Bankruptcy prediction models on Swedish companies.

Bankruptcy prediction models on Swedish

companies

Analysis of the performance of the Altman, Ohlson and Zmijewski bankruptcy

prediction models in Sweden.

Jocelyn Charraud, Adrian Garcia Saez

Department of Business Administration Master's Program in Finance

Master's Thesis in Business Administration III, 30 Credits, Spring 2021 Supervisor: Jesper Haga

[THIS PAGE LEFT BLANK INTENTIONALLY]

Abstract

Bankruptcies have been a sensitive topic all around the world for over 50 years. From their research, the authors have found that only a few bankruptcy studies have been conducted in Sweden and even less on the topic of bankruptcy prediction models. This thesis investigates the performance of the Altman, Ohlson and Zmijewski bankruptcy prediction models. This research investigates all Swedish companies during the years of 2017 and 2018.

This study has the intention to shed light on some of the most famous bankruptcy prediction models. It is interesting to explore the predictive abilities and usability of those three models in Sweden. The second purpose of this study is to create two models from the most significant variable out of the three models studied and to test its prediction power with the aim to create two models designed for Swedish companies.

We identified a research gap in terms of Sweden, where bankruptcy prediction models have been rather unexplored and especially with those three models. Furthermore, we have identified a second research gap regarding the time period of the research. Only a few studies have been conducted on the topic of bankruptcy prediction models post the financial crisis of 2007/08.

We have conducted a quantitative study in order to achieve the purpose of the study. The data used was secondary data gathered from the Serrano database. This research followed an abductive approach with a positive paradigm. This research has studied all active Swedish companies between the years 2017 and 2018.

Finally, this contributed to the current field of knowledge on the topic through the analysis of the results of the models on Swedish companies, using the liquidity theory, solvency and insolvency theory, the pecking order theory, the profitability theory, the cash flow theory, and the contagion effect. The results aligned with the liquidity theory, the solvency and insolvency theory and the profitability theory. Moreover, from this research we have found that the Altman model has the lowest performance out of the three models, followed by the Ohlson model that shows some mixed results depending on the statistical analysis. Lastly, the Zmijewski model has the best performance out of the three models. Regarding the performance and the prediction power of the two new models were significantly higher than the three models studied.

Keywords: Bankruptcy prediction models, Ohlson model, Altman model, Zmijewski model,

Sweden, logistic regression, probit regression, AUC, Youden index, Liu index.

[THIS PAGE LEFT BLANK INTENTIONALLY]

TABLE OF CONTENTS

1. Introduction 1

1.1 Background 1

1.2 Research Gap 3

1.3 Purpose 4

1.4 Research Question 5

1.5 Delimitations 5

1.6 Contributions 5

1.7 Disposition 6

2. Theoretical Methodology 8

2.1 Preliminary Understanding and Choice of Subject of Study 8

2.2 Research Philosophy 9

2.3 Research Assumptions 11

2.3.1 Ontological Assumption 11

2.3.2 Epistemological Assumption 13

2.3.3 Axiological Assumption 14

2.4 Research Approach 14

2.5 Research Design 16

2.5.1 Methodological Choice 17

2.5.2 Research Strategy 18

2.5.3 Time Horizon 18

2.6 Literature Search 19

2.7 Source Criticism 19

2.8 Ethical and Societal Considerations 20

3. Theoretical Framework 22

3.1 Theories 22

3.1.1 Liquidity Theory 22

3.1.2 Solvency and Insolvency 24

3.1.3 The Pecking Order Theory 24

3.1.4 Profitability 25

3.1.5 Cash Flows 26

3.1.6 The Contagion Effect 27

3.1.7 Connection to Previous Studies 27

3.2 Models 29

3.2.1 Altman Z’’-score model 29

3.2.2 Ohlson model 31

3.2.3 Zmijewski model 32

3.2.3 Conclusion 33

4. Practical Methodology 34

4.1 Sample Data and Data Collection 34

4.2 Statistical Analysis 36

4.2.1 Logistic Regression 36

4.2.2 Probit Regression 37

4.2.3 Receiver Operating Characteristic (ROC) curve 37

4.2.4 Area Under the Curve (AUC) 39

4.2.5 Accuracy Ratio 40

4.2.6 Cut-off point optimization: Liu index and Youden index 40

5. Empirical Findings 42

5.1 Logit and Probit Model Estimation 42

5.1.1 Logit Model Estimation 42

5.1.2 Probit Model Estimation 43

5.2 Descriptive Statistics 45

5.2.1 Altman’s Z’’-score 45

5.2.2 Ohlson’s model 47

5.2.3 Zmijewski’s model 49

5.2.4 New logit model 51

5.2.5 New probit model 53

5.3 Receiver Operating Characteristic (ROC), Area under the Curve (AUC) and Accuracy

Ratio (AR) 56

5.3.1 Altman’s Z’’-score 56

5.3.2 Ohlson’s model 56

5.3.3 Zmijewski’s model 57

5.3.4 New logit model 58

5.3.5 New probit model 58

5.4 Cut-off point optimization with Youden and Liu indexes 59

6. Discussion of Results 60

6.1 Summary of the Empirical Results 60

6.2 Connection to Theories 62

6.3 Comparison with Previous Studies 63

7. Conclusions 65

7.1 Concluding Remarks 65

7.2 Theoretical and Practical Contributions 66

7.3 Ethical and Societal Implications 66

7.4 Truth Criteria 67

7.5 Further Research 68

References 69

Appendix 76

LIST OF FIGURES

Figure 1: Different answers to the primary ontological question on the nature of reality. 12

Figure 2: Epistemology. 12

Figure 3: Two opposing sets of answers to the axiological questions. 14

Figure 4: The three ways of approaching research. 15

Figure 5: The research “onion”. 17

Figure 6: Example of a ROC curve with AUC. 39

Figure 7: Histogram of the Z-score for sample A. 45

Figure 8: Histogram of the Z-score for sample B. 46

Figure 9: Histogram of Ohlson’s probability for sample A. 47

Figure 10: Histogram of the O-score probability for sample B. 48

Figure 11: Histogram of Zmijewski’s probability for sample A. 50

Figure 12: Histogram of Zmijewski’s probability for sample B. 51

Figure 13: Histogram of new logit model for sample A. 52

Figure 14: Histogram of new logit model for sample B. 53

Figure 15: Histogram of new probit model for sample A. 54

Figure 16: Histogram of new probit model for sample B. 55

Figure 17: ROC curve and AUC for the Altman model. 56

Figure 18: ROC curve and AUC for the Ohlson model. 56

Figure 19: ROC curve and AUC for the Zmijewski model. 57

Figure 20: ROC curve and AUC for the new logit model. 58

Figure 21: ROC curve and AUC for the new probit model. 58

LIST OF TABLES Table 1: Summary of the main research paradigms and their components. 10

