Big Data Analytics in the Financial Statement Audit
A critical examination of the possible value to the auditors
Bachelor thesis Accountancy & Control
Ivar van den Boogert
10562079
29
thof June 2016, final draft
Professor Brendan O’Dwyer
Statement of Originality
This document is written by Ivar van den Boogert who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Abstract
Currently the position of the financial statement auditor is under pressure, several reports such as the Green Paper of 2010 pressure for change in the audit profession. In this thesis I will answer the question whether big data analytics could be beneficial to the auditors, by examining in which stages of the audit big data analytics could improve the audit in terms of efficiency, cost reduction and quality. Big data analytics is a hype of the last five years, and while interesting applications have found in several fields, there does not yet exist an application for the assurance service industry. My contribution to the existing literature is twofold. First of all I synthesize existing literature concerning big data in a fashion suitable to the auditing profession. Second, in the Big Data forum of the journal Accounting Horizons (2015) authors were highly positive about the possibilities of big data analytics in the
financial statement audit; however, the authors neglect to argue where the possible benefits could be realized. In my thesis I will try to specify the areas where big data analytics could indeed prove to be beneficial. I will answer this research question by performing a literature review. The main finding is that, despite of the potential of big data analytics, the perceived value for financial statement auditors is ambiguous in terms of efficiency and quality. Cost reduction is most certainly not achieved at the moment.
Samenvatting
Momenteel staat de positie van de auditor ter discussie, zoals kan worden afgeleid van onder meer het Groenboek 2010. De druk voor verandering is aanzienlijk hoger dan voorheen, daarom onderzoek ik in deze thesis of big data analytics een waardevol
hulpmiddel kan zijn voor auditors in de financial statement audit. Big data is een hype waar veel over wordt gespeculeerd en tevens zijn er interessante toepassingen waargenomen in verschillende sectoren. Daarom wordt in deze thesis onderzocht, door middel van een literatuurstudie, in hoeverre big data analytics waardevol is voor auditors in termen van efficiëntere, goedkopere en kwalitatief betere audits. Mijn bijdrage aan de bestaande literatuur is tweeledig. Ten eerste vat ik de bestaande literatuur met betrekking tot big data samen, waarbij de toepassing op het accountantsberoep centraal staat. Ten tweede toets ik de zeer positieve houding van auteurs die schreven in het forum Big Data van het journal
Accounting Horizons. De auteurs waren uitermate positief over de mogelijkheden van big
data analytics in de financial statement audit, maar beargumenteerden niet in welke fase(n) van de audit deze potentiële waarde wordt gerealiseerd. Ik probeer dit in deze thesis wel te specificeren. Uit mijn onderzoek blijkt dat het niet eenduidig is dat er efficiëntere en
kwalitatief betere audits worden gerealiseerd met big data analytics. Lagere kosten worden hoogstwaarschijnlijk niet gerealiseerd.
Table of Contents
1. Introduction ... 6
2 Overview of the theory ... 8
2.1
General context of auditing ... 8
2.2
Performing an audit ... 10
2.3
Big Data ... 11
3 How to assess the value of big data analytics for auditors ... 16
3.1
The Iron Triangle ... 16
3.2 Audit Quality ... 17
3.2.1 Audit input ... 17
3.2.2 Audit Process ... 18
3.2.3 Audit output ... 20
4 Examination of the possible value of big data analytics ... 22
4.1
Big Data Analytics and Efficiency ... 22
4.2
Big Data Analytics and Audit quality ... 24
4.3
Potential challenges and hurdles ... 28
5 Discussion ... 30
References ... 33
1.
Introduction
The auditing profession has been under considerable pressure the last fifteen years, as can be concluded from the Green Paper issued in 2010 and from the paper of Power (2003, p. 379). First, the major fraud committed by the management of Enron, which became known in 2001, causing billion-‐dollar damage and the bankruptcy of Arthur & Anderson (Brickley, 2003, p. 1). Arthur & Anderson used to be one of the big 5 auditing firms, but after their failure with regard to the Enron scandal became public knowledge bankruptcy followed soon in 2002 according to Brickley (2003, p. 2).
