Modeling and Analysis of Business Failures: Application to Moroccan SMEs
Jamel L and Derbali A
Published on: 20191008
Abstract
Business failure is undoubtedly one of the most raised issues in the field of business management. Small and mediumsized enterprises (SMEs) play an important economic role in many countries, especially in developing countries. In Morocco, for example, they make up more than 90% of Moroccan companies and contribute around 50% to job creation and considerably to valueadded. But despite this, most of them are at risk of default and there is very little research and empirical research on this topic in Morocco, as in many developing countries. This work aims to contribute to the understanding of the determinants of business failure using the chisquare independence test. The goal is to explain but also to explain a methodology that can be applied to explain business failures by determining their internal factors. Our empirical results seem to support the claim that the main cause of such failures lies in the absence of a strategic vision and managerial knowledge and culture among SME managers.
Keywords
SME failure; Chisquare test; Gharb Chrarda Beni Hssen; Kenitra ProvinceIntroduction
Business failure is undoubtedly one of the most raised issues in the field of business management. Small and mediumsized enterprises (SMEs) play an important economic role in many countries, especially in developing countries. Their contributions to the job creation and added value are considerable. But despite this, most of them are at risk of default and there is very little research and empirical research on this topic in Morocco, as in many developing countries. In the case of Morocco, recent years have witnessed a worrying increase in the number of defaults given the difficult economic situation faced by Moroccan SMEs and in view of the slowdown in growth, the tightening of margins due to competition, the lack of funding, etc., in addition to internal dysfunctions related to management as a whole. In 2014, the country broke its default record and is listed in the international ratings among the countries with the largest failures Hermes. For this reason, the present paper proposes to analyze and to determine the factors which are at the origin of these failures and this, through a modeling applied in the Moroccan context, in particular with the SMEs of the region of Gharb Chrarda Beni Hssen (Kenitra province). In this sense, it is an extension of the similar work applied to other contexts, European and AngloSaxon, including Belgian SMEs [16]. Given its empirical vocation, our work aims to be, on the one hand, an empirical verification of failure factors sufficiently understood in the existing literature, and, on the other hand, an empirical validation of the failure factors found in the literature other contexts. Because some determinants may be found in some regions and not in others, for some categories of business, but not necessarily for others, depending on the characteristics of each context. The choice of the Moroccan context is necessary because of the very high number of failures in the country, which makes such a study urgent. The goal is both to enrich scientific research given the originality of our work and to attract the attention of SME leaders on a set of determinants and factors that can alert them to the symptoms of failure. In both cases, our contribution is of definite interest both academically and professionally. Therefore, the problematic that we will try throughout this research to bring elements of answers: “How could one explain the failure of SMEs in the Moroccan context?”. Thus, some secondary and constitutive questions of the problematic are necessary: ??What are the theoretical explanatory approaches of the failure? And what are the most explanatory factors of the failure of SMEs in Morocco? To do this, we propose to follow the following plan. As a first step, we will return to the existing literature on business failures. In a second step, we will explain our methodological approach. Finally, we will present the results and discuss their findings and interpretations.
Literature Review and Research Hypotheses
The analysis of business failures has been the subject of a great deal of work since authors and researchers became interested in this phenomenon, which became a field of investigation in its own right after the crisis of the 1930 [7]. There are numerous factors which, from a theoretical point of view, seem to be the most explanatory of failure. In what follows, we will focus our attention particularly on the factors related to the “business management” dimension. Indeed, a review of the literature on business failure has allowed us to identify some of its determinants. These will be presented successively with their respective hypothesis.
Hypothesis 1: "The stable job coverage rate of permanent funds (X1) may explain the failure of the company (Y)"
Indeed, one of the most frequently cited causes of failure in the literature is the existence of an unbalanced financial structure that reveals the need for stable resources to finance stable jobs in the firm [8]. In our case, we consider that the Fixed Asset Coverage Rate reflects the company's ability or inability to finance stable employment.
Hypothesis 2: “The liquidity of the company (X2) is an explanatory factor of the default (Y)”
Such a situation is often the result of a persistent imbalance between resources and costs from the operating cycle. This explanation is advanced by several authors including [913]. Thus, we can assess the general level of liquidity through the ratio of general liquidity that reflects a need (greater than 1) or surplus working capital (less than 1) of the company.
Hypothesis 3: “The debt capacity (X3) is likely to explain the failure of the company (Y)”
The impossibility for the company to obtain external funds because of the mistrust of its lenders leads it to default. This is found in the works of Argenti [1415]. This capacity is largely dependent on the financial autonomy of the company, that is to say, the weight of the debt in its financial structure. Reason why, we consider the leverage ratio can be of significant relevance in the translation of such capacity.