Table 2: Confusion matrix. 38

Table 3: New logit regression. 42

Table 4: New probit regression. 43

Table 5: Descriptive statistics for the Altman model. 45

Table 6: Descriptive statistics for the Ohlson model. 47

Table 7: Descriptive statistics for the Zmijewski model. 49

Table 8: Descriptive statistics for the new logit model. 51

Table 9: Descriptive statistics for the new probit model. 53

Table 10: Youden and Liu indexes. 59

[THIS PAGE LEFT BLANK INTENTIONALLY]

1

1. Introduction

1.1 Background

After five years of very low bankruptcy levels in Sweden, the number of bankrupt companies have started to increase again in 2017 and have seen a sharper increase in 2018 (Tillvaxtanalys.se, 2018; Tillvaxtanalys.se, 2019). In 2018 the Swedish Agency for Growth Policy Analysis saw a jump in corporate bankruptcies (Tillvaxtanalys.se, 2019).

Corporate bankruptcies have increased by 13% in 2018 in comparison to 2017, respectively 6714 and 7599 companies filed bankruptcy in 2017 and 2018. When a company files bankruptcy it is important to keep in mind that their employees will lose their jobs, the number of employees affected by bankruptcies from 2017 to 2018 rose by 11% to 19.466 individuals. Broken down by industries, the industry which has been the most affected in 2018 by bankruptcy was the hotel and restaurant sector with almost 1.8%

of all the companies filed bankruptcy (Tillvaxtanalys.se, 2019).

From Credit safe, one of Sweden's leading credit information companies mentions that the biggest bankruptcy that happened during 2018 was in the airline sector with NextJet AB with its subsidiary NextJet Aktiebolag counting more than 858 million SEK (2018) in turnover and more than 500 employees. Following the second-largest bankruptcy with Ecotrans Kontinent AB working in the transport industry accumulated a turnover of over 550 million SEK (2018) with over 50 employees. Furthermore, it is important to note that Örebro county has seen its bankruptcy level increase by over 40% in the year 2018 (Mynewsdesk, 2021).

Within business research, perhaps one of the questions of greatest importance is the identification of the main drivers behind the demise of corporations because it concerns the very issue of their survivability, and by extension, also the welfare of the large majority of the population that directly depends on the business fabric for a living.

Bankruptcy prediction models seek to give an answer to this crucial question.

Additionally, another related question of undeniable relevance within business research, namely, that of anticipating or forecasting bankruptcies, is also directly answered by bankruptcy prediction models. A reasonable question may be raised on the accuracy of the currently available models for bankruptcy in recent times. Since the models studied in this research were created from 1968 to 1984 and the economy, regulations and companies’ financial structures were different back then.

Over the past 50 years, the topic of bankruptcy prediction has developed to become a

major research domain of corporate finance (Balcaen & Ooghe, 2006, p. 3). Bankruptcy

prediction research has been heavily studied with the main goal to develop bankruptcy

prediction models, those studies have been conducted in both developed and developing

countries from all around the world. Therefore, the academic research on bankruptcy

prediction is very extensive (Balcaen & Ooghe, 2006, p. 4). The definition of bankruptcy

from the Dictionary of Finance and Investment Terms (2014) reads “state of insolvency

of an individual or an organization, in other words, an inability to pay debts.” In the many

different legal jurisdictions across the world, each country and administration possess its

own bankruptcy law. Within corporate finance, bankruptcy is understood to happen when

2 a company files for bankruptcy within the legal system of the country in which it is established.

In the context of research, ultimately the legal definition of bankruptcy does not usually represent an issue, since, in most jurisdictions a legal, or de jure, bankruptcy entails an equivalent (if not equal) effective, or de facto, financial status of insolvency. In this way, economic reality and legal status are not in disagreement. Under Swedish bankruptcy law, an insolvent debtor shall by their own or their creditor's application be declared bankrupt unless otherwise mandated (Verksamt, 2021).

There are multiple reasons regarding this increase in interest towards the topic of bankruptcy prediction and avoidance of bankruptcy. The main reason is that when bankruptcy happens it affects multiple parties and, accordingly, has a large cost as it generates an interest in both private agents and governments.

For the private agents, this interest is mainly related to the urge to find a precise bankruptcy model to be able to prevent or eventually correct a possible failure in a firm.

The governmental interest is due to their willingness to distinguish companies that are performing badly and then be able to take corrective actions with the goal to prevent the bankruptcy of the company (Charitou et al., 2004). From previous studies, we know that a company in financial distress will see its market value considerably decrease, which will potentially severely affect its stakeholders (Balcaen & Ooghe, 2006, p. 4).

Furthermore, a company bankruptcy may harm all of its stakeholders meaning that the total cost of the bankruptcy may be large economically and socially. Meaning that if one company goes bankrupt, it can have a negative impact on its internal and external stakeholders which can be employees, debit and credit suppliers, clients and even the Government meaning that the whole economy of a country can be affected by a bankruptcy (Doumpos & Zopounidis, 1999, p. 72-73).

Balcaen and Ooghe (2006) show some other reasons for this increase in interest towards bankruptcy. Such as how the economy of developed countries has changed at a fast pace and how companies' operations have also changed with, for example, the increase in competition and companies operate more and more on an international basis (Van Caillie

& Dighaye, 2002, p 7-11). This interest could be also due to an increase in publicly available data about companies’ financial citation, which allows researchers to conduct more research and more reliable research since the quantity of data has highly increased over the years (Ball, 2016, p. 549-551).

Stoškus et al. (2007, p. 26) present that corporate bankruptcy can be caused by mainly

two reasons, the first being the effect of external factors on companies (Stoškus et al.,

2007, p.26). In their research Stoškus et al. (2007) present the effect of external factors

on a market and on bankrupt companies, starting with the effect of the overall economy

of companies (bankrupt companies in this case). An economic cycle is mainly constituted

of an expansion phase and a recession phase, during the expansion phase consumers tend

to spend more and be more confident but during the recession phase, consumers tend to

be less confident and then spend less. Which will directly affect companies, especially

companies involved in a niche market and smaller/weaker companies.