Second, since the global financial crisis, users of financial statements and regulators of the auditing profession have become more sceptical about the role that the auditor fulfils in the economy (Green Paper, 2010, p. 6). During and after the peak of the recent financial crisis many financial institutions went bankrupt without any warning from the auditors (Green Paper, 2010, p. 9). The auditor has the obligation, since 1989, to evaluate the viability of the auditee (entity being audited) for a reasonable time, which is outlined in the
Statements on Auditing Standards (SAS) No. 59 (AICPA, 2002). The financial crisis made it evident that auditors were failing to fulfil this obligation. Consequently, several measures have been put in place: for example in the United States, the Sarbanes-‐Oxley Act has been implemented and in Europe the Green Paper 2010 has been issued.
Pressure on the audit profession is not only coming from regulatory bodies, increased competition in the audit profession is another important factor. The margins on financial statement audits are extremely tight, because the auditing market is highly concentrated and regulatory bodies have introduced more competition by mandating firm rotation (Knechel, 2007, p. 387; AICPA, 2013). In order to remain competitive and credible, the auditing profession would therefore benefit from new cost-‐reducing and quality-‐ enhancing techniques in financial statement audits.
An important innovation in the business environment the last couple of years is Big Data. A survey performed by Gartner (2015) showed that 75% of the responding companies expected to invest or were already investing in big data and analytical tools that could be used to process big data. In an earlier survey, also performed by Gartner (2012), it was estimated that the total amount invested in big data would reach 232 billion dollar by 2016. An example of the application of big data analytics is described by Zang, Yang and Appelbaum where the researchers successfully predict the change of the Dow Jones
Industrial Average stock exchange, using the mood on Twitter as explanatory variable (2015, p. 425). Another example comes from Walmart, the retail corporation used weather forecast
information to guide their advertising of flashlights (Dezyre, 2013). By successfully using information about storms and tornado’s and anticipating on a higher demand for flashlights, Walmart’s was able to sell more flashlights. More examples can be found in the medical and the insurance industry.
As shown above, there seems to be a variety of possibilities for big data analytics in the business environment. However, by my knowledge no successful application can as of yet be found in the assurance service industry, or more specifically, within the auditing profession. As mentioned earlier, the auditing profession is under pressure and new techniques might bring some reprieve to the profession. Therefore, I will research whether big data analytics could be such a technique, by answering the question whether big data analytics is beneficial for financial statement auditors in a financial statement audit.
The research will take the form of a literature review. My contribution to the literature is twofold. First I examine whether the several studies that claim that that big data analytics will be beneficial for the auditor are correct. Second, I synthesize the existing literature covering the topic of big data with respect to the auditing profession, which can serve as reference for further research.
Based on the literature study, it can be concluded that the value of big data analytics for financial statement auditors is not as obvious as originally thought by authors of the Big Data forum edition in Accounting Horizons. It is ambiguous whether more efficient audits are achieved, and cost reductions are definitely not realized with the current competition for data scientist. Furthermore, increased quality of financial statement audits is ambiguous. Big data analytics has potential to increase quality, but currently the proven positive effects of big data analytics are not in the scope of financial statement audits. Rather, more specialized assurance services such as forensic audits could benefit from these new techniques.
The remainder of this paper is structured as follows. In the second chapter background information is provided on the financial statement audit and how a financial statement ought to be performed. Furthermore, chapter 2 will describe big data and several analytical techniques to analyse big data. In chapter 3 the criteria to assess the value for auditors of big data analytics are presented, which will be the basis for the analysis in chapter 4. In chapter 5, a discussion of the results of chapter 4 is presented, followed by a conclusion.
2
Overview of the theory
2.1
General context of auditing
The business environment has become more complex over the years, especially with the increased amount of data available (Gray, 2002, p. 9). Therefore, the demand for assurance services has increased to reduce the information risk associated with the more complex business world (Arens, Elder & Beasley, 2014, p. 26). Arens et al. define information risk, as the possibility that the information presented is not entirely truthful and could result in wrong decisions by internal and external users of the information (2014, p. 26). The (external) financial statement audit is one of those demanded assurance services and in general when referred to auditing, this type of assurance service is meant (Arens et al., 2014, p. 29). The ultimate purpose of the audit is to improve the level of confidence placed in the financial statements by the users of the financial statements (IAASB, 2012).
Arens et al. present the following definition of auditing: “Auditing is the
accumulation and evaluation of evidence about information to determine and report on the degree of correspondence between information and established criteria. Auditing should be done by a competent, independent person” (2014, p. 24). Specifically, the financial
statement audit is performed to verify that the statements are in agreement with criteria such as general accepted accounting principles (GAAP) (Arens et al., 2014, p. 34).
The above-‐presented definition of auditing will be used to explain which role auditors fulfil in the business environment and how they fulfil it. According to Arens et al. the auditor must obtain reasonable assurance about whether the financial statements are free from material misstatements, and thus present a fair view of the underlying economics of the entity (2014, p. 164). However, the auditor’s assurance concerns the historical financial statements, which is termed in literature the ‘rear-‐view window check’ (AIPCA, 2015, p. 53). To increase the relevance of the financial statement audit, auditors are required to make an assessment whether the auditee is likely to continue as an entity for a certain period of time (AICPA, 1989, p. 2048).
The assurance, however, is given to the shareholders. So even though the auditee orders and pays for the audit, it is actually executed for the shareholders of the auditee, which is a rather unusual construction (Teeter, 2014, p. 2).
To enable the auditor to express an opinion about the financial statements, s/he has to evaluate the auditee following a structured plan that can be referred to as the audit
approach (Arens et al., 2014, p. 441). There are four general phases in the audit identified by Arens et al., but each audit firm has the liberty to develop their own specific methodology that ultimately could lead to a competitive advantage (Jeppesen, 1998, p. 520). In the next section these four phases will be discussed in depth.
Several important terms from the definition of auditing will be discussed in the remainder of this section. Evidence as defined by Arens et al. is any form of information used by the auditor to test assertions made by the management of the auditee (2014, p. 24). Two aspects are important when discussing evidence within auditing, which are
appropriateness and sufficiency. Appropriateness consists of the relevance and reliability of the evidence collected (Arens et al., 2014, p. 196). A more thorough explanation can be found in the third section of this thesis. The auditor has several techniques to collect evidence, for example physical examination (for inventory items) and analytical procedures such as financial ratios for risk assessments (Arens et al., 2014, p. 199).
Sufficiency is about the question how much evidence the auditor should gather. The method used by auditors to determine the amount of evidence that should be aggregated, is the audit risk model (AICPA, 1983). The following equation adopted from Arens et al. is the basic form of this method; Planned Detection Risk (PDR) =!"!!"!#$ !"#$ !" × !"#$%"& !"#$(!")!""#$ !"#$ !"#$% !"#$ (!!")
The outcome PDR, which indicates the risk that audit evidence fails at detecting misstatements, is inversely related to the amount of evidence the auditors have to gather (Arens et al., 2014, p. 279). Hence, a lower PDR requires more evidence. The next
component, AAR, reflects the risk the auditor (in general the managing partner) is willing to take that the financial statements contain material misstatements after the audit is
completed (Arens et al., 2014, p. 280).
IR refers to the chance the auditor imputes to the possibility of material
misstatement before taking the internal controls into account (Arens et al., 2014, p. 279). CR refers to the chance that the internal control system of the auditee is unable to detect material misstatements. The model as presented above is described in the auditing standards, which characterizes the auditing profession. These standards contain outlines that dictate, for la large part, how the audit should be performed. Compared to other professions, auditing is highly regulated.
Another important term is ‘reasonable assurance’. Auditors do not guarantee that financial statements are free from material misstatements, since it would not be
assurance is said to be at least 95% sure that the financial statements do not contain material misstatements.
Related to reasonable assurance is the term material misstatement. Auditors have the responsibility to detect material misstatements and not every misstatement. Materiality is highly subjective and is defined in the following manner: something is considered material when omission or misstatement of the information is likely to change the decision of a reasonable person (Chewning, Pany & Wheeler, 1989, pp. 80-‐81). Materiality can vary per auditee, obviously the monetary material level of an organisation such as Apple Inc. is different from the local fruit retailer.
2.2
Performing an audit
In this section the different phases of the audit are briefly discussed. As emphasized earlier, while the precise methodology followed by an audit firm can differ from what is outlined below, the content will generally be similar.
The first phase is the planning phase. In the planning phase the auditors examine whether to accept the client by analysing the industry of the auditee and evaluate the reasons for the audit (Arens et al., 2014, p. 231). Furthermore, the auditor achieves a sufficient understanding of the business and the industry of the client in order to make a proper business risk assessment, which will determine the AAR and the risk of material misstatements (Arens et al., 2014, p. 239). Using the information obtained in the planning phase, a materiality level is determined. Often a percentage of the net income is used as value to classify irregularities as either material or immaterial (AICPA, IAS 320). At the end of the first phase, based on the analysis of the client and its industry, an overall audit approach is designed.