Hypothesis 4: “The financial profitability (X4) is a factor explaining the failure of the company (Y)”
The lack of financial profitability is synonymous with the lack of profit for the capital providers. The latter can no longer ensure by the success of the company their own enrichment can decide the dissolution of it. This factor seems more discriminating according to the work of [16]. To assess the level of profitability, we will stick to the financial profitability ratio.
Hypothesis 5: “Economic profitability (X5) explains the failure of the company (Y)”
The lack of profitability of the economic asset is often mentioned as a factor of failure. Thus, this factor seems more significant in discriminant analyzes since healthy companies report good profitability, in contrast to companies in a state of default. We then consider that the economic profitability ratio is best placed to reflect the level of this profitability.
Hypothesis 6: “Commercial profitability (X6) can explain business failure (Y")”
Indeed, there is a set of commercial factors that can explain the failure of the company as the poor understanding of the needs and expectations of customers, the inadequacy of supply, the inadequate positioning on the market, the strength sales, etc. These factors, therefore, are at the root of misbehavior in the field of commercial policy and may explain the lack of commercial profitability and therefore the failure of the enterprise [1724].
Hypothesis 7: “Customer Delays (X7) May Explain Business Failure (Y)”
It happens that some companies grant too many debts in order to satisfy its customers. This is obviously done to the detriment of its cash flow and its financial balance, which reveals the need for financing on the one hand, and the risk of default related to the failure of a customer (domino effect) on the other hand. Thus, granting too many receivables can result from a weak bargaining power of the company which puts it in a weak position visàvis its partners (customers). In our model, this assumption will be tested through the receivables ratio [2529].
Methodological Approach
In this part, we will explain the methodology followed which we will decline in three points, namely: sampling, variables of the model and statistical test adopted.
Sample
Our methodological approach is to study a sample of SMEs in the Gharb Chrarda Beni Hssen region (Kenitra province). Specifically, we observed a sample of 62 SMEs located mainly in the Kenitra province. Half of the sample is composed of companies in a state of legal default and companies in activity. We have also ensured that the sample does not include startups with less than five years of exercise and a natural risk of failure, to avoid the "age effect" [3038]. We have also tried to ensure that the sample is as representative as possible of the economic fabric of the region to eliminate the "sector" effect. For each of the companies observed, we took data on the ratios considered as variables of our statistical test. This data was collected from accounting firms, auditors and legal advisers [3942].
Variables
It is, first, the variable that we seek to explain, namely the failure. It is a qualitative variable that takes a binary form depending on the presence or absence of the "default" state for each of the companies in the sample. Then, we have used as independent variables seven balance sheet ratios which we try to test their explanatory power of failure. These ratios are also the most representative of the economic and financial health of the company. They are summarized in the (Table 1) below:
Table 1: List of ratios taken as explanatory variables.
Ratio 
Definition 
Type 
Formula 
R1 
Fixed Asset Coverage Rate by Permanent Funds (TCF) 
Financing Ratio 
Permanent capital / Fixed assets 
R2 
Ratio of General Liquidity (RLG) 
Liquidity Ratio 
Current assets / Circulating liabilities 
R3 
Financial Leverage Ratio (RLF) 
Financing Ratio 
Term Debt / Equity 
R4 
Return on Investment Ratio (ROE) 
Ratio of Profitability 
Net income / equity 
R5 
Economic Ratio of Ratio (RoA) 
Ratio of Profitability 
Net income / total assets 
R6 
Ratio of Commercial Profitability (RoS) 
Ratio of Profitability 
Net Result / Turnover 
R7 
Customer Credit Ratio in days (CC2) 
Management Ratio 
(Accounts receivable / sales tax incl. VAT) × 360 
Statistical tests adopted
We tested the research hypotheses by following, initially, a bivariate model in which we test the significance of each of the variables X (the ratios R1, ..., R7) compared to the variable Y (failure). To do this, we applied the chisquare independence test. In a second step, we tried to confirm our results by a multivariate model notably by applying the binomial logistic regression model [4351]. At the end of this one, we arrive at the following equation under which will be expressed the results of this test of modeling applied to the failures of companies:
Y’’ = Log [P(Y)/1P(Y)] = β0 + β1X1 + β2X2 + · · · + βzXz (1)
Analysis of the Results
After explaining our methodology, we will present in this part the results of our modeling.
Results of the Chisquare independence test
The link Rate of Coverage by Permanent Funds and Failure: Hypothesis H1 postulates that: "The stable coverage rate of jobs by the permanent funds (X1) can explain the failure of the company (Y)". The Chisquare independence test is used to study the link between the "default" variable and the "Permanent Fund Coverage of Assets" variable. Dependence is significant at the 5% level. Thus, the theoretical χ2 is 0.0039, which is less than the calculated χ2 28.18. This result suggests that the variable TCF explains the variable "failure". The values ??of Phi and V of cramer indicate that this bond is quite strong (67%). As a result, the null hypothesis (H0.1) is rejected and the hypothesis H1 is verified (Table 2).
Table 2: Khi2 test.

