TABLE OF CONTENTS
LIST OF TABLES
LIST OF FIGURES
LIST OF ABBREVIATIONS
1.1 General overview
1.2 Background of study
1.3 Background of problem
1.3.1 Economic groups of Iran
1.4 Problem Statement
1.5 Research Objectives
1.6 Research Questions
1.7 Significance of the research
1.8 Scope of Study
1.9 Thesis Organization
1.10 Summary of the Chapter
2 LITERATURE REVIEW
2.2 Bankruptcy and Financial Theory
2.2.1 M-M Theory
2.2.2 Signalling theory
2.3 Credit Risk and Capital Requirement
2.3.1 Probability of Default
2.3.2 Loss given Default
2.3.3 Exposure at Default
2.3.4 The distribution of potential loss
2.4 Global Financial Crisis and Default Prediction
2.5 Probability of default: a literature on concepts and models
2.5.1 Classification of corporate default predic- tion models
188.8.131.52 Statistical Models
184.108.40.206 Theoretical Models
220.127.116.11 Artificial Intelligence Models
2.6 Empirical Evidences of the Developed and Devel- oping Countries
2.7 Determinants of Probability of default
2.7.1 Firm-specific Determinants
18.104.22.168 Financial ratios
2.7.2 Macro-economic Determinants
2.7.3 Industry determinants
2.8 Theoretical framework
2.9 Summary of the chapter
3 RESEARCH METHODOLOGY
3.2 Research Design
3.3 Source of Data and Sample Population
3.3.1 Formulation of Variables
22.214.171.124 Dependent Measure of the study
126.96.36.199 Independent measure of the Study
188.8.131.52 Data pre-processing (Data Cleaning)
184.108.40.206 Diagnostic Test
220.127.116.11 Implementation and Tools
3.4 Data Mining for Corporate Default Prediction
3.5 Data Mining Algorithms
3.5.1 Logistic Regression
3.5.2 Decision Tree
3.5.3 Artificial Neural Networks
3.5.4 Support Vector Machines
3.6 Multiple Classifiers Framework
3.7 Summary of Previous Studies
3.8 Comparing Classification Methods
3.9 Predictive performance metrics
3.9.1 Root Mean Squared Error
3.9.2 Mean Absolute Error (MAE)
3.9.3 Cross validation
3.9.4 Contingency table
3.9.5 ROC curve
3.9.6 Kappa Statistic- Measure of Agreement
3.10 Evaluating the Sign of Model Coefficients
3.10.1 Faulty Theory or Domain knowledge
3.10.2 Interpretation Errors
3.10.3 Model Development Problems
3.10.4 Data Problems
3.11 Summary of the Chapter
4 DATA ANALYSIS
4.2 Descriptive Summary and Correlation Matrix analysis
4.2.1 Descriptive summary for Firm-specific and Industry Determinants
4.2.2 Descriptive summary for macroeconomic Determinants
4.2.3 Correlation Matrix
4.2.4 Data Visualizing
4.2.5 Attribute Selection
4.3 Probability of default modeling
4.3.1 Models building without industry effect
4.3.2 Model building with industry effect
18.104.22.168 Gross Domestic Product (GDP)
22.214.171.124 Lending Interest Rate
126.96.36.199 HH Index
4.3.3 Model building across industries
4.4 Benchmark Classification Techniques
4.4.1 Model Development Procedure
4.4.2 Artificial Intelligence Models Analysis
188.8.131.52 CART Classification Tree
184.108.40.206 Neural Network-Multilayer Perceptron
220.127.116.11 Support Vector Machine
4.4.3 Multiple classifiers
4.4.4 Different Economics Periods Effect on Firm’s Probability of Default
5 CONCLUSION AND RECOMMENDATION
5.2 Findings and Discussions
5.2.1 Research Objective 1: To investigate the significant determinants of probability of default
5.2.2 Research Objective 2: To explore the effects of industry on probability of default
5.2.3 Research Objective 3: To compare the accuracy of statistical and data mining techniques to predict default among Iranian companies
5.2.4 Research Objective 4: To examine the effects of different economic periods (pre- and post-global financial crisis) on the determinants of the probability of default of Iranian firms
5.3 Contribution of study
5.4 Research Limitation
5.5 Future Research Recommendations
LIST OF TABLES
2.1 Emperical Evidence from Developed, Emerging and Devel- oping Markets
2.2 Summary of Significant Financial Ratios to Probability of Default
2.3 Financial Ratios
2.4 Summary of Empirical Evidence on Firm-specific and Macroeconomic Determinants of Probability of Default
3.1 Formulation of Independent Variables
3.2 Summary of Default Prediction Techniques
3.3 Contingency Table or Confusion Matrix
3.4 An overview of Performance Metrics for Probability of Default
3.5 Summary of main reasons for the different sign problem
4.1 Summary of Descriptive Statistics of Probability of Default Determinants on Overall Dataset
4.2 Summary of Descriptive Statistics of Probability of Default Determinants on Chemical products Industry
4.3 Summary of Descriptive Statistics of Probability of Default Determinants on Cement Industry
4.4 Summary of Descriptive Statistics of Probability of Default Determinants on Food Industry
4.5 Summary of Descriptive Statistics of Probability of Default Determinants on Sugar Industry
4.6 Summary of Descriptive Statistics of Macroeconomic Determinants
4.7 Correlation Matrix of Firm-specific Determinants of Proba- bility of Default
4.8 Hosmer and Lemeshow Test
4.9 Correlation Matrix from Principal Component Analysis
4.10 Eigenvectors for selected variables from Principal Compo- nents Analaysis
4.11 Maximum likelihood estimates of regression coefficient for logistic regression (without industry effect)
4.12 Detailed accuracy for logistic regression
4.13 Maximum likelihood estimates of regression coefficient for logistic regression (with industry effect)
4.14 Detailed accuracy for logistic regression
4.15 Maximum likelihood estimates of regression coefficient for logistic regression and expected sign
4.16 Correlation Matrix of Significant Firm-specific Determinants of Probability of Default
4.17 Maximum likelihood estimates of regression coefficient for logistic regression after eliminating Ca/CL
4.18 Detailed accuracy for modified model
