Consequences of adopting an expanded auditor’s report in the United Kingdom

When the auditor concludes that the financial statements comply with the applicable financial reporting framework, the auditor’s opinion is called unqualified (FRC 2013a; PCAOB 2017) or unmodified (IAASB 2015a). For simplicity, we use the term “unqualified.”

Although the auditor is permitted and, in some circumstances required, to modify the wording of the auditor’s report while still expressing an unqualified opinion, modifications are relatively narrow in nature. For instance, Czerney et al. (2014, p. 2116) note: “Additional explanatory language typically refers to the adoption of new accounting standards, changes in accounting methods, restatement of prior financial statements, reliance on other auditors to complete the audit engagement, or any other matters the auditor wishes to emphasize.” Moreover, Czerney et al. (2014) demonstrate that the most common type of explanatory language mentions the adoption of new accounting standards and changes in accounting methods. These modifications are qualitatively different from the new expanded report disclosures.

Academics (Bedard et al. 2016) and regulators (FRC 2015; PCAOB 2017) refer to the report including additional client-specific information as the extended or expanded auditor’s report. For simplicity, throughout our study we use the term “expanded auditor’s report.”

Under the new rule, the length of the auditor’s report for premium companies has tripled, increasing from approximately one to three pages. The auditor describes typically a total of four risks, including one unique risk not mentioned in the audit committee report. The average materiality is approximately 0.5% of total assets.

Reviews of the first two years of reports conducted by the FRC involving discussions with investors and auditors suggest that there is room for improvement, for example, by increasing the granularity of risk reporting and providing more explanations regarding the materiality, scope, and risk disclosures (FRC 2015, 2016). A potential threat to the relevance of risk disclosures is the inclusion of too many and boilerplate risks that act as an auditor’s disclaimer of responsibility (Brasel et al. 2016; Kachelmeier et al. 2017).

These experimental studies investigate the effect of critical audit matters following the PCAOB’s rules (CAMs, which are similar to risks of material misstatement in the UK context) on investors’ decisions and jurors’ perceptions of the auditor’s responsibility and legal liability in case of a misstatement. The external validity and the generalizability of these studies to the United Kingdom is an open question.

In Section 5, we explain in detail how our analyses and findings complement those of Lennox et al. (2017), Reid et al. (2017), and Reid et al. (2015). Moreover, we conduct several additional analyses aimed at mitigating concerns that the differences between our findings and those reported by Reid et al. (2015, 2017) are the result of sample selection or research design choices. We find no compelling evidence that our conclusions are significantly influenced by differences between our study and those by Reid et al. (2015, 2017), except for the results on discretionary accruals (audit quality), which are sensitive to the calculation of this proxy and sample years. There is also another related study by Smith (2017), but its main focus is on machine-based textual analysis of the auditor’s report. Using audit report readability and tone as proxies for communication value, this study finds that, following the passage of the new United Kingdom rules, audit reports have become easier to read and better reflect the risk-related nature of financial statement audits.

The Main Market is the LSE’s principal market for listed companies from the United Kingdom and overseas. Premium listing is only open to equity shares issued by trading companies and closed- and open-ended investment entities. Premium companies must comply with the United Kingdom’s highest standards of regulation and corporate governance. In the United Kingdom, the new rules apply to premium companies with listing of equity shares, regardless of whether they are incorporated there or elsewhere. In Ireland, this includes Irish incorporated companies with a primary or secondary listing of equity shares on the Irish Stock Exchange. In our analyses, we focus on companies incorporated in Great Britain.

These disclosures include (1) a description of the significant accounting issues that the audit committee (AC) considered in relation to the financial statements and how these issues were addressed, (2) an explanation of how the AC assessed the effectiveness of the external audit and the approach to the appointment of the auditor, (3) information about auditor tenure and last tender, and (4) how the AC ensures the objectivity and independence of auditors who provide non-audit services (paragraph C.3.8 of the UK Corporate Governance Code FRC 2012).

The US Securities and Exchange Commission approved the standard on October 23, 2017 (SEC 2017).