Value 
ddl 
Asymptotic significance (bilateral) 
Khi2 of Pearson 
28,182^{a} 
1 
,000 
Correction for the continuity^{b} 
25,434 
1 
,000 
Report of likelihood 
32,828 
1 
,000 
Nominal per nominal Phi V of Cramer 
0,6742 0,6742 

,000 
Number of valid observations 
62 


a. 0 cells (0.0%) have a theoretical size of less than 5. b. Calculated only for a 2x2 board 
Table 3: Khi2 test.

Value 
ddl 
Asymptotic significance (bilateral) 
Khi2 of Pearson 
27,395^{a} 
1 
,000 
Correction for the continuity^{b} 
24,588 
1 
,000 
Report of likelihood 
35,032 
1 
,000 
Nominal per nominal Phi V of Cramer 
0,665 0,665 

,000 
Number of valid observations 
62 


a. 0 cells (0.0%) have a theoretical size of less than 5. b. Calculated only for a 2x2 board 
The link general liquidity and default
This test hypothesis H2 for which: "The liquidity of the company (X2) is an explanatory factor of the failure (Y)". Here, the Chisquare test intends to test the link between the "default" variable and the "General Liquidity" variable. It appears that the dependence is not significant at the 5% threshold and that the theoretical χ2 is 0.0039, which is less than the calculated χ2 53.6. In addition, this association is quite strong with association strength of 66%. We can conclude that the variable RLG, too, explains the "failure" of the company. Therefore, the null hypothesis (H0.2) is rejected and the hypothesis H2 is verified.
RLF and Failure
This hypothesis (H3) assumes that: "The debt capacity (X3) is likely to explain the failure of the firm (Y)". It appears that the result confirms the existence of an addiction. Indeed the value of p (= 0.000) is quite below the risk threshold of 5%. On another plane, the theoretical χ2 is 0.0039 and is less than the calculated χ2 32, which favors the confirmation of the hypothesis in question. In addition, the strength of association is high and is significant. So, we can conclude that the variable RLF is in turn explains the variable "failure". Therefore, the null hypothesis (H0.3) is rejected, the hypothesis H3, for its part, is validated (Table 3).
ROE and Failure
We now test hypothesis 4 which states that "Financial profitability (X4) is a factor explaining the failure of the company (Y)". The present test intends to verify this hypothesis by studying the independence between the variable "default" and the variable "Financial Profitability ROE". The result shows that this link is significant since the value of p (= 0.00) is lower than the risk threshold (Table 4). Thus, the theoretical χ2 is 0.0039, which is less than the calculated χ2 13.37. This result shows that the variable ROE explains the variable "failure". Therefore, the null hypothesis (H0.4) is refuted and hypothesis H4 is admitted.
ROA and Failure
This is test hypothesis H5 postulating that "The economic profitability (X5) explains the failure of the company (Y)". It appears that the result confirms the existence of an addiction. Indeed the value of p (= 0.00) is significant at the risk threshold of 0.05. In addition, the theoretical χ2 is 0.0039 and is less than the calculated χ2 14.88. This result leads us to conclude that the variable ROA explains the variable "failure".
Therefore, the null hypothesis (H0.5) is refuted, the hypothesis H5, for its part, is accepted.
ROS and Failure
We test the last hypothesis H6 for which "The commercial profitability (X6) can explain the failure of the company (Y)". The Chisquare test shows a dependency between failure and commercial profitability since the asymptotic significance is below the threshold of 0.05. The theoretical χ2 (= 0.0039) is much lower than the calculated χ2 (= 18.083). So, hypothesis H6 is validated (Table 5).
Accounts receivable and Default