4.19 Maximum likelihood estimates of the logistic regression parameters are used in classifying default and non-default firms across industries
4.20 Summary of Results of Hypothesis Tests
4.21 Comparison of models in classifying default and non-default firms across industries
4.22 Contingency Table
4.23 Classifying default and non-default firms using support vector machine
4.24 Comparison of models in classifying default and non-default firms among industries
4.25 Maximum likelihood estimates of the logistic regression parameters are used in classifying default and non-default firms (Y=1) across industries before and during crisis
4.26 Comparison of models in classifying default and non-default firms across industries before and during crisis
4.27 Comparison of Adaboost ensemble in classifying default and non- default firms across industries before and during crisis
5.1 The most significant determinants of probability of default across industries
5.2 Ranking of classification models for classifying firms to default and non-default
LIST OF FIGURES
1.1 The trend of GDP during 2004-2014, Iran
1.2 The trend of inflation rate during 2000-2014, Iran
1.3 The trend of interest rate during 2000-2014, Iran
1.4 Industry Share of GDP of Iran
2.1 The probability of default distribution (Guthner, 2012)
2.2 Loss Given Default distribution ((Guthner, 2012)
2.3 Theoretical framework
3.1 A graphical research design
3.2 Tehran Stock Exchange’s De-listing procedure (TSE.ir)
3.3 A graphical depiction of the methodology
3.4 Structure of a decision tree (Han and Kamber, 2006)
3.5 Single layer neural network (Han and Kamber, 2006)
3.6 Multi-layer Perceptron neural network (Han and Kamber, 2006)
3.7 Support Vector Machine
3.8 The framework of AdaBoost algorithm
3.9 Proportion of Models Employed by Past Studies (Aziz and Dar, 2006)
3.10 10-fold Cross Validation (Han and Kamber, 2006)
3.11 ROC curve (Han and Kamber, 2006)
4.1 Probability of Default Indicators within overall data, default (blue marks) and non-default (red marks)
4.2 Probability of Default Indicators within overall data, default (blue marks) and non-default (red marks)
4.3 Probability of Default Indicators within overall data, default (blue marks) and non-default (red marks)
4.4 Visualized data for classifying default (blue marks) and non- default (red marks)
4.5 ROC curve for final models of classifying default and non- default firms (without industry effect)
4.6 ROC curve for final model of classifying default and non- default firms (with industry effect)
4.7 ROC curve of Logistic Regression model of classifying default and non-default firms for chemical products, Cement, food and Sugar industry.
4.8 ROC curve of CART model of classifying default and non- default firms for chemical products, Cement, food and Sugar industry.
4.9 ROC curve of Random Forest model of classifying default and non-default firms for chemical products, Cement, food and Sugar industry.
4.10 ROC curve of Neural Network model of classifying default and non-default firms for chemical products, Cement, food and Sugar industry.
4.11 ROC curve of Support Vector Machine model of classifying default and non-default firms for chemical products, Cement, food and Sugar industry.
4.12 Z-plot obtained from support Vector Machine
4.13 Classification Error for Single Classifiers, Logistic Regres- sion (LR), CART Classification Tree (CART), Random Forest (RF), Neural Network (NN) and Support Vector Machine (SVM)
4.14 Classification Error for Single Classifiers, Adaboost Logistic Regression (Ad-LR), Adaboost CART Classification Tree (Ad-CART), Adaboost Random Forest (Ad-RF), Adaboost Neural Network (Ad-NN) and Adaboost Support Vector Machine (Ad-SVM)
LIST OF ABBREVIATIONS
Abbildung in dieser Leseprobe nicht enthalten
First and foremost I would like to express my special appreciation and thanks to my supervisor Dr.Suresh Ramakrishnan. He has been a tremendous mentor for me. I would like to thank him for encouraging my research and for allowing me to grow as a research scientist. The good advice, support and friendship of my second supervisor Dr. Mahmoud Bekri have been invaluable on both an academic and a personal level, for which I am extremely grateful. I appreciate all his contributions of time and ideas, his advice on both research as well as on my career, which have been invaluable. It is a great honour for me to become one of his students.
My special thanks to Prof. Dr Zainab Khalifah, Dean of Business and Management department, who never failed to support and encourage me during my PhD journey. I am profoundly grateful to Dr.Sharollah bin AbdulWahab and Prof Madya Dr.Khalil Bin MD Nor, for all their support. I also like to express my gratitude to Prof Dr. Wolf-Dieter Heller for the opportunity to visit Karlsruhe Institute of Technology and his good advice and support to improve my research knowledge during my stay in Germany. I am very grateful to all the staff members in the department of Management, UTM.
An especial thanks to my family. Words cannot express how grateful I am to my mother, for all of the sacrifices that she have made on my behalf. Her prayer for me was what sustained me thus far. I would also like to thank my beloved Mahmoud for being supportive and compassionate through out this long jouney. This thesis would have remained dream without your constant support and encouragement. I would also like to thank my lovely sisters Jamileh and Galavij, and my great brother, Payam. I cannot thank you enough for encouraging me throughout this experience.
Other past and present group members that I have had the pleasure to work with or alongside of are my colleagues at the Faculty of Management, UTM; Muhammad Naveed, Hishan, Parvaneh, my lovely friends Rokhsare and lili. I am thankful for their kind comments, discussion, encouragement and the pleasant working environment.