3 The second reason which could cause bankruptcy is poor management within the company. This could happen by improper financial forecasting, frauds, inexperienced management teams and not following the customer preferences (Ramana et al., 2012, p.

41). Ramana et al. (2012, p. 44) continue on developing that healthy or not companies need proper management and proper planning for their future furthermore to achieve successful growth and not bankruptcy a company needs a professional management team, a flexible organization and finally proper systems and controls.

1.2 Research Gap

As there are many different bankruptcy prediction models and each of those models has had its own results and effects on different industries, environments, and times (Kumar

& Ravi, 2007), the researchers have not found a clear validation of the three models tested in Sweden on Swedish companies. In other words, the research has not found any paper that scientifically shows that the Altman, Ohlson and Zmijewski models are accurate or able to predict bankruptcy in Sfindweden in our time (2017 and 2018 in this case).

The research problem or research gap that we have identified during the course of the literature review is that of the lack of research, to the best of our knowledge, in the field of bankruptcy prediction in our current time (2017 and 2018) for Swedish companies.

Most of the research and literature available testing the performance of one or all three models used in this research have been conducted in the USA and European countries as a whole but not Sweden specifically. There are several ways in which Sweden is different from the USA (which was the focus country for the development of the Altman and Zmijewski models). For instance, the legal system, as well as the macroeconomic environment, is very different. These differences serve as arguments in favour of conducting a new study focusing on Sweden.

Timmermans (2014), has conducted a similar research as the one conducted in this paper, using the Altman (1968), Ohlson (1980) and Zmijewski (1984) models. In his research he used two samples, the first sample has 63 bankrupt companies and the second one of 63 non-bankrupt companies, where the companies are American listed companies.

Timmermans (2014) used data from 2005 to 2007 to avoid any misleading data that could have occurred during the financial crisis of 2008. From Timmermans (2014) research, he concluded that the models were not accurate and needed recalibration which he conducted and from this recalibration, he found an increased inaccuracy of the models.

In Avenhuis (2013) research, he tested the accuracy of the Altman, Ohlson and Zmijewski

models for Dutch listed and large non-listed firms using two samples. The first sample

includes 15 bankrupt companies and 476 non-bankrupt companies which are used as an

estimation sample and the second sample includes 14 bankrupts and 326 non-bankrupt

companies which were used as a validation sample. This research used data from post-

2008 to 2010. As a result of this research Avenhuis (2013, p. 38) concluded that

practitioners should be careful using those three models especially in a different economy

and country of when and where they were created because the type I errors were high

with the Ohlson and the Zmijewski models and the accuracy rate of the Altman model

was low. And Avenhuis (2013, p. 38) recommends re-estimating the coefficient of these

bankruptcy prediction models with a specific and bigger sample to be able to increase the

accuracy of the models.

4 Altman et al. (2017) have researched the Altman z-score bankruptcy prediction model in an international context using a sample of companies from thirty-two European countries and three non-European countries (Colombia, China, and the USA). Except for the USA and China, all the companies used in the sample are privately owned and non-financial companies across all types of industries. The results from this research show that the Altman z-score bankruptcy prediction model in an international context has a high accuracy level (above 75%) and some countries stand out with an accuracy level above 90% (name them). The authors also mention that improving the variables with specific country estimation have a significant positive effect on the accuracy of the model Altman et al. (2017, p. 1). Furthermore, they add that adding other variables to the model does improve the performance and classification of the model, but the improvement is not strong enough and the effects vary from one country to another Altman et al. (2017, p.

19).

Lastly, most of the research conducted on the topic of bankruptcy prediction models have been done through the study of listed companies (Timmermans, 2014; Alyhr &

Holmberg, 2012). Which is a serious research gap since only a few studies have been done on the private companies or a mix of private and listed companies, and private companies represent the majority of the companies in Sweden. Furthermore, the body of knowledge on bankruptcy prediction models is missing an overview of both listed and non-listed companies.

The importance of research in this area can hardly be overstated, given that part of the business fabric (and with it, many other societal stakeholders such as employees and families too) is jeopardized by such troublesome in bankruptcy periods, and therefore a less uncertain knowledge on the matter would benefit businesses and society.

1.3 Purpose

The purpose of this thesis is to test the performance of three bankruptcy models, the Altman z-score, the Ohlson o-score and the Zmijewski score for all companies in Sweden.

It is interesting to explore the predictive abilities and usability of those three models in Sweden. In other words, the researchers are testing if these three models are still accurate and usable to forecast or predict bankruptcies for the Swedish companies in their decision process. In order to investigate this topic, the authors have created a quantitative study where two samples of Swedish companies will be studied, the first sample will be of companies that filed bankruptcy in 2017 and 2018 and the second sample will be a random sample of companies that did not file bankruptcy in either 2017 or 2018.

Afterwards, the authors will run the bankruptcy prediction models on the basis of previously available accounting and market information. Next, we intend to test the performance of the models by looking at how many times each bankruptcy prediction model was correct in its forecast and then calculate Receiver Operating Characteristics (ROC) curve following by calculating the Area Under the Curve (AUC) and finally by conducting an accuracy ratio (AR). Furthermore, a logit and probit regression models will be created to find two optimal models based on the variables of the three models studied.

These statistical analyses have for aim to test the accuracy of the three models on the two

selected samples and to get a deeper understanding of the models on Swedish companies.

5

1.4 Research Question

From the information gathered within the sections above, we create our research question with the aim of contributing to the knowledge of bankruptcy prediction in our current time with a focus on Swedish companies.

“What is the performance of the Altman, Ohlson and Zmijewski models in Sweden?”

1.5 Delimitations

This research is limited to Swedish companies. The researchers have decided to focus on Sweden for the reason that both of them are living and studying in Sweden. Furthermore, as mentioned earlier in this paper the authors have found only limited research on bankruptcy in Sweden and none of them was using the same models used in this research.

Also, Sweden has seen a rise in the number of companies filing bankruptcy in 2017 and an even sharper increase in 2018 (Tillvaxtanalys.se, 2018; Tillvaxtanalys.se, 2019). This makes Sweden a case worthy of special attention. Furthermore, it is possible that a country-wide approach for a pioneering study like this one is sufficiently broad for the resolution of a novel research question like ours. In the future, with an already more mature field, subsequent research may start to benefit from wider approaches.

