Until the finalization of the standard, the disclosure responsibility doesn’t reside with the entity’s management according US Gaap, but resides with the auditors. However in common practice the GC guideline of SAS. 59 was applied by companies and FASB when it came to GC disclosure.

2.2 Auditors Responsibility (AICPA/PCAOB)

As noted above, US Gaap forces the responsibility regarding disclosure about the ability of an entity to continue as a going concern to the auditor. This leaves the auditor with 2 issues that are somewhat outside of his routine business. 1) As mentioned earlier, the auditor has to make predictions about the future, where his routine business is backward looking.1_{2) The auditor himself has to make the}

assessment about disclosing information about GC, it is not initially an assertion made by management which is subsequently reviewed by the auditor.

SAS 59. prescribes the following: ‘The auditor has a responsibility to evaluate whether there is

substantial doubt about the entity's ability to continue as a going concern for a reasonable period of time, not to exceed one year beyond the date of the financial statements being audited (hereinafter referred to as a reasonable period of time). The auditor's evaluation is based on his or her knowledge of relevant conditions and events that exist at or have occurred prior to the date of the auditor's report’.

The auditor is not required to design specific GC procedures before commencing the audit. According to §3.a the auditor is required to consider whether the results of his audit indicate whether there could be substantial doubt about the entity’s ability to continue as going concern for a reasonable period of time. Only if there is a possibility for substantial doubt further procedures need to be performed.

Conditions and events that could lead to substantial doubt are outlined in §6

• Negative trends—for example, recurring operating losses, working capital deficiencies, negative

cash flows from operating activities, adverse key financial ratios

• Other indications of possible financial difficulties—for example, default on loan or similar

agreements, arrearages in dividends, denial of usual trade credit from suppliers, restructuring of debt, noncompliance with statutory capital requirements, need to seek new sources or methods of financing or to dispose of substantial assets

• Internal matters—for example, work stoppages or other labor difficulties, substantial

dependence on the success of a particular project, uneconomic long-term commitments, need to significantly revise operations

1_{In §4 of Sas 59: States that auditor is not responsible for predicting the future conditions or events, however this} seems somewhat inconsistent with the requirement of overall GC disclosure: The auditor is not responsible for

predicting future conditions or events. The fact that the entity may cease to exist as a going concern subsequent to receiving a report from the auditor that does not refer to substantial doubt, even within one year following the date of the financial statements, does not, in itself, indicate inadequate performance by the auditor. Accordingly, the absence of reference to substantial doubt in an auditor's report should not be viewed as providing assurance as to an entity's ability to continue as a going concern.

(7)

• External matters that have occurred—for example, legal proceedings, legislation, or similar

matters that might jeopardize an entity's ability to operate; loss of a key franchise, license, or patent; loss of a principal customer or supplier; uninsured or underinsured catastrophe such as a drought, earthquake, or flood

§7 states that the auditor should consider management's plans for dealing with the adverse effects of the conditions and events. The auditor should obtain information about the plans and consider whether it is likely the adverse effects will be mitigated for a reasonable period of time and that such plans can be effectively implemented. If the substantial doubt remains after assessment of management plans, the auditor should include an explanatory paragraph to reflect that opinion.

When reviewing the conditions and events (C&E) noted by the AICPA in §6 of SAS 59 I note that these C&E could be identified by the user of the financial statement himself, (Negative trends), or could be identified by professionals who have access to management (External Matters). For Other indications of possible financial difficulties and Internal Matters it is possible that the auditor has more information than other parties. However I note that the Credit Rating Agencies (CRA’s) are securities analyst with special access to entities management, protected by law, refer to paragraph below. It seems likely that CRA’s will use this access to obtain this information as well.

2.3 Going Concern Decision

In some of the earliest research done on the auditors GC opinion, it was already noted by Kida (1980) and Mutchler (1984) that auditors are quite good at predicting if a company is in financial distress or not. It is the final decision on whether not issue a GC opinion for the financial distressed companies which attracted most academic attention in recent years (Carson et. al. 2013). See for the GC opinion framework as developed by Carson et. al. appendix B.

Mutchler (1985) studied the information content of the GC opinion. She held questionnaire interviews with 2 auditors of each big 8 firm2_{, in which the auditors had to choose a set of ratios for the calculation} of a possible GC issue. Purely based upon the indicators, as identified by the auditors, the model had a predictive accuracy of 89,9% for al GC opinions and 83% for companies that received a GC opinion for the company for the first time. These results indicate that without the use of auditors, financial statement users could make their own prediction of a company receiving a GC opinion with fair accuracy.

Lasalle & Anandarajan (1996) did a similar questionnaire amongst audit partners as Mutchter did. They found the following top 5 ratios as indicated by the audit partners, (1) net worth/total liabilities, (2) cash flows from operations/total liabilities, (3) current assets/current liabilities, (4) total liabilities/total assets, and (5) change in net worth/total liabilities. Next to the financial indicators, Lassalle & Anandarajan also asked the auditors on what they found the most important element in the internal control environment. The competence of the company’s management was the most important internal control element according to the auditors. Several other factors besides internal control and financial ratios were discussed. Important outcome of the study is that auditor’s value bad news regarding GC (in regardless of what source) more than good news possible mitigating the GC issue, indicating an

asymmetric information process by the auditors, meaning that auditors need more good news to reverse a GC opinion than they need bad news to issue one.

Defond, Raghunandan & Subramanyam (2002) (amongst others) studied the relationship between (non) audit fees and the auditors’ propensity to issue a GC opinion. They didn’t find a relationship indicating

2_{In 1985 the Big 4 was still the Big 8, Arthur Anderson, Arthur Young & Co, Coopers & Lybrand, Ernst & Whinney,} Deloitte Haskins & Sells, Peat Marwick Mitchell, Price Waterhouse, Touche Ross.

(8)

that auditors perform their GC evaluation independent from the total fee they receive. Geiger and Rama (2003) even find a positive relationship between audit fee and the auditor’s propensity to issue a GC opinion. They argue that this indicates that market-based incentives like losses from possible litigation and possible damage to reputation may be strong factors in influencing the auditor’s decision.