Finally, I thank my God, for helping me through all the difficulties. I have experienced his guidance day by day. Thanks my Lord.
Corporate default prediction has acquired prime importance in academic research, business practice and government regulation. As a result of internal and external economic shocks, unexpected corporate insolvencies had lead to severe damage to the economy. This highlights the crucial importance of an accurate corporate default prediction model. This study aims to obtain new insights on how to establish and validate predictive probability of default models. The first section examines the industry impact on default prediction. It is plausible that probability of default may differ for firms, due to distinctive nature of each industry. The second section deals with proposed key idea to assess and compare different default prediction models among Iranian listed firms across industries to examine how good a real-life probability of default can be predicted. Finally, in order to mitigate the severe consequences of the global financial and economic crisis, the current study underlines the differences of default prediction determinants during different economic periods (pre- and during global financial crisis periods). In order to achieve the proposed objectives, logistic regression and four supervised models were employed. This includes Decision Tree, Neural network, Support Vector Machine and ensemble Adaboost classifiers. A variety of different performance metrics were applied to investigate the models accuracy. The findings of the current study are as follows: In the course of investigating the industry’s characteristics, the indirect impact is clearly traceable due to changing of sign and magnitude of determinants across industries. The results reveal that supervised models yield higher performances than traditional linear techniques. The use of different data mining techniques improves the prediction power of the models to forecast probability of default across industries. The mechanism between variables and the probability of default is dependent on economic conditions of the country. The results indicate that the impact of different economic periods varies across industries. In addition, the results of this study may be valuable for any financial institution performing credit risk models to estimate their minimal capital requirements and to reduce the costs of risk management.
Ramalan lalai korporat telah mencapai kepentingan utama dalam penyelidikan akademik, amalan perniagaan dan peraturan kerajaan. Hasil daripada kejutan ekonomi dalaman dan luaran, insolvensi korporat yang tidak dijangka telah membawa kepada kesan buruk kepada ekonomi. Ini menunjukkan betapa pentingnya kesan daripada model ramalan lalai yang tepat. Kajian ini bertujuan untuk mendapatkan wawasan baru untuk membentuk dan mengesahkan kebarangkalian model ramalan lalai. Bahagian pertama adalah mengkaji kesan industri pada ramalan lalai. Ada kemungkinan bahawa kebarangkalian lalai itu mungkin berbeza bagi firma yang telah bertapak di dalam industri yang berbeza kerana sifat tersendiri bagi setiap industri. Bahagian kedua pula memperkatakan cadangan idea utama untuk menilai dan membandingkan model- model ramalan lalai yang berbeza di kalangan firma-firma Iran yang tersenarai dan mengkaji sejauh mana kebarangkalian sebenar mungkin boleh diramalkan. Akhir sekali, untuk mengurangkan kesan buruk krisis kewangan dan ekonomi global, kajian semasa menggariskan perbezaan penentu ramalan kegagalan berdasarkan tempoh ekonomi yang berbeza iaitu sebelum dan semasa tempoh krisis kewangan global. Bagi mencapai objektif yang dicadangkan, regresi logistik dan empat model telah dijalankan. Ini termasuk,(decision tree, neural network, support vector machine dan ensemble adaboost classifiers). Oleh itu, pelbagai metrik prestasi yang berbeza telah digunakan untuk menyiasat ketepatan model itu. Dapatan kajian semasa adalah seperti berikut: Dalam usaho menyiasat ciri-ciri industri, kesan tidak langsung adalah jelas dikesan akibat perubahan tanda dan magnitud penentu di kalangan industri. Keputusan menunjukkan bahawa model diselia menghasilkan keputusan yang lebih tinggi daripada teknik linear tradisional. Penggunaan teknik perlombongan data yang berbeza memperbaiki kuasa ramalan model untuk meramalkan kebarangkalian lalai di kalangan industri. Mekanisme antara pembolehubah dan kebarangkalian lalai adalah bergantung kepada keadaan ekonomi di negara ini. Keputusan menunjukkan bahawa kesan daripada tempoh ekonomi yang pelbagai adalah berbeza di kalangan industri. Di samping itu, hasil kajian ini mungkin bermanfaat kepada mana-mana institusi kewangan dalam melaksanakan model risiko kredit untuk menganggarkan keperluan modal minimum mereka dan untuk mengurangkan kos pengurusan risiko.
CHAPTER 1 INTRODUCTION
”Risk varies inversely with knowledge.”
-Irving Fisher (American economist, 1867-1947)
1.1 General overview
In the face of internal and external economic shocks, the issue of credit risk modelling has become a crucial topic among financial institutions, banks and regulators. These crises have severe impact on the banking sector of developed, emerging and developing markets (World bank, 2010). Recalling that risk is sum of two parts, namely; volatility and sensitivity. However, these are not under discretion of banks. Many banks went bankrupt and many are in distress due to their sensitivities to financial risks enlarged by the crisis (Eken, et al., 2012). In line with economic and financial turmoil, the issue of credit risk has received serious attention in the banking sector across the world.
As the corporate firms are mostly based on external financing, one of the key decisions, which lending institutions have to make is, how to establish the assessment metrics while issuing loan to firms. Suffice to say, the firms losses and financial distress is an important issue among all businesses. Consequently, its rate of occurrence in the aggregate have attentive impact on the outcomes of economic growth. Evaluating the probability of a company being able to pay back its financial obligations is of fundamental significance to providers of capital, academics and economists as well. In this context, it is important for banks to predict the potential loss of new loans in order to determine the minimal required capital to act as a safety cushion in case of companies’ defaults.
