How difficult is to understand the extended audit report?

Abstract The external audit report is a guarantee of financial information reliability that companies provide to the market for stakeholders’ decisions. However, some firms did not disclose relevant information on company risks for users due to managers’ pressure. Few years ago, auditors of listed companies in Europe have had to break down firms’ risks using the Key Audit Matters (KAMs) in a new format called extended audit report. This article conducts an empirical analysis to examine the content of this report, to know how difficult to read is and its differences between the first and the second year of its implementation. Using all Spanish listed companies, our results show that extended audit reports are difficult to understand. Furthermore, the evidence corroborates that KAMs are sticky for a firm year to year as there are no statistically significant differences in the readability indices, neither in the narrative nor linguistic style, comparing the first two years of its implementation.


Introduction
The accounting scandals that have taken place during the last two decades calls for a greater concern by the accounting profession. The impact of fraud was estimated at $360 billion in the US (Awolowo et al., 2018), in addition to the social consequences for thousands of employees and workers who lost their savings, their retirement funds, their jobs, etc. Despite management errors, there was unethical behavior that destroyed the credibility of their managers and, therefore, the reliability of economic and financial information issued by companies in the market. Moreover, these scandals affected not only the financial sector but were widespread (telecommunications, energy, health care, etc.). In this context, the following question must be asked: how was it possible that no one noticed the risks?
As a double-check, external auditors in advanced economies are the guarantee of reliability of firms' financial information as "a cornerstone of confidence in the world's financial systems" (PWC, 2017). Auditors are independent professionals hired by firms' managers to issue an opinion about the financial position and the statement of profit or loss of the company every year. The audit opinion can be positive (unqualified), negative (qualified), disclaimer or adverse as to whether the firms' financial records comply with the accounting standards of the country in which the company operates. Moreover, the report should refer to any material uncertainty that may cast significant doubt about the firm's ability to continue as a going concern. The going concern is a standard qualification generally issued when a firm's financial viability is in question. Thus, auditors are required to mention any evidence found during the audit processes regarding the risk of failure (McKee, 2003). Furthermore, users should rely on audit opinion when deciding whether to invest in business according to its financial condition (Geiger et al., 2005). However, some users are dissatisfied with the ability of auditors to warn about imminent corporate failures, overall during the global financial crisis. Users complained that auditors did not alert adequately in their opinions about impending bankruptcies (Geiger et al., 2014;Sikka, 2009). Another question arises at this point: how was it possible that companies with a clean audit opinion could go bankrupt without the auditor being aware of their situation? By way of example, the Enron case not only led to the company's bankruptcy, but also dragged down the auditor Arthur Andersen and, as a result, the credibility of all audit firms suffered a severe blow. Another example, from the analysis of the audit reports of bankrupt Spanish companies, 13% of those reports did not include any comment from the auditor (Muñoz-Izquierdo et al., 2018). Some of these 13% reports on bankrupted Spanish companies have received some penalty or been investigated for irregular professional practices (for example, 1 Pescanova, Bankia and Abengoa). Although improving the clarity of the reports does not necessarily lead to improving their intrinsic quality, it may help the readers to identify weak reports (this point could be stressed).
As a result of the above-mentioned financial scandals and with the increasing complexity of financial reporting, there has been a demand for more informative audit reports and for auditors to provide more relevant information on companies' risks to users of financial information (Abad et al., 2017;IAASB, 2016). In this context, the International Auditing and Assurance Standards Board (IAASB) approved a new international standard that incorporates significant changes to the information contained in audit reports (IFAC, 2013). In the European Union, the IAASB's regulatory framework has been adopted by Directive 2014/56/EU and EU Regulation No. 537/2014(of the European Parliament and Council, 2014. In the US, the new audit report has been implemented in 2020. In Spain, as a first step in the process of adopting the European regulations, in July 2015, Law 22/2015 on the Auditing of Accounts (LAC) was published, and it was applied to the financial years beginning on 17 June 2016.
Historically, the audit report has been described as a pass/fail model because both opinions contain highly standardised wording (Cipriano et al., 2017). One of the main new features of the extended audit report is the introduction of Key Audit Matters (hereinafter "KAMs"), which are those issues that, in the auditor's opinion and selected from among the matters reported to the audited entity's management, have been of the greatest importance for the audit of the financial statements for the period. KAMs arise from those matters that required significant attention from the auditor, for example, areas of significant audit risk or matters where management and directors were required to make important judgements or estimates in the preparation and fair presentation of the financial statements. They may also relate to relevant circumstances or transactions that have taken place during the period. In fact, KAMs are expected to enhance the reporting capability of the auditor's report by, among other things, helping "investors and other users of financial statements to focus on those aspects of the companies' financial statements that the auditor has identified as deserving special attention" (PCAOB, 2013, p. 6) and, providing "a roadmap that makes complex financial reports more accessible to users and helps them focus on the issues that are likely to be most important for decision-making" (IAASB, 2016, p. 36).
This study aims to explain how difficult to understand the extended audit report is and whether there are differences between the content of the new extended audit report for the first and the second year of the implementation, especially focused on the KAMs. Previous investigations into the effects of extended audit reports and on investor reaction to KAMs have been conducted mainly in Anglo-saxon context (Kend & Nguyen, 2020;Velte & Issa, 2019) and have focused on three areas of analysis: the number, type, level of detail (granularity) of KAM (Sirois et al., 2018;Filipović et al., 2019;Gambetta et al., 2019a and b;Gold et al., 2020;Moroney et al., 2020;Smith, 2021). Moreover, the current paper examines the KAMs narratives, focusing on the readability and understandability. In this context, we adopt the concept of audit quality in terms of the readability of the KAM section. The quality of the extended audit report is a crucial research question due to its recent implementation in many countries around the world. In addition, the readability and understandability of the financial information, and the extended audit report, in particular, is relevant to check whether the aim of the reform is achieved. Hence the audit firm's style and the way to communicate the information affect the quality of the KAMs and the informative value, the main objectives of the reform.
Our work complements previous studies in the field because it is the first study that analyses the content of the audit report, using a linguistic analysis and readability indices, in a non-Englishspeaking country, Spain, focusing on KAMs in the second year of experience after the audit reform. It is also noted that Hsieh et al. (2021) examine textual similarity of KAM in Spain for two years, focusing on Cosine similarity index. There are, however, important differences between both studies. First and foremost, whereas Hsieh et al. (2021) do not provide a detail descriptive of KAM, considering the number, type, granularity, market, recurrence, and characteristics of KAMs. Second, whereas Hsieh et al. (2021) use an index to compare the similarity between two years, our study conducts a content analysis focusing on three readability indices and a detail linguistic analysis.
We believe that regulators, auditors, investors, managers, and analysts can benefit from the results of this study, as it contains evidence on how informative the report is after the IAASB reform. Therefore, this study contributes to the advancement of the line of research on the effects of regulatory changes on auditing from a critical point of view, answering the following questions: Where are we going? Is it the desired direction? Is there room for improvement? What are the control mechanisms?
The rest of this document is organized as follows. Section 2 examines the theoretical framework and formulates the research questions. Section 3 describes the sample and methodology. Section 4 presents the results and discussions of this analysis. Finally, the conclusions are presented in section 5.