In the second phase, auditors carry out test of controls and substantive tests on transactions (Arens et al, 2014, p. 442). By testing the specific internal controls of the auditee, the auditors can determine the control risk, which is the CR in the audit risk model. In the case of weak internal control more evidence has to be gathered to verify the
monetary amounts of transactions and balance sheet items in the subsequent phase (Arens et al., 2014, p. 442).
The third phase of the audit consist of two main activities, which are analytical procedures and tests of details of balances (Arens et al., 2014, p. 184). The analytical procedures are used to find patterns and plausible relationships between different balance
sheet items. For example, a ratio of accounts receivable to sales is assumed to remain stable, when deviations are found large enough the auditor should proceed with a test of detail of balances. Those tests of detail consist of contacting customers of the auditee to confirm certain accounts receivable amounts. Conversely to the second phase, evidence is mostly retrieved from third parties (Arens et al., 2014, p. 184).
After the auditors have completed all procedures and acquired all the evidence to meet the objectives of the audit, an overall verdict is reached. The auditors draw an overall conclusion in the final phase of the audit, whether or not the financial statements are free from material a misstatement, which is referred to as the auditor’s opinion (Arens et al., 2014, p. 70). For simplicity’s sake one of two opinions can be expressed: either a clean opinion or a modified opinion. When no material misstatements are detected the auditors will express a clean opinion (Arens et al., 2014, p. 68). When material misstatements are detected, the auditor will modify his/her opinion. While there are several different types of modified opinions, for this thesis the broad distinction above will satisfy.
2.3
Big Data
Big data and the analytics performed on them have been quite the hype in numerous industries for the past few years (Deloitte, 2013, p. 2). But, as with any hype, its true value is not as evident as people might think. In the introduction, two applications of big data analytics were mentioned. To make an assessment of the possibilities of big data analytics in the auditing profession, it is paramount to define big data as well as analytics in a fashion that suits the auditing profession.
Different professionals in different industries use different definitions of big data (Alles & Gray, 2015, p. 8). The Mckinsey Global Institute employs the following definition: as soon as data cannot be captured, analysed and stored by the traditional information
systems, it should be labelled as big data (2011, p. 1). Using this definition any firm is capable of generating big data if, for instance, when trends on Facebook are used as input for decision-‐making. This type of information falls outside the scope of traditional
information systems according to Yoon et al. (2015, p. 431). The Mckinsey Global Institute deliberately established a subjective definition, so that every industry has the liberty to come up with a specific definition that is most suitable for their particular industry (2011, p. 1). Such a vague definition does make it questionable whether big data is fully understood by anyone. Nevertheless, by synthesizing what is currently known of big data, I try to
establish an accurate description of big data and provide examples of data analytics relevant for financial statement audits that can be performed with big data.
In the existing literature, definitions of big data can be divided into two broad categories. The first category of definitions focuses on specific examples of big data (Alles & Gray, 2015, p. 8). This definition, however, requires specific examples of big data that can be used in auditing. Due to the lack of research, specific examples are not available, which makes this definition unusable. In this thesis we will therefore rely on the second category of big data definitions. This category, according to Alles and Gray, is based on specific
characteristics of big data (2015, p. 8). Those characteristics are commonly known as the 4 V’s. It must be noted that, since big data is a current issue it is likely more definitions and characterizations will follow. For example, at the Big Data Summit in Boston two additional V’s were presented (Normandeau, 2013).
Interestingly, the 4 V’s definition is derived from a blog (META group), now a part of Gartner, that came up with the taxonomy in 2001, which was before the big data hype actually started (ACCA, 2013, p. 11). META group defined the first 3 V’s, which are: volume, velocity and variety, as cited by Alles and Gray (2015, p. 8). The fourth V, veracity, was later added to these 3 V’s. Big data distinguishes itself from ordinary data due to the 4 V’s (McAfee & Brynjolfsson, 2012, p. 62). The first 3 V’s will be explained in this section, the fourth V will be explained in the analysis section.
The first V is volume, which refers to the size of the data, as shown in the figure below. Moffit and Vasarhelyi argue that traditionally information was generated by the information system of the auditee, but an increasing amount of information is generated by other sources (2013). The lower left square of the figure below, which represents
transaction data, is currently the most important information for the auditor (Alles & Gray, p. 10). However the auditee’s information system is not the only data-‐generating system. Connely identifies the following two additional sources of information: human-‐sourced information, such as the social medium Facebook, and machine-‐generated information, which is information from data sensors and mobile tracking sensors (2012). These sources are an alternative classification of ‘interaction’ and ‘observation’ used in the figure below (Alles & Gray, 2015, p. 9). Volume also refers to the growth rate of information. According to Deloitte, the amount of world data increased from 2,5 zettabytes (21 zero’s) to 8 zettabytes in a five year time span (2013, p. 6). The auditors, when searching for information to test assertions of the management, might want consider other forms than transactional data.
There will be a more elaborate discussion of the results of this characteristic as well as for the other characteristics in the analysis section.
The second V, velocity, refers to the rapid pace at which data changes, which means that information is continuously updated (Alles & Gray, 2015, p. 9). The third V, variety, is related to the different forms of information that are included in big data. These forms range from structured internal information, such as transaction history, to unstructured external information such as social media information (Deloitte, 2013). This unstructured type of information could be useful for financial statement auditors as described below. The wide variety of information is a logical consequence of the different information generators that were identified earlier.
Currently, auditors depend on structured financial information (GAAP-‐compliant information) as evidence to support the opinion about the financial statements (Cao, Chychyla & Stewart, 2015, p. 427). Therefore, the ‘new’ information big data adds to the information currently used by the auditor is unstructured non-‐traditional information (Moffit & Vasarhelyi, 2013, p. 2). However, without techniques to analyse the new data the value to auditors derived from big data will be equal to zero. As stated by Alles and Gray, value from (big) data is determined by the analytics performed with them (2015, p. 13). Therefore, several data analytical tools are considered below that might be useful to auditors to analyse big data and hence indicate the relevance of the 3 V’s as explained above.
Data analytics (also termed business intelligence/artificial intelligence) have been divided into three levels by Chen, Chiang and Storey in an often-‐cited article. The first level consists of simple regression techniques on structured databases such as ERP systems of enterprises (Chen et al., 2012, p. 1166). The second level has been largely developed under the influence of the Internet, according to Chen et al. (2012, p. 1167). The authors argue that with the Internet new kind of information came available, which required new techniques and tools to analyze (2012, p. 1167). The third level is still in its developmental stage, which incorporates the different information made available by smartphones and other devices equipped with GPS and other applications (Chen et al., 2012, p. 1167).
Data analytics is the practices of selecting and cleaning data, modelling, and finding patterns in datasets using data mining tools, which can be used to gain certain insides and aid the auditor in, for example, risk assessments (Sharma & Panigrahi, 2012, p. 38). Below several data mining techniques discussed in auditing literature are presented, note that the list is by no means not exhaustive.
The first tool is neural network (NN); in contrast to standard logistic models, NN uses non-‐linear models to analyse datasets (Sharma & Panigrahi, 2012, p. 40). By incorporating complex algorithms multiple pieces of information can be evaluated at the same time (Calderon & Cheh, 2002, p. 205). In terms of big data, neural network might be able to link financial and non-‐financial data to find certain patterns or discrepancies (Chen et al., 2012, p. 1170). A simple example of a discrepancy is higher reported sales, while the amount of stores decreases. Assuming that Internet sales remain the same, it could indicate suspicious accounting (Yoon et al., 2015, p. 435). Further, when considering social media, decreasing popularity, indicated by likes and re-‐tweets, could be an indicator of going concern issues.
In contrast to neural networks the second tool, text mining, is a technique analysing ‘soft’ data rather than financial ‘hard’ data. Different approaches exist to analyse plain text: searching for specific words, searching for specific word combinations or identifying any other abnormality in plain text, which is termed text analysis (West & Bhattachrya, 2016, p. 55). This tool might be valuable when considering using social media, emails, management letters etc. as information source to the auditors, since it consist largely of textual data.
The next tool discussed is process mining, which refers to analysing transactions and event logs (West & Bhattachrya, 2016, p. 55). When a certain transaction has to be
completed, firms normally have certain protocols that should be followed (Jans, Alles & Vasarhelyi, 2013). Most mid-‐sized and large firms have Enterprise Resource Planning (ERP) systems that automatically record the steps taken to complete the transaction. By analysing this data, auditors could verify whether the actions taken are indeed the actions that should have been taken (West & Bhattachrya, 2016, p. 55).
Another tool is Benford’s law, which is an example of how suspicious accounts are identified. The theory is about the probability that certain numbers appear in a certain order, for example the ‘9’ appears only in 5% of the cases as the first number (Durtschi, Hillison & Pacini, 2004, p. 19). Benford’s Law, however, has been established in 1938, but has not been widely accepted as a proven theory, which made it until now a controversial technique. According to Durtschi et al., Benford’s Law is merely an addition to existing analytical techniques used by auditors today, without consensus that it actually aids the auditor in mapping suspicious accounts (2004, p. 21). But its relative ease makes it appealing to use, one can simply choose an account on the balance sheet/ income statement, which should be analysed and let the ‘app’ do the work (Cleary & Thibodeau, 2004, p. 6).
All of the above indicates that big data is not easy to define, with all the
the process of identifying big data accurately by using the Hindu analogy of the giant elephant. The analogy is about blind men trying to size up a giant elephant, but all of them have only a limited area they can explore due to natural limitations (2014, p. 98). Restricted by a limited perspective, each blind man will come to a different conclusion of what they think they have in front of them (a wall or a tree are examples of the conclusions drawn). Big data for now can be seen as the giant (growing) elephant, which we are trying to define. Furthermore, Wu et al. acknowledge that currently no tools exist to fully analyse big data, the aforementioned techniques only have the potential to analyse elements of big data (2014, p. 102). The majority of the existing literature is therefore based on expected future progress in analytics.
3
How to assess the value of big data analytics for auditors
In its most basic form, big data analytics can be seen as a tool for the auditor when conducting the audit. An audit tool is any technique, manual or computerised, used in the audit (Curtis & Payne, 2008, p. 105). This section will describe the considerations for the auditors when they adopt a new audit tool. The umbrella criterion wills that big data analytics should provide benefits to the auditors in some form. Therefore, this section tries to define what ‘beneficial’ is for the auditors.
3.1
The Iron Triangle
In the article of Vasarhelyi and Romero, the Iron Triangle is used to evaluate whether new audit technology is beneficial to auditors, and hence should be adopted (2014). The Iron Triangle will be used as a starting point in this thesis. It consists of three components: efficiency/effectiveness, cost reduction, and quality (Vasarhelyi & Romero, 2014). As
demonstrated below, efficiency/effectiveness and cost reduction are rather straightforward; quality, however, is a controversial topic when put in the auditing context (Fischer, 1996, p. 220) and will be discussed in more detail.
Efficiency and effectiveness are often used as complements of one another, which comes down to the following definition: the degree to which established goals are realized, and the amount of resources used to do so (IPPF, 2010, p. 2). This is a fairly general
definition and needs further specification in order to be useful for evaluating external financial audit tools. Efficiency is defined in terms of the resources that are used (Rosenfeld). One feature of auditing is the labour intensity of the job, with other words the resources used. Therefore, decreasing the labour hours needed to achieve the same level of assurance is a good way of defining efficiency without impairing effectiveness, which is maintaining a certain assurance level. The IAASB further differentiates resources in qualitative and quantitative resources (2012). For example, hours worked by a managing partner are different in terms of quality than hours worked by a staff assistant.
Cost reduction is to some extent the logical consequence of fewer resources that are used. However, there are more considerations with respect to cost reduction, which can be derived from the diffusion of innovation theory (DOI). For instance, does the new tool supersede other tools, hence can it replace current tools used (Rosli, Yeow & Eu-‐Gene, 2013, p. 5). Moreover, education is required to enable auditors to use specific tools (Romero & Vasarhelyi, 2014), which depends partly on the complexity of the audit tool (Rosli et al.,
2013, p. 5). The cost reduction should be seen in the long term, but the future is often uncertain. Therefore, the pay-‐off, less resources used, and the cost, for instance of education, can be hard to estimate (AICPA, 2015, p. 72).
3.2 Audit Quality
The last component of the Iron Triangle is quality. As opposed to other services, financial statement audits are not transparent. This means that assessing the quality of the audit is difficult when the audit report is the only outcome to go on (IAASB, 2012). The first
definition of audit quality is from DeAngelo, which underlies many of the other definitions of audit quality established after DeAngelo (Al-‐Khaddash, Al Nawas & Ramadan, 2013, p. 207). DeAngelo argues that quality is the joint probability that an auditor will both discover and report a breach in the client’s accounting system, as cited by Al-‐Khaddash et al. (2013, p. 207).
Furthermore, each stakeholder of financial reporting (auditor, investors, regulators etc.) will determine the quality of the audit on different criteria (Knechel, Krishman, Pevzner, Shefchik, Velury, 2013, p. 386). Auditors value the perceived quality of their work by various stakeholders as well as actual quality, since both will determine the relevance of the
auditor’s work (Al-‐Khaddash et al., 2013, p. 211). Therefore, ‘quality’ should be assessed from multiple perspectives and not only from the auditor’s perspective.
In the remainder of this section a framework will be presented that tries to capture a balanced view on different measures of audit quality. The framework will distinguish input of the audit, the audit process, and the output of the audit when considering audit quality. Overall, quality is defined by the PCAOB as meeting customer demand (2013).
3.2.1 Audit input
First, the input of the audit. Input refers to what audit firms employ to perform the audit and achieve the desired result (PCAOB, 2013). From the different inputs for the audit, the IAASB identifies ‘people’ as most influential on audit quality (2012). Ultimately, the skills and the personal qualities of audit partners and staff determine the quality of the work
performed (FRC, 2008) as cited by Knechel et al. (2013, p. 388). So what qualities are perceived as ‘good’ in the auditing literature?
According to Knechel et al., the financial statement audit consists of many judgement calls that have to be made in the audit process, which in turn determine the quality of the audit (2013, p. 390). To enable the auditor to make proper decisions, several personal qualities should be present. One of the most important qualities is professional scepticism (PCAOB, 2013). Specific examples of qualities of a professional sceptical auditor are a questioning mind set and the ability to critically evaluate the obtained evidence (ICAEW, 2013).
Moreover, knowledge, which determines expertise to a large extent (Ashton, 1991, p. 220), of the industry and the auditee are identified by Knechel et al. as important
contributors to higher quality decision making and hence, higher quality audits (2013, p. 392). Knechel et al. argue that industry-‐specific knowledge could enhance judgement calls that have to be made by the auditors (2013, p. 392). Hence, being able to analyse industries more thoroughly has the potential to enhance the quality of the audit.
Furthermore, personnel should be independent and competent to perform high quality audits (PCAOB, 2013). The independence of the auditor is determined by the objectivity of the auditor (ICAEW, 2003). One of the threats to the objectivity of auditors is when they provide other non-‐audit services to a client (Reynolds, Deis, & Francis, 2004, p. 31). Other qualities found in independent auditors are integrity and impartiality of the auditor (Arens et al., 2014, p. 56). Moreover, personnel should possess the technical capabilities to execute audit procedures (Khaddash et al., 2013, p. 210).
Another important input factor according to the PCAOB is tone at the top (2013). Specifically for audit partners and firm managers, a positive relation has been found between tone at the top and audit quality (PCAOB, 2013). When the top strives for
innovative and high quality audits, it is more likely that staff will do the same (PCAOB, 2013).
3.2.2 Audit Process
The next component of the framework is the ‘process’ of the audit, which refers to the four phases described in section 2.2. Several important judgements in the audit process
according to auditing literature are: risk assessment, obtaining and evaluating evidence and review of the work, as cited by Knechel et al. (2013, pp. 393-‐397). Judgements made by the auditor are structured, semi-‐structured or unstructured, with structured judgements requiring almost no judgement and unstructured decisions requiring a high level of judgement (Arens et al., 2014, p. 190).
Knechel et al. identify two potential hazards that impair the auditor’s judgement. The two hazards are anchoring and adjustment, and representativeness (Knechel et al., 2013, p. 396). Anchoring and adjustments happen in the ordinary course of the audit. The expectation is that adjustments to the anchor value are made in the ‘correct’ direction (Kinney & Uecker, 1982, p. 56). For example, the auditor has an idea about certain book values, the anchor, and during the audit the auditor finds evidence supporting or contradicting this expectation, which underlies the adjustment. However, according to Kinney and Uecker, it does happen that the ‘anchor’ is not sufficiently adjusted because of sample outcomes (1982, p. 57).