Li (2009) did a follow up study on Defond et. al. and Geiger & Rama. He compared a pre-Sox (2001) and post-Sox (2003) data set. No significant association was found between the audit fee and the auditor’s propensity to issue a GC opinion in the pre-sox data, in line with the results of Defond et. al. The post-sox data showed a positive relation between the audit fee and the GC opinion, in line with Geiger & Rama.

Independence might be threatened by the client possibly switching of auditor after the issuance of a Going concern opinion. Matsumura, Subramanyam & Tucker (1997) find that auditors will be less likely to issue a GC opinion if it is possible that another auditor wouldn’t issue a GC opinion. The risk exists that the client will go ‘opinion shopping’. Carcello & Neal (2003) find support for this, showing that auditors are more likely to be dismissed in the year following a qualified opinion. Hoitash & Hoitash (2008) find that this relation weakened in the years following the implementation of Sox.

Sengupta & Shen (2007) find that accrual quality is negatively linked to the likelihood of receiving a GC opinion, indicating that auditors include their trust in the client in their evaluation. Similar results are found by Goh, Krishnan & Li (2011) who find that the reporting of Material weaknesses under S404 is positively linked to the issuance of GC opinions.

2.4 Going concern opinion’s and the market’s reaction.

As prescribed in the regulation, management and the auditor have to assess the ability of an entity to continue as a going concern (SAS 59). The entity’s management can be seen as a priori biased as they are strongly connected with the entity. From this view the review by the accountant is more valuable, as the auditor is an independent third party reviewing the company’s financial statements (independence is debated in literature see paragraph on auditor decision). As described above the auditor has to assess from the information available to him if there is a going concern issue at the date of reporting for the entity under audit.

In doing this the auditor faces the risk of making a type 13_{or a type 2}4_{error. In common practice a type} 1 error is viewed as more severe than a type 2, due to the possible outcomes for the entity receiving the modified audit opinion. By wrongly issued GC opinion Investors might be triggered to divest, lenders might be less willing to issue new debt or it can be part of debt covenants. A GC opinion which is not issued although it should, could lead to the opposite reaction by investors and lenders. Geiger and Rama (2006) note in their study on Going-Concern reporting accuracy that both errors can be costly for the auditor. Carcello and Neal (2003) show that auditors are in threat of being replaced if they don’t want to issue an unmodified opinion, when the company is facing a possible going concern issue. Due to this the auditor faces the threat of losing the client/revenue. On the other hand Carcello and Palmrose (1994) show that auditors face serious litigation threat when they didn’t issue a modified audit report in the year before that a company went bankrupt. In their study they show that auditors face higher litigation costs for a type II error than a type 1 error. As both errors have a possible negative outcome for the auditor it is an area where an auditor should be cautious not to make a mistake.

Also the possible self-fulfilling prophecy of the going concern opinion is a reason for auditors to be cautious. As discussed by Vanstraelen (2003) the results of prior empirical studies are mixed. Cintron &

3_{Type 1 error: When the auditor issues a modified opinion due to going concern issues, however there is actually} no going concern issue. The auditor incorrectly identified a going concern issue.

4_{Type 2 error: When the auditor issues a clean opinion, however there is a going concern issue. I.e. a company} with a clean audit report in the prior year goes bankrupt. The auditor has failed to identify the going concern issue.

(9)

Taffler (1992) and Louwers et. al. (1999) find no supportive evidence. George et. al. (1996) do find supportive evidence for the self-fulfilling prophecy. Louwers et el. (1999) elaborate on this problem indicating that is also due to the difficulty in disentangling the disclosure from other indicators (ratios, stories in popular press) of financial distress. Furthermore there is the problem that you cannot measure if a company that went bankrupt would still have existed if there hadn’t been a going concern opinion, or vice versa.

Prior literature isn’t unambiguous about the market’s reaction on modified audit report. Fleak and Wilson (1994), Jones (1996) and Menon & Williams (2010) find a negative reaction of the market to the initial issuance of a modified audit opinion. Taffler et. al. (2004) find in their UK based study on the medium long term also a negative reaction, however on the short term the markets seems to be in denial of the bad news incorporated in a modified audit opinion. Herbohn et al. (2007) (Australia based) and Blay and Geiger (2001) don’t find a significant market reaction at all. Raghunandan & Subramanyam (2003) find that a model which incorporates financial statement and market information outperforms the going concern opinion of auditors. After controlling for the aforementioned information the going concern opinion however does have incremental predictive value for bankruptcy. Ogneva and

Subramanyam (2007) oppose the results of Taffler et al (2004) with their US & Australia based research, using factor models or after controlling for momentum, they don’t find negative abnormal returns. This indicates that the market values good news or bad news from a going concern opinion equally. As a reaction to this Kausar et al (2009) did a new study with a US database. They found, in line with Taffler et al (2004), a market anomaly of -14% for initial going concern opinions. They argued that the different results of Ogneva and Subramanyam were due to issues of the research model used by O&S. Possible explanation given by Kauser et. al., for the market underreacting to GC announcements is that high trading costs result in arbitrageurs lacking incentives to exploit this first-time GC-firm mispricing

anomaly, thereby preventing such information from being fully impounded into stock prices on a timely basis

Holder-Webb & Wilkins (2000) and Tan (2002) show that the Going Concern opinion reduces the bankruptcy surprise by the market, however it must be noted that Tan found that only the surprise was reduced for entities with no other sources that were more timely with communicating the financial distress.

Although results are mixed, there are indications that markets do value the information content of GC opinions.