Hypothesis 7 postulates that: "Customer delays (X7) can explain business failure (Y)". Thus, the present test intends to verify this hypothesis by studying the independence between the "default" variable and the "CC2 Receivables" variable. The result shows that this link is not significant since the value of p (= 0.00) is less than the significance level of 5%. Thus, the theoretical χ2 is 0.0039, which is much lower than the calculated χ2 23.9, which confirms the existence of a link between these two variables. Thus, the strength of association (= 61.3%) is significant. So, variable CC2 explains the variable "failure". As a result, the null hypothesis (H0.7) is rejected. In contrast, hypothesis H is confirmed.
Results of Logistic Regression
In this case and as we seek to regress the variable Y on several independent variables, there is the risk that they present colinearities between them and, therefore, reflect the same information (redundancy of information), which could limit the logistic regression model. However, to overcome this limitation, the correlation matrix makes it possible to evaluate the degree of bilateral dependence between the variables. This matrix, symmetrical, is in the form of a containing the different possible correlation coefficients between said variables. The closer the coefficients are to the extreme values ??1 and 1, the higher the risk of multicollinearity (redundancy of information). On the other hand, the closer these coefficients are to 0, the less correlated they are. In our case, the test results are as follows: These results show that some variables are strongly linked and others are moderately dependent. In particular, this is a correlation (in the opposite direction, anticorrelation) and mean between CC2 and ROA (49.1%) and ROE and CC2 (52.7%). It also appears that the variable CC2 presents a risk of colinearity at the 5% threshold at which we will test our hypotheses. For this reason, we will exclude from the model the variable CC2 to which we also add RFL since it is perfectly decorrelated with all the variables of the model, which could, of course, affect the quality of the model (Table 6). Therefore, the hypotheses H3 and H7 will not be tested. As for the other variables, they have weak correlations and, therefore, will be kept in the implementation of the model. In addition to the multicollinearity test and after eliminating two variables, it seems equally important to test the validity of the model with the variables selected. In fact, in logistic regression and in order to be able to explain the studied phenomenon effectively, it is necessary to have a large sample. In the opposite case, the results will be limited in scope, which could be a limit to the model used. In practice, the rule is that for each explanatory variable, at least 5 to 10 observations will be needed. In our case, we selected five variables, a minimum of total observations between 25; 50. And since we performed 62 observations, the minimum number required for logistic regression modeling was satisfied (Table 7). By now analyzing the variables according to the logistic regression model, in particular the backward method which removes nonsignificant variables according to their likelihood ratio, the results are as follows: Based on the elimination of the variables on the likelihood ratio, it follows from these results that it is always the two variables namely TCF and ROS that significantly explain the failure at the 0.05 threshold. The equation of the model is as follows:
Y’ = 0.163 + 0.364 TCF + 0.097 ROE + 0.024 ROS
Like conditional statistics, these variables have a negative effect on the Y variable. Exp (B) of the parameters suggests that, on the one hand, the increase in the coverage rate by the permanent funds (R1) of a unit leads to decreasing the probability of failure by 0.364 times, and increasing the commercial profitability of a unit reduces the probability of failure by 0.024. This result corresponds in part to the ones we presented above, which confirms the hypotheses H1 and H6, and invalidates the hypotheses H2, H4 and H5.
Table 4: Khi2 test.