Another limitation of this paper is the temporal range under study. The decision of focusing exclusively on the years of 2017 and 2018 was made because of the sharp increase in the number of bankrupt companies after 5 years of low bankruptcy levels in Sweden. Furthermore, the data available on the Serrano database for the years 2019 and 2020 was missing some information that supports the choice of studying the years 2017 and 2018.

1.6 Contributions

The potential outcomes from this study are twofold, and in any of both cases, it will be possible to derive relevant implications. On the one hand, the main conclusion could be that the bankruptcy prediction models turn out to be still reliable in 2017 and 2018 as they were during their creation economic periods. On the other hand, a discovery might be made in that the bankruptcy prediction may turn out to perform less than satisfactorily under more recent periods such as 2017 and 2018 in this research. Whether the main conclusion is that the usual models are vindicated or not, the implications would be manifold and far-reaching for several stakeholders.

The main contributions of the research we set out to undertake span all three,

practitioners, academicians, and society as a whole. For practitioners like business owners

or investors, we will be able to identify the bankruptcy prediction model with the highest

performance among the three we put to the test, thereby allowing for more confidence in

our current economy. Furthermore, we will also be able to give business managers

information on the main drivers of bankruptcies, thus allowing for more certain

identification, and thereby recommending closer monitoring of them.

6 For academicians, our study would contribute by opening new avenues for research in the direction of further refinement of the model or models that perform most satisfactorily according to our analysis. Furthermore, our research contributes to the growing body of work studying the performance of bankruptcy models in our current economy and more specifically in Sweden. We contribute to the literature of the three models studied (Altman, 2014; Ohlson, 1980; Zmijewski, 1984) but also to most recent research conducted across the world and in Sweden (Timmermans, 2014; Avenhuis, 2013; Altman et al., 2017).

Even society as a whole stand to benefit from the study we seek to undertake. The reason is none other than the fact that most people rely on working as employees or business owners in ongoing businesses for a living, and a piece of improved knowledge on bankruptcy prediction could make it possible for earlier anticipation on the demise of corporations, thus allowing for the necessary measures and decisions in order to guarantee the continued existence of the corporation to be taken in advance, thereby potentially permitting its ulterior survival. In this way, jobs, and with them, the welfare of a majority of the population, would be better preserved. In a broader perspective, we saw earlier that a company going bankrupt can have a strong negative impact on its stakeholders but not only it also can have a negative impact on the country's economy. Therefore, by studying bankruptcy predictions we could bring up the knowledge needed to our society to avoid bankruptcies and all their possible negative impacts on our society and economy.

1.7 Disposition

This research paper is constructed in seven main chapters which will be introduced and presented in this subchapter. The first chapter of the paper is the introduction chapter where background information is presented, then the researchers present the research gap which this research aims to cover. Furthermore, the purpose of the study and the research question is explained to the reader. Lastly, the delimitations and the contribution of the paper are introduced and explained in detail.

The second chapter of this research is the methodology chapter. In this chapter, the researchers discuss the research philosophical assumptions, approach and design that have been built during the time of this research. First, a presentation of the preliminary understanding and choice of the subject of the study is developed. Following with a detailed presentation of the research philosophy, then the research assumptions followed by the researchers. Later, the research approach and the research design of the paper are presented and explained. Finally, a literature search, a source criticism and the ethical and societal considerations are introduced and deeply explained.

The third chapter of this research is the theoretical framework. In this chapter, the

researchers start by presenting the six theories used in this research. Which are the

following ones: liquidity theory, solvency and insolvency theory, pecking order theory,

profitability theory, cash flows theory and lastly contagion effect theory. The second

subchapter is the presentation and explanation of the three bankruptcy prediction models

used in this research, the three models used are the revised Altman z-score (2014), the

Ohlson o-score (1980) and the Zmijewski score (1984).

7 The fourth chapter of this research is the practical methodology of this paper. First, a presentation and explanation of the population of the research and the two samples used.

A deep explanation of the data collection process with the reasoning of each step conducted by the researchers while collecting and selecting the data used in this research.

Furthermore, a presentation of the statistical analyses used in the research will be conducted.

Then the statistical analysis conducted in the research was introduced and presented. The following are the statistical tests used in this research, the receiver operating characteristic (ROC) curve, the area under the curve (AUC), the accuracy ratio (AR), a logit and probit regression and finally two cut-off point optimization.

The fifth chapter of this research is the empirical finding. In this chapter, the results of the three bankruptcy prediction models with the results of the three statistical tests will be presented. Following with the two new logit and probit models and the results of their statistical tests.

In the sixth chapter of this research, a discussion of the results will be presented followed by a connection of the results found with the theories presented in this thesis.

Furthermore, the findings will be compared with the findings of previous research.

Finally, the reliability and the validity of the results will be presented.

The last chapter aims to conclude on the findings of this research and the practical and

theoretical contributions and compare them to the expected contribution which has been

presented earlier on. Moreover, further research suggestions will be given to the reader

on the topic of bankruptcy prediction models.

8

2. Theoretical Methodology

In this chapter, we discuss the research philosophical assumptions, approach, and design we have built during the time of this research. First, a presentation of the preliminary understanding and choice of the subject of the study will be developed. Following a detailed presentation of the research philosophy, then the research assumptions followed by the researcher. Later the research approach and the research design of the paper will be presented and explained. Finally, a literature search, a source criticism and ethical and societal considerations will be introduced.

2.1 Preliminary Understanding and Choice of Subject of Study

In the elaboration of this thesis, two students of the Master of Finance from Umeå University worked together. Considering this circumstance, in addition to the fact that both of the authors also completed their respective bachelor’s degrees prior to the start of the aforementioned master (and also in addition to the fact that the topic of bankruptcy is related to both levels of education received by the authors) allows for an increased understanding of the topic on the part of the authors from the beginning. Despite the relatively extensive education of the authors in matters related to the topic of the thesis, the work experience possessed by the authors in related fields is scarcer.

The topic or subject of study that was chosen for this thesis is within the field of finance.

The choice was not only influenced by the more in-depth knowledge that the authors possessed of topics within the field of finance as opposed to some other topics, which, albeit within the general discipline of Business Administration, were outside of the specific topic of finance, but also it was driven by several other factors. One of them is that the authors were able to convince themselves of the contributions a study of this kind would entail for society as a whole as well as other parties, as we discussed in more detail in the namesake subchapter. Moreover, the contributions are more direct with this specific topic than it might be with topics such as the one we also considered back during the process of topic selection for the thesis, whose contributions to society and to other parties were deemed by us to be much more vicarious in comparison to the topic of this thesis.