2.5 Credit Rating Agencies

Securities analysts are described by Coffee (2004a) as one off the gatekeepers in the securities markets, similar to auditors. Gatekeepers are defined as following: ‘Independent professionals who pledge their

reputational capital-to protect the interests of dispersed investors who cannot easily take collective action.' So phrased differently, Gatekeepers help the investors in making their investment decisions. For

my study I focus particularly on credit rating agencies (CRA). The business of a CRA is to assess the creditworthiness of an organization under review as a whole (issuer credit ratings), or assess the credit quality of an individual debt issue and the relative likelihood that the issue may default, such as a corporate or municipal bond (debt issue rating). In my study I use the issuer credit ratings as these ratings give an assessment of the organization as a whole. In making the issuer credit ratings, CRA’s make a prediction about the entities ability to continue as a going concern, what actually is required by US Gaap from the auditor. This is defined as following by S&P in its credit ratings guide: Credit ratings

(10)

are opinions about credit risk. Standard & Poor’s ratings express the agency’s opinion about the ability and willingness of an issuer, such as a corporation or state or city government, to meet its financial obligations in full and on time. S&P also states in the same guide: Ratings are provided by credit rating agencies which specialize in evaluating credit risk. Furthermore S&P states that these ratings are based

upon analysis made by experienced professionals. Summarizing this: dedicated and trained specialist make predictions about the ability of an entity to meets its future obligations. Similar explanations of the business of a CRA are to be found on the websites of Moody’s and Fitch, which are the 2 other major CRA’s, all 3 having a combined market share between 90-95%.

Unlike the GC opinion, which is present or not in the auditor’s opinion, a credit rating is made on a gradual scale. This scale is somewhat different per CRA, but mostly AAA indicates very good quality and D default (Refer to appendix A). Within this gradual scale a difference between investment grade and non-investment grade. Ratings below BBB- are considered non-investment grade/speculative. There is however a significant difference between CRA’s and auditors. S&P states in its guide the following: ‘Unlike other types of opinions, such as, for example, those provided by doctors or lawyers,

credit ratings opinions are not intended to be a prognosis or recommendation. Instead, they are primarily intended to provide investors and market participants with information about the relative credit risk of issuers and individual debt issues that the agency rates.’ Investors cannot rely upon the

credit ratings from CRA’s in their buy or sell decision, it only helps them. This is different for an auditor who gives an opinion on a financial statement, giving assurance that the FS can be relied upon. The auditor is required to do so by law. White (2010) describes in his paper how the CRA’s claim that their rating is merely an opinion protected by the First Amendment.

Although argued as being merely an opinion by CRA’s, credit ratings play an important role in today’s financial system. Since 1936 banks are only allowed to invest in bonds that are investment grade, BBB or higher. Initially only ratings of CRA’s that were ‘recognized rating manuals’ were allowed to provide the ratings. In 1975 the SEC introduced ‘nationally recognized statistical rating organization’ (NSRO). The SEC declared that only the ratings of NRSROs were valid for the determination of the broker-dealers’ capital requirements5_{. So although a credit rating is only an ‘opinion’, a negative rating can have some}

importance for an organization6_.

The introduction of the Dodd-Frank act in the US limited the playing field of credit rating agencies. Article 939A of the act requires federal to review/remove any references to or requirements in such regulations regarding credit ratings, one year after the enactment date July 21, 2010. Basel III however puts new reliance upon credit ratings as issued by CRA’s. Credit ratings are specifically introduced as measurements of credit risk in the global regulatory standard for the banking sector as issued by the Basel Committee of Bank Supervision, effective from the 1st_{of January 2013. The US adopted Basel III,} however substituted the credit rating of CRA’s, by country risk classifications from the OECD, for country credit risk (PWC 2013).

5_{The idea is that banks and other financial institutions should not need to keep in reserve the same amount of} capital to protect the institution (against, for example, a run on the bank) if the financial institution is heavily invested in highly liquid and very "safe" securities, such as U.S. government bonds or commercial paper from very stable companies. The safety of these securities, under this approach, is reflected in their credit ratings, as determined by certain highly respected CRAs.

6_{Greighton, Gower and Richards (2007) summarize the following US acts which place reliance upon CR’s: the}

Investment Company Act ensures money market funds invest only in securities rated in the two highest categories;

and the investment grade distinction is important in the Federal Deposit Insurance Act and the Secondary

Mortgage Market Enhancement Act, where corporate debt is only ‘investment grade’ if rated in one of the four

highest categories, and mortgage-related securities must have a rating in the highest two.

(11)

2.6 Credit Ratings and the market’s reaction

Holthausen and Leftwich found in their 1986 study based upon US data, that the market does show negative abnormal returns after a credit downgrade, but almost doesn’t show positive abnormal returns after a credit upgrade. Matolscy and Lianto (1995) find similar results in an Australia based study. Both studies indicate that credit ratings do contain new information for the market, however markets reaction is limited. Goh and Ederington (1993) make a distinction between 2 types of negative

downgrades: those due to deterioration in the firm's financial prospects and those due to an increase in leverage. They observe a negative equity market reaction to the first group of downgrades but no reaction to the second. Their results indicate that credit ratings are not a homogeneous population. Ederington and Goh (1998) made a comparison between market analyst and CRA’s. Their results indicate In line with other studies that the market reacts to a negative credit rating announcement, but almost doesn’t react to a positive announcement. Furthermore they noted that the market reacts quicker and more efficient to credit rating downgrade announcements than analysts do. Ederington and Goh also noted that market is already declining together with the analyst before the credit rating announcement.

As described above the credit ratings issue, although being only an ‘opinion’ can be of significant importance due to regulation requiring the use of credit ratings. To overcome this Creighton et. al (2006) did a study in Australia on the market’s reaction to credit ratings. They chose for Australia as credit ratings only play a limited role in the countries regulation. They found that the market does react to credit ratings, although being significant the price impact is relatively limited. The effects of a credit rating announcement were larger for 1) small firms, 2) downgrades from investment to speculative grade and 3) where agencies have not previously indicated the rating is under review.

The regulatory impact was further studied by Jorion, Liu and Shi (2005). In 2000 the SEC implemented Regulation Fair Disclosure, which requires US companies which disclose nonpublic information to a select group also disclose it simultaneously to the Public. Exemption to this requirement are credit rating agencies (October 2010 this exemption was removed by SEC). Jorion et al. found that after the

implementation of the regulation the market’s reaction to credit rating changes on stock prices was more pronounced. This indicates that the market does value to the information that is known to the CRA’s.

All studies mentioned above indicate that the market incorporates the news conveyed in credit ratings announcements; however the information is already partly anticipated by the market in advance of the announcement triggering only a small market reaction. Market reaction is stronger if the announcement is negative.