The entries and exists of companies in today’s corporate world is considered as a general phenomenon in economic system of countries. The global business world has increased competition among all the stakeholders. As a result, every firm is struggling to gain competitive advantage to ensure its long-term breadth and solvency. Credit risk models attempt to predict the sentiments of firm’s default. If used correctly, credit risk models are valuable tools for any credit manager to assess risk and make good financial decisions or avoid potential losses. Thus, credit risk is one of the most serious and significant of all risks, which banking system may face. It can be definite as the potential for loss due to default of a borrower to meet its contractual obligation to clear a debt in accordance with the established terms in contract. The importance of model accuracy and credit risk management is rapidly growing. The regulatory changes and risk evolving management practices have led to banks and other financial institutions to look at this credit risk prisma more closely.
1.2 Background of study
Default is one of the most important events in a company’s life which can place stakeholders in financial trouble. The company’s default gives great impact to the stakeholders. Probability of default (PD) prediction is one of the important tasks of rating agencies in credit risk assessment as well as of banks and other financial companies to measure the default risk of their counterparties.
After the recent firm bankruptcies, such as Enron and Worldcom, in the United States, the risks involved in corporate liability and bankruptcy prediction has become a very important concern for the various stakeholders in firms including shareholders, managers, creditors and business partners, as well as government institutions accountable for maintaining stability of financial markets. Similarly, in 2008, Lehman Brothers collapsed under its debt load of 144 billion dollars. A good account of the collective damage can be found in Standard’s and Poor (2011). Between 2007 and 2011, a total of 496 rated mostly US institutions have defaulted, representing over one trillion dollars of debt outstanding and dwarfing anything seen previously. The impact has been felt not only across the corporate and sovereign but also on the consumer side. Almost like an overreaction, these gloomy conditions led banks to suppress lending on a large scale, which resulted in the credit crunch. Questions may arise as to why all this happened. But perhaps more important than the detection of one potential cause of the crisis is the resolve to design devices that could prevent such distress from spreading further. It is essential to avoid any repetition in the future. According to Stiglitz (1998), the fear of the latter may be founded in the light of the seemingly increasing occurrence of financial crisis. But, in addition to crises that usually catch most of the world off-guard, another peril that mostly goes unnoticed is caused by the increasing amount of debt. It potentially causes harm in a more continuous way. In the worst case, it consequences in evermore powerful shocks.
To guarantee a functioning lending system, it is imperative to rein in the possible damage. It might result from a firm’s default in repaying the debt in the predetermined manner. To do so, strong and capable methods have to be introduced and reinforced to direct the lender’s attention to the inherent risks of firm’s debt positions. These evidences clearly show that credit risk is one of the most (if not the most) important issues which all businesses are struggling with.
Since the study of Fitzpatrick (1932), default prediction becomes a challenging issue in corporate finance. In this study, the value of financial ratios was compared to recognize the firm status. In the beginning, most of the studies on default risk focused on firm-specific indicators including financial ratios as a predictor of firms default across United States (FitzPatrick, 1932; Smith and Winakor, 1935; Merwin, 1942; Chudson, 1945; Jackendoff, 1962; Meyer and Pifer, 1970;Deakin, 1972; courtis, 1978). Smith and Winakor (1935) in a study of 183 failed firms from a variety of industries, found that Working Capital to Total Assets was a far better predictor of financial problems than both Cash to Total Assets and the Current Ratio. In 1962, Jackendoff compared the ratios of profitable and unprofitable firms. He reported that the following two ratios are higher for profitable firms than for unprofitable firms: the Current Ratio and Net Working Capital to Total Assets. There has been some fluctuation in the range of the number of firm-specific indicators used in the studies over the last 40 years. However, the most common indicator among previous studies, is the ratio of Net Income to Total Assets. The second most common determinant is the ratio of Current Assets to Current Liabilities (Refer to Table 2.2). Hence, majority of the studies in past literature, use the firm-specific variables. However, due to the relationship between general economic and bankruptcy rates, later some attempts have been made to predict default based on macroeconomic variables. In this regard, some macro-economic indicators have been used in the past literature, including: interest rate, stock index return and GDP. Many surveys as that are conducted for U.S. firms, revealed that macroeconomic indicators affect default prediction (Merton, 1974; Black and Scholes, 1973; Rose, et al., 1982; Hennawy and Morris, 1983; Goudie and Meeks, 1991; Vassalou and Yuhang, 2004; Liou and Smith, 2007; Reisz and Perlich, 2007;Bharath and Shumway, 2008; Zhou, et al., 2012; Tinoco and Wilson, 2013; Bhattacharjee and Han, 2014). These studies also attempt to predict default using different techniques.
Based on the past literature, the initial studies related to default prediction have employed univariate models. These traditional methods of interpreting financial ratios have been used since 1930s (FitzPatrick, 1932; Smith and Winakor, 1935; Merwin, 1942; Chudson, 1945; Jackendoff, 1962). Jackendoff (1962) compared profitable and non-profitable U.S firms, using univariate technique. According to this study, the current ratio and net working capital to total assets were higher for profitable than unprofitable firms. This technique is easy to use. However, it has certain weaknesses when examining a compound matter of financial position, using just one particular ratio. According to Esmister (1972), there exist various determinants that can explain the financial status of the firm. Therefore, a single financial ratio could not represent the business default.