Yet another driver for our choice of the subject or topic of the thesis was the advice we received pointing towards its direction instead of towards the other topic we jointly considered initially.

The decision of the subject of study was also driven by the suitable relevance and

adequacy of the topic for the particular level of study, both of which are quite important

requirements to keep in mind during the selection process of any topic for a thesis in

Umeå University. Regarding the relevance of the topic, besides the societal benefits

mentioned earlier, we found that the general topic of bankruptcies has received

considerable attention in scholarly research, which helps establish the relevance of the

topic. Regarding the adequacy of the topic, we shall point out its complexity, which

matches the required level of study in the view of our supervisor, as well as our own view.

9 An important preliminary consideration relates to the issue of preconceptions. In order for this study to be valid, the authors of this thesis, as any other researchers, have the duty to remain neutral, and therefore, free of preconceptions. Two main measures were taken to this end. The first one is that, as we explain in the necessary detail in subsequent methodological sub-chapters within the current chapter, we chose to conduct a quantitative study, which, according to our research philosophy and assumptions, ameliorates bias through an increase in objectivity in the data analysis and discussion. In case there still existed some bias, the second measure taken was the analysis, in a separate sub-chapter, of the potential ethical and societal issues, which would ease the uncovering of any underlying existing bias.

As an additional third measure against preconceptions (which was not only available to us as a team of researchers, but also to other teams), several work-in-progress drafts of this thesis were reviewed by fellow students and by thesis supervisors, thus with both groups, as external and independent observers, being able to identify and point to us any preconceptions or other types mistakes that we had incurred upon, or that we could have incurred upon, and that had gone unnoticed up to that point.

Despite the efforts that we have previously outlined in these last paragraphs towards the goal of neutrality, we recognize the fact that the human being is always imperfect and so flaws such as preconceptions may always exist. As discussed, we have strived to eliminate any mistakes and we recognize the duty that we have, as researchers, to remain unbiased.

2.2 Research Philosophy

Collins and Hussey (2014, p. 43) refer to a research paradigm as the philosophical framework that will guide how the research is operated. Philosophy is defined as a specific belief system that originates from someone’s view of the nature of existence and knowledge. According to Salma (2015), a research philosophy or research paradigm encompasses at least four components: an ontology, an epistemology, a research approach and one or more methods. The two most common and opposing research paradigms are called positivism and interpretivism but the reader should be aware that multiple paradigms exist in between those two opposing research paradigms (Marsh & Furlong, 2002, p. 22-30).

The following table summarizes the components of five of the most common research

paradigms:

10 Table 1: Summary of the main research paradigms and their components. Source:

Salma (2015).

First, the positivism paradigm took its origins from the natural sciences scientific field and its key beliefs are that the researchers are totally independent of our reality. The focus of conducting a research with a positivism paradigm is mainly to discover theories by doing observations and experimentations. Furthermore, the found knowledge can then be derived and then verified using mathematical and logical models (Walliman, 2011, cited in Collis & Hussey, 2014, p. 44).

The researchers who are using the positivism paradigm in their research promote and highlight logic and make sure to be precise and objective in their research. Collis and Hussey (2014, p. 44) show that positivists see reality to be independent of the researchers, it considers that the researchers and their process of conducting research cannot affect or modify the specific subject of the research. Under this paradigm, theories are used as the basis for explanations and the anticipation of the studied phenomena. By doing so an individual may predict the likelihood of occurrence of the phenomena which will allow them to take steps to control the outcome of the phenomena. The concept of explanation consists of finding the relationship between the research variables by establishing causal laws together with connecting them to deductive or integrated theory. Therefore, the social and natural world are bound by certain fixed rules. Then since the positivist think that society can be measured, the studies done in this paradigm are done with a quantitative research approach with quantitative data and statistical testing (Collis &

Hussey, 2014, p. 44).

The opposite paradigm to positivism is the interpretivism paradigm. The interpretivist

paradigm has emerged from the criticism of positivism, with the rise of social sciences,

it would appear that the positivist approach may not be always suitable for social scientists

(Collis & Hussey, 2014, p. 44). There are mainly five criticisms of the positivist approach

11 which are important to understand to be able to explain why interpretivism was created.

First, it is not possible to remove a person from the social world they live in. Secondly, an individual cannot fully understand the subject of their study without investigating the subject’s perceptions of its actions. Thirdly, a highly structured research design may lead the research to ignore other valuable findings and place restrictions on the results.

Fourthly, the research may not be entirely objective, the researcher takes part in the research and brings their own beliefs and interest to the research. Lastly, it is delusive to capture complex phenomena with one single variable. For example, it is not possible to capture an individual’s intelligence with only numerical values (Collis & Hussey, 2014, p. 44-45). Interpretivism is based on the belief that social reality is not objective but instead, it is highly subjective because it is formed by our own perceptions. When the researcher conducts research, it interacts with the subject of the research because it is not possible to separate what exists in the social world from what is in the research mind (Collis & Hussey, 2014, p. 44-45). Interpretivist investigating a social reality will influence and alter that reality. The positivists are trying to measure a social phenomenon, the interpretivist is focused on exploring the complex dynamics of a social occurrence with a goal to gain interpretive knowledge. Therefore, interpretivist tend to not use quantitative methods of research, but they rely on methods that have the goal to describe and understand a specific meaning, instead of looking for a frequency behind a social phenomenon. As a final point, interpretivist study tends to study results gathered from qualitative research (Collis & Hussey, 2014, p. 45).

As this study focuses on examining the performance of bankruptcy models, it will have a positivist approach since the authors will collect data and draw conclusions based on the data collected. Thus, the authors believe there is only one objective reality which is independent from the researchers and the research itself. Furthermore, there is only one reality that can be measured, and it is bound by certain laws that must be followed. In this case an interpretivist paradigm would be inappropriate since it does not allow to measure reality as it is subjective and open to multiple interpretations, which goes against the intent of this study and the beliefs of the authors. As the intention of this study is to generate undeniable findings and not open to interpretations. Consequently, if an individual were to replicate the study following the same methodology it should find the same results and conclusions. The authors believe that they are not interacting with the research subject and will not influence the results throughout the process of inquiry thus, the interpretivist paradigm is inappropriate following the authors fundamental beliefs.

Then, in conclusion the positivist paradigm allows for answering the research question and is the only reasonable paradigm.

2.3 Research Assumptions

Three main types of research assumptions can be identified, namely, ontological, epistemological, and axiological (Saunders et al., 2019, p.133).

2.3.1 Ontological Assumption

In research terminology, the answers a researcher or a team of researchers chooses to give

in their study to the question of the nature of reality (which is generally regarded as the

primary ontological question) and to other related questions are called ontological

12 assumptions (Saunders et al., 2019, p.135). The importance of the ontological considerations for business research can hardly be overstated, as Pessu (2019, p.38) puts it: “epistemology and methodology are driven by ontological beliefs”.

Several ontological assumptions or answers with respect to the primary ontological question about the nature of reality are possible. According to Moon & Blackman (2014, p.3), two main answer or assumption categories can be identified, namely, realism and relativism. Realism holds that there exists only reality (and the different subtypes of realism further specify whether this unique reality is knowable or not), while relativism holds that several realities exist (and subtypes within it answer the question of whether these realities are delimitable in time and space, or not).

Figure 1: Different answers to the primary ontological question on the nature of reality.

Source: Moon & Blackman (2014, p.3) In this study, we have chosen to keep a naïve realist position towards the nature of reality as our primary ontological assumption. This choice is consistent with the wider paradigm of positivism as previously discussed.

The rationale for this choice of primary ontological assumption lies in our understanding that the present study we seek to undertake would benefit from it since our main goal is to elucidate the performance (or lack thereof) of an assortment of bankruptcy prediction models and afterwards generalize the result to other similar periods in order to draw significant conclusions, an action disallowed by relativism as an ontological position, as such a broad inference would conflict with the principle of multiple realities.

Furthermore, the naïve subtype is more appropriate than the other subtypes of realism given the academic consensus on the matter that points in the direction of implicitly assuming that reality (in regard to bankruptcies) can be understood with the appropriate methods, exactly agreeing with the naïve realist position. Thus, for the purposes of this study, we choose to ontologically assume that reality is unique and understandable, two properties that specify its nature.

One should note that besides the ontological assumption par excellence (this is, the

primary one with regards to the nature of reality), some authors have mentioned other

ontological assumptions. Al-Saadi (2014, p. 6) lists not one, but several assumptions of

the ontological kind that are claimed to be held by objectivists and constructivists

(equivalent to realists and relativists in the previous classification).

13 2.3.2 Epistemological Assumption

Figure 2: Epistemology. Source: (Moon & Blackman, 2014, p.3) The epistemology assumption is interested in what can be accepted as valid knowledge (Collis & Hussey, 2014, p. 47). This will require an examination of the relationship between the researchers and the research object. As presented earlier in the ontology sub- chapter the positivists have a goal to stay independent from their research, therefore, positivists only accept knowledge that can be measured and observed. In contrast, interpretivists tend to reduce the distance between themselves and the research, and interpretivists might get involved in some form of participative inquiry (Collis & Hussey, 2014, p. 47-48).

MacIntosh and O’Gorman (2015) present four types of epistemological positions:

positivist, interpretivism, action research and critical realist. First, let us have a look at the positivist view of epistemology, which usually consists of methodological procedures of natural sciences which can be customized to the specific research of social interactions.

The results found from the research on social science appear in causal laws and will be value-free. Secondly, MacIntosh and O’Gorman (2015, p. 62) present interpretivism as it acknowledges multiple realities depending on the individuals’ perspectives, the context of the research being conducted, the understanding and the contextual interpretation of the data gathered and finally the nature and to which extend the researcher participates in the research. Then we can understand that interpretivism focuses on the understanding of the observations gathered during the research instead of just measuring those observations (MacIntosh & O’Gorman, 2015, p. 60-65). Action research is not only a single approach since it covers multiple disciplines within the research (MacIntosh &

O’Gorman, 2015, p. 63). The common definition for action research is that it involves working with organizational members over a matter that is a real concern to them and in which they need to take action. The critical realist assumption is derived from both objective and subjective ontologies, also it assumes that there is only one reality that is independent from our perception (the human perception), but the human access is consistently limited and altered by the human perception. Furthermore, an individual's perception both physically limited for example an individual cannot see into the past nor the future and ideologically limited for example an individual is biased by their own past personal experiences.

Bryman and Bell (2015) highlighted an important issue in their paper, being whether or not social studies should or can be studied using procedure, principles and ethos used in the study of natural sciences. Meaning that if the researcher answer is yes, then the researcher should follow a positivist approach and if its answer is no then it should follow an interpretivist approach (Bryman & Bell, 2015, p. 27-31).

This study aims to study and test the performance of bankruptcy prediction models.

Furthermore, the authors conducting this research believe that there is only one objective

reality, and this reality is independent of the research and from the researchers. Since this

reality is bounded by specific laws which can be measured, then it is possible to measure

a specific social phenomenon. Considering that all the observations of this study are

14 external to the researchers and can be viewed objectively, the researchers believe that a positivist paradigm is the most suitable and the authors have also decided to take a positivist stance towards what to accept as a piece of valid knowledge.

2.3.3 Axiological Assumption

In research terminology, the answers a researcher or a team of researchers chooses to give to the question of the role of values and ethics in their study (which is the primary axiological question) and to other related questions are called axiological assumptions (Saunders et al., 2019, p.134-5). Saunders et al. (2019, p.135) propose two main types of assumptions or answers to axiological questions (including the primary one), namely, the objectivist and subjectivist assumptions:

Figure 3: Two opposing sets of answers to the axiological questions. Source: Saunders

After the necessary consideration was given to this matter, the researchers decided to choose objectivism as the answer (or assumption) given to the axiological questions, including the primary one. This decision was driven both from the researchers current understanding that it is possible and even desirable to remain unbiased when conducting research, and also from the fact that the choice is consistent with the wider paradigm of positivism (Saunders et al., 2019, p.144). Thus, this research is intended by us to be the subject of no moral influence on our part and we as researchers choose to stay detached, neutral, and independent from it (Saunders et al., 2019, p. 144).

2.4 Research Approach

Saunders et al. (2019, p.153) present three contrasting approaches in a scientific research

project for the purpose of theory development: deduction, induction, and abduction.

15 Figure 4: The three ways of approaching research. Source : Saunders et al. (2019,

The deductive approach took its origins in the research of natural sciences (Saunders et al., 2019, p. 155). According to Bryman (2016, p. 21), the researchers draw upon what is already known on the topic of the research by first investigating the already existing theories, following they deduct hypotheses from the existing theories which will be either accepted or rejected by the empirical findings. Blaikie (2010, cited in Saunders et al., 2019, p. 154) present six steps in which a deductive approach will progress: (1) examination of the previous theories, (2) the deduction of possible theories, (3) collecting the data, (4) the analysis of the collected data, (5) the rejection or the confirmation of the hypotheses and finally, (6) the possible revision of the theory.