Financial information and narrative corporate disclosures
The financial information comprises a set of numerical expressions and quantitative data, as well as narrative disclosure reported in annual reports and other management documents. The narratives complement the quantitative information. Taken together, the corporate information should allow managers to show a true and fair view of firm's financial situation and results. A growing literature examines the narratives of financial information, including the levels of complexity and simplicity, readability, legibility, and linguistic styles (Asay et al., 2018;Draeger et al., 2020;Jones & Shoemaker, 1994;Lennox et al., 2022;Merkl-Davies & Brennan, 2007;Merkl-Davies et al., 2011;Nyman et al., 2021;Smith, 2021;Zhang et al., 2019).
Managers can use narratives to confer additional information to users, that is, narratives can be useful to overcome information asymmetries and increasing information usefulness. But it also possible that managers use their discretion to engage in opportunistic behavior, although audited companies have a lower level of earnings manipulation (absolute discretionary accruals) than the non-audited ones (Huguet & Gandía, 2016). Agency theory postulates that parties' interests (the principal and the agent) are not always aligned. The information asymmetry and the different interests can motivate managers to engage in dysfunctional behaviour to maximize their utility and their interest at the expense of the other groups (Jensen & Meckling, 1976). Impression management, obfuscation hypothesis or incomplete revelation also explains managers motivation to use narratives to mask certain events such as exposure risks, weak results (obfuscation hypothesis), and intentionally alter users' expectations (Courtis, 1998;Merkl-Davies & Brennan, 2007). Indeed, several studies show that the information on bad news is less readable and more complex to understand than the information of good news (Asay et al., 2018) and managers have incentives to associate bad results with external factors and good results with internal factors subject to their control (Merkl-Davies & Brennan, 2007). Leary and Kowalski (1990) argue that the alteration of the language intentionally contains two elements: the motivation (understanding the incentives of individuals to engage in such behavior) and the construction (the type of narrative chosen by the individual).
The legitimacy theory also explains managers' decisions in order to confer organisational legitimacy to the firm. Firms are social agents and play an important role in the society. Then, managers choices can be motivated to confer firm legitimacy and avoiding legal actions, such as litigation and sanctions (see e.g., Deegan, 2002). In recent years, managers have increased the narratives disclosed in annual reports, but the objectives pursued can be different. Badawy and Ibrahim (2019) review the readability of narrative corporate disclosures over the last two decades and call for more empirical papers and linguistic techniques to capture better the manager narratives. They also document that the readability of the corporate disclosures could differ across countries (as firm´s location is a factor to control).