2.7 Hypothesis development

As described in chapter 2, paragraph 2, the auditor is required (amongst others) to evaluate Negative

trends, Other indications of possible financial difficulties, Internal Matters and External matters that have occurred, if the auditor has substantial doubt about the ability of the entity to continue as a going

concern. A credit rating or a credit rating downgrade can provide information for 2 of the items which need to be evaluated by the auditor.

• Other indications of possible financial difficulties: a default on a loan or denial of trade credit

can be 2 indications according to SAS. 59 §6 of financial difficulties. Debts covenants with 3rd parties can be based upon credit ratings, with for example increasing interest for issued debt in presence of a negative credit rating, or denial of loan/credit need for financing of operations. Furthermore banks are until recently only allowed to invest in bonds that are investment grade. Companies falling below this grade have a much smaller capital market from which they can

(12)

subtract funds at their disposal, probably causing an increase in the cost of capital. I argue that this can be a trigger for an auditor to proceed in disclosing his doubt about the ability of an entity to continue as a going concern.

• External Matters that have occurred: a credit rating can be an external matter of influence on

the entity as already discussed above, where I already discussed its possible indication of financial difficulties. I expect that the credit rating (downgrade) provides reassurance to the auditor. A stated earlier I deem the CRA to be more a specialist in the field of assessing the ability of GC of an entity than the auditor. A negative credit rating (downgrade) can give the auditor its reassurance that his doubt about the ability of the entity to continue as GC is substantial and needs to be disclosed in his opinion.

Based upon the above I expect that the auditor will include a credit rating in his assessment because of 3 reasons.

• 1 Real Impact: A credit rating can have negative impact on the actual operations of an entity due to the company violating debt covenants, which are linked to credit ratings. Furthermore a decrease below investment grade reduces the size of the capital market available for an entity, due to the restrictions for financial institutions; see Creighton et al. (2007) for their summary of US acts mandating CR’s. Also Basel III introduces credit ratings as measurement of credit risk for banks (White, 2010).

• 2 Reassurance: A CRA is more specialized in assessing the ability of an entity to meets its future debt obligations (i.o.w. its ability to continue as GC). A negative credit rating (downgrade) gives an auditor reassurance that his assessment is correct and decreases his risk of misjudgment. Prior literature shows that CRA’s base their evaluation mainly upon financial indicators; as such they evaluate information that is already known to auditors. They add value by interpreting the same information with their knowledge of credit evaluation. CRA’s pride themselves in their expertise in evaluation. From S&P Guide on ratings: ‘Standard & Poor’s aim is to balance

quantitative measures with qualitative analysis based on the judgment and analytic skills, training, and experience of our credit ratings analysts.’

• 3 Undisclosed information: Credit rating agencies have access to information which is not publicly available. Jorion et al. (2005) noted the increase of the stock price reaction after the implementation of the Regulation Fair Disclosure by the SEC. Not the rating itself, but the information embedded in the rating i.e. the information not known to the auditor, but known to the CRA is relevant to the auditor.

Hypothesis 1a & 1b test if a credit rating and if a credit rating event (downgrade) have impact on the auditor’s assessment.

Hypothesis 1a: Lower credit ratings are positively associated with the probability of a company

receiving a GC opinion.

Hypothesis 1b: A greater credit rating downgrade in the year prior to the auditor report is positively

associated with the probability of a company receiving a GC opinion.

The following hypotheses are specifically aimed at the conditions of the credit rating, or credit rating event.

Hypothesis 2a: The magnitude of A credit rating downgrade is positively associated with the probability

(13)

As discussed earlier, financial instructions were since 1936 not allowed to hold any investments below investment grade, making this a sensitive area for entities. This because the capital market available to the entity decreases in size. Besides the real impact there is also the reputational signal of being below investment grade. As can be seen in appendix A, the cut-off between investment and non-investment is between BBB- and BB+. To follow up on this the following 2 hypothesis:

Hypothesis 2b: A speculative grade credit rating is positively associated with the probability of a

company receiving a GC opinion.

Hypothesis 2c: A first-time downgrade below investment grade in the year prior to the auditor report is

positively associated with the probability of a company receiving a GC opinion.

3 Research Method

3.1 Model

To test my hypothesis I developed the timeline as depicted in figure 1. Moment t is defined as the moment when the auditor’s opinion is published in the financial statement of the audited entity. Before publishing of the opinion a company has received a credit rating from a credit rating agency. This rating can be given within one year before the opinion, but can also be published at a date earlier (H1a). Between the moment t, the audit opinion, and t-1, 1 year before the opinion, the event of a credit change can happen. A credit change be positive (upwards), or negative (downwards). As described in my hypothesis development I argue that this credit rating event in itself contains information for the auditor (H1b & H2a). Information contained in a credit rating issued before t-1 will be likely already processed in the assessment before the prior year opinion. A credit rating issued close to moment t is expected to be more of relevance than as it was close to t-1 (H2b).

Figure 1.

I use the following logistic regression model to test my hypothesis:

GCO = α0 + α1CR + α2CRD + α3CRDS + α4CRI + α5CRDI + α6M_Return + α7Curr_Loss + α8Prior_Loss +

α9NetW_TotLib + α10CashF_TotLib + α11CurrAss_CurrLib + α12TotLib_TotAss + α13ChNetW_TotLib +

α14Prob + α15Comp_Size + α16Default + α17Ind_Spec + α18Rep_Lag + α19Audit_Fee +ɛ

The variables and the control variables which I will use are described below. I will apply the variables on a standalone basis and in a combined model to test my hypothesis.

3.2 (In)dependent Variables

Dependent

(14)

The main dependent variable in this study is the going concern opinion (GCO) as issued by the auditor or not. This opinion is either present or not, making it a dichotomous variable. I assign a 1 if the opinion is present or 0 if it’s not.

Independent

CR = Credit Rating as given by S&P

The creditworthiness of a company is rated with a letter as described above in chapter 3. This rating is given on an ordinal scale with 22 levels, with AAA being the highest and D the lowest possible (See Appendix A). I assign all ratings an equivalent number, AAA = 1, AA = 2, etc. D =22. This rating can be assigned in the year preceding the audit opinion, but can also be given before this timeframe

CRD = Credit Rating Downgrade in the year preceding the audit opinion.

Dichotomous variable indicating if there was a credit rating downgrade between t-1 and t. 1 is assigned if there is a credit downgrade event, 0 is there is no such event.

CRDS = Credit Rating Downgrade Steps.

Ordinal variable indicating the number of number of steps a credit rating is downgrade in the event there is a credit downgrade between t-1 and t. S&P indicates that a CR only indicates the agency’s opinion on the probability of default. A BB is more likely to default than BBB in S&P’s opinion; however the relative difference in probability cannot be measured.

CRDI = Credit Rating downgrade below investment grade.

Dichotomous variable indicating in the event of a downgrade, if the downgrade is a downgrade from investment to speculative grade between t-1 and t. 1 is assigned if there is a credit downgrade event below investment, 0 is there is no such event.

3.3 Control Variables

Carson el. al. (2013) did an extensive synthesis of existing literature relating to auditors reporting on going-concern. Chapter 4 of their study comprises a summary of client related factors associated to GC as documented and used in prior studies. I only considered control variables for which is consensus in prior literature, or for which is little evidence supporting the contrary. For this reason I excluded auditor size, i.e. Big-4 or not. It is included by some studies as a control variable (Mutchler et. al. 1997) and some weak evidence has been found by Boone et. al. (2010). Contrary results were found Defond et. al (2011), Defond and Lenox (2011) and Numan and Willekens (2011). As such I deemed there was no sufficient support for this variable to be included as a control variable.

I sub-divided the control variables into 4 different segments, dependent upon were information is gathered from.

• Financial Statements. The information in a FS is publicly available for other users than the auditor. Prior studies have shown that information from a FS is a reliable source for bankruptcy prediction. Mutchler (1985) and LaSalle & Anandarajan (1996) found in their questionnaires amongst auditors that use this in their information as well.

• Company. Prior research showed that company specific variables can be of influence as well. Several studies found that larger companies are less likely to receive a GC opinion. Furthermore companies in default are more likely to receive a GC opinion.

• Auditor. Auditor specific characteristic are of importance as well as they are the entity issuing the GC opinion. Industry specialization, audit fee and audit report lag are all found to be positively related to the probability of a GC opinion.

(15)

Control Variables

Abbreviation Description Expected

Relation Reference

Market

M_Return Firm's stock return over the year Negative Dopuch et. Al (1987) Defond et. Al. (2002)

Financial

Statements

FS_Curr_Loss Current Year Loss. Dummy variable, 1 if there is a CY loss, 0 if there is no CY loss. Positive Dopuch et. Al (1987)

FS_Prior_Loss Loss in prior year. Dummy variable, 1 if there was a LY loss, 0 if there was no LY

loss. Positive Bruynseels & Willekens (2012)

FS_NetW_TotLib Net worth / Total Liabilities. Nr. 1 ratio as found by L&A Negative LaSalle & Anandarajan (1996)

FS_CashF_TotLib Cashflow from operations / Total liabilities. Nr. 2 ratio as found by L&A Negative LaSalle & Anandarajan (1996)

FS_CurrAss_CurrLib Current Assets / Current liabilities. Nr. 3. ratio as found L&A Negative LaSalle & Anandarajan (1996)

FS_TotLib_TotAss Total Liabilities / Total Assets Nr. 4. ratio as found by L&A Positive Dopuch et. Al (1987) LaSalle & Annandarajan (

Defond et. Al. (2002) FS_ChNetW_TotLib Change in net worth / Total liabilities Nr 5. ratio as found by L&A. (Current Net

Worth - Prior year Net worth) / Total liabilities. Negative LaSalle & Anandarajan (1996)

FS_Prob Zmijewski (1984) probability of default score. Positive Dopuch et. Al (1987), Defond et.al (2002), Bruy

Willekens (2012)

Company

C_Comp_Size Company Size. Larger are less likely to receive a GC opinion. Log (Total Asset). Negative Mutchler et.al (1997), Geiger & Raghunandan (

Geiger & Rama (2006) C_Default If the company is in payment default or technical default. Dummy variable, 1 if the

company is in payment default or technical in default. Positive Chen & Church ( 1992) Bruynseels & Willekens

Auditor

A_Ind_Spec Industry specialist auditors are more likely to issue GC opinions. Dummy variable, 1

if auditor has a 50% market share for an industry SIC code. Positive Reichelt and Want (2010), Lim and Tan (2008) A_Rep_Lag Auditor report lag. Number of days between fiscal year end and the date of the audit

report. Positive

Carcello et. Al. (1995) Defond et.al (2002). Geig (2005)

A_Audit_Fee Increase in audit fee leads to higher propensity to issue a GC opinion. Possible due

(16)

3.4 Data selection

My research focusses on the years 2000 to 2013. In my data request from Compustat & AuditAnalytics I also requested the year 1999 as I also included the control variables prior year loss and Change in Net

Worth in my model which are dependent upon prior year information. Furthermore I included data up

until 2014 in my data request as opinions for the year 2013 are issued in the year 2014.

The initial date request contained 237,351 Firm/Years for 48,380 individual firms. All off the firms included in the dataset are firms who reported their annual report to the SEC. Based upon the individual firm code I requested from Compustat Ratings, rating information for the same years. For only 8047 individual firms a credit rating was available during 2000-2013. Credit rating information was mostly available for US and Canadian firms. With the remaining dataset I calculated the variables needed for the model. I filtered out all firm/years for which one or multiple variables were missing. Resulting in 14,754 firm/years for which data is available.