Value 
ddl 
Asymptotic significance (bilateral) 
Khi2 of Pearson 
32,904^{a} 
1 
,000 
Correction for the continuity^{b} 
29,980 
1 
,000 
Report of likelihood 
38,523 
1 
,000 
Nominal per nominal Phi V of Cramer 
0,728 0,728 

,000 
Number of valid observations 
62 


a. 0 cells (0.0%) have a theoretical size of less than 5. b. Calculated only for a 2x2 board 
Table 5: Khi2 test.

Value 
ddl 
Asymptotic significance (bilateral) 
Khi2 of Pearson 
13,373^{a} 
1 
,000 
Correction for the continuity^{b} 
11,052 
1 
,001 
Report of likelihood 
17,641 
1 
,000 
Nominal per nominal Phi V of Cramer 
0,464 0,464 

,000 
Number of valid observations 
62 


a. 0 cells (0.0%) have a theoretical size of less than 5. b. Calculated only for a 2x2 board 
Table 6: Khi2 test.

Value 
ddl 
Asymptotic significance (bilateral) 
Khi2 of Pearson 
14,880^{a} 
1 
,000 
Correction for the continuity^{b} 
12,503 
1 
,000 
Report of likelihood 
19,544 
1 
,000 
Nominal per nominal Phi V of Cramer 
0,49 0,49 

,000 
Number of valid observations 
62 


a. 0 cells (0.0%) have a theoretical size of less than 5. b. Calculated only for a 2x2 board 
Table 7: Khi2 test.

Value 
ddl 
Asymptotic significance (bilateral) 
Khi2 of Pearson 
18,083^{a} 
1 
,000 
Correction for the continuity^{b} 
15,592 
1 
,000 
Report of likelihood 
23,551 
1 
,000 
Nominal per nominal Phi V of Cramer 
0,49 0,49 

,000 
Number of valid observations 
62 


a. 0 cells (0.0%) have a theoretical size of less than 5. b. Calculated only for a 2x2 board 
Table 8: Khi2 test.

Value 
ddl 
Asymptotic significance (bilateral) 
Khi2 of Pearson 
23,290^{a} 
1 
,000 
Correction for the continuity^{b} 
20,903 
1 
,000 
Report of likelihood 
25,026 
1 
,000 
Nominal per nominal Phi V of Cramer 
0,613 0,613 

,000 
Number of valid observations 
62 


a. 0 cells (0.0%) have a theoretical size of less than 5. b. Calculated only for a 2x2 board 
Comparison and Summary of the Results
Throughout this section on business failure modeling, we used two tests and therefore obtained two results that should be compared to draw the final conclusions (Table 8). The table below shows all the results obtained according to the methods followed.
Our results show that the lack of commercial profitability has an important explanatory factor in the analysis of the failures of SMEs in our region (Table 9).
Table 9: Correlation Matrix.