The extended audit report
The main objective of the extended audit report is to increase the quality of the traditional audit report and its informative value. Unlike the old pass and fail audit report, the extended audit report contains more information as shown in Table 1.
The regulatory reinforcements are ultimately aimed at improving the content and disclosure of the audit report and at improving auditors' practice in assessing companies' accounting practices and risks. As described in the previous section, the new audit report is one of the elements that can contribute significantly to reducing the audit expectation gap, that is, the difference between what users expect from the audit report and what it actually expresses.
As a result of the consequences of the economic crisis that began in 2007, many users of companies' financial information have complained about the process of auditing and verifying such information. The literature review has identified several measures to dismiss the gap in audit expectation: increased awareness of audit responsibilities (Adeyemi & Marte Uadiale, 2011;Salehi, 2011), increased auditor independence (Gassen & Skaife, 2009;Ruhnke & Schmidt, 2014;Toumeh et al., 2018) and improved audit reporting models (PWC, 2019;Ratzinger-Sakel & Gray, 2015;Vanstraelen et al., 2012). In this paper, we will focus on the third mechanism related to the extended audit report and, on the impact of the content of the audit report and, in particular, on KAMs. In that sense, some studies conclude that KAMs have positive effects in the reduction of the audit expectations gap (Manoel & Quel, 2017;Pinto & Morais, 2019;Taslima & Fengju, 2019;Trpeska et al., 2017) while others show the opposite (Boolaky & Quick, 2016;Fakhfakh, 2016;Gold Litjens et al., 2015). Some other studies provide mixed results (Coram & Wang, 2021;Lennox et al., 2022). However, KAMs are relevant as a recent study posits (Chang et al., 2022).
The empirical research conducted so far has focused on three relevant issues: the number, type and granularity of KAMs. As regards the number of KAMs, Gambetta et al. (2019a) describe that the number of KAMs broken down in the UK in the first year of adoption of ISA 701 (2013) was 4.4 issues, while in 2016 it was 4.5, so it cannot be shown that the number of KAMs in audit reports increases as experience is gained. Moroney et al. (2020) provide evidence that investors perceive the audit to be more valuable when KAMs are disclosed than when KAMs are absent. Sirois et al. (2018) conducted a study of the effect of KAMs' communication on the auditor's report on users of financial information when analysing financial statements. One of the main conclusions of the study is that when auditors break down several KAMs in audit reports, users of financial information pay less attention to the remaining paragraphs of the report. Suttipun (2022) document that the average of KAM per company in Thailand is 1.63 KAMs, and the number of issues is similar during the first years of the extended audit report.
As for the type of KAMs, Filipović et al. (2019) show that the most frequently KAMs broken down referred to accounting items of income, impairment, and valuation of assets and provisions. Gambetta et al. (2019b) analyse the influence of auditor and client characteristics on the number and type of KAMs broken down in the audit reports of FTSE 100 companies in the United Kingdom during the period 2013-2016. They conclude that auditor and client characteristics are determinants of the number of KAMs described. Furthermore, these factors determine the type of KAMs in the audit reports. Abdullatif and Al-Rahahleh (2020) examine the application of ISA 701 in Jordan for the first two years of its implementations (2017 and 2018). The results show that firms report a low number of KAMs, being the valuation of accounts receivable the most common. It is also noted that the number of words per KAMs is similar in both years, and, in general, the narratives are not extensive.
Considering the granularity in the risk description, Smith (2021) provides empirical evidence that narratives in the disclosure of KAMs are more effective in enhancing the informational value of auditors' reports for investors than the mere presence of sections of KAMs. In a companion study, Gold et al. (2020) finds that managers are more conservative behavior when received an extended report with KAMs. This conduct can be a signal of less aggressive financial reporting decisions when anticipating a KAM. Using a sample of non-sophisticated users, Moroney et al. (2020) find that KAMs enhance the audit report (in terms of perceived value and credibility) when the financial statements are audited by Non-Big 4 firms. It is also detected that the inclusion of KAMs distracts the attention from the main information provided in the audit report. Recently, Chang et al. (2022) highlight that companies with KAMs that contain more client-specific information are perceived as having lower reporting quality and this is due to the risk-related description in KAMs. Overall, their results provide evidence supporting the information value of KAMs in the Taiwanese audit market.
It should be noted that the empirical literature on KAMs is scarce and no-conclusive yet (Minutti-Meza, 2020;Velte & Issa, 2019) and may be because this topic is new in the US and recent in Europe (with the UK and Dutch auditors in 2014 being the first to adopt these changes in the audit report). One Anglo-saxon study (Kend & Nguyen, 2020) and three specific studies (FRC, 2015;Mazars, 2018;Auditanalytics, 2019) have analysed the longitudinal implementation of KAMs from a continental institutional setting. Kend and Nguyen (2020) provide the first evidence on whether auditors used the same or different disclosures related to audit procedures when reporting on the same KAM in the second year in Australia, an Anglo-saxon institutional setting. All of three others are descriptive and adopt a professional approach. Nguyen and Kend (2021) document KAMs disclosures are perceived differently amongst market participants. Hence, only some stakeholders consider that KAMs enhance audit quality. Our study responds to this gap by providing a critical approach by empirically comparing the first and second years of KAMs implementation holistically, and from the point of view of a non-Anglo-Saxon country. Many studies show differences between regulations due to their national context as it also happens with KAMs (Minutti-Meza, 2020; La Porta et al., 1998).

KAMs narrative and readability
The narrative of the audit report with the incorporation of KAMs is an area of special interest for researchers since some previous studies have shown that the introduction of KAMs in the audit report reduces its readability (Carver et al., 2017). Moreover, KAMs does not increase the information for investors (Lennox et al., 2022), only some stakeholders consider that KAMs enhance audit quality (Nguyen & Kend, 2021) and the readability of the extended audit report is greater compared to the previous audit report without KAMs although after the first year of implementation of KAMs its readability is not significant (Smith, 2021).
Most previous research of the field uses readability indices and alternative measures based on the number of words in the text, words per sentence, number of complex words (measured by the number of syllables in the words), number of pages of the document, etc., to measure the complexity of the text (see, for example, Courtis, 1998). However, the conclusions can be more precise considering additional variables related to linguistic characteristics. The frequency of positive and negative words or the tone (optimistic, neutral, or pessimistic) is a proxy to measure how the message is communicated to users (see for instance, Lennox et al., 2022;Merkl-Davies & Brennan, 2007;Smith, 2021). For example, Smith (2021) uses readability and tone language in pre and post ISA 700 audit reports and finds that the last one is easier to read and better reflect the risk-related nature of the financial statement. Overall, these results show that expanded audit disclosures can be communicated easier to process and potentially meaningful to the financial statement user, but "auditors must be intentional in maintaining the benefits of expanded disclosure" (p. 2). Zhang et al. (2019) also combine the readability indices (Fog index) and the narrative of the management report (specifically, the use of words related to causal arguments) to determine whether or not there is a relationship with the analysts' earnings forecast. The results show that this linguistic dimension provides additional information and therefore improves analysts' projections, i.e. the dimension associated with causal reasoning is positively associated with analyst following and earnings forecast accuracy. Fakhfakh (2016) examines the linguistic performance of the illustration on consolidated financial statements under ISA 700. The author detects linguistic problems and linguistic imperfection of international accounting standardization which reduce the relevance, the comprehensibility, and the reliability of the audit report. Indeed, the evidence shows that the audit report is not readable by all users. Kabuye et al. (2019) find that the tone used by top management is associate with risk management practices. Hsieh et al. (2021) focus on content analysis of KAM in Spain for two years using Cosine similarity index. The evidence indicates similarities in KAM disclosures, driven mainly by specific KAM topics for clients within specific industries. (2018) find differences in disclosure readability when performance is bad (less readable) than when performance is good (more readable), motivated mainly by manager self-enhancement. The linguistic analysis shows that firms with bad news use fewer personal pronouns in the first person (I, me) and more passive voice tenses to transmit distance between the preparers or managers and the message (accounting report) presented to users of financial information. The use of personal pronouns in the first person as well as verb tenses in the active voice focuses attention on the preparers of the message and communicate the message more directly. In addition, companies with poor results present more justifications or causal arguments ("because", "therefore", etc.) and use more future verb tenses compared to past verb tenses, which supports the thesis that attempt to report more information to reduce the asymmetry of information, as well as to focus attention on the future projection of the firm rather than the performance of the year. Brennan and Merkl-Davies (2018) argue that readability is not conditioned by external factors. Jones (1996) explains that readability only focuses on the content of the text and the semantic difficulty. However, complex language, ambiguous sentences, duplicate information, and other factors, can make difficult its fully comprehension.
In summary, there are controversial conclusions about the effectiveness of the new extended audit report in reducing the audit expectation gap and, in particular, about the impact of KAMs on this process. This issue reinforces the relevance of the analysis of KAMs and their evolution over time and across jurisdictions.

Research questions
Firstly, we examine the number, type, granularity, market, recurrence and characteristics of KAMs. Secondly, we focus on the KAMs narratives. Specifically, our research questions are as follows: Research Question 1 (RQ1): Are KAMs easy to understand? Are there differences in the KAMs readability between the first and the second year of its implementation?
Our expectations are that KAMs are easy to understand because they should explain the risks firms' must face in their future. However, not only the readability is important but also the linguistic analysis of the KAMs. Therefore, our second question is defined as follows: Research Question 2 (RQ2): How are the narratives of KAMs? Are there differences in the linguistic analysis of the KAMs between the first and the second year of its implementation?
Prior critics to the utility of audit reports were based on its standardization although recent studies highlight exactly the opposite that "the new audit report model has contributed to its destandardization" (Ferreira & Morais, 2020, p. 262). As KAMs are new information for auditors they should be not standardized. For this controversy, we posit the next research question: Research Question 3 (RQ3): Does the extended audit report tend to the standardization?
The aim is to empirically tests whether the extended audit report, in terms of the information disclosure tends to be standardized. In this context, if KAM disclosure achieves the stated goals, the audit report should be more informative. From a managerial point of view, a potential standardization could be a risk for firms' stakeholders as auditors highlight the same KAMs all the time without managers' actions. In other words, if auditors disclose the same KAMs from one period to other period, some stakeholders could think some inactions from managers. Taken together, we examine in detail the audit report, with a focus on KAMs, including a handmade analysis, readability indices, linguistic analysis and its potential standardization. Finally, we propose some improvements.

Sample
The sample is composed of the whole population of the companies listed on Spain's main stock exchange on 31 December 2017 and 2018 (see, Table 2). The sample includes listed companies on IBEX-35, the benchmark stock market index of the Bolsa de Madrid and comprise the 35 most liquid Spanish stocks, and Continuous Market (CM) with consolidated financial statements, all stocks trade simultaneously on the Madrid, Bilbao, Barcelona and Valencia stock exchanges. The final sample is obtained after applying certain criteria: restructuring and dissolution companies are excluded, companies transfer to Euronex, companies with non applicable UE Reform in 2017, and companies without consolidated financial statements. The final sample, thus, is composed of 131 companies (27% of the IBEX-35 and 73% of the CM), during the period 2017-2018. Our sample is grouped into 8 sectors: construction and real estate (20.6%), consumer and distribution (6%), energy and natural resources (12.2%), financial (10.7%), pharmaceutical (1.5%), industrial (23%), services (17.6%) and technology and telecommunications (8.4%).
In this paper, we use the first and the second years of the implementation because our objective is to examine the first implementation of KAMs in the expanded audit report. Prior research has found recurrent KAMs and textual similarities year over year. For example, Kend and Nguyen (2020) found that around 70% of Australian listed firms have the same KAMs disclosed in both years 2017 and 2018. Consistently, using Thai listed companies, Suttipun (2022) show that, although the word count of KAMs reported fluctuated during the three-year period studied (2016)(2017)(2018), the volume of matters reported is similar each year. Thus, due to the international adoption of the audit reporting regulation, we expect a similar number of KAMs disclosed in the Spanish market in 2017 and 2018, so we believe that the first years of implementation of this new regulation are enough for the purpose of our investigation.

Variables
The variables of this study have been obtained from the analysis of the extended report and have been grouped into the following categories: Paragraph type. Three variables have been identified based on the type of audit opinion paragraph in accordance with ISA 701: "audit opinion" (OPIAU), which may be favourable or qualified; "emphasis of matter" (EMPA) if it reflects a matter that is presented or disclosed properly in the financial statements and which, in the auditor's opinion, is material to users' understanding of the financial statements; and "going concern" (GCPA) if it describes a material uncertainty related to the entity's ability to continue as a going concern.
Type of KAMs. KAMs are broken down into two categories: Accounting KAMs and General KAMs, according to the classification of the Spanish Institute of Chartered Accountants (ICJCE), ISA 701, Lennox et al. (2022) and Pérez-Pérez et al. (2019). Accounting KAMs. These are those related to a specific heading in the annual accounts. Six variables have been identified corresponding to KAMs associated with each of the headings in the financial statements: Non-current assets (NCA), which include intangible assets, investment property, financial investments and investments in Group companies and associates; Current assets (CA), which include inventories and other current assets; Liabilities (LI), which include bank borrowings, pensions, other financial liabilities, provisions, and contingent liabilities; Income recognition (IR); Goodwill and business combinations (BC); Deferred taxes (DT).
General KAMs. Those that are related to more general issues and are generally conditioned by external factors. Four variables have been identified corresponding to KAMs that are not associated with any specific heading in the financial statements, but rather represent general risks that affect the financial statements globally: Legal and regulatory compliance (LRC), including compliance with the terms of control; Unusual or significant transactions (UST); Going concern (GCPA); and Information systems (IS).

Methodology
Firstly, a detailed analysis was conducted, consisting of the reading, analysis and classification of the content of the extended audit report of the 131 companies listed on the Spanish market. From the reading and analysis of this extended report, a codification of its content was developed, based on the study by Muñoz-Izquierdo et al. (2016) which classifies the variables of the analysis considering the types of opinion paragraphs and the types of KAMs, as described in section 3.2. Specifically, to classify each KAM presented in the extended audit report, dichotomous variables are created that takes the value from 1 if the corresponding category appears in the report, and 0 otherwise. As a result of this process, dichotomous variables are generated that include both the type of opinion paragraph and the KAMs included in them. That is, we transform the qualitative nature of the audit reports into a set of quantitative or dichotomous variables for subsequent statistical analysis. Then, we analyse the number, type and granularity of KAMs in our sample.
Subsequently, a content analysis of the KAMs was performed. For responding to RQ1, we calculate the readability indices. The objective is to test whether it is easy to understand and whether there are differences in the readability indices in the two years analyzed. In particular, we aim to test the improvement in the second year compared to the first year. In English, the most common readability indices are Fog Index, Flesch Kincaid and Flesch Reading (see, Badawy & Ibrahim, 2019;Courtis, 1998;Hasan, 2020;Zhang et al., 2019). Most common languages have adapted these indices or created new indicators to measure the readability of texts. In Spanish, Fernández-Huerta made the first adaptation of the Flesch index in 1959, calling it the Readability formula. Subsequently, several authors have proposed variations and adaptations of this index as Szigriszt Pazos (IFSZ), and Escala Inflesz which is a reinterpretation of the Szigriszt Pazos index. These indices have been used to measure the readability of Spanish financial reports (Melón-Izco et al., 2021) and also in audit standards (Novejarque Civera, 2012). Based on the previous works, this study measures the readability of the extended audit report using the Flesch-Fernández Huerta index, the Flesch-Szigriszt index (and their reinterpretation, the Escala Inflesz), and the mu index, which is based on the variance of the characters in a text and the average number of characters. Subsequently, for responding to RQ2, a linguistic analysis of the KAMs was conducted. The objective is to examine the narratives of the KAMs and whether there are differences in the dimensions in the two years analyzed and in particular, to test the improvement in the second year compared to the first year. We used the LIWC program, which is a tool developed by Pennebaker et al. in 2001, 3 to assess linguistic and cognitive styles across a range of psychological categories and subsequent studies applied in management, audit or finance (Asay et al., 2018;Draeger et al., 2020;Jones & Shoemaker, 1994;Merkl-Davies & Brennan, 2007;Merkl-Davies et al., 2011;Zhang et al., 2019). Draeger et al. (2020 point out that the LIWC program is an alternative to manual text coding and analysis in qualitative research, avoiding errors and biases in data processing, and therefore advises researchers to use it. The Spanish version of the LIWC 4 (equivalent to the English version, and validated by; Ramírez-Esparza et al., 2007 and also used in several papers such as; Salas-Zárate et al., 2014), contains 72 linguistic dimensions (based on 7,515 words and word roots) that include standard language categories (e.g., articles, prepositions, pronouns, etc.), affective processes (e.g., positive and negative emotions, cognitive processes), words related to relativity (e.g., time, verb tense, space, etc.), and are organized hierarchically so that they export the percentage of use of each dimension. In sum, the categories selected from LIWC are: general dimensions (words per sentence, words of more than 6 letters and numbers), affirmations and negations, verbal tenses (past, present and future), psychological processes (positive and negative emotions), cognitive processes (causation, which contains words related to "because", "effect", "by", etc.; insight, which contains words related to "consider", "know", "think", etc.; discrepancy, which refers to "should", "could", etc.; differentiation such as "but", "else", etc.; tentative such as "maybe", "perhaps", etc., and certainty such as "always", "never", etc.) and relativity (inclusions and exclusions). We also calculate two additional measures, obtained manually, which are the number of pages of the audit report and the number of pages of the KAMs (considering the difference between the initial and the final number of pages).
Subsequently, following Merkl-Davies et al. (2011), we calculated the indicator of analysis of the linguistic style of the reports (self-representation dissimulation) using the following formula: Self-representation dissimulation indicator = -z number of words-z references to the company -z references to others + z positive emotions + z negative emotions-z cognitive processes. 5 The indicator is composed of four dimensions obtained directly from the LIWC software (the logarithm of the number of words in the document and the percentage of positive emotions, negative emotions and cognitive processes in the document) and two dimensions created ad-hoc, which are the references to the company or self-references (first person plural, we, our, the group and the name of the company) and the references to others (industry, sector, competitor and rival). Both are also measured in percentages. When the indicator takes higher values, it is understood that the company is showing a more truthful and accurate image of the company according to the financial information (Merkl-Davies et al., 2011). For the second objective, we determine whether there are differences in the value taken by the index in the two years analyzed. If the index takes higher values in the second year, we could interpret it in favour of the audit report, and specifically the KAMs, containing more accurate or truthful information about the company's risks.
Finally, for responding RQ3, we based on the results of the handmade analysis, as well as the readability indices and linguistic analysis. The content analysis conducted in the paper allows us to know if the extended audit report tends to the standardization.
Several steps have been taken to analyse the content of the audit reports: the readability indices and linguistic analysis. Firstly, the reports in pdf format obtained from the CNMV website have been manually reviewed, as most of them are presented in a protected format that does not allow manipulation. Secondly, KAMs information has been extracted from the audit reports, eliminating other types of information such as the company logo, number of pages, headings concerning the company name, references to the auditing firm, etc. Subsequently, the KAM information in pdf format has been converted into txt files, manually checking that the text recognition is done properly. In those cases where text recognition has not been possible or has given rise to reading errors, the report has been removed from the sample.

Results and discussion
In this section, we present the results of the content analysis of the extended audit reports issued by the auditors in Spain first time. Those are the annual reports of entities listed on the main Spanish stock exchanges ("IBEX-35" and "CM"), after the entry into force of the enhanced auditing information standards (i.e. for the financial statements as of 31 December 2017 and 2018).
We have divided the results into two sub-sections related to our research questions. First, we analyse the main descriptives of the audit reports manually with a focus on KAMs. Second, we calculate the readability indices and the linguistic analysis of the KAMs and its evolution. Finally, we propose some improvements.

Analysis of KAMs descriptives
In this section, we present a deep analysis of KAMs, including the number of risks, type of risk and granularity (Tables 3 and 4). Table 3 shows that the audit reports of the Spanish listed companies broke down at least one KAM, but not more than 8 (in the UK, the highest number is 10 according to the FRC, 2015). Classifying the KAMs according to the market, the extended audit reports show 8 and 7 KAMs for the IBEX-35,   The evidence shows that KAMs were generally consistent across the two exercises analysed in this paper. Abdullatif and Al-Rahahleh (2020) obtained similar conclusions in the application of ISA 701 in Jordan considering the first two years of its application. The removal or addition of a KAM was often caused by one-time or unusual, non-recurring transactions. However, it is notable that the uncertainty about hyperinflation in Argentina has been added as a new KAM in several cases.

Number of risks
Further exploration shows that the changes between 2017 and 2018 per firm in terms of number of KAMs are between 1 and 2, except in two specific cases which are: "Adveo Group International, S.A." which has reduced the number of KAMs from 5 to 1, because in 2018 the auditor has issued an adverse opinion by reducing the number of KAMs and "Distribuidora Internacional de Alimentación, S.A." which has increased the number of KAMs from 2 to 7, due to the operating company paragraph which implied many accounting areas to focus on. Table 4 provides additional information for other companies such as Repsol, Caixabank, Abengoa, among others.
Taken together, the results indicates that the removal or addition of a KAM is often due to oneoff or unusual, non-recurring transactions. However, it is noteworthy that there are two cases in which KAMs identified by the current auditor are recurrent, as is the case of Caixabank and Liwe Española, S.A., with the change of auditor between the Big Four and the second between the Big Four and other corporations (see , Table 4). Table 5, last row, reveals that the total number of risks are 373 and 358 in 2017 and 2018. Most of the KAMs identified in large companies in Spain include non-current assets, goodwill impairment, revenue recognition, and deferred tax recovery. Since these are often critical areas of "management judgment", and of particular importance for the valuation of companies, this is perhaps not surprising. These four highest-rated risks represent 69% (67% in 2018) of the total number of risks reported by auditors. This is in line with some similar studies conducted in the context of the 2017 UK audit reports (FRC, 2015;Auditanalytics, 2019). In Jordan, Abdullatif and Al-Rahahleh (2020) obtain that the valuation of accounts receivable is the most common KAM in Jordan, followed by the valuation of inventory, investment property, and revenue. It is also noted that goodwill impairment is not frequently reported.

Types of risks
Considering the nature of KAMs, some conclusions caught our attention. Concerning accounting KAMs, the most important variation comes from business combinations, as explained in the following section when analysing unusual transactions. As for general KAMs, the most significant variations are found in (i) Legal and regulatory compliance, since in 2017 some KAMs were related to the application of IFRS 9 and Bank of Spain Circular 4/2017 of January 2018, which financial institutions were required to implement in 2017 and, therefore, were not applicable in 2018 and (ii) unusual or significant transactions and others due to specific transactions carried out during each year that are not recurring. This is also related to business combinations, with the total increase in this KAM being an exceptional 4%. This could also be due to the better Spanish economic cycle that could promote these new transactions, mergers, acquisitions, etc.