In conjunction with above strand of arguments, the majority of the discussion related to default prediction develops around the decisive works of Altman (1968), Ohlson (1980), Zmijewski (1984) and Shumway (2001). It was Altman (1968), who applied Multivariate Discriminant Analysis (MDA) for the first time to classify the failed and non-failed U.S firms. Researchers still use his model as a benchmark to predict firm default. Altman’s Z-score model is a linear analysis of five ratios and this score is a basis for firm classification. Besides this, Blum (1974) employed the same MDA technique for default prediction, some years prior to failure. Similarly, to assess the predictive accuracy of accounting ratios, Libby (1975) measured the prediction achievement of a selected set of accounting ratios for U.S firms. In the past literature, there are numerous studies, which applied this method as a benchmark for default prediction in U.S. firms (Deakin, 1972; Altman, 1973; Benishay, 1973; Blum, 1977; Norton and Smith, 1979; Rose et al., 1982; El Hennawy and Morris, 1983; Taffler, 1984; Gilbert, Menon, and Schwartz, 1990; Goudie and Meeks, 1991; Hellegiest, et al., 2004).
In a subsequent study, Ohlson (1980) introduced the Logit technique in default prediction, using firm-specific indicators for the first time. He noted that the analytical power of any model relies on the availability of financial information (financial ratios, especially) is assumed to be available. Some other researchers employed Ohlson’s O- score model (Mensah, 1984; Gentry, et al., 1985a; Hellegiest, et al., 2004). Likewise, Zmijewski (1984) employed probit analysis, which is a static approach for default prediction in U.S firms. These studies have been done in U.S to evaluate default prediction determinants (firm-specific), using static methods.
On the contrary, Shumway (2001) introduced a dynamic analysis of corporate default. In general, default prediction models were successful in classifying companies as ”failed” or ”non-failed”, using multivariate statistical techniques. Alternatively, artificial methods were applied to the default prediction. According to Atiya (2001), many banks and financial institutions developed this method for default prediction. In the past literature, some studies applied support vector machines to the prediction of default such as, (Hardle, et al., 2005; Chen, et al., 2011; Shiyi, et al., 2011). In addition, these studies mostly focused on default prediction for U.S firms.
A number of studies have diverted their attention to other developed countries. In conjunction with that, Taffler (1984) highlighted the need for separate models for manufacturing and distribution companies in UK. Alternatively, Boritz, et al. (2007) examine the bankruptcy prediction models for Canadian firms. In another study, Boritz et al. (2007) designed the default model for Canadian firms. Similarly, Appiah and Abor (2009) used relevant financial information of private failed and non- failed manufacturing firms in Great Britain (UK) to determine default. They predict default for UK firms by developing a Z-score model, employing firm-specific and macroeconomic variables. Furthermore, some studies were conducted based on the specific non-US developed countries (Levallee and Altman, 1980; Izan, 1984; Keasey and Watson, 1986; Legault and Vronneau, 1986; Peel and Peel, 1987; McNamara, et al., 1988; Messier and Hansen, 1988; Liou and Smith, 2007, Appiah, 2009; Chen, et al., 2011; Gordini, 2014).
Another strand of the empirical literature focused on individual country studies within the emerging and developing nations, i.e., Skogsvik (1990), Sweden; Laitinen (1991), Finland; Altman, et al., (1994), Italy; Kiviluoto (1998), Finland; Unal (1988), Turkey; Bandyopadhyay (2006), India; Yap, et al., (2010), Malaysia; Chen (2012), Taiwan; Moradi, et al. (2013), Iran. These studies further demonstrate the exclusivity of the default prediction models that vary across nations due to their different business environments. For instance, Sandin and Porporato (2007) estimated the usefulness of financial ratio analysis to predict bankruptcy in a period of stability of an emerging economy, such as the case of Argentina in the 1990s. Yildiz and Akkoc (2010) provided an alternative for Turkish bank bankruptcy prediction with Neuro Fuzzy. In a similar work, Alifiah, et al., (2011) conducted a study to predict default risk in the development sector in Malaysia based on accruals and financial ratios. They found debt ratio as the most practical ratio for the prediction of financial distress companies in the development sector in Malaysia. Consistent with past literature, it seems that default prediction determinants vary across different countries and based on different techniques, which by previous researchers have been employed (Sandin and Porporato, 2007; Yap, et al., 2010; Yildiz and Akkoc, 2010). According to Wennekers et al. (2005), in cross country equations, it is important to control for the level of country development as this affects the likelihood of starting a business, the institutional environment and insolvency. Therefore, the level of development is a factor which may differentiate the probability of default across countries. Consistent with above spat of discussion, the subsequent section provides a comparative overview of emerging and developing economies.
The economists classify countries around the world based on their level of economic and industrial development. The emerging economy was a term coined by World Bank economist Agtmael (1981) in reference to nations undergoing rapid economic growth and industrialization (World Bank, 2010). The level of foreign investment is also critical for an emerging economy. Generally, an increased foreign investment implies that the economy has potential. The injection of foreign currency into local economy helps long-term investment in the industrial sectors (World Bank, 2010). Moreover, the emerging economies involve structural and policy reforms, and capital market development. On the other hand developing economies, also called less-developed countries, are markets with under-developed industrial base and Human Development Index (HDI) and have low Gross national Income (GNI) per capita relative to emerging and developed economies (World Bank, 2012). The fundamental difference is that emerging economies are growing rapidly and becoming more important on the world economic stage, while developing markets struggle in comparison and still need help from trade partners around the world. Furthermore, the emerging economies differ from developing countries in that they have made impressive gains in infrastructure and industrial growth, and are experiencing increasing incomes and quick economic growth (Economy Watch, 2010). Based on country’s national income as well as the development of its market infrastructure, the Financial Times Stock Exchange (2011) has further classified markets into advanced emerging markets and secondary emerging markets. Economy Watch (2010) provided another classification that emerging economies tend to experience a shift from agricultural to industrial and services sectors. Whereas, the agriculture sector is often seen as a vital component of developing economy’s GDP.
