Faculty of economics and business administration a future in accounting without human intervention

Background: In order for automation to replace the accountant, technology needs to provide useful financial information; it needs to be relevant, represented faithfully, comparable, ver

Master’s Dissertation submitted to obtain the degree of:

Master of Science in Business Economics - Accountancy

Title page

Title: A future in accounting without human intervention

Author: Mélanie Simon

E-mail address: Melanie.Simon@UGent.be

Student number: 01615385

University: University of Ghent

Faculty: Faculty of Economics and Business Administration

Study: Master of Science in Business Economics – Accountancy

Study year: 2017 – 2018

Promotor: Prof Dr Patricia Everaert

Email address: Patricia.Everaert@UGent.be

Document status: Final version

Abstract

Objectives:

The aim of this study is to examine the impact of automation on the accounting profession, in order to answer the question if a future in accounting without human intervention is possible

Background:

In order for automation to replace the accountant, technology needs to provide useful financial information; it needs to be relevant, represented faithfully, comparable, verifiable, timely and understandable

Systematic literature study:

32 articles were selected, main subjects identified were: consequences on accounting, moral decision-making, future role, implications on small accounting firms, implications

on the labour market and solutions

Methods:

Semi-structured interviews were conducted with accountants from eight different companies in Belgium and Luxembourg Professionals have been interviewed regarding their use of technology and their future perspective on the accounting profession

Results:

Eight interviews have been conducted, main subjects identified were: the use of automation, qualitative characteristics, skills and implication on small accounting firms Results show that the accountant will be using automation for routine tasks, rather than being replaced by it Tasks that require critical-thinking and creativity seem to be more difficult to be automated In the coming years, the technology will be able to assist accountants in non-repetitive tasks The business model of accounting firms will change and accountants who are not ready for automation will be at risk of being replaced by automation Specific skills will already need to be acquired before starting to work

Relevance for practice:

Accountants will shift to either advisory or consultancy IT-, tax- and analytical skills will have to be developed Universities will need to change their education programs in order for future accountants to be ready to work alongside automation

4.6 Consequences on the accounting profession 25

4.9 Implications on small accounting firms 32

6 Results 41

6.3 Qualitative characteristics of the financial information 47

Acknowledgments

The completion of this thesis has been very fascinating and instructive, but also hard work These seven months of work have been very challenging for me and I would like to thank the people who contributed directly or indirectly to the completion of the work presented in this thesis

This accomplishment would not have been possible without them

I would first like to thank my promoter, Prof Dr Patricia Everaert The door to her office was always open whenever I had questions about my research I thank her for advising and guiding me to the right direction during these last couples of months

I would like to express my gratitude to all the respondents of the interviews for taking time to participate in this study

My sincere gratitude also goes out to Inge van der Veen and to Casper van den Berg for carefully and critically reviewing my thesis

I would especially like to express my gratitude to my partner for providing me moral and intellectual support throughout the process of researching and writing this dissertation Without him I would not have come this far

List of abbreviations

AI Artificial Intelligence

BDO Binder Dijker Otte

CEO Chief Executive Officer

CFO Chief Financial Officer

CO Controlling

FI Financial Accounting

GDPR General Data Protection Regulation

IASB International Accounting Standards Board

IBM International Business Machines

IFRS International Financial Reporting Standards

IT Information Technology

KPI Key Performance Indicator

KPMG Klynveld Peat Marwick Goerdeler

LPL Lo Presti Ludovic

PS Project System

At Google's annual Input/Output developer conference, Google affirmed its desire to integrate Artificial Intelligence into people’s daily life through a smart personal assistant (Staff, 2017) Other technology giants as Apple, Facebook, Microsoft and Amazon are also interested in AI and invest heavily in this technology The use of Artificial Intelligence and automation can reduce the need for human labour This leads to uncertainty concerning certain professions, such as accounting (Manjoo, 2017)

According to a study conducted by Frey & Osborne (2017), 702 job titles are at risk of automation Among these professions, accounting is on top of the list with 94 percent probability of being computerized in the next two decades (Nagarajah, 2016) Artificial Intelligence can be integrated into accounting processes and thereby replace humans In fact, an artificial agent called Amelia has already started at Shell and Baker Hughes (two

of the biggest gas groups) to take over the duties of accountants and call centre agents The system has the ability to understand natural language that allows to interact with humans It does not only recognize words, it also understands the meaning of them These are tangible signs that the employment of white-collar workers could be threatened by the rise of Artificial Intelligence (Twentyman, 2017)

Recently, International Business Machines (IBM)'s AI has demonstrated its exceptional ability to replace humans in performing tasks previously reserved for human intelligence The software can answer any question asked by a human in natural language, orally or

in writing, in eight different languages (Tual, 2017)

1.2 Problem Statement

According to an analysis provided by Accenture, 40 percent of transactional accounting work could be automated by 2020 (Seek, 2017) As an authority on the profession, the Association of Chartered Certified Accountants is sceptical about the future of the accountants The skills that accountants nowadays apply may not be relevant anymore

in the next coming years (Galarza, 2017)

The aim of this study is to research the impact of automation in the accounting field In

order for automation to be able to replace accountants, useful financial information still

has to be provided Therefor, the attributes that make financial information useful also

have been researched

1.3 Research question

The following research question is formulated: is a future in accounting without human

intervention possible?

In order to answer this research question, a systematic literature review and interviews

have been conducted First, all the literature regarding the impact of technology on the

accounting profession has been gathered Later, several interviews have been

conducted with eight different Belgian and Luxembourgish companies

1.4 Structure of the thesis

This thesis consists of eleven chapters The first chapter is the introduction in which the

context, the problem statement and the research question is described The second

chapter consists of background information regarding Artificial Intelligence and

automation The third chapter consists of definitions of the accountant, the auditor and

the management accountant The qualitative characteristics will also be addressed in this

chapter The systematic literature review can be found in chapter four including the

search strategy and the results The methodology of the empirical study can be found in

chapter five, where the study design, the data collection and the reliability and validity

The sixth chapter consists of the results from the semi-structured interviews The

discussion can be found in chapter seven, in chapter eight the conclusion and in chapter

nine the limitations Suggestions for future research are described in chapter ten Finally

in chapter eleven, the management and policy implications of this study are given

2 Artificial Intelligence and automation

In order to answer the question if a future in the accounting profession without human intervention is possible, in this chapter background information about the subject will be provided Four things are mainly important: 1) what is Artificial Intelligence and automation 2) definitions of the accountant, the auditor and the management accountant 3) a brief history of the accounting profession and 4) objectives of the financial reporting and when automation is useful in light of the accounting profession

Artificial Intelligence is the theory and development of computer systems that are able to

perform tasks that normally require human intelligence, such as visual perception, speech recognition, decision-making and translation between languages (Oxford Dictionary, n.d.a) The system can perform functions that a human brain does, like learning and problem solving

The term Artificial Intelligence was first introduced in 1956 by John McCarthy (Smith, McGuire, Huang et al., 2006), but in 1950 Alan Turing already wrote a paper about the ability of machines to do intelligent things The purpose of Turing’s paper was to consider the question “if machines can think”? He replaced this question by testing if a machine could replace a human being in the game of imitation The purpose of the test was to ask

a person to make the distinction between answers given by a machine and those provided by a human, by communicating via an old teleprinter Turing predicted that in the year 2000, the machine would be able to fool 30% of the respondents in a five-minute test

Familiar definitions are automation, big data, machine learning and Expert Systems

Automation is “the technique, method or system of operating or controlling a process by

automating devices, reducing human intervention to a minimum Automation has a single purpose: to let machines perform repetitive, monotonous tasks” (Dictionary, n.d.)

“Automation is a technology that actively selects data, transforms information, makes decisions and controls processes” (Lee & See, 2014)

Big Data is “a set of extremely large data that may be analysed computationally to reveal

patterns, trends and associations, especially relating to human behaviour and interactions The more data the machine collects, the more it will be able to learn and the better it will function” (Oxford Dictionary, n.d.b) “Big Data has the ability to scan large volumes of data and perform analytics with sophisticated algorithms to facilitate decision-making in the accounting function” (Brands & Smith, 2016)

Machine learning is “the ability of a computer to learn from experience, for instance to

modify its processing on the basis of newly acquired information It is a type of AI that focuses on the development of computer programs that can access data and use it to learn for themselves” (Oxford Dictionary, n.d.c) According to Tynan (2017), “machine learning systems essentially code themselves, developing their own instructions by generalizing from examples The classic example is image recognition, where the machine learning will identify what is on the image without a human ever telling the machine what is on the picture.”

Expert system is “a computer system that can provide information and expert advice on a

particular subject The program asks users a series of questions about their problem and gives them advice based on its store of knowledge” (Oxford Learner’s Dictionary, n.d.) Quinn (1990) defined Expert System, as “an interactive computer program that asks the same questions a human expert would ask, and from the information given to it by the user, provides the same answer the expert would provide If a body of knowledge can be codified into a set of questions and answers, it can be incorporated into an Expert System software program.”

3 The Accountant, Auditor and Management Accountant

The field of accounting involves the study of accountancy, auditing, finance, financial management and tax (Wohlner, n.d.) This thesis will only focus on the accountant, the auditor and the management accountant In this chapter, these definitions are given, as well as a brief history of the accountant Finally, the financial reporting will be addressed

3.1 Definitions

The accountant

An accountant is a qualified person who is trained in bookkeeping and in preparation, auditing and analysis of accounts Accountants prepare annual reports and financial statements for planning and decision-making and advise on tax laws and investment opportunities (Business Dictionary, n.d.) Accounting is the process of measuring and summarizing business activities, interpreting financial information and communicating the results to management and other decision makers (Minnesota Libraries, n.d.)

The auditor

When the accounting process ends, auditing begins

Auditing is the purpose of determining the true and fair representation of the financial statements The auditor examines the financial report of an organization

There are four main steps in the audit process (PriceWaterhouseCoopers, n.d.) The first step is to determine the auditor’s role and the terms of engagement, which is a letter signed by the client The second step is to plan the audit, which includes details of deadlines and the departments the auditor covers Once the auditor is aware of the company's sector and its internal control and has identified the risks, it is necessary to analyse the accounts more precisely in order to identify the risks of any fraud or errors This is the third step The last and most important element of an audit is reporting the results The results are documented in the report of the auditor, including the justification

of the auditor’s opinion This opinion is the conclusion of all the work carried out during the audit

The management accountant

The management accountant is a professional who assists managers by helping them making decisions The management account hence prepares and analyses the financial statements of a company and analyse its financial performance in order to advice the managers in the decisions-making process (Institute of Management Accountants, 2008) The management accountant provides the necessary information and advices to the decision-makers To forecast the future and to monitor the performance of the company, managers will need information provided by the management accountants (Certified Practising Accountant Australia, 2012)

3.2 A brief history of the accountant

The accounting profession has recently been recognized, but the art of accounting is nearly 6,000 years old (Mason, 1953) The Romans and Egyptians were the first to use accounting in commercial life In the Roman Empire, one of the purposes of accounting was to present the economic situation of the merchants to their customers The purpose was to record on the left side of the notebook the use or consumption of resources, while the right part was used for the origin or production of resources (Alexander, 2002)

Although accounting has indeed been used since the Stone Age, the most significant development with regards to modern accounting occurred during the Renaissance period

in Italy Brother Luca Bartolomes Pacioli, an Italian born in 1445, is the inventor of modern accounting Pacioli was the first person to publish a work on double-entry bookkeeping system (Mason, 1953)

The main principles of accounting that are currently used in companies were found at that time Modern accounting techniques are based on double entry bookkeeping, defined as “debit” and “credit” Three books were necessary to keep adequate records for every business: a ledger, a journal and a memorandum book A trial balance has to

be made at the end of each year (Mason, 1953)

Nowadays there are standards set by the International Accounting Standards Boards (IASB) to guide and harmonize the accounting practices The IASB develops and approves International Financial Reporting Standards (IFRS), a set of international accounting standards to specify how to report the accounts, so that everybody can understand the business and the reporting from companies situated in different countries The Board regularly updates the Conceptual Framework for Financial

Reporting to facilitate the use of IFRS Standards The Conceptual Framework includes qualitative characteristics of the financial information to help accountants decide what information to provide and how to present it

In the next subchapter, the objectives of the financial reporting and the qualitative characteristics will be described In accounting, automation is useful when it is able to perform the same tasks as accountants do In other words: AI needs to provide useful financial information Useful financial information, as described by the Conceptual Framework (Ernst & Young, 2010), will be explained

3.3 Financial reporting

The conceptual framework for financial reporting sets out the concepts for the preparation of the financial statements for external users The conceptual framework is structured according to the following hierarchy:

- The objectives of the financial reporting are stated;

- The qualitative characteristics of the information contained in the financial statements;

- The definition, recognition and measurement of the elements from which financial statements are constructed (assets, liabilities, equity, income and expenses) The following chapter explains the objectives of the financial reporting and the qualitative characteristics of useful financial information (Deloitte, n.d.)

3.3.1 Objectives of the financial reporting

The purpose of financial reporting is to provide useful financial information about an entity to potential investors, lenders and other creditors who use that information to make decisions about buying, selling or holding equity or debt instruments and providing or settling loans or other forms of credit (IFRS, n.d.)

To achieve this objective, the financial reports must provide information on the economic resources of the entity, their counterparty and the transactions and other events and circumstances that affect them The degree of usefulness of financial information depends on the qualitative characteristics

3.3.2 Qualitative characteristics of useful financial information

The qualitative characteristics of financial information, as set out in the Conceptual Framework of the IASB are fundamental to identify the types of information that are most likely to be useful for the purpose of making decisions about the reporting entity based

on the information presented in its financial report (Ernst & Young, 2010)

The revised Framework distinguishes two types of qualitative characteristics that are necessary to provide useful financial information: fundamental qualitative characteristics and enhanced qualitative characteristics

3.3.2.1 Fundamental qualitative characteristics

By fundamental qualitative characteristics is meant: the relevance and the faithful representation of financial information

Relevant information is capable of influencing the decision made by users It is capable

of making different decisions if it has predictive value, confirmatory value or both

Predictive value helps users in predicting or anticipating future outcomes Confirmatory value enables users to check and confirm earlier predictions or evaluations

The Financial reports represent economic phenomena in words and numbers To give a

perfectly faithful picture, the financial information must have three characteristics: it must

be complete, neutral and free of errors The revised Framework acknowledges limitations

in achieving a faithful representation; financial information might not be totally free from errors However, a faithful representation is achieved if no errors or omissions affect the description of economic phenomena and the process applied to produce reported information has been selected and applied without errors

3.3.2.2 Enhancing qualitative characteristics

The usefulness of financial information is enhanced when it is comparable, verifiable, timely and understandable The purpose is to enhance the relevant and faithfully

financial information

Comparable:

User decision-making implies making choices between various options, such as selling

or holding an investment or investing in one entity over another Therefore, information is more useful if it can be compared with other items, in different periods within a set of financial statements and across different reporting entities

Verifiable:

Verifiability assumes that different well-informed and independent observers could come

to a consensus -but not necessarily to a complete agreement- on whether a particular depiction of an event or transaction is a faithful representation

Understandable:

The information is understandable when it is classified, characterized and presented in a clear and concise manner A company's financial information should be presented in such a way that a person with a reasonable knowledge of business and finance, and the willingness to study the information, should be able to comprehend it

4 Systematic literature study

The fourth chapter discusses the methodology of the literature study and the obtained results from the systematic literature study

4.1 Databases and search strategy

The databases that were searched for relevant studies are: ABI/INFORM Collection, Accounting, Tax and Banking Collection and Web of Science The searches were conducted in January 2018 (Week 2) The following search strategy was used:

(accountancy OR audit OR accounting OR auditing) AND (automation OR technology

OR artificial intelligence OR robots) AND (Future)

In the database ABI/INFORM, this search strategy resulted in 764,477 articles, in Accounting, Tax and Banking Collection it resulted in 81,490 articles and finally in Web of Science it resulted in 5,695 articles Due to the number of articles found in ABI/INFORM Collection and Accounting and Tax and Banking Collection, the following filters were used: full-text, peer reviewed, scholarly journals and articles These filters led to 75,236 articles in ABI/INFORM and 7,159 articles in Accounting, Tax and Banking Collection No filters were used in the database Web of Science The total number of articles found by this searching strategy was 88,090 In annex 1, a screenshot of all databases can be found

4.2 Study selection

One reviewer searched for relevant studies using the search strategy described above The selection of studies was determined by two steps: the studies were first filtered on relevance of the title (n=164) and after that, the studies were filtered on relevance of the abstract (n=92) Duplicates and studies written in another language than English were excluded (n=44) The remaining studies (n=48) were read full-text and articles not related

to the different technologies as described in Chapter 2 and not related to the future of accounting were excluded The 32 selected articles underwent quality appraisal

The flow chart of the selection process can be found on the next page

Figure 1: Flow chart of the selection process

4.3 Quality appraisal

The quality assessment of the selected articles was conducted by using Hawker, Payne, Kerr (et al., 2002)’s framework This framework includes nine questions, regarding the following domains: abstract and title, introduction and aims, method and data, sampling, data analysis, ethics and bias, results, generalizability and implications and usefulness Each of the domains can be scored from 1 (very poor) to 4 (good), with a maximum score of 36 High quality is defined as 30-36 points and medium quality as 24-29 points

Of the 32 articles, 13 articles had a high-quality score and 13 articles had a medium quality score 6 articles had a lower quality score, but the findings were yet seen as interesting for this research

The framework for quality appraisal can be found in table 1, in table 2 to 5 the quality assessment of the selected articles can be found

1 Abstract and title Did they provide a clear description of the study?

2 Introduction and aims Was there a good background and clear statement of the aims?

3 Method and data Is the method appropriate and clearly explained?

4 Sampling Was the sampling strategy appropriate to address the aims?

5 Data analysis Was the description of the data analysis sufficiently rigorous?

6 Ethics and bias

Have ethical issues been addressed, and what has necessary ethical approval gained? Has the relationship between researchers and respondents been adequately considered?

7 Results Is there a clear statement of the findings?

8 Generalizability Are the findings of this study transferable to a wider population?

9 Implications and usefulness How important are these findings to policy and practice?

Table 1: Framework for Quality Appraisal

Htaybat

Al-et al.,

2017

Anonymous,

Kokina

et al.,

2017

Liu et al.,

4.4 Data extraction

From the included articles, the data was gathered using a standard data extraction form: author, title, year, country, study objective, method and findings The data of the articles used can be found in annex 2

4.5 Description of the studies

The included studies vary in method of study, location of study and time of study

Regarding the study methods, the 32 studies that are included can be categorized in sixteen review studies (Kokina & Davenport, 2017, Oschinski & Wyonch, 2017, Chelliah,

2017, David, 2017, Coyne, Coyne & Walker, 2017, Rattunde, 2016, Gamage, 2016, Henry & Hicks, 2015, Liu & Vasarhelyi, 2014, Moudud-Ul-Huq, 2014, Baldwin, Brown & Trinkle, 2006, Silverman, 1996, Omar, 1993, Chase & Shim, 1991, Anonymous, 1987 & Blum, 1986), eight survey studies (Özdoğan, 2017, Sorgner, 2017, Gonzalez, Sharma & Galetta, 2012, Parham, Noland & Kelly, 2012, Beaman & Richardson, 2007, Sangster,

1994, Tuzhilin, 2004 & Wilson & Sangster, 1992), four interview studies (Al-Htaybat & Von Alberti-Alhtaybat, 2017, Marcello, Ray, Carmichael et al., 2017, Herbert, Dhayalan & Scott, 2016 & Zarowin, 1994), three quantitative studies (Kim, Kim & Lee, 2017, Arntz, Gregory & Zierahn, 2017 & Frey & Osborne, 2013) and one case study (Cole & Hales, 1992)

Of the 32 studies, 19 were conducted in the United States (Coyne et al., 2017, Kokina et al., 2017, David, 2017, Marcello et al., 2017, Rattunde, 2016, Henry et al., 2015, Liu et al., 2014, Frey et al., 2013, Parham et al., 2012, Gonzalez et al., 2012, Beaman et al.,

2007, Baldwin et al., 2006, Tuzhilin, 2004, Silverman, 1996, Zarowin, 1994, Cole et al.,

1992, Chase et al., 1991, Anonymous, 1987 & Blum, 1986), seven were conducted in the United Kingdom (Al-Htaybat et al., 2017, Kim et al., 2017, Özdoğan, 2017, Chelliah,

2017, Herbert et al., 2016, Sangster, 1994 & Wilson et al., 1992), two in the Netherlands (Arntz et al., 2017 & Omar, 1993), one in Canada (Oschinski et al., 2017), one in India (Moudud-Ul-Huq, 2014), one in Australia (Gamage, 2016) and one in Russia (Sorgner, 2017)

The year the studies took place varies from 1986 to 2017 Ten studies took place in 2017 (Coyne et al., Sorgner, Al-Htaybat et al., Marcello et al., Kokina et al., Kim et al., Arntz et al., Özdoğan, Oschinski et al & Chelliah), three studies in 2016 (Gamage, et al., Dhayalan & Scott), two studies in 2015 (David & Henry et al.), one study in 2014 (Moudud-Ul-Huq), one study in 2013 (Frey et al.), one study in 2014 (Liu et al.), two studies in 2012 (Parham et al., & Gonzalez et al.), one in study 2006 (Baldwin et al.), one

study in 2007 (Beaman et al.), one study in 2004 (Tuzhilin), one study in 1996 (Silverman), two studies in 1994 (Sangster, Zarowin), one study in 1993 (Omar), two studies in 1992 (Cole et al & Wilson et al.) one study in 1991 (Chase et al.), one study in

1987 (Anonymous) and one study in 1986 (Blum)

Main subjects identified were: 1) consequences on the accounting profession, 2) moral decision-making, 3) future role, 4) implication on small accounting firms, 5) implications

on the labour market and 6) solutions Each of these topics will be discussed in the next paragraph

4.6 Consequences on the accounting profession

The first step is to distinguish routine tasks (which can easily be automated) and routine tasks (which are more difficult to be performed by machines or software) Jobs that require critical thinking and human contact will not be automated soon (Oschinski et al., 2017) These occupations need high-level creativity and training Jobs that generally consist of routine tasks do not require a level of high education and only little human interaction is needed compared to non-routine tasks Non-routine tasks can be divided into manual occupations and intellectual occupations Manual occupations generally require lower qualifications than cognitive jobs that generally require a high level of education (Oschinski et al., 2017)

non-Herbert et al (2016) explored the possibilities for transforming the way professional work

in the future, by using automation The study describes that since automation is used to eliminate routine and repetitive tasks, it will allow employees to focus on more creative, non-structured tasks that require more thinking While focusing more on creative, non-structured tasks, the value of the accountant`s contributions will increase Kim et al (2017) examined the relative quantities of jobs that are susceptible to become computerized in the future and concluded that jobs that require little creativity or complex training (routine occupations), are most likely to be replaced Jobs that require critical thinking and human contact will not be easily automated These occupations require high-level creativity and training Tuzhilin (2004) found the same results by examining current trends in the technology-driven automation and the effect that it will have on different jobs The author describes that repetitiveness, stability and structure are the characteristics of jobs that can be automated In other words: routine production jobs can

be performed by automation Arntz et al., (2017) demonstrated that empirical

assessments are wrong saying that half of all jobs in western industrialized countries are

at risk of automation in the next 10 to 20 years According to the authors, many accounting tasks are already automated in firms, such as invoicing, payroll and book-keeping, which involve the processing of large amounts of data and consists of repeated, stabile and structured tasks In general, the process of accounting information has already become largely automated According to Liu et al (2014), automation will constantly develop and make some tasks -like bookkeeping- disappear and at the same time create new ones

Wilson et al (1992) examined the use of computer technology by the UK accounting profession and why the accounting profession should be aware of automation By conducting a survey under the population of members of the Institute of Chartered Accountants of Scotland, the authors asked the respondents to determine which factor could be a motivation for technological change The majority identified the need to meet accounting deadlines as the most significant motivation, just as the importance to provide better information to clients According to Al-Htaybat et al (2017), by using technology the quality and relevance of the accounting information will improve Big Data reduces the time of reporting, since technology can provide real-time updates The computerized accounting systems are able to convert accounting data into valuable information, which reduces work-time and improves the quality of the financial information Companies are looking for efficiency as well as productivity and profitability As a consequence, companies will prefer using technology rather than human intervention (Wilson et al., 1992) According to Anonymous (1987) the benefit of using an Expert System is that the system will not leave the company or retire unlike employees Cole et al (1992) published an article describing the benefits of automation According to the authors, automation might eliminate some tasks and by reducing the number of employees the labour costs of the company will decrease The system could be useful in situations where a human intervention is costly or in situations where a human expert is not available Problems, which were solved only by the people with a specific expertise, can now be solved by the system and enables the firm to enter a new market (Anonymous, 1987) Furthermore, Expert Systems can train new inexperienced accountants, who otherwise needed intervention from an employee of the company The employee productivity can hence improve by keeping the experienced personnel working on important tasks and by giving the new accountants enough training to become effective Gonzalez et al (2012) were interested in the general adoption of the technology and

explained that, due to the pressure coming mainly from clients and competition, companies need to invest in automation To become more efficient, companies need to satisfy their clients’ expectation regarding the decrease in prices By automating the processes of certain tasks, companies become more efficient and therefor stay ahead of the competition

The study of Gamage (2016) explored the latest developments in Big Data and its impact

on accounting education According to these findings, the decision-aid is one of the greatest benefits of Big Data Accountant researchers have already been using automation for the decision-making process The measurement of the data has been enhanced and the information is better understood (Liu et al., 2014) Expert Systems have the benefit that it can assist the accountant during the analysis of complex data The systems can provide help to professionals to make sure that the right questions are addressed, and the right decisions are being made (Anonymous, 1987) The system can help to identify plausible issues and guide the accountant to find the best solution to those problems (Anonymous, 1987) Data accessibility has been enhanced by technologic innovation, such as the financial information provided to accountants increased in effectiveness and efficiency (Liu et al., 2014) Hence, the decision-making process will be improved by providing more accurate and detailed data (Al-Htaybat et al., 2017)

According to Herbert et al (2016), 90 percent of the errors or accidents are caused by humans Machine learning would be a solution to reduce these errors In the coming years, 40 percent of the companies will use automation to avoid those human errors

According to Anonymous (1987), due to the fast improvement in technology, Expert Systems are likely to become part of some accounting tasks, for instance:

- Account attribute analysis The Expert System provides guidance concerning different accounts;

- Quality review Expert Systems are able to improve the quality of accounting such as in

compliance and annual report disclosures, by fulfilling or judging the quality review of the firm;

- Accounting decisions The treatment of complicated transactions can be facilitated with

the use of Expert Systems, which are able to help accountants in making decisions;

- Tax planning Experts Systems are able to make complicated analyses and provide the

appropriate documents to accountants;

- Management consulting Management consulting is another area where Expert

Systems are introduced to identify patterns and relationships The system will provide solutions based on the situation and evaluate the effect of changes in consulting engagements

Auditors are also able to take advantage of automation Baldwin et al (2006) reviewed the nature of accounting and auditing problems and the need for the application of automation According to this study, auditing is a field that is intrigued by the use of automation Auditors have to deal with uncertainties and incomplete financial information, but the decisions that are being made are often repetitive Chase et al (1991) studied the use of Expert Systems in large accounting firms Firms generally use Expert Systems mostly to reduce time and costs while auditors have more time to take important audit decisions Expert Systems bring more accuracy and consistency in the audit procedures and auditors can work faster without having to ask questions to a senior auditor (Moudud-Ul-Huq, 2014)

A study from Moudud-Ul-Huq (2014) showed that automation was not suitable for every audit task The audit tasks in which, according to this study, automation is useful to help the auditor during the decisions-making process are:

- Audit planning Expert Systems are able to help evaluating risks, establishing audit

objectives and giving prescription regarding the audit steps and procedures The Expert System not only increases reliance on the decision aids, it also enhances the decision during the audit planning

- Analytical review procedures The use of artificial neural network improves the

analytical review procedures because the system provides objective information about a client company

- Materiality assessment The Expert Systems are able to identify plausible

misstatements in the financial statements

- Materiality judgments The Expert Systems can be used during the materiality

judgements to assist the auditor during the formulation of judgements

- Internal control evaluation The Expert Systems are able to increase effectiveness and

efficiency during the evaluation of the internal controls Accounting firms could benefit from the use of to discern potential weaknesses

- Risk assessment The Expert Systems alerts the auditor when audit tests are

unnecessary to perform on firms free of misstatement This should improve the efficiency

as well as effectiveness of the audit The Expert Systems is able to provide several benefits during the risk assessment procedure including: advisory, consistency of the decisions and increase in productivity

- Going-concern decisions The Expert system assists and supports auditor’s judgment

about a client’s going concern Advices from an auditor specialist are not needed anymore, which represents a cost advantage for the companies

Nevertheless, relying only on the data provided by technology and not using the experience and knowledge of accountants could be dangerous Accountants are sceptic regarding the reliability of the financial information provided by automation (Al-Htaybat et al., 2017) The lack of sufficient knowledge of Big Data and the analysis of the data could generate inappropriate results, accountants may not be able to analyse and interpret the results correctly Marcello et al (2017) held a roundtable discussion on the past, present, and future of the auditing profession One of the professionals from the roundtable discussion believes that accountants and auditors need to be careful when using Artificial Intelligence According to this respondent, human intelligence exceeds machine learning The professional is sceptical about the use of Artificial Intelligence and does not trust machine learning concerning the decision-making The paper of Sangster (1994) considered the way in which organisations develop The results show that 60 percent of the respondents would not use the Expert System in the most effective manner and 67 percent of the respondents indicated that this type of technology would not be trusted Frey et al (2013) examined how susceptible jobs can be computerized This study indicates that the current technology is not yet sufficiently developed to perform tasks that require human thinking In the future, some non-routine manual tasks will probably

be performed by automation The technology evolves so fast that the machine learning is nowadays in a position to replace many of the decisions that a human used to make

3) Freedom from disturbing passion;

4) The ability to vividly imagine the feelings and circumstances of the parties involved

According to this study, Artificial Intelligence is able to perform the three first conditions However, the last condition involves emotions, which AI is not able to fulfil Furthermore, the machine does not know what happened in the past, neither what is happening in the present nor what will happen in the future

Zarowin (1994) examined the computer revolution in the accounting profession by interviewing the Chief Executive Officer (CEO) of the Computer Associates International The respondent recognized that until now, Artificial Intelligence could not perform accountants’ most valuable functions: interpreting and analysing financial information Accountants do not need to worry about being replaced by technology yet Kokina et al., (2017) discussed the current capabilities of cognitive technologies and the implications these technologies will have on human auditors and the audit process According to an interview conducted during this study, senior accountants in large firms stated that the need for human accountants would not go away anytime soon At least over the next couple of years, accounting is one of the many business fields that are likely to be augmented by technology, rather than fully automated

4.8 Future role

Beaman (et al., 2007) studied the role of the management accountants in the future and state that the accountant’s role is dominated by scorekeeping and other requirements Accountants need to develop their skills regarding the use of AI if the employees want to keep adding value to the firm The authors concluded that management accountants who continue spending much time in scorekeeping activities (instead of providing decision support services to managers), risk losing their jobs Management accountants need to know what the critical pieces and the outcomes of data are, in order to add value to the business (Gamage, 2016)

Young accountants coming into the profession need to understand what are the skills needed to work alongside automation The need to acquire and develop these skills is crucial to avoid job loss Many of the jobs that will persist in the future will require interpersonal interaction, flexibility, adaptability and problem solving (David, 2015)

Future accountants will be required to have diversity of experience, curiosity and the ability to learn continuously The profession does not only need students who understand audit standards, the students also need to know how to solve problems and how to think critically (Marcello et al., 2017)

Parham et al (2012) examined which skills are important for the future career of accounting students, these skills are: written and oral communication, motivation, decision-making, financial analysing and professional judgement

Accounting companies are looking for employees who are not afraid of technology, but who are creative and open-minded These employees also need to know how to work with and how to use the data provided by technology (Al-Htaybat et al., 2017)

Universities will have to work with companies to make sure the students learn the required skills to work with Big Data Accountants that are able to work with Big Data, extract the necessary information and make the information useful at the right time will

be needed in the accounting field Silverman (1966) explored the effect of automation and came to the conclusion that automation destroys old skills, but at the same time creates new skills that require the knowledge of how to use complex machines Accountants will have a more proactive role in the business and will be required to stay

in contact with employees working in different areas -like Information Technology (IT)- (Coyne et al., 2017; Gamage, 2016)

According to Kokina et al (2017) the following types of activities will exist in accounting jobs:

- Working with machines to improve performance and results of the company;

- Overseeing the use of intelligent machines and determining if a different automation tool is necessary;

- Working with vendors to develop Artificial Intelligent systems and to maintaining the existing ones;

- Performing tasks that are still impossible to perform with automation;

- Performing accounting tasks, in which the use of an automated system would not

be efficient

4.9 Implications on small accounting firms

Automation will be integrated in daily operations, therefor companies will have to invest

in technology to remain relevant and to survive against the competition in the twenty-first century (Chase et al., 1991) Small companies do not always have enough capital to invest in the technology, since the technology can be capital-intensive This can create problems for small accounting firms, as these firms cannot keep up with the bigger firm`s investments in AI Since AI is more efficient in performing routine tasks than human intelligence, small firms are consequently less efficient than bigger firms that are able to invest in AI and are therefore less competitive

4.10 Implication on the labour market

Rattunde et al (2016) examined the impact of automation on the employment in the United States Specifically, the authors analysed how computer-based technologies and robotics have contributed to job polarization by reducing the number of “middle-skilled” jobs, while reinforce employment in both low-and-high skilled jobs This paper indicated that automation did not reduce the overall employment Automation has replaced some tasks, but also complemented other tasks Formalized and codified tasks have already been automated, since machines represent less labour costs for the companies and are more accurate and more productive Tasks that require flexibility, judgment and common sense are more complex to automate Humans still have the advantage of being able to make decisions in a situation of uncertainty Some low-skilled manual jobs that require language recognition, social interactions and situational adaptability are difficult to automate, just as the high-skilled professions that require creativity, critical thinking and problem-solving skills Blum (1966) also reported that there has been an increase and a decrease in the number of jobs requiring less skills, as well as both an increase and a decrease in the number of jobs requiring more skills Automation can replace some tasks, but at the same time also create new tasks The amount of blue-collar jobs decreased compared to the white-collar jobs, due to the different skill requirements and training According to Blum (1966), technological unemployment affects especially young workers, old workers, low-skilled and low educated workers Workers who possess only one skill in that specific occupation will have difficulties to develop new skills in other occupations Sorgner (2017) provided an overview of current trends and developments

on the labour markets due to the automation of jobs This study also describes the most recent dynamics of self-employment related to the risk of the automation of jobs The author reported that middle-skilled workers in routine jobs are more susceptible to

automation, while people with low and high levels of education are less likely to have changes in their work occupation Middle-skilled workers looking for a job will have to possess or develop skills that are hard to automate, such as creativity or social interactions This paper indicated that people who are willing to take risks (like starting a completely different job or developing new skills) have less probability to be unemployed

it represents less salary Therefore, sharing work time with machines is only a temporary solution

Another solution is to propose social programs to the jobs that could be replaced by technology The susceptible jobs that could be replaced by automation are the jobs that are relatively low paid and do not require much creativity Hence, these unemployed workers could be helped by programs provided by the government to motivate these persons to work in another field However, these programs are often expensive and would be covered by taxes from citizens, entrepreneurs and capitalists Thus, it would help unemployment only for a short period

The creation of new jobs proposed by the government could be a long-term solution For instance, accelerating the creation of new jobs by stimulating business through tax benefits Unfortunately, government programs are often seen as national embarrassments and it represents a huge investment

The most reasonable solution would be a change in education The future employees should have the necessary skills to work alongside machines and the current employees should develop their skills to stay important in the company The education system might change by focusing on critical and system thinking and developing students’ creativity skills Students with the right qualifications are able to work with technology, instead of being replaced by it

5 Methodology

Now all the literature regarding the future of the accounting is gathered, it is important to test these results in an empirical study First, the research design will be mentioned Then, the data collection methods will be discussed Furthermore, the reliability and the validity of this study will be described Finally, the ethical considerations will be discussed

5.1 Research design

This study is a qualitative study In quantitative studies, the results are derived from numbers collected and statically analysed In qualitative studies, verbal data and experiences are necessary for completeness (Verhoeven, 2010) The benefit of a qualitative study is that during the collection of the data, emotions, opinions as well as human experiences are explored, which can contribute to a better understanding of the subject

5.1.1 Interviews

During this qualitative study, interviews have been conducted Eight professionals have been interviewed regarding their use of technology and their future perspective on the accounting profession The duration of the interviews differed from 25 minutes to one hour and ten minutes, depending on the respondents and their available time

5.1.2 Population

Since accountants have 94 percent of probability of losing their jobs within the next two decades (Nagarajah, 2016), those professionals were selected to study the question if a future in accounting without human intervention is possible As mentioned, a total of eight interviews have been conducted, in which eight accountants were interviewed One respondent stated to remain anonymous, in order to protect the clients’ information shared during the interview The names of the companies of other respondents have been given

5.1.3 Place

As mentioned earlier, interviews with professionals from Belgian and Luxemburgish companies have taken place

5.1.4 Description of the respondents

Respondents Experience Years of Title When Where How Duration

Binder Dijker

Senior Manager / Accounting

& Reporting services

12/03/18 Brussels In person 1h06

Partner / Accountancy

& Advisory

6/03/18 Ghent In person 1h10

Director / Accounting

& Corporate Services

Director / Accounting department

9/05/18 Ghent In person 36 min

Director / Tax & Legal Services

22/05/18 Ghent In person 39 min

University of

Luxembourg 20

Head of Finance and Accounting department

31/05/18 Luxembourg Telephone 58 min

Table 6: Description of the respondents

The following topics take a central role in each interview: the use of automation within the firm, the role of the accountant, the qualitative characteristics of the financial information, the skills of the accountant and the small accounting companies

It is important to show the interview results as detailed as possible, without omitting or modifying the text and thus the context The interviews have been analysed in a consciously way After all, it is a qualitative research in which qualitative analysing is necessary

The result of the interviews have been processed in the following way: The interviews are audio-recorded and afterwards transcribed, meaning that the interviews have been written down on a document using the actual words of the respondents (Saunders et al., 2009) Then, the answers are per respondent divided in small fragments This process is necessary to assign a code to the fragments with a term By coding the fragments, an overview has been created (Boeije, 2010)

In annex 3 the interview guideline can be found

5.3 Reliability and validity

In order to provide sufficiently consistent and relevant evidence, the quality of a research must be both reliable and valid How the reliability and validity of this study have been ensured, will be clarified in this paragraph.

5.3.1 Reliability

Mistakes during a qualitative study can happen accidentally The reliability of the research results indicates to what extent the research is free of mistakes To test the reliability of the study, the study should lead to the same results if another researcher would repeat the study If it leads to the same results, the research is reliable (Verhoeven, 2010)

The reliability of the semi-structured interviews has been guaranteed by holding trial interviews The advantage of holding trial interviews is that the interview questions can

be tested In this way, the questions asked to the respondents of the empirical study have been adequate

During the interview unprepared questions have been added, which allowed the interviewer to add and expand on answers given to get a more realistic and honest response

To avoid unnatural responses to questions due to unavoidable effects -like anxiety- the transcribed interview has been sent shortly after to the respondent to guarantee that the transcription represents a true reflection and is free of errors

The reliability of the study has been increased by the usage of a recording device Recording the interview avoids the risk of interpretation of notes and enables the interviewer to listen to the interview again and thus to report correctly However, if the respondent would not have felt at ease to mention something due to the recording, the recording would have been ended

5.3.2 Validity

The validity of a research shows to what extent the research is free of any error (Saunders et al., 2009) There are three types of validity: internal validity, external validity and construct validity

5.3.2.1 Internal validity

The internal validity implies that the collected data represents the reality (Verhoeven, 2010) To increase internal validity of the data, various steps are taken Hence, the choice of interviewing experts who have experience in the accounting field increases the credibility of the results Internal validity can also be established by increasing the number of interviews The greater the number of interviews, the more reliable is the data collected Furthermore, the recording of the interviews enables to transcript the whole

conversation and avoids errors or missing data The transcribed interviews have been returned to the respondents for reviewing and approving

This study uses triangulation Triangulation is the use of multiple data collecting methods, in this case a systematic literature study and semi-structured interviews The advantage of triangulation is that data provided by multiple sources increases the accuracy and validity of the research (Verhoeven, 2010)

5.3.2.2 External validity

By external validity is meant if the chosen sample gives the correct reflection of the (accounting) population If that is the case, then the data collected can be generalized (Verhoeven, 2010) When talking about qualitative research, the size of the chosen sample is often insufficient to be representative for the whole population In this case, the group of respondents is too small to claim the representativeness of all accounting companies in Belgium and in Luxembourg However, a lesser external validity does not necessary mean that nothing can be done with the results of a research (Verhoeven, 2010)

5.4 Ethical considerations

It is important to follow ethical considerations when it comes to dealing with interviews During the collection of data, the identity of the respondents has been protected through anonymity, unless the respondent agrees on publishing personal information A reminder

of the research subject has been mentioned before each interview The consent of the respondent to an audio-recorded interview has been requested Each respondent has been given a written version of the discussion The information obtained during the interview has been reported in all honestly The company of the respondents have been kept confidential, unless indicated otherwise

The signed confidentiality agreement can be found in annex 4