Granularity
ISA 700 requires that "to be useful to users of financial statements, explanations of matters to be included in the auditor's report should be described". We have evaluated the descriptions of KAMs in each of the 131 audit reports in our study. In general, the ratio between granular risks citing specific circumstances and generic risks written in a more standardised language is similar between 2017 and 2018, at around 40% and 60% respectively, so there is room for improvement in this regard. Compared to the United Kingdom, for example, this percentage is worse, as in this jurisdiction the granularity increases from 50% to 80% in the second year. This conclusion is in line with Smith's study (2021) on the extended audit report in the UK and the FRC report (2015) which indicated that further improvements were still needed, particularly about increasing client-specific risk reporting. KPMG and PWC are the companies that have made a major effort to increase granularity in the descriptions, aggregating granularity in KAMs by almost 10%. The company that still offers more specific descriptions of KAMs is Deloitte with 50% of all KAMs described as specific. Abdullatif and Al-Rahahleh (2020) also noted that Big 4 audit firms generally report more information about the nature of the risk (although they attempt to avoid entity-specific matters and prefer to show industry-specific matters) compared to non-Big 4 audit firms.
Moreover, investors particularly value information when it is presented concisely. This creates a particular challenge for auditors, as they are required to convey granular information on highly technical matters in a concise and easily understandable manner. This, therefore, led us to review the volume of breakdown, as well as its content. We have analyzed the average word count for the risk descriptions in each report (see, Table 6). The average number of words for IBEX-35 companies is around 450 and 400 for CM companies in 2017 and 2018, with a slight increase of 7% and 3% respectively. On average, risk descriptions tend to be longer for IBEX-35 companies than for CM companies. This may be because they are generally larger and more complex companies. Concerning audit firms, PWC and Deloitte tend to include longer risk descriptions in all their reports. This reflects how these audit firms have sought to provide greater transparency and detail in risk reporting but highlights the challenge of meeting the potentially conflicting expectations of report users. Table 6 shows slight differences between the two years. Abdullatif and Al-Rahahleh (2020) find that the wording in Jordan is lower compared to Spain, and Big 4 audit firms tend to report more information than non-Big 4 audit firms. Moreover, as we can see in the table, going concern KAMs have the highest number of words as it is a vital issue for companies. As going concern issues could be totally correlated with bankruptcy (  well, firms could disappear. It is a high-risk issue that conditions the future viability of companies and inevitably increases the risk of litigation towards auditors. Related to tax KAMs, their number of words is lower, as it is a more objective issue due to the existence of specific fiscal regulation and tax controllers by governments. Unlike going concern KAM, which is a subjective issue, an estimation for auditors, tax issues are objective based on strict and concrete regulation. Table 7 presents the readability indices for answering RQ1. The first section, the general dimension, shows that the extended audit report on average is 7.41 pages. The number of pages of the KAMs in the extended audit report (considering the starting and the final page in the pdf) is approximately 47% of the total number of pages in the audit report. It should also be noted that words longer than 6 letters are frequent, which is one of the variables used in many of the readability indicators of the texts in English, and it increases in 2018 compared to 2017. The use of numbers and figures in the audit report is not common (0.19 in the two years under review). In this regard, empirical papers in psychology show that the combination of text and numbers increases the readability of documents. However, the position and the exact location of the figures and graphs in the text are also matters.   Table 8 shows the results of the linguistic analysis for answering RQ2. The results show that the frequency of negative words ("no", "never", etc.) is higher than the frequency of positive words ("yes", "of course", etc.). In the verb tenses dimension, a preference for the present compared to the past or future is observed. Indeed, expressions in future verb tenses, which may inform issues in a near time horizon, are not very frequent. Concerning cognitive processes, the frequency of words associated with insight and causation categories are the most recurrent, followed by tentative. Less recurrent ones are words associated with certainty ("yes", "always", etc.) or discrepancies ("should", "could", etc.). In the last dimension, words referring to "with", "and", "including", etc. (category of inclusions) are more recurrent than the words referring to "but", "without", etc. (category of exclusions).

Content analysis: readability indices and linguistic style
In the last column of Table 8 and for all categories, the results show that there are no statistically significant differences between the years 2017 and 2018: general dimensions (p > 0.05), affirmations and negations (p > 0.05), verb tenses (p > 0.05), affectivity (p > 0.05), cognitive processes (p > 0.05), relativity (p > 0.05). Although the most relevant variables for the audit report have been examined, it should be noted that there are no statistically significant differences either in the other dimensions available in the LIWC program.
Taken together, the handmade content analysis, readability indices, and linguistic analysis, that correspond with our RQ3, indicates that the extended audit report tends to the standardization. We can conclude that the information on KAMs is difficult to understand and in the first two years of adoption of the new regulations is similar in number, nature, and the variations are due to non-

Discussion and proposals for improvement
While the standards provide guidelines for determining whether an audit issue constitutes a KAM, there is relatively little guidance regarding the number of KAMs to be reported and how to report them. Indeed, which and how many KAMs to report is a matter of professional judgment. Although the new requirements were designed to allow the auditor to provide users with more information about the audit process, following the results of the previous analyses, we believe that there are still opportunities for improvement to achieve greater transparency and to help to reduce the audit expectation gap.
Since the issuance of ISA 701, several accounting/audit regulatory bodies have issued various documents or guidance on the breakdown of KAMs to provide further clarification on their wording and approach. Similarly, market supervisors play a very important role in reviewing the compliance of companies and their auditors with accounting regulations and have also issued documents in this regard. On the other hand, some countries, such as the United Kingdom, have gone beyond  Table 9. Additional information contemplated in the audit reports in the United Kingdom versus the Report for the Audit Committee in Spain

Report for the Audit Committee in Spain
An explanation of the concept of materiality in planning and conducting the audit.
• Description of the scope and timing of the audit.
• Description of tasks in case of joint audits.