1.3 Background of problem
In corporate finance, credit risk is an essential area of research. There exists a broad market interest in disaggregating the components of credit risk. In the viewpoint of background of study, most of the research on default prediction to date has remained focus to firm-specific indicators, which have the power to influence the probability of default of the firms. Another strand of literature concentrated on macroeconomic factors and institutional differences across developed and emerging economies, which affect the firms’ capability of repayment (Rajan and Zingales, 1995; Booth et al., 2001). Hence, most of the research concentrates on the effect of firm- specific and macroeconomic indicators on probability of default. This argument is debatable as other factors related to the industry may affect the probability of default determination. For the financial and economic development of any country, the industries are considered as generators of economic growth. They provide obstinate base, which is essential for the sound financial structure of the firms. It is obvious that the firms are related to the industries, which tend to have distinctive nature. Every industry is subject to different level of maturity, growth, competitive environment, riskiness and facing different challenges. Hence, higher level characteristics of industries may differently affect the firms’ characteristics. Eventually, it may affect firms capability to repay their obligations. Accordingly, the industry characteristics are crucial in explaining the risk of business (Kale, et al., 1991). According to Nishat (2001) industry plays an important role in explaining the risk volatility as firms are related into industries, which are supposed to take different level of riskiness.
Kale, et al. (1991) argue that industry characteristics are important in estimating the risk of business. A single information can affect the industries differently across whole market. Therefore, industries tend to have different issues and challenges which, could differently influence the default probability of firms. For that reason, industry plays significant role in explaining the national market volatility (Nishat, 2001). According to Nishat (2001), financial reforms and changing industry specific policies may cause the change in industries risk level. Furthermore, Grinold, et al. (1989) argue that some industries are internally more volatile than others. Thus, level of riskiness may vary due to the industries’ structure. Moreover, the major industries are more likely to raise financing. According to Opler and Titman (1994), the industrial environment affects the firms’ performance. They revealed that by identifying the riskiness, changes and investment opportunities in their operating environment, firms can generate more value and financial returns. So, firms operating in more competitive industries tend to face more dynamic, uncertain and complex environment. Regardingly, Kayo and Kimura (2011) state that the firms tend to have similar properties that operate within a particular industry. Therefore, it is expected that they have similar environment. Likewise, Ramakrishnan (2012) highlighted the sectors’ importance to capital structure decision making of Malaysian firms. He argues that the sector’s behavioural affects the firms’ leverage mechanism. It is demonstrated that firms are normally very dependent on their internal funds as their main source of financing but as the internal resources reserves, they will rely on debt financing and, subsequently, external equity as a last resort for financing. Many studies evident that firms follow leverage targets ( Graham and Harvey, 2001; Fama and French, 2002; Flannery and Rangan, 2006). It can be said that any wrong decision about capital structure may lead the firm to financial distress and eventually to bankruptcy (Eriotis, et al., 2007). Thus, the impact of external factors should be considered. Firms operate under different industries, with specific level of dynamics, growth and competitive environment. Moreover, the policies and reforms vary across industries. Therefore, exploring how the specific environment of each industry could differently affect the probability of default of firms is an essential issue. In line with these arguments, it is crucial to figure out the significant effect of each industry’s distinctive environment on firms’ probability of default.
In conjunction with the above argument, Kale, et al. (1991) proposed that the intra industry variation in financial structure in competitive industries is well explained by industry-specific factors. He argues that the industry characteristics are important in explaining the risk of business. According to Kayo and Kimura (2011) firms tend to have similar properties that operate within a particular industry. A number of empirical evidence highlights the impact of industry on probability of default. In this regard, Opler and Titman (1994) and Maksimovic and Phillips (1997) support the significance of industry effects on probability of default. Likewise, Berkovitch and Israel (1998) model the decision to reveal default as a strategic variable. They show that the fraction of firms that go default is higher for firms operating in mature industries than firms in growth industries. The reason could be higher under-investment problems in growth industries in compare with mature industries. In related research, Lang and Stultz (1992) reveal competitive intra-industry effects of bankruptcy announcements. According to Acharya et al. (2003), industry conditions at the time of default affect the recovery rate. In general, there are variations of behaviour across industries that may indirectly affect the probability of default. This impact, indirectly provides insights about the nature of industry and its impact on the firm capability of repayment. Though, few past studies have used dummy variables to characterize the industry. Such techniques do not provide a clear depiction which shows the existent effect of a particular industry (Kayo and Kimura, 2011). Moreover, these studies do not take into account institutional settings within developing countries.
In the context of developing countries, the importance of an industry on probability of default is under-explored. A number of studies tend to ignore the significance of industry in their model specification. In addition, researchers face problem in establishing the industry-specific variables due to data limitations across developing countries. Therefore, most of the past studies removed the industry effect by including an industries dummy (Chava and Jarrow, 2004). Based on the previous discussion, the argument between firm-specific and industry effect is indecisive across developing countries. However, there exist enormous institutional differences. As well, the effect of sectoral behaviour on probability of default determinants may differ across different markets. Despite that, the unique behaviour of each industry varies between countries due to different financial settings. It is affirmed that, the lack of developed bond markets is often one of the reasons for the intensity of the financial crisis across developing countries. As the financial system in most emerging and developing economies is centred on banks, an important aspect of the development of bond markets is the impact on the banking system. Since the Asian financial crisis of 1997-98, attention has increasingly focused on the relative roles of the banking sector and of the capital market in developing economies. In many instances, the domestic bond market, where it exists, is generally under-developed, in both breadth and depth, compared to the banking system and the equity market. It has been argued that, over- reliance on bank lending for debt financing exposes an economy to the risk of a failure in the banking system.