TCF 
RLG 
RLF 
ROE 
ROA 
ROS 
CC2 
CC2 
1,000 






CC2 
,430 
1,000 





CC2 
,000 
,001 
1,000 




CC2 
,066 
,051 
,000 
1,000 



CC2 
,194 
,322 
,000 
,366 
1,000 


CC2 
,000 
,001 
,000 
,002 
,000 
1,000 

CC2 
,334 
,253 
,000 
,527 
,491 
,001 
1,000 
This could be due to the decline in turnover of these companies which itself could be explained by:
 The intensity of competition (Daubie, 2005);
 Loss of market share;
 The bad positioning on the market;
 The lack of accuracy of forecasts (Argenti, 1976; Sheppard, 1994);
 Demotivation of the sales team (Ooghe and Waeyaert, 2004);
 Price policy (Brilman, 1982);
 Customer infidelity due to competitive intensity;
 Customer dissatisfaction;
 Poor understanding of expectations needs (Ooghe and Waeyaert, 2004);
 The absence of a commercial policy.
As well as other factors that can be summarized in the absence of good conduct in terms of policy, or management, commercial. In addition to commercial profitability, the lack of permanent funds (reflected in the TCF) to cover the investment cycle of the company, and possibly its operating cycle in the event of excess working capital, explains significantly the failure of the company as this obviously causes a persistent financial imbalance in these companies. Such a situation could also be explained by (Table 10).:
 The difficulties related to external financing (Ben Jabeur, 2011);
 The tightening of credit conditions (guarantee requirement for example) and the reluctance of Moroccan banks to finance SMEs;
 The absence of a compartment dedicated to SMEs in the financial market (note that a project to launch this compartment is underway);
 The manager's reluctance to use external / banking financing for religious and cultural factors (risk aversion);
 Balance sheet imbalance (Crutzen and Van Caillie, 2010);
 The absence of a financial and accounting policy (Crutzen and Van Caillie, 2010).
Although these explanations are not exhaustive, it seems that an explanation holds our attention. This is essentially the absence of a business management culture (Table 11).
Table 10: Variables in the equal.

A 
E.S. 
Wald 
Ddl 
Sig. 
Exp(B) 
Step 1^{a} 
TCF 
,812 
,501 
2,623 
1 
,105 
ROE 
3,138 
2,437 
1,658 
1 
,198 