Saunders et al. (2019, p. 154) present multiple characteristics of the deductive approach to research. The first one being that the research needs to have a methodology that is highly structured to ease the process of replication since deduction is linked to the positivist approach and it has the aim to generalize the process as mentioned earlier in the paper. The second characteristic is the generalization process, to be able to generalize the finding of a research the sample of the research needs to be selected carefully and needs to have a sufficient size. Lastly, the methodology of the research needs to be conducted in a way that enables the facts to be measurable and quantifiable.

On the other hand, the inductive approach is particularly concerned with the context or

even the research takes place (Saunders et al., 2019, p. 155). With this approach, it is

more common and appropriate to use a smaller sample size in opposition to the deductive

approach where the sample size needs to be large. With an inductive approach, it is more

common to use qualitative data and to use a high variety of methods to collect this data

to be able to establish different views of a phenomenon, but even if it is more common

with qualitative data it is also possible to use quantitative data while following an

inductive approach.

16 The abductive approach can be seen as a combination of both inductive and deductive approach by shifting back and forth by moving from theory to data (deduction) and from data to theory (induction) (Suddaby, 2006 cited in Saunders et al., 2019, p. 155). Since the abduction approach is more flexible than the deduction and induction approach it can be used with multiple different research philosophies. Saunders et al (2019, p. 156) argue that it is a complicated task (or even impossible) for the research to follow a pure deductive or inductive approach, then most researchers use some elements of the abduction approach.

As reflected in figure 4, in the abduction approach, existing theories can be incorporated where appropriate, and new theories or can be generated or modified. Additionally, instead of moving exclusively from data to theory (induction), or from theory to data (deduction), the abductive approach moves back and forth between theory and data (Suddaby, 2006, cited in Saunders et al., 2019, p.155). Since the authors of this thesis incorporate existing theoretical knowledge into the performance analysis through the use of three extant bankruptcy prediction models (which, as explained before, is characteristic of the abductive approach), and since we also intend to compute two new bankruptcy prediction models, thus contributing with new theoretical knowledge (which is also a characteristic of the abductive approach as previously mentioned), we decided to employ the abductive approach in this thesis.

2.5 Research Design

Following Saunders et al. (2019, p.172), the researchers have divided the formulation of

the research design into the formulation of three components: methodological choice,

research strategy or strategies, and time horizon choice. The following figure by Saunders

et al. (2019, p.174) summarizes the whole of research methodology, including research

design:

17 Figure 5: The research “onion”. Source: Saunders et al. (2019, p.174) In this figure, the three previously mentioned components of research design are represented in different shades of blue as ellipses. Each blue ellipse (one for each component of research design) includes as text a brief list of some of the choices a researcher can make in the design of that particular area of research. More generally, each ellipse, regardless of colour, includes as text a brief list of the possible choices that a researcher can make for that methodological area. In addition, the ellipses corresponding to some areas of research methodology (including the three areas of research design) are confined within others if the authors believe the larger ones encompass the others in methodological scope (Saunders et al., 2019, p.174).

2.5.1 Methodological Choice

According to Al-Saadi (2014, p. 5), there are two main types of methodological choices

that can be made in the process of designing research: quantitative and qualitative. To

this classification, Saunders et al. (2019, p.174) add the mixed method’s type, which

combines elements of both the quantitative and qualitative methodological choices. The

quantitative and qualitative choices are traditionally distinguished by their use of numeric

data (Saunders et al., 2019, p.175). While the quantitative methodology is associated with

the use of numeric data, qualitative methodology is most strongly associated with the use

of all other types of data.

18 For the purpose of the study of testing bankruptcy prediction models, the researchers have set out to undertake in this thesis, we have decided to apply a quantitative methodology for our research design. This choice was driven by the association between positivism and a quantitative research design (Saunders et al., 2019, p.176) and by the numeric nature of the data we would be working with.

2.5.2 Research Strategy

In every study, the researcher or team of researchers seek to answer the research question, and consequently, they follow, either explicitly or implicitly, at least one research strategy in order to accomplish such a goal of answering the research question. As Saunders et al.

(2019, p. 189) explain it: “In general terms, a strategy is a plan of action to achieve a goal.

A research strategy may therefore be defined as a plan of how a researcher will go about answering her or his research question.”. Saunders et al. (2019, p.190) discuss a total of eight main research strategies available for business research: survey, experiment, case study, narrative inquiry, action research, ethnography, grounded theory, archival and documentary research.

After giving the necessary preliminary consideration to the goal by seeking to achieve (that is, to the goal of answering the research question about assessing the performance of the three bankruptcy prediction models in Sweden) and considering the available strategies provided by Saunders et al. (2019, p.190), the researchers arrived at the conclusion that the best fitting strategy that could be followed given these factors is conducting an archival and documentary research. This choice was driven as a result of the central role that the Serrano database plays as our only source of data for the quantitative analysis. The Serrano database contains financial information on Swedish businesses (as identified by corporate ID) on yearly basis (Weidenman, 2016, p.2). The information contained in the Serrano database was originally collected by an assortment of public entities in Sweden (Weidenman, 2016, p.2) primarily for tax purposes, thus the database itself can be considered an archive (as it is secondary data), and the individualized information for each specific company in it, documents. This archival and documentary nature of the Serrano database, together with the fact that Serrano database is our only source of information for the analysis (thus driving our research), are arguments in favour of the choice of archival and documentary research as the thesis research strategy.

2.5.3 Time Horizon

Every study can be classified according to its time horizon as either cross-sectional or longitudinal (Saunders et al., 2019, p.212).

Bryman and Bell (2011, p. 53) define cross-sectional design as entailing: “the collection

of data on more than one case [...] and at a single point in time in order to collect a body

of quantitative or quantifiable data in connection with two or more variables [...], which

are then examined to detect patterns of association.” On the other hand, a longitudinal

design is defined by Bryman and Bell (2011, p. 175) as: “a research design in which data

are collected on a sample (of people, documents, etc.) on at least two occasions”.

19 The data used in this research is obtained all at once from only one sample and the same time period. Therefore, the time horizon of our research determines that it is cross- sectional in design.