We note that, although generally similar, the FRC, IAASB, and PCAOB requirements regarding significant risks of material misstatement (KAMs and CAMs) do not have identical definitions. Please refer to the online Appendix D for a summary of the most recent regulatory developments, including details about the definitions of KAMs and CAMs.

On the other hand, an Australian study (Ogneva and Subramanyam 2007) failed to find an association between going-concern opinions and investors’ reaction, potentially reflecting the fact that the incidence of going-concern opinions in Australia is significantly higher than elsewhere and that the market there could anticipate the opinions (Simnett et al. 2016).

There is also a stream of recent experimental research that examines the consequences of the expanded auditor’s report, focusing particularly on the effect of critical audit matters (CAMs) on investors’ decisions and on jurors’ perceptions of the auditor’s responsibility and legal liability in case of a misstatement. These studies generally support the notion that the expanded report provides useful information to sophisticated investors and lawyers (Christensen et al. 2014) and affects jurors’ perceptions of auditor liability following a misstatement (Gimbar et al. 2016; Brasel et al. 2016; Kachelmeier et al. 2017). For additional discussion of the archival and experimental literature on the auditor’s reporting model, please see the reviews by Bedard et al. (2016) and Mock et al. (2013).

For instance, The National Audit Office considered “the proposed new requirements as illustrated do not provide genuine insight into the audit nor the most useful ‘hooks’ for investors to use in subsequent discussions with management.”

The FTSE 100 index is maintained by the FTSE Russell Group This index includes the 100 companies listed on the LSE with the highest market capitalization (representing approximately 80 percent of the total LSE capitalization). This index is a widely used summary indicator for the UK stock market. For additional details, please see http://​www.​ftserussell.​com. In untabulated analyses, we also examined the trends in closing prices and trading volume for the FTSE 250, a similar index including the 101st to the 250th companies listed on the LSE with the highest market capitalization. The trends are qualitatively similar to those for the FTSE 100. In 2013, other European indexes also experienced an upward trend, reporting their best annual gains since 2009. France’s CAC-40, Spain’s IBEX, and Germany’s DAX increased by 18, 20, and 26 percent, respectively. The FTSE 250 closed at 15,935 and gained 28.8 percent over 2013 (Allen 2013).

For example, see the announcement by ITE Group plc on December 24, 2013, in RNS that its annual report is available at http://​www.​londonstockexcha​nge.​com/​exchange/​news/​market-news/​market-news-detail/​11816292.​ html. The FRC requirements are described in detail at http://​fshandbook.​info/​FS/​html/​handbook/​DTR/​6/​3. We were unable to identify reliable report filing dates in Worldscope, Bloomberg, or other online open sources. We find that the annual report filing dates in Worldscope and Bloomberg often correspond to the earnings release date, as per the RNS announcements. The LSE RNS website can be accessed searching each company at http:// www.​londonstockexcha​nge.​com/​prices-and-markets/​markets/​prices.​htm. The NSM can be accessed searching each company at http://​www.​morningstar.​co.​uk/​uk/​NSM. An alternative source is the website http://​www.​investegate.​co.​uk. For a small number of observations, we are unable to identify the specific annual report release date. In these cases, we use the Annual General Meeting announcement date, if available, or the Annual General Meeting date.

We calculate market variables by aggregating companies’ individual data. We obtained the constituents of the LSE from the WSCOPEUK list from Worldscope for securities on LSE that are major and primary (EXDSCD = ‘LN’ MAJOR = ‘Y’ and ISINID = ‘P’) and that are neither investment trusts nor closed-end funds (TYPE = ‘EQ’). This list includes both Premium and AIM listed companies. Our measure of abnormal volume is consistent with those of DeFond et al. (2007) and Miller (2010).

Discretionary accruals are estimated using an annual cross-sectional model for each SIC two-digit industry (Kothari et al. 2005) for all the companies on the London Stock Exchange. To obtain information about these companies, we use the WSCOPEUK list from Worldscope for securities on LSE that are major and primary (EXDSCD = ‘LN’ MAJOR = ‘Y’ and ISINID = ‘P’) and that are not investment trusts or closed-end funds (TYPE = ‘EQ’). Discretionary accruals are the residual of the following model: TACCR_i,t = α + β₁ROA_i,t + β₂ΔSales_i,t + β₃PPE_i,t + ε_i,t, where, for firm i and year t, TACCR_i,t is income before extraordinary items - cash flow from operations)/average total assets; ROA_i,t is (net income before extraordinary items)/average total assets; ΔSales_i,t is (sales in year t - sales in year t-1)/average total assets; and PPE_i,t is (gross property, plant, and equipment)/average total assets. The model is estimated by industry (SIC two-digit), excluding industries with less than 20 observations.

In the context of our study, it is difficult to implement analyses of other common audit-quality proxies. Going-concern opinions have a very low incidence; we find only 21 going-concern opinions among 1212 firm-year observations used in our audit fee analyses. Financial restatements are also rare and arguably have limited consequences. Each year, the FRC typically examines 300 annual reports from all listed companies and asks about 15 to 20 companies either to restate their accounts or to change their practices for the following year. However, some observers claim that “such an FRC requirement will not attract a wider audience as it will be mentioned only briefly in a company’s report and accounts” (Smith 2014). In our additional analyses section, we discuss results using other dependent variables for audit quality, including meeting or beating the analyst consensus forecast and earnings response coefficients (ERCs). However, we note that, despite our best efforts, all our proxies for audit quality are inherently indirect.

We do not explicitly assess the effect of scope disclosures, apart from examining the effects of the total length of the auditor’s report, because they are difficult to quantify, in many cases discussing the group structure of the audit without providing additional details.

We consider the risk section language a separate issue. Two similar companies can have standardized or boilerplate language in their risk disclosures, but one company reports four risks and the other reports only one. There is still variation in disclosure; arguably four risks are a more severe threat than one. We believe it is reasonable to expect that the number of risks matters, despite the language used to explain each risk. A concurrent study by Smith (2017) examines variation in readability and tone of the auditor’s report language.

Although large companies, being in most cases relatively more complex, tend to report more risks than smaller ones, these risks are widely scattered across companies of various sizes. For example, a descriptive report by KPMG (2014, p. 18) notes: “auditors of many FTSE 300+ companies are reporting just as many risks as the auditors of some FTSE 50 companies.”

Excluding financial firms is a typical practice in audit fees and audit quality research. Banking and other financial companies (e.g., mutual funds) face high levels of regulatory scrutiny and have additional mandatory disclosures. There are no straightforward reasons to expect that the expanded auditor’s report will be incrementally informative for these companies. We reviewed a number of reports for these companies and found that the most common risk mentioned is valuation of assets, focusing on determining the fair value of investments.

The data used in our analyses are provided in online Appendix C.

The companies listed in the LSE are available at http://​www.​londonstockexcha​nge.​com/​statistics/​companies-and-issuers/​companies-and-issuers.​htm. We used the list available on April 2015. We identified only three early adopters in our sample. Vodaphone Group included an auditor’s report based on the FRC’s exposure draft for fiscal year-end March 31, 2013. Sky plc included an auditor’s report based on the revised FRC standard issued in June 2013 for fiscal year-end June 30, 2013. Ashmore Group included an auditor’s report based on the revised FRC standard issued in June 2013 for fiscal year-end June 30, 2013. Vodaphone Group and Sky plc are included in our sample. For these companies, consistent with the September 2013 cut-off, we considered March 31, 2014, and June 30, 2014, as the first regular year of adoption. We eliminated from our sample a few cases in which companies did not disclose a specific materiality amount, audit fees were not disclosed, or we were unable to extract the auditor’s report from the annual report. We used the list of LSE AIM companies available on July 2016.

We also examine whether the auditor’s report is potentially more informative for investors when companies have large earnings surprises. Arguably, there is comparatively higher investor attention in these cases. For premium companies, we estimate our pre-post model using the subsample with large (i.e., above the median) absolute returns on the earnings announcement day. For this group of companies, we also do not find evidence of a pre-post adoption effect on our market reaction variables.