•
Indication of the quantitative level of materiality for the financial statements as a whole and, where appropriate, for particular categories of transactions, balances and information and indication of the qualitative factors which have been considered in determining the level of materiality.
A summary of the scope of the audit, in the case of group audits, and how materiality has influenced it.
• Scope of consolidation and, where appropriate, the exclusion criteria applied by the audited entity, indicating whether such criteria are consistent with the financial reporting framework.
• Indication of the parts of the work performed by third country auditors. (Continued)

Report for the Audit Committee in Spain
Exceptions by the auditor in relation to the section of the annual report where the Audit Committee (Audit Committee Report) describes the significant issues discussed in relation to the financial statements and how they were addressed • Description of the methodology, indicating, among others, which items have been verified directly and which through system testing and compliance testing, explaining the variations in the weight of evidence, with respect to the previous year and the state of resolution by management of the significant deficiencies in internal control and/or the accounting system identified.
• Evaluation of the valuation methods applied to the various items in the financial statements, including the possible effects of changes in such methods

•
Argumentation of the judgments relating to facts detected during the audit that may raise doubts about the company's ability to continue as a going concern, indicating whether they constitute material uncertainty, together with the support measures that have been taken into consideration. Source: Own elaboration the requirements of ISA 701 by incorporating through their local legislation the publication in the audit report of additional aspects related to the work performed by the auditor (see , Table 9). In Spain, following Article 11 of EU Regulation No. 537/2014 through the Article 36 of Law 22/2015 on the Auditing of Accounts (LAC) auditors must issue the additional report to the Audit Committee, in writing, on the date on which the audit report is presented and they should explain, among other aspects, the results of the audit procedure. Table 9 shows the additional information presented by the public audit reports in the United Kingdom and the report for the Audit Committee presented in Spain. In this case, this document is not public but only issued for this Committee in the private sphere of the company.
Based on the analysis of all the previous information, our proposal to advance with the development of the new audit report is summarising, providing greater value to the interested parties, in three initiatives, to be considered by the regulatory bodies that may take a stand on this matter: update the current guidelines on KAMs, increase the information to be issued in the audit report; and grant greater supervision to the audit reports. In addition, an executive summary with the main results of the auditing report can help investors.

Conclusions
The extended audit report requires that auditors provide new information about firms' risk to make financial information more effective for the decision-making process of their users. In this context, the IAASB adopted a new international standard that incorporates significant changes to the information contained in audit reports. One of the main changes is the obligation for listed companies to include KAMs in the audit report, which is the subject of this paper. The auditors of listed companies in Spain have had to report on KAMs since the financial years beginning on 17 June 2016.
Our research empirically analyses the content of the extended audit reports for the whole listed companies in the second year after the reform, from a critical perspective. The results show that more than half have presented the same nature and number of KAMs as in the first year (2.7 on average), with variations due to non-recurring transactions. The audit reports describe at least one KAM, but not more than eight. 69% (80% in 2018) of the IBEX-35 and 96% (98% in 2018) of the CM companies break down between 1 and 4 KAMs. The most commonly reported KAMs were current assets, goodwill impairment, revenue recognition and deferred tax recovery. Moreover, the description of KAMs is more extensive in the second year, although there is room for improvement in providing more specificity of these risks.
According to the readability indices, KAMs are difficult to understand. This result opens the debate about the usefulness and informative value as it pushes auditors to more effort, time and consequently costs to issue that report. The results of the content analysis show that there are no differences in the readability indices of the KAMs or in their narrative in the first two years of implementation of the extended audit report. Our evidence corroborates that KAMs are sticky for a firm year to year as there are no statistically significant differences in the readability indices, neither in the narrative nor linguistic style, comparing the first two years of its implementation. This raises a crucial question of whether the extended audit report improve the information about firm risk areas to be useful for the users or tends to be a standardised report for the next years. This should primarily be viewed in light of the informational value associated with disaggregation sections, and the cost of implement the new reports for preparers. Managers could see the standardization as a risk for firms' stakeholders as auditors highlight the same KAMs all the time without managers' actions. If auditors disclose the same KAMs from one year to other, some stakeholders could think that managers do not do anything to solve those critical material misstatements for their firms.
The evidence contributes to the debate and discussion of the effects of the new accounting standards, and in particular of the audit report which is a cornerstone in the financial information published by listed companies. In this regard, we consider that there are still opportunities for improvement to achieve greater transparency and contribute to reducing the gap in auditing expectations, and it is necessary to incorporate some actions. Among these, we propose the preparation of specific guidelines by local accountancy/auditing bodies that allow auditors to break down risks in financial information in a more concrete way and/ or explore the possibility of incorporating additional information such as in UK reports (materiality, etc.) into these new reports, as well as incorporating the analysis of these KAMs into the reports already issued by supervisors about financial information and the audit report.
Our study extends the existing literature by emphasizing the importance of risks on financial reporting in the extended audit report and to improve the extended audit report, focusing on the breakdown of KAMs to reduce the audit expectation gap. We hope that our study will encourage discussion of additional mechanisms to improve the content of the audit report not only in Spain but in other jurisdictions. For example, in the United States, the new extended audit report on KAMs is mandatory for large accelerated filers as of accounting closures after 30 June 2019 and 15 December 2020 for all other companies. Future contributions could also focus on qualitative research based on stakeholders' perceptions of the information contained in the new audit report. It would be worth adding more variables as type of auditor, industry, audit fees, corporate governance variables, among others, to test whether there are systematic differences in the extended audit report readability.
Finally, this study is not without its limitations. The research is based on Spanish companies listed in the second year of its implementation, but a comparative analysis between different countries could be an interesting avenue for future research. Also, studies can continue increasing the number of years and investigating the COVID-19 pandemic effects.