Table 1 Sample Selection

Procedures Observations

Firms/Years available in Compustat AuditAnalytics 1999-2014 237,351

Less:

Firms/years for which no credit rating is available (214,197)

Firms with missing data in Compustat (8,400)

Final Sample 14,754

Out of the 14,754 firm/years for which data is available a GC opinion was issued for 328 firm/years as can be seen in Table 2. This is only 2,2% of the total population. This is much lower than the 17,47% over the years 2002/2009 as reported by Carson et.al. (2013). this is due to a combination of 2 reasons. 1) As described in prior literature, small companies are more likely to receive a GC opinion, 2) larger

companies are more likely to receive a credit rating. Due to combining audit opinions with available credit ratings, a bias for larger firms was created in my sample. As described larger firms are less likely to receive a GC opinion. Of the 328 firms/years for which a GC opinion was issued, 162 were first time GC opinions. This is only 1,1% of the data. The relative low amount of GC opinions compared to the total sample is due to that I didn’t identify distressed firms on forehand. Studies like Defond et. al. (2002), Hopwood et. al. (1994). I do acknowledge that the GC opinion decision is most salient for financial distressed firms. However I deem the chosen approach of the whole sample more fit for the purpose of this study as this assesses the behavior of the credit rating agency and the auditor for the whole population.

(17)

4 Results

4.1 Univariate tests

Table 2, listed below, shows the distribution of the sample over the different years selected. 10 out of 14 years have more than 1,000 opinions issued for the year. The lack of data available in 2000 is likely due to the lack of comparative informative information for 1999. The decrease in number of opinions in the year 2011, 2012 and 2013 is likely due to audit reports not yet issued. As can be seen in table 3 some audit reports lag for 857 days, which is more than 2 years. Especially for the year 2013, many of the audits are still in process and as such for a large number of companies some data is missing. Over years there is some variance in the amount of GC opinions per year. In the beginning of 2000’s the average is above 3 % in the aftermath of dotcom bubble burst. 2004 – 2007 show lower averages in the relative amount of GC opinions issued, this is likely due to the stabilized economy at that moment and growing stock markets. In 2008 the percentage of GC opinions is above 3% again, most likely due to the effects of the subprime mortgage crisis with some noteworthy chapter 11 filers like Lehman Brothers and

Washington Mutual (others like Bear Stearns, AIG, Freddie Mac and Fannie Mae were saved by government or taken over by competitors). After 2008 the percentage of GC opinions per year

decreases. Opinions which are prone to a qualification (GC or other) are most likely the ones which are delayed as auditors want reassurance before taking such a decision. Therefore the amount of GC opinions decreases over time.

Table 2

Distribution of Opinions across years Firm/Years 14.754

Firms 1.628

Distribution Total Non GC GC Percentage GC Opinion

2000 424 411 13 3,1% 2001 1.078 1.048 30 2,8% 2002 1.269 1.223 46 3,6% 2003 1.428 1.381 47 3,3% 2004 1.393 1.357 36 2,6% 2005 1.318 1.290 28 2,1% 2006 1.240 1.213 27 2,2% 2007 1.134 1.120 14 1,2% 2008 1.092 1.058 34 3,1% 2009 1.061 1.040 21 2,0% 2010 1.026 1.018 8 0,8% 2011 997 986 11 1,1% 2012 963 951 12 1,2% 2013 331 330 1 0,3% 14.754 14.426 328 2,2%

Table 3 contains the descriptive statistics for the sample population. The average and median credit rating are both rounded 11, which corresponds with a BB+ credit rating. This is the highest possible speculative rating. This implies that on average the credit ratings in the sample are almost equally distributed between investment & speculative ratings, with a slight tendency towards speculative. This is

(18)

in line with mean for speculative rating or not which is 0,52 with 1 being speculative and 0 investment grade. The median rating for a firm/year receiving a GC opinion was CCC-, and average 18,4 which is between CCC and CCC-. This indicates that both the auditor and the credit rating agency identify similar companies a priori for which they have doubts about the ability of the entity to continue as a going concern. The highest credit downgrade in magnitude was an 18 steps downgrade from AA- to D for Franks Nursery and Crafts as the company filed for chapter 11 in May 2002, subsequently a GC opinion was issued by the auditor in the same year. The median for downgrade & downgrade steps is 0 for both, indicating that most firm/years there is no downgrade event. Furthermore the downgrade mean is 0,13 indicating that downgrades don’t happen often. Although speculative ratings are more than half of the credit ratings, there are very few ‘first time speculative’ ratings in my data set. This is possible due to the severity of a downgrade below this level. Due to regulatory requirements discussed above the capital markets for companies shrink after receiving a speculative credit rating.

The mean for GC of first time GC are both very close to 0, as can be seen in table 1 as well. This is possibly explained that mostly larger corporations receive a credit rating, as seen in prior studies larger companies less likely receive a GC opinion.

FS_Curr_Loss = Dummy variable, 1 if company makes a net loss in the current year, 0 if not FS_Prior_Loss = Dummy variable, 1 if company makes a net loss in the prior year, 0 if not FS_NetW_TotLib = (Total Assets – Total Liabilities) / Total Liabilities

FS_CashF_TotLib = Operating Cashflow for the year / Total Liabilities FS_CurrAss_CurrLib = Current Assets / Current Liabilities

FS_TotLib_TotAss = Total Liabilities / Total Assets

FS_ChNetW_TotLib = Current year (Total Assets – Total liabilities) – Prior Year (Total Assets – Total liabilities) / Current Year

Total Liabilities.

FS_Prob = Probability of bankruptcy score (Zmijewski [1984]) C_Comp_Size = Natural Logarithm of Total Assets

C_Default = Dummy variable, 1 if company is defaulting on its debt obligations, 0 if not

Table 3

Descriptive Statistics for 14,754 companies from 2000 to 2013 (including 162 firms with GC Opinion) with available information for the variables in the model

Full Sample (n= 14754)

Variables Mean Median SD Minimum Maximum Variables Mean Median SD Minimum Maximum

Control Independent FS_Curr_Loss 0,78 1,00 0,41 0,00 1,00 Credit Rating 11,04 11,00 4,04 1,00 22,00 FS_Prior_Loss 0,21 0,00 0,41 0,00 1,00 Downgrade 0,13 0,00 0,34 0,00 1,00 FS_NetW_TotLib 0,95 0,63 2,62 (0,97) 263,00 Downgrade steps 0,10 0,00 1,27 0,00 18,00 FS_CashF_TotLib 0,23 0,14 4,78 (7,14) 580,00 Speculative grade 0,52 1,00 0,50 0,00 1,00 FS_CurrAss_CurrLib 1,91 1,50 3,30 0,02 264,00