The FRC argues that this is a possibility in the discussion paper “Auditors and Preliminary Announcement” in April 2017 (FRC 2017).

A visual inspection of figures 4 through 7 suggests that the parallel trend assumption requirement for difference-in-differences holds for all our dependent variables in the two years before the expanded auditor’s report requirements.

We repeated these analyses using a 75th percentile cut-off in defining the indicator variables for long reports, large number of risks, large number of unique auditor risks, and large materiality. These alternative definitions aim to capture significant or more extreme departures than a 50th percentile cut-off. We find (1) a negative and statistically significant coefficient (p < 0.05) for the LENGTH(1/0)_i,t interaction in the absolute abnormal returns model, (2) a positive and weakly significant coefficient (p < 0.10) for the NRISKS(1/0)_i,t interaction in the audit fees model, (3) a negatively and weakly significant coefficient (p < 0.10) for the LENGTH(1/0)_i,t interaction in the discretionary accruals model, (4) and a negative and significant coefficient (p < 0.05) for the NRISKS(1/0)_i,t interaction in the discretionary accruals model. It is difficult to interpret the absolute returns results without a confirmation from the abnormal volume analyses. Moreover, the combined results from the fees and discretionary accruals do not reveal a consistent pattern. The fee results suggest a positive relation between the number of risks and auditor effort, but the discretionary accruals results suggest that clients with a large number of risks (and longer reports) have lower audit quality.

An interesting aspect of the paper by Lennox et al. (2017) is that, for half of their sample, they performed a manual search of risks disclosed by the auditor in year t in the existing company’s disclosures in year t-1 (i.e., annual reports, earnings announcements, and conference calls) and concluded that investors were already informed about the majority of risks addressed in the auditor’s report. These preliminary analyses suggest that the content of the risks’ disclosures is not a significant new source of information.

Our discussion is based on the last available updates of these manuscripts, specifically July 29, 2015, for Reid et al. (2015), September 2017 for Reid et al. (2017), and February 2017 for Lennox et al. (2017). Please refer to the online appendix for a more detailed analysis of the differences between our study and Reid et al. (2015, 2017).

We calculate total accruals (TA) using the following equation: TA = (change in current assets – change in cash – change in current liabilities + change in current debt – depreciation) / lag of total assets. Hribar and Collins (2002) show that there are important shortcomings in calculating accruals using this approach. In our main analyses, we follow Kothari et al. (2005) and calculate total accruals as net income – cash flow from operations.

We note that the inclusion of variables that require analyst forecasts (e.g., meet or beat and unexpected earnings) significantly reduces our sample size. We obtained the latest forecast date from the I/B/E/S detail file and consensus forecasts from Datastream.

In addition, we calculate a pre-post model in which the interaction between POST_i,t and unexpected earnings captures the effect of the expanded auditor’s report for premium companies after September 2013 and find a statistically insignificant pre-post effect.

For instance, among the 302 companies with available risk data in year t and t + 1, 77 increased and 99 decreased the number of risks in year t + 1.

Abnormal return volatility in the period surrounding the filing of annual report date is calculated using the formula \( \frac{\sum_{d\in \left(-1,+1\right)}{u}_{it,d}^2}{3{\sigma}_{it}^2} \), where \( {u}_{it,d}^2={Ret}_{it,d}-{\widehat{\alpha}}_{it}-{\widehat{\beta}}_{it}{Mkt}_{t,d} \) is the abnormal return on day d, \( {\sigma}_{it}^2 \) is the variance in abnormal returns over the 80-day non-announcement period and \( {\widehat{\alpha}}_{it} \) and \( {\widehat{\beta}}_{it} \) are estimated over the 80-days non-announcement period (i.e., −45 to −6 with respect to the earnings announcement and from +6 to day +45 with respect to the filing date).

In our main analyses, most of our continuous variables are winsorized at the top and bottom 99 and 1%, the exceptions being the investors’ reaction dependent variables (i.e., abnormal returns and volume). This is typical practice in capital market studies (see Leone et al. 2016), but we also find results similar to those found in our main analyses after winsorizing the investors’ reaction variables at the top and bottom 99 and 1%.