Tversky and Kahneman originally established the definition of representativeness in 1974, as cited by Aston (1984, p. 80). The theory behind representativeness is that auditors attach a higher probability to uncertain events that are more in line with expectation (Ashton, 1984, p. 81). The expectation is based on certain resemblance between the
uncertain event and the population, i.e. the representativeness of the uncertain event of the population (Ashton, 1984, p. 81). For example if we have item A and we want to assess the probability that it comes from a population A or B, looking at the resemblances with the population A or B is a logical step to take (Schroeder, Reinstein & Schwartz, 1996, p. 18). Furthermore, Ashton argues that several factors will influence the likelihood that the heuristic representativeness occurs (1984, p. 82). The first factor is the correspondence between the sample and the parent population. When the auditor draws a sample in which essential properties are more similar with the population, representative of the population, the auditor will deem this scenario more likely (Ashton, 1984, p. 82). The opposite is true as well. Therefore, the sample drawn is critical for the judgement of the auditor. In addition, sample size has an influence on the auditor’s judgement, since in smaller samples extreme values are more likely (Ashton, 1984, p. 81). However, larger samples, termed
protectiveness, do not guarantee that the above heuristics are prevented (Schroeder et al., 1996, p. 19).
Another important process in the audit is obtaining and evaluating obtained audit evidence (Mcknight & Wright, 2011, p. 194). As synthesized by Smith and Kida, audit evidence has an influence in multiple ways on the auditor’s judgement (1991). Francis argues that the financial statement audit is only as good as the evidence obtained (2011, p. 135). Arens et al. describe two measures of evidence quality, which are sufficiency and appropriateness of audit evidence (2014, p. 196). Sufficiency has been explained in section
2.1, whereas appropriateness of evidence deserves further elaboration. According to Arens et al., appropriateness consists of relevance and reliability of the evidence (2014, p. 196). Relevance of the evidence refers to the intended use of the evidence in the audit. For example, when the auditor is testing whether all sales are invoiced, tracking back from the invoiced sales to shipping records is irrelevant. Relevant evidence would be to consider a sample of sales shipped and track them back to invoiced sales to see or all are indeed invoiced. The auditor should therefore obtain evidence that can be used to meet certain audit objectives (Arens et al., 2014, p. 196).
Reliable evidence has several characteristics as described by Arens et al. (2014, p. 197). First of all, evidence directly obtained by the auditor is considered more reliable than information obtained indirectly (Arens et al., 2014, p. 197). Second, when information is obtained indirectly, the source should be qualified to do so. If so, evidence obtained from outside the firm is regarded as higher quality evidence (Arens et al., 2014, p. 197). Third, evidence that needs little judgement is regarded more reliable, hence higher quality of evidence (Arens et al., 2014, p. 197). Finally, timeliness of audit evidence is an indicator of quality. Arens et al. argue that evidence obtained for balance sheet accounts close to the balance sheet date is more reliable than evidence obtained a considerable time before the balance sheet date (2014, p. 197). Timeliness for the income statement is slightly different, Arens et al. emphasize that evidence should be obtained from the whole year, rather than only at the end (2014, p. 197).
The last component of the process part of the quality framework is review and control of the work that has been done, since it is positively related to audit quality according to Knechel et al. (2013, p. 396). To achieve a high quality review, the PCAOB argues that the technical competence of the reviewer should be of a considerable level (2013). Furthermore, reviewers should have enough time to properly review the work that has been done, which makes a proper planning of the audit essential (PCAOB, 2013).
3.2.3 Audit output
Output is the last component of this quality framework, which will be considered most important with respect to audit quality by users of financial statements and regulators of the audit profession (IAASB, 2012). Most indicators are only quantitative and difficult to
transform to criteria. For example, the amount of false positives and false negatives (errors that arise with going concern opinions) are empirically testable, but hard to define in
qualitative terms. In the end, it depends on how effectively the audit process is executed, which is determined by the people performing the audit. However, the IAASB found an important qualitative aspect of ‘output’, which is transparency of the audit performed (2012). They argue that when the audit is more transparent, it could be considered of higher quality by different stakeholders (2012). Further, the time-‐lag between the year-‐end and the issuance of the audit report decreases the relevance for decision making, which decreases the perceived usefulness of the financial statement audit (Chan & Vasarhelyi, 2011, p. 152). Therefore, decreasing the time-‐lag would increase the quality, since customer demands are better satisfied.