First time speculative 0,02

0,00 0,14 0,00 1,00 FS_TotLib_TotAss 0,64 0,61 0,44 0,00 32,32 FS_ChNetW_TotLib 0,97 0,04 108,95 (16,51) 13.237,00 Dependent FS_Prob (0,83) (1,01) 3,28 (212,91) 179,60 GC 0,02 0,00 0,15 0,00 1,00 C_Comp_Size 7,97 7,89 1,62 0,05 12,84 FT_GC 0,01 0,00 0,10 0,00 1,00 C_Default 0,03 0,00 0,17 0,00 1,00 A_Ind_Spec 0,39 0,00 0,49 0,00 1,00 A_Rep_Lag 61,46 58,00 35,09 0,00 857,00 A_Audit_Fee 14,39 14,37 1,22 7,45 18,78

(19)

A_Ind_Spec = Dummy variable, 1 if company is Industry Specialist for a CIK industry code based upon if an auditor has a market

share of 50% or more in sample year based upon audit fees.

A_Rep_Lag = Number of days between fiscal year end and auditor sign off date A_Audit_Fee = Natural Logarithm of Total Audit Fee

CR = Credit rating for a company 1 for AAA, 23 for D

DG = Dummy variable, 1 if there is a credit rating downgrade event in the year preceding audit opinion, 0 if not DG_Steps = Number of downgrade steps if there is a downgrade event

Spec_Gr = Dummy variable, 1 if the credit rating is speculative: below BBB-, 0 if not

FT_Spec_G r= Dummy variable, 1 if credit rating downgrade event is a first time speculative rating: below BBB-, 0 if not GC = Dummy variable, 1 if the company receives a Going Concern opinion, 0 if not

FT_GC = Dummy variable, 1 if the company receives for the first time a Going Concern opinion, 0 if not

4.2 Multivariate results

I hypothesized that the information embedded in the credit ratings was incorporated in the going concern opinions. The results are reported in Table 4. I used several proxies for the information embedded in the credit ratings which are reported in model 2 to 6. Model 1 reports the results for the baseline model with only the control variables included. Subsequently in Model 2–6 introduce one individual variable to the model, to verify if a variable adds any value to the model on a standalone basis. In general, the results are consistent with the hypotheses.

More specifically model 2 reports the findings for the credit rating proxy. Here, the coefficient on credit rating is positive and significantly different from zero, which implies that firms with a more negative credit rating face a greater likelihood of receiving a going concern opinion (p<0,01). Model 3 examines the role of credit rating downgrades. Here the coefficient on downgrade is positive and significant (p<0,1), which suggest that companies receiving a credit rating downgrade in the year preceding the issuance of the audit opinion are more likely to receive a GC qualification.

In model 4 the effect of the severity of the downgrade is examined. The coefficient in downgrade steps is positive and significant (p<0,1). For each extra step in downgrade in the year preceding the audit opinion the likelihood of receiving a GC opinion for a company increases. Model 5 examines the role of a speculative rating. Here, the coefficient on Speculative rating is positive and significant (p<0,01).

Companies with a speculative credit rating are more likely to receive a GC opinion. This is line with model 2, were I noted that companies with a lower credit rating are more likely to receive a GC opinion. Speculative ratings are of lower degree than investment ratings. In my final model I examined the role first time speculative CR’s. A positive relation was noted, however not significant. This result is

somewhat at odds with the outcomes of model 5 as that models shows that speculative rating or not is of significant influence on the likelihood of a company receiving a GC opinion. The event of a first time speculative rating is however not of significant influence, indicating that possible changes in the available capital market (due to regulation for banks/ investor, discussed earlier) in the year preceding the opinion are of less importance than the fact that a rating is speculative or not. Possible explanation is that a credit rating downgrade from investment to speculative is often a minor downgrade from a rating just above the line to a rating just below the line. As noted earlier the average and median rating in the sample is already speculative at BB+. As show by model 4 the likelihood of a company receiving a GC opinion increase in line with the number of steps of the downgrade. Likely most downgrades from investment to speculative are only 1 or 2 step downgrades and as such not significantly increasing the chance of a company receiving a GC opinion.

In table 5 the independent variables are combined into one model. Credit rating, Downgrade steps and speculative grade continue to have a positive coefficient which is significant (p<0,01). Downgrade is not significant anymore. Likely this is due to that the information provided by Downgrade steps overrules the information provided by Downgrade. First time speculative is not significant, which is in line with the

(20)

standalone model. The R2 of model 1 (model with only control variables) was 98,2%, with all the

variables added into one model this is still at 98,2%.However in the GC category the prediction increased from 38,1% in the control variables only model to 40,9% including the independent variables.

Overall, the findings suggest that the information imbedded in credit ratings is also used in going concern opinions. With regard to variables, the results are generally consistent with prior research. 3 control variables were not significant in any single model, Total Assets, Industry specialization and audit fee. Default was significant in all standalone models at p<0,05, strangely this control variable is not significant in the combined model in table 5.

(21)

Independent

Variables Coefficient Estimates

Wald chi-square

P-value Coefficient Estimates Wald

chi-square

P-value Coefficient Estimates

P-value Coefficient Estimates Wald

chi-square

Wald chi-square P-value Constant 1,001 3,123 ,077 1,149 30,372 ,000 1,013 4,191 ,041 1,026 4,440 ,035 1,032 5,821 ,016 1,002 3,134 ,077 Control FS_Curr_Loss ,199 39,568 ,000 ,200 21,655 ,000 ,202 30,530 ,000 ,202 29,092 ,000 ,200 31,442 ,000 ,199 37,970 ,000 FS_Prior_Loss ,174 17,795 ,000 ,173 6,024 ,014 ,174 16,558 ,000 ,176 18,453 ,000 ,173 13,675 ,000 ,174 18,616 ,000 FS_NetW_TotLib ,268 23,336 ,000 ,235 17,838 ,000 ,273 20,241 ,000 ,272 20,642 ,000 ,266 19,724 ,000 ,269 23,335 ,000 FS_CashF_TotLib ,448 59,689 ,000 ,446 49,248 ,000 ,451 58,677 ,000 ,459 55,033 ,000 ,446 60,316 ,000 ,449 59,561 ,000 FS_CurrAss_CurrLib ,095 25,297 ,000 ,094 25,584 ,000 ,096 23,561 ,000 ,096 22,811 ,000 ,097 28,340 ,000 ,095 25,277 ,000 FS_TotLib_TotAss ,447 1,152 ,283 ,414 2,123 ,145 ,452 ,274 ,601 ,455 ,128 ,720 ,443 1,251 ,263 ,448 1,056 ,304 FS_ChNetW_TotLib ,022 72,983 ,000 ,022 61,265 ,000 ,022 70,046 ,000 ,023 66,041 ,000 ,022 73,043 ,000 ,022 72,784 ,000 FS_Prob ,060 5,700 ,017 ,057 7,594 ,006 ,059 3,569 ,059 ,060 2,596 ,107 ,059 6,399 ,011 ,059 5,528 ,019 C_Comp_Size ,068 4,471 ,034 ,067 ,028 ,866 ,070 7,481 ,006 ,070 8,771 ,003 ,068 2,315 ,128 ,068 5,051 ,025 C_Default ,171 265,904 ,000 ,276 4,562 ,033 ,174 264,025 ,000 ,178 212,842 ,000 ,171 247,537 ,000 ,171 266,356 ,000 A_Ind_Spec ,151 ,068 ,794 ,151 ,000 ,986 ,152 ,024 ,876 ,153 ,082 ,774 ,150 ,020 ,888 ,151 ,105 ,746 A_Rep_Lag ,001 27,574 ,000 ,001 21,174 ,000 ,001 24,514 ,000 ,001 19,431 ,000 ,001 25,827 ,000 ,001 26,341 ,000 A_Audit_Fee ,087 ,042 ,838 ,087 ,144 ,705 ,088 ,130 ,719 ,089 ,315 ,575 ,087 ,004 ,951 ,087 ,054 ,816 Independent Credit Rating ,034 80,926 ,000 Downgrade ,160 20,719 ,000 Downgrade steps ,036 38,421 ,000 Speculative grade ,306 10,110 ,001

First time speculative ,382 2,425 ,119

N 14.754 14.754 _14.754 _14.754 _14.754 _14.754

Likelihood Ratio 1.617,041 1.525,910 1.596,926 1.578,405 1.605,071 1.614,873

Pseudo R2 (%) 98,2% 98,2% 98,2% 98,2% 98,2% 98,2%

No GC 99,6% 99,6% 99,6% 99,6% 99,6% 99,6%

(22)

4.3 Additional analysis First Time GC opinion

In prior literature most emphasis is put on first time GC opinions, as they are genuinely considered the most value relevant for the market. I repeated the analysis as performed on GC opinions. In table 6 below I included the model with only the control variables in model 1 and added subsequently an independent variable on an individual basis in the subsequent models 2-6.

The independent variables that had positive coefficients and were significant for GC opinion are this also for First GC opinions (all p<0,01). Difference is that now the first time speculative (model 6) has a

positive coefficient and is significant as well (p<0,01). This indicates that the information embedded in a Table 5

Logistic Regression for the Association all Variables

Model 1

Independent Variables

Coefficient

Estimates Wald chi-square P-value

Constant 1,135 27.951 .000 Control FS_Curr_Loss ,207 19.239 .000 FS_Prior_Loss ,179 7.721 .005 FS_NetW_TotLib ,242 16.733 .000 FS_CashF_TotLib ,451 45.892 .000 FS_CurrAss_CurrLib ,094 21.108 .000 FS_TotLib_TotAss ,435 0.894 .344 FS_ChNetW_TotLib ,022 56.258 .000 FS_Prob ,059 4.676 .031 C_Comp_Size .071 1.033 .310 C_Default .308 1.006 .316 A_Ind_Spec .155 .021 .885 A_Rep_Lag .001 15.529 .000 A_Audit_Fee .089 .001 .974 Independent Credit Rating .043 64.047 .000 Downgrade .246 .108 .742 Downgrade steps .058 6.906 .009 Speculative grade .421 8.283 .004

First time speculative .542 .075 .785

N 14.754

Likelihood Ratio 1.499,654

Pseudo R2 (%) 98,2%

GC 99,5%

(23)

first time speculative down grade is relevant for the auditor in his GC decision making. This is in line with the expectation in hypothesis 2c. Possible explanation is that for a first time GC the auditor needs a triggering event which pushes to make the GC decision. The move from investment to speculative can be such a triggering event due to the regulatory restrictions discussed earlier. For non-first time GC opinions the move from investment to speculative is of no value if there is already a GC opinion issued. In table 7 the 5 independent variables are combined in one model. Credit rating and Downgrade steps remain significant at p<0,01, first time speculative at p<0,1. Downgrade & speculative grade are not significant in the combined model. Likely the downgrade steps and first time speculative overrule downgrade & speculative grade in relevance. Results indicate that for an auditor to issue a first time GC opinion a severe triggering event is required, either a downgrade of multiple steps or a downgrade from investment to speculative grade.

Compared to the model with only the control variables the pseudo R2 increased from 98,2% to 99%. The number of GC opinions predicted correctly increased from 8,6 for model 1 to 23,5% for the model with all the variables included.

Findings suggest that the information impounded in credit ratings is also used in going concern opinions. Strange movement in control variables is that Zmijewski’s probability of default score is not significant in any of the models on first time GC, opposite to the models on GC opinions. This indicates that the auditor’s decision to issue a GC opinion is not strictly based upon financial indicators. Combining this with the 3 independent variables still significant in the combined model (credit rating, downgrade steps & first time speculative) indicates that the auditor needs a triggering event before issuing a GC opinion. The credit rating is of a very low level, the downgrade is of a high magnitude or the downgrade is for the first time speculative.

The influence of credit ratings on auditor’s going concern decision