In conjunction with repeated debt crises over the last three decades, the risks of excessive reliance on foreign currency borrowings have been exposed by developing countries. To avoid these risks, there is a strong case for governments to raise long- term resources from the domestic debt market. The government has been recognized as a key player in the initial stage of bond market development in its role as an issuer, regulator and promoter in developing markets. Additionally, regulation can take many forms and the form of regulation policy adopted in developing countries has shifted over time (Minogue, 2005). From the 1960s to the 1980s, market failure was used to legitimize direct government involvement in productive activities in developing countries, by promoting industrialization through import substitution, investing directly in industry and agriculture, and by extending public ownership of enterprises. Moreover, the government economic policy affects the level of growth and opportunities of different industries which may affect the firm’s probability of default operating in different industries. Accordingly, these arguments are highly related to one of the developing countries, Iran, due to its unique financial settings.
Consistent with the above argument, Iran has its own unique financial characteristics, capital market, and political and economical ties (Noravesh, et al., 2007). Being in developing markets, Iranian firms are often exposed to different financial crisis, i.e., bankruptcy, manipulation of resources and agency conflicts due to the lack of stakeholders’ protection (Yahyaabadi et al., 2013). By controlling the majority of businesses in Iran, the Iranian government has made significant efforts to expand the capital structure. According to International Monetary Fund (IMF) (2011), Iranian banking system is unique in that all banking activities must follow the principles of Islamic Law (Sharia). Moreover, Islamic banking in Iran is regulated by the law of 1983 on interest and free banking. Whereas, other countries hosting Islamic banking are calling on regulatory level to introduce provisions for specific requirements of Sharia, especially the prohibition of interest. Financial assets that comply with the Islamic law can be classified in accordance with their tradability and non-tradability in the secondary markets. A market which, has not yet developed in Iran, as it has in Malaysia and Gulf Cooperation Council (GCC) countries. This unique system may affect the firm’s capability to repay their obligations, due to the government policies and lending system in Iranian banks. According to Dalal-Clayton and Bas (2009), the economic environment of developing countries is very different from the developed countries. As a developing country, Iran, has traditional market economy. By the end of the 20th century, Iran’s economy faced many obstacles, including: market forces, global financial crisis and international sanctions (Ebrahimi, 2009). Notwithstanding, the international sanctions relating to nuclear program of Iranian government, as well as the global financial crisis in 2008, the value of Iranian companies has kept the growth (Yahyaabadi et al., 2013). According to International Monetary Fund report of 2010 and World Bank statistics of 2011, Iran’s economy was the eighteenth largest economy in the world with regard to purchasing power parity. The financial factors (i.e., growth, profit and tax rates) made differences in the data set of Iran in comparison with developed countries (Safdari, 2009). Over the past three decades, an increasingly changing market-oriented corporate sector has driven growth in the Iranian economy. According to economic report of 2009, the annual growth rate of industrial production in Iranian companies was ranked 39th in 2008.
Additionally, Iran is the second largest economy in the Middle East and North Africa (MENA) region after Saudi Arabia, with an estimated Gross Domestic Product (GDP) of USD 366 billion in 2013-14 (Figure 1.1). Iran ranks second in the world in natural gas reserves and fourth in proven crude oil reserves. Aggregate GDP and government revenues still depend to a large extent on oil revenues and are therefore intrinsically volatile. In recent years, Iran’s economic growth has been hampered by double-digit rates of inflation. Although high inflation is widespread among the oil- exporting countries in the Middle East and Central Asia, Iran has one of the highest rate (IMF, 2008). Iran’s average Consumer Price Index (CPI) inflation level was above 25 percent at year-end 2008 (Figure 1.2). Through 2009, the CPI inflation level dropped, but remained above 13 percent. As a developing economy, Iran tends to have volatile business cycles and experience economic crises more frequently than developed economies. According to Neumeyer and Perri (2005), emerging and developing economies face volatile and highly interest rate. In Iran, the central bank of Iran does not use the benchmark interest rate. Instead, the Central Bank of Iran sets the Bank Profit rates for lending and borrowing. Figure 1.3 shows the trend of interest rate of Iran during 2000-2014.
Abbildung in dieser Leseprobe nicht enthalten
Figure 1.1: The trend of GDP during 2004-2014, Iran
Abbildung in dieser Leseprobe nicht enthalten
Figure 1.2: The trend of inflation rate during 2000-2014, Iran
Abbildung in dieser Leseprobe nicht enthalten
Figure 1.3: The trend of interest rate during 2000-2014, Iran
The global financial crisis is set to depress oil producing economies. As the crisis is already pushing down oil prices, a firm response to the fallout of the crisis from governments and central banks is expected. Though, the Iranian government does not seem to be greatly concerned about the ongoing global situation, although many Iranian economists believe that sanctions and the international financial crisis have taken their toll on Iran’s economy by unfavourably affecting oil, trade, and trade financing. Even if the relative isolation from the world’s economy may seem to protected Iran from the negative impact of the global financial crisis to a certain extend, plunging oil prices and a massive credit deterioration suggest otherwise. As one of the biggest oil producing countries, Iran has become increasingly dependent on oil earnings since the 2005. Therefore, it is plausible to assume that Iranian industries are affected by the global financial crisis due to the highly dependency of Iran’s economy on oil income and as a result of oil prices plunging .