2.6 Literature Search

A literature search is a systematic process with the aim to identify the already existing body of knowledge on a specific topic. This process starts at an early stage of the research since it starts when the researchers have their first thoughts on the topic, and it will last until the end of the research. While conducting the literature search the researchers will not only gain knowledge on the subject of the research, but also on the previous methodology used in the specific field of research (Collis & Hussey, 2014, p.76).

This research relies only on secondary sources, meaning that the sources used are secondary information that is retrieved from current sources. Secondary sources include academic journals, databases, and books (Collis & Hussey, 2014, p.76). Considering that this paper aims to build on already existing knowledge, the use of secondary sources can be justified and not using secondary sources, in this case, could be seen as a disadvantage.

The peer-reviewed articles used in this research have been accessed through the Umeå University Library, Google Scholar, DiVA. To find relevant articles we used terms such as “bankruptcy models”, “bankruptcy”, “bankruptcy prediction”, “pecking order theory”,

“liquidity theory” and many more. From this primary stage many articles have been found on the topic of bankruptcy, the following step was to do a systematic literature review to be able to identify the previous contributions of the already done research, to then analyse and summarize findings (Bryman & Bell, 2015, p. 96-98). This research also used finance-related books and articles which have been accessed via Umeå University Library and via the course material provided by Umeå School of Business. The researchers also had to use some official Swedish websites to get legal information regarding bankruptcy such as Riksbank and Verksamt. Finally, the researchers have decided to use original sources (when it is possible), by doing so the research avoided any interpretations made from another author.

2.7 Source Criticism

For this research, the authors have decided to work with data collected from secondary sources. Cowton (1998, p. 427-429) presents some of the positive and negative sides of using secondary data which are the following: The first positive aspect of using secondary data is that it is cheaper than collecting primary data since primary data can be time and economically consuming. Then by using a secondary source of data the researchers save time and money that they can allocate to the analysis and the statistical testing of the data.

The main negative side of using secondary data is that if the researcher is not in charge

of collecting the data, then they can run the risk of losing control of the data since other

individuals were in charge of collecting the data. Moreover, a researcher can face the

issue of the property of the data and the risk of the information of the data which might

not be a perfect fit with the research and the research question. Finally, a researcher needs

to be aware of the possible exposure to bias when using secondary sources of data.

20 Concluding on Cowton (1998, p. 427-429) point of view on secondary data, if a researcher is well aware of all the advantages and disadvantages of using secondary sources of data, it will allow the research to be more reasonable when using this kind of data.

For this research, the Serrano database will be used which is the first database with the financial history of Swedish companies at the company level. The data available on Serrano is gathered from the Swedish Companies Registration Office (Bolagsverket), Statistic Sweden (SCB) and from the Bisnode’s group register (Serrano Database, 2021).

The Serrano database is a secondary source of data that was accessed via the Umeå University.

2.8 Ethical and Societal Considerations

The term ethic or more specifically research ethics refers to the moral values and the principle that design the basis of a code of conduct a research should follow (Collis &

Hussey, 2015, p. 30). The sub-chapter on ethics does not only focus on how the research is conducted but also on how the findings are reported in the paper. Bryman and Bell (2011) have created a list of ethical and societal principles that need to be respected in a research, this framework will be used in this paper to show the researchers considerations while writing the research:

The first consideration is harm to participants (Bryman & Bell, 2011). Is the need to avoid potential harm during the research process and the need to make sure of the physical and psychological wellbeing of all the research participants, the researchers, and others.

Since this case will not have any contact with participants due to the fact that this research will only use secondary data. Consequently, any possible risk of causing harm towards individuals connected to this research is totally avoided. The second consideration is dignity (Bryman & Bell, 2011). Is the requirement to respect the dignity and to avoid any kind of discomfort or anxiety of the research participants, the researchers, and others. To be able to respect the dignity of the companies we will not disclose any names to protect the company from any potential negative exposure. For the respect of the researchers and the research, the researchers made careful considerations concerning the presentation of their work and to be fair and accurate.

The third consideration presented by Bryman and Bell (2011) is informed consent. It

refers to the need to ensure that the research participants are fully informed. Since this

research is only using secondary data which is available to the public this principle is not

a concern. The fourth consideration is privacy (Bryman & Bell, 2011). Is the need to

protect the privacy of research subjects or avoid the invasions of privacy. As already

presented earlier, in this research no individuals will be involved or named in this

research. Then the issue of privacy is not a concern in this paper. The fifth consideration

is confidentiality. It refers to the requirements to ensure the confidentiality of the research

data presented in the research relating to any individuals, groups, or companies. Since the

data used in this research is publicly available, therefore there is no risk of causing any

confidential issues.

21 The anonymity consideration is the requirements to protect the anonymity of organizations and individuals. As mentioned earlier in this subchapter the researchers will not name any companies in this paper and the data used is publicly available. The seventh consideration presented by Bryman and Bell (2011) is deception. It refers to any potential deception that can occur during the research process, it could be for example lies, misleading behaviour, etc. Since the data used in this research is secondary there is no risk of deceiving research participants. During the whole research, every step will be clear and as transparent as possible to hedge against the possible risk of deceptions or misleading behaviours. Affiliation is the next consideration to take into account while writing a thesis. It refers to the need of declaring any professional or personal affiliations that may influence the research, including sponsorship and conflict of interest, and the information on the source of the research findings. This research has not been sponsored by any companies nor from personal affiliations which could affect the results of this research which eliminate any possible conflicts of interest.

Honesty and transparency are the next consideration presented by Bryman and Bell (2011). This consideration refers to the need for honesty and openness in the presentation of the information in the research to all the parties interested, also including the need for trust. To be able to be the most transparent possible the authors will present and explain every step taken in this research. Followed by the reciprocity consideration which refers to that the research should be a mutual benefit between the researchers, the research participants and any other kind of collaborations or any active participation should be involved. Since this research will only use secondary data and the research will not have any research participants or any other external help the reciprocity principle should not be a concern in this research. But it is important that this research aims to contribute new knowledge to academic researchers, companies, and any individual with an interest in bankruptcy.

The last consideration is misrepresentation Bryman and Bell (2011) refers to it as the need

to avoid any misleading, misrepresentation, misunderstanding or false reporting of the

research findings. Since the research aims to be transparent and clear in presenting their

methodology and how the data will be analysed. By doing so the researchers should

minimize the risk of misleading the reader. The researchers aim to fill a research gap and

bring new knowledge to this field of research meaning that the researchers have no

meaning on falsifying the data nor the findings of this research.