Search

This thesis presents an investigative study of the detection of fraudulent financial statement by applying machine learning techniques. It specifically focuses on the processes of computational intelligence. My research aims to compare the applicability of a relatively inclusive group of machine learning techniques to enable financial statement fraud prediction. More specifically, first, we examine which can utilise algorithms the most given the variegated assumptions about the fraud's classification costs. Second, we discuss which predictors are essential in algorithms for the discovery of fraudulent financial statements. Furthermore, this thesis examines whether the utilisation of creative accounting decreased after adopted IFRS. In particular, Chapter 1 is the introductory chapter of the research. Chapter two contained a literature review of the accounting environment in Greece. Also, chapter two examines the causes that led to the establishment of Greek legal and accountancy systems. It addresses the association between the accounting and the taxation systems. Moreover, this chapter offers information on the differences between the IFRS and the Greek GAAP. Chapter three comprises a literature review and theoretical analysis of creative accounting-chapter four refers to the research methods and empirical studies utilised and undertaken, respectively. We present a comprehensive classification and using the critical aspects of the algorithm detection used. We investigated the fraud type and the detection methods' performance for financial statement fraud, analyse the existing fraud detection literature. Specifically, we study the implementation of machine learning techniques, like the Naïve Bayes, Random Forest, Support Vector. Machine (SVM), Decision Tree, Logit Regression, and K-NN. We followed a comprehensive classification framework of machine learning techniques' application in fraudulent financial statements detection. Furthermore, Chapter six refers to the effect that IFRS implementation had on earnings management. It emphases the utilisation of creative accounting in the periods before and after the IFRS is adopted. Chapters five and seven address the experimental results of this research. Finally, Chapter Eight is the concluding chapter, and it refers back to the research questions and objectives. Also, it states the contributions and the findings of the study again.

Private firms are the predominant form of incorporation in both the developed and the developing world. They were, however until recently under-researched, mainly due to the lack of data availability. To overcome this difficulty, both researchers and academics approximate private entities through public equivalents. One of the inputs that is needed to do that is the discount rate, by which the cash flows, from investing in a private firm, need to be adjusted to account for the risk that such investments entail. The main focus of this thesis is to carefully examine the variables that have been identified throughout the literature, as being impactful to the discount rate. These factors are examined through the scope of a novel methodology and will allow appraisers to have a framework of reference when valuing a private company. The purpose is not only to highlight the determinants of the Price to Earnings (P/E) ratio in the private companies’ valuation however, but to do so in an international setting, as both the UK and the US are examined, in an attempt to document differences in the risk profiles of investors from these countries. To determine the level at which each of the factors, affects the discount rate, Principal Component Analysis is employed, on public companies, selected to proxy private firms. This methodology is used to reduce the size of extensive datasets, while retaining most of their variability. The components produced, are linear combinations of the original variables, devoid of any multicollinearity issues inherent in large datasets. These components are then regressed against a valuation proxy, a Price-to-Earnings ratio, calculated initially for the public companies’ dataset and later, for a private companies’ set, the latter being adjusted to account for the illiquidity discounts exclusive to private enterprises during the Mergers and Acquisitions process. The results indicate that a private company’s discount rate can be approximated best with the inclusion of the Free Cash Flows (FCF), the Debt-to-Equity, Assets, Cost of Debt, Total Beta, Z-Score, Auditors and Jensen’s Alpha for the UK and Earnings Before Interests Taxes Depreciation Amortization (EBITDA), external shocks (Financial Crisis), FCF, Debt-to-Equity, Return on Capital, Percentage of Insiders Holding Stock, Beta, Tax Rate, Marginal Profit (MPK) and Jensen’s Alpha for the US. Investors from the UK, appear to be more risk-averse than their US counterparts, as they seem to value traditional variables more, than the ones focused on profitability, earnings and debt.

The subject of corporate governance and corporate performance has been widely discussed and examined over the last two decades. A great deal of change has developed within British Boardrooms since the emergence of the Cadbury Committee Report in 1992. UK Corporate Governance reforms over the years have been consistently developed where an increase in the number of non-executive directors on board, their roles and their effectiveness, was evident throughout the development of these reports. For instance, the Cadbury Report set a minimum of three non-executive directors for each company. Also, the independent non-executive director has become the catalyst for better performance since it has been recommended by the 1998 Combined Code that at least one-third of the board is to be independent and increased to half by the 2003 Combined Code. Although it has been evident that the level of compliance by companies has increased, the relationship between firm performance and corporate governance has been mixed and inconclusive in previous research. A large number of empirical works found no clear link between firm performance and corporate governance. There is an argument posited by scholars that better firm performance is achieved in well-governed firms. Therefore, the main question addressed in this thesis is whether a relationship exists between internal corporate governance mechanisms and performance of FTSE All shares non-financial firms listed on London Stock Exchange for a period 2005 to 2010. It specifically looks at the link between firm performance (measured as Tobin’s Q and Return on Assets (ROA) and board characteristics (board size, independent directors, CEO duality and Audit committee), managerial ownership, executive remuneration and financial policies (Debt and Dividend) as governance mechanisms. This study will draw upon the agency theory to test whether the hypothesised relationships exist between firm performance and corporate governance mechanisms in the UK. Ordinary Least Square (OLS) regression analysis produced mixed results. According to OLS regressions, the results provide some evidence of a relationship between some governance mechanisms and firm performance. In general, based on market measures (Tobin’s’ Q), some governance mechanisms (independent directors, board size (apart from in 2006), role, managerial ownership (apart from in 2008), executive remuneration and debt (apart from in 2005)) positively relate to firm performance, while audit (apart from in 2010) negatively relates to firm performance. However dividend pay-out produced mixed results. Based on accounting measures (ROA), independent directors (apart from in 2010), role (apart from in 2006), managerial ownership and executive remuneration positively relate to firm performance. Board size, audit and debt negatively relate to firm performance. However, dividend pay-out produced mixed results. Further analysis using two-stage least square regressions indicate that any causal effect runs from governance to firm performance rather than in the opposite direction. Overall, the findings of the research are period specific. Variables showing significant explanatory power at the start of the sample period may cease to be significant or change sign at the end of the sample period. In the corporate world corporate governance has been a growing issue and it has contributed to becoming a key business discipline in the management of companies. This study contributes to the increasing number of research studies on the link between firm performance and corporate governance. The lack of clarity, mixed and permanent relationships provided, show that the association between performance and different governance is complex and dynamic: optimal governance arrangements may differ from firm to firm in relation to board characteristics.

The main objective of this study was to develop an algorithmic financial model to determine and examine the characteristics of key value drivers, earnings, net income, EBITDA, sales, and book value, that formulate the value aspects of a company to compute raw value multiples using multi-linear regression analyses of scaled value driver, Price-to-Earnings (PE), Price-to-Net_Income (PX_Earn_Com), Price-to-EBITDA (PEBITDA), Price-to-Sales (PS), and Price-to-Book (PB), against a comprehensive list of independent proxy variables. The resulting spectrum of raw value multiples is utilised in further computation that encompass the triangulation of the spectrum raw value multiples in a weighted process based on the adjusted coefficient of determination measurement, which would synthesise a raw market share price of the company (Adj. Vs_PX) comparable to Bloomberg-based share prices (PX). Effectively, the multi-linear regressive algorithmic financial model would be used for assessing market value signalling a buy or sell based on the position of synthesised market share price relative to current market share prices. The amalgamated data sample for this study comprises of the market indices representing the Anglo-Saxon and European markets, namely the FTSE-All-Share (ASX) of UK, S&P 500 (SPX) of the USA and STOXX Europe 600 (SXXP) of Europe with a data availability ranging from 2001 to 2011 obtained from Bloomberg. The main objective was successfully completed by the analysis of 170 regression models based on 5 scaled dependent variables regressed against 56 independent proxy variables for 8,851 company-years out of 14,340 company-years representing the 3 market indices, ASX, SPX, and SXXP. The descriptive statistics measures of the computed raw value multiples and share prices relative to the Bloomberg-based values have overall generated robust and significant results. Generally reflecting a low standard error, consistent standard deviation and yielding sample means that are very similar. Relating the computed raw value multiples of PE, PX_Earn_Com, PEBITDA, PS, and PB, against the respective Bloomberg-based multiples has mostly shown similar company values for ASX and SPX, signifying that the listed companies are efficiently valued. Whereas for the companies listed on the SXXP index, the results highlighted that there were differences in values observed between the synthesised raw multiples and the Bloomberg-based multiples, implying that companies are either over-valued or under-valued. Overall the corresponding PS and PB multiples displayed the most consistent and explanatorily significant results compared to the three earnings multiples. However, the observed discrepancies in the synthesised values relative to the Bloomberg-based values would mostly be offset collectively between PE, PX_Earn_Com, and PEBITDA, thus presenting consistent and significant results. This study concludes that the cross-sectional relative valuation analysis of any fully-listed company in the Anglo-Saxon and European markets in an identical process to be achievable. Hence, the process of valuation analysis using independent proxy variables can be standardised for the Anglo-Saxon and European markets and the triangulation of value multiples to synthesise comparable market share prices. The various aspects of the methodologies applied are founded on multi-linear regression analysis and relative valuation using a standardised database for all the data obtained from the three market indices: ASX, SPX, and SXXP. Thus, the multi-linear regressive algorithmic financial model is capable of computing cross-sectional valuation, as well as cross-market valuation for any fully-listed company, to compute value multiples that can be triangulated to synthesise respective share prices premised on standardised proxy variables.

The twelve refereed publications which form the basis of this PhD each deal with the impact of regulatory change on the perceptions and behaviours of key participants. The publications are drawn from three collaborative research projects exploring different aspects of financial reporting, auditing and corporate governance regulation. The first project considered the way in which a new regulatory enforcement body, the Financial Reporting Review Panel interpreted its role and the impact of its actions on preparers and auditors of financial statements. The second project reviewed the developing perceptions of preparers, regulators and auditors following the European Commission announcement that IFRS would be adopted for the consolidated accounts of all EU listed companies with effect from 2005. The third project explores the impact of the re-regulation that followed the Enron and Worldcom scandals. In particular it examines auditor / client financial reporting and auditing interactions and presents evidence on the role taken by the newly empowered audit committee in that process. All of the publications present the results of empirical evidence. While a wide range of research methods are used, interviews and questionnaire surveys dominate. All of the publications are policy relevant and recommendations are made in each paper. Although all the research has been undertaken in a UK setting, the issues are explored within the framework of international literature, and most of the findings and recommendations are of international significance. The first part of this commentary provides an overview of the contribution and quality of the work being submitted. The publications are then located within the broad sweep of regulatory theory to demonstrate the coherence of the submission. Each of the three projects is then considered in turn and the contribution to knowledge of each of the publications is then discussed in detail, and summarised in the conclusion.

The ten-refereed papers underlying this thesis primarily focus on the measurement of fraud and corruption, counter fraud and corruption strategies and the policing of fraud and corruption. The refereed papers deal with these issues in different contexts e.g. health care, sport, the FTSE 100 etc and illustrate that while different sectors there are common themes and strategies needed to prevent and reduce fraud and corruption. All of the publications in this thesis therefore make suggestions – some implicit and some explicit – on how to tackle fraud and corruption and are thus not only an academic contribution to knowledge, but also have practical relevance too. This contribution is based on actual research, rather than literature based work, and therefore includes a variety of research methods and the publication of primary research. Furthermore, many of these publications are international in their approach drawing on literature from around the world. Therefore, to ‘set the scene’ the contemporary relevance and importance of the research is explained with a focus on the contribution to knowledge these papers produced. This is followed by the contribution these papers have made in exposing the limitations of present policing of fraud and corruption strategies. This is then supported by the usefulness of fraud risk measurement exercises, and then the development of an anti-fraud culture, which is a common theme in the majority of these papers. In the conclusion all the core elements of these papers are combined to illustrate the contribution this research has made to current academic debate regarding fraud and corruption.