ROA 
2,148 
2,934 
,536 
1 
,464 

ROS 
2,644 
1,940 
1,856 
1 
,173 

RLG 
,228 
1,149 
,039 
1 
,843 

Constant 
1,620 
1,176 
1,897 
1 
,168 

Step 2^{a} 
TCF 
,791 
,538 
2,166 
1 
,141 
ROE 
2,545 
2,151 
1,401 
1 
,237 

ROS 
3,354 
1,847 
3,298 
1 
,069 

RLG 
,667 
1,825 
,134 
1 
,715 

Constant 
1,132 
1,222 
,857 
1 
,355 

Step 3^{a} 
TCF 
1,010 
,450 
5,040 
1 
,025 
ROE 
2,328 
2,036 
1,308 
1 
,253 

ROS 
3,749 
1,694 
4,896 
1 
,027 

Constant 
1,816 
,783 
5,374 
1 
,020 

a. Variable (s) entered in Step 1: TCF, ROE, ROA, ROS, RLG. 
Table 11: Summary of results.
Hypotheses 
variables 
Bivariate model (Tχ2) 
Multivariate Logistic Model (R.Likelihood) 
H1 
TCF 
Validated 
Validated 
H2 
RLG 
Validated 
Rejected 
H3 
RLF 
Validated 
Not tested 
H4 
ROE 
Validated 
Rejected 
H5 
ROA 
Validated 
Rejected 
H6 
ROS 
Validated 
Validated 
H7 
CC2 
Validated 
Not tested 
Indeed, most SME managers still manage in the traditional way and do not have a managerial culture or a strategic vision of their business. Hence the lack of financial and accounting information, monitoring of the financial and economic state of the company, sectoral and market studies to adapt to changes in the environment, etc.
Conclusions
After this empirical examination of the problem of failure, we can conclude that all the variables tested explain the failure. This is due to the fact that the failing companies report misbehavior in the area of ??commercial policy (poor market positioning, poor understanding of the needs and expectations of customers, the problem of accurately forecasting the evolution of demand, prices, etc.) but also in financial and investment do not generate enough profitability, which joins and confirms the work of the authors cited. In response to the problem, we can say that the failure is mainly attributed to the lack of a business management culture among SME managers [5256]. They still manage in the traditional way without tools, indicators, or even financial and accounting systems that can feed into their decisionmaking system or, at least, inform them in their choices. Thus, the main contributions of our work are at two levels: academic and professional. With regard to the first, our work proposes to contribute to the understanding of a phenomenon little apprehended in the academic world and whose existing contributions, few in fact, are limited to the theoretical aspect of the question without as much to go to an empirical verification applied to the Moroccan context. Regarding the latter, this work would be very useful for business leaders, consulting firms and consultants wishing to arm themselves with forecasting tools and / or prevention of business failures. In this sense, the factors we identified would be used to develop dashboard indicators to alert managers to symptoms of failure. In any case, the present work is of a certain contribution as well on the academic level, as professional. However, such contributions should not overshadow the inherent limits, or even specific to any contribution that is meant to be scientific. First and foremost, we mention the unavailability of certain information (lack of free access databases on this subject) and the confidential nature of the data we needed (access to corporate balance sheets), which explains why the refusal of certain institutions to give us access to their documentation. Added to this are the limited size of our sample, especially by sector, which would have obviously played on the statistical tests conducted, and the nonexhaustive nature of the ratios taken as variables in this work. Reasons why, we consider that these limits are remediable by proposing in particular some ways of future research on this subject. For improvement purposes, it would be interesting to increase the size of this sample to include several sectors and regions of the realm. For deeper purposes, we propose, on the one hand, to widen the choice of independent variables by integrating, in addition to those used, other quantitative variables such as productivity, margin and valueadded ratios, rotation, etc. but also qualitative such as the death of the principal partner, conflicts between partners, etc. On the other hand, it is important to deepen the issue of defaults by focusing on a particular sector, the sector where the failures are the most, for example.
References
 Altman EI. Financial ratios, discriminant analysis and the prediction of corporate bankruptcy. J Fina. 1968; 23: 589609.
 Altman EI. A further empirical investigation of the bankruptcy cost question. J Fina. 1984; 39: 10671089.