It can be said that, the mixed economic performance of the Iranian economy has resulted from several domestic and external factors: (1) Iran is a country where 50% of the economy is centrally planned. The Iranian government policy affects macroeconomic factors, unemployment rate, the inflation rate, and the rate of economic growth. Hence, the Iranian state continues to play a main role in the economy. However, owning large public enterprises, which partly dominate the manufacturing and commercial sectors. The financial sector is also dominated by public banks. Moreover, the business environment remains weak with the country ranking 152nd out of the 189 countries (Doing Business Report, 2014). Between 1999-2000 and 2012-2013, the Iranian economy and in particular the banking industry went through considerable ups and downs. (2) Iran’s economy relies heavily on oil revenues. Thus, aggregate GDP and government revenues still depend to a large extent on oil revenues, and are therefore intrinsically volatile. This revenue has been used to implement a range of policies (i.e., government subsidies). According to the economic report 2011, Iran has been ranked 39th for producing $23 billion of industrial products in 2008. From 2008 to 2009 Iran has leaped to 28th place from 69th place in annual industrial production growth rate. The government of Iran has plans for the establishment of 50-60 industrial parks by the end of the fifth Five-Year Socio- economic Development Plan by 2015. In addition, the industries have a significance role in the Iran economy with the highest share of GDP, 44.47%. (3) Economic activities suffered from a high level of uncertainty that was caused by government’s monetary and fiscal policy. This resulted therefore, in large fluctuations in the real rate of growth of GDP. Moreover, a large number of public banks suffered from a massive level of over-due loans and pushed them to the edge of bankruptcy. The devastating effect of the high level of uncertainty can still be felt not only in the banking sector but also across non-financial industries.
1.3.1 Economic groups of Iran
According to the article 44 of the Constitution of Iran, the economy of Iran consists of three sectors: (1) the state, (2) the cooperative and (3) the private sectors. Accordingly, the growth and development of Iranian firms are strongly related to their organizational sector, economical state and governmental system. For a country’s economy, non-financial sector is an important segment (Noravesh, et al., 2007). It provides steady and healthy industrial base, which is essential for economic well- being and populace of a country. According to Central Bank of Iran, the non-financial sector in Iran represents a diversified nature of economic groups, which include; oil and gas, chemical products, steel, textile, and automotive manufacturing accounted for an estimated 45 percent of the Iran’s GDP. The services sector, including financial services, represented about 44 percent of Iran’s economy. Agriculture constituted about 11 percent of Iran’s economy. Iran’s economic sectors remain heavily dominated by the state, but there are some privatization efforts under way. Figure 1.4 depicts the trend of industry share of GDP of Iran. As it can be observed, the industry share of GDP increases after 2000, which affirms the importance of industries in GDP growth of Iran during last decade.
Abbildung in dieser Leseprobe nicht enthalten
Figure 1.4: Industry Share of GDP of Iran
Under the new privatization act, the ownership and management of most major industries, including upstream oil and gas, banking and insurance, power distribution, radio and television, aviation, harbor, water distribution, and military production, will remain with the government. The privatization of the economy and acquisition of public assets by private investors will undoubtedly require a robust bankruptcy and insolvency regime entailing the rights and obligations of creditors and debtors. Recently, the government handed out loans to private enterprises. A number of these private companies in different industries went default due to financial hardship or economic crisis i.e., international sanctions, government policies and interest rate.
For years, the Iranian economy has suffered from international isolation for pursuing a nuclear program, and the international community, including the United Nations (UN), the United States, and the EU, has imposed severe economic and financial sanctions against Iran. Many Iranian companies became bankrupt, because of the rising costs of doing business and the inability to export goods to international markets. Furthermore, Iranian businesses could not access foreign currency or open letters of credit that are needed for conducting everyday banking transactions. In sum, many factories and production units have ceased operations, and Iranian businesses such as construction, cement, textile, agriculture, carpet, chemical products and food industries have suffered setbacks from the international sanctions (Yahyaabadi et al., 2013). Furthermore, Iranian government subsidies -particularly on food and energy- have influenced the national economy for more than 30 years. Recently, the government decided to reduce subsidies (Majlis, 2010). The cut of subsidies have a dramatic effect on many Iranian industries. Moreover, a junction of inappropriate factors including global financial crisis, political instability, high interest rate and high inflation consequences in slow Gross Domestic Product (GDP) growth of Iran. This risky situation differently influenced the Iranian firms across industries.
The government had been the dominant force in the operation of major industries, including oil, gas and petroleum, telecommunications, banking and insurance, automotive and transport, and construction. Despite the approval of privatization program in the first (1990-1995), second (1995-2000) and third (2000- 2005) plans. The privatization of the economy and acquisition of public assets by private investors will undoubtedly require a robust bankruptcy and insolvency regime entailing the rights and obligations of creditors and debtors.
According to Noravesh, et al. (2007), Iranian industries are heavily dependent on bank financing, which has been shown to affect the investment decisions of firms. Consequently, during uncertain macro environment, banks are resistant to advance long-term loans to the private industries, and often retreat to short-term lending. However, a huge part of financial lending are allocated to the major industries due to the government policies. Therefore, the companies, which do not operate under the major industries are more probable to face default during difficulties periods. Based on significance of industries in the performance of firms, it reveals the importance to investigate the industry effects on probability of default of firms in developing countries. Keeping in view the aforementioned research gap, the literature on credit risk modeling in Iran mainly remained concentrated on the effect of financial ratios on the probability of default (Salehi and Abedini, 2009 and Moradi et al., 2013). Therefore, this spat warrants the need to examine the impact of industry characteristics on probability of default among Iranian listed firms as a developing market. Consistent with background of study and background of problem, the starting point is the argument that the literature on default prediction in the context of significance of industry effects remained untapped in developing markets. In this regard, the research investigation is likely to be more effective as this study adopts a country from the developing market, where industry significance remained unexplored in the context of default risk. Hence, this is the first research gap in this study.
- Quote paper
- Maryam Mirzaei (Author), 2015, Corporate Probability Default Prediction With Industry Effects Using Data Mining Techniques, Munich, GRIN Verlag, https://www.grin.com/document/306404