 Amor SB, Khoury N, Savor M. Forecast model of the default of Quebec SME borrowers. J Small Bus Entrepreneurship. 2009.
 Argenti J. Corporate collapse: the causes and symptoms. Holsted press, McGrawHill, London, 1st Edition. 1976.
 Aziz A, Emanuel DC, Lawson GH. Bankruptcy prediction: An investigation of cash flow based models. J Manag Stud. 1988; 25: 419437.
 Back B, Laitinen T, Sere K, Wezel MV. Choosing bankruptcy predictors using discriminant analysis, Logit analysis and Genetic Algorithms. Turku Center for Computer Sience, Technical Report. 1996; 40: 118.
 Beaver W. Financial Ratios as Predictors of Failure. J Accou Res. 1966; 6: 71 87.
 Bescos PL. Failure and recovery of P.M.I Search for indices and causes of failure. CEREG Research Paper, 8701, University of ParisDauphine 1987. 1989.
 Jabeur BS. Bankruptcy Statute in Financial Theory: Theoretical Approaches and Empirical Validations in the French Context, Ph.D. in Management Sciences, University of South ToulonVar; University of Sousse. 2011.
 Jabeur BS. Modeling Corporate Failure: A PLS Logistic Regression. Federal Committee on Statistical Methodology Research Conference, Washington. 2009.
 Boisselier P, Dufour D. Scoring and anticipation of business failure: A logistic regression approach. Open Archives HAL. 2003.
 Boujelben S, Hassouna F. The relevance of operating cash flows and traditional financial information in predicting the financial distress of Tunisien compagnies. Accounting and environnement. France. 2007.
 Brilman J. The recovery of companies in difficulty. Editions Hommes et Techniques. 1982.
 Casta JF, Zerbib JP. Predicting Business Failure?. French Accounting Review. 1979; 97: 506526.
 Charalambous C, Charitou A, Kaourou F. Comparative analysis of artificial neural network models: application in bankruptcy prediction. Annals Operations Res. 2000; 99: 403425.
 Cohen E. Financial Analysis. 6th Edition, Economica, Paris. 1997.
 Cohen E. Financial management of the company and financial development. Francophone University, Edition EDICEF. 1991.
 Crutzen N, Caillie VD. A taxonomy of distinctive explanatory business failure patterns amongst small firms: a qualitative approach. Cahier de Recherche du CEPE. 2010.
 Crutzen N, Caillie VD. The sequence of business failure factors: A reconciliation of organizational and financial approaches. 2007.
 Crutzen N. Dynamical models representative of the failure of the company: a state of theories in the literature, Working paper written and Presented in order to obtain the Diploma of Advanced Studies. HECSchool of Management of the University of Liege, June. 2006.
 Daubie M, Meskens N. Business Failure Prediction: A Review and Analysis of the Literature. 2001; 115.
 Desjardins J. Logistic regression analysis, University of Montreal Tutorial in Quantitative Methods for Psychology. Flight. 2005; 1: 3541.
 Coulibaly DA. The failure of Belgian SMEs: Analysis of determinants and statistical modeling. Theses of the Catholic University of Louvain (UCL). Edition University Presses of Leuven. 2004.
 Fitzpatrick PJ. A Comparison of ratios of successful industrial enterprises with those of failed firms. Certified Public Accountant. 1932.
 Gentry JA, Newblod P, Whiteford DT. Classifying bankrupt firms with funds flow components. J Accou Rese. 1985; 23: 146160.
 Gerard P, Windey J, Regoire GM. The Judicial Concordat and Bankruptcy  Laws of July 17 and August 8, 1997, The Courts Journal Records. Editions De Boeck And Larcier, Brussels. 1998; 221.
 Gresse C. Companies in difficulty. Paris, Economica. 1994.
 Guilhot B. Business failures: seventy years of theoretical and empirical analysis. Revue Française de Gestion. 2000; 130: 5267.
 Jaminon R. Factors Explaining Bankruptcy, Review of the Faculty of Law of the University of Liege. 1986; 3:197207.
 Jones S, Hensher DA. Predicting firm financial distress: a mixed Logit model. The Accounting Rev. 2004: 79: 10111038.
 Keasey K. Watso NR. Nonfinancial symptoms and the prediction of small company failure: A test of the Argenti hypotheses. J Bus Fina Accou. 1987; 14: 335354.
 Keasy K, Mcguinness P. The failure of UK industrial firms for the period 19761984: logistic analysis and entropy measures. J Bus Fina Accou. 1990; 17: 119135.
 Koenig G. Companies in difficulty: from symptoms to remedies. French Rev Manag. 1985; 8492.
 Khoufi W, Feki R. The macroeconomic determinants of firm’s failure the case of Tunisian small and medium size industries. 2004.
 Laitinen EK. Financial ratios and different failure processes. J Bus Fin Accou. 1991; 18: 649673.
 Lemerle M. Business failures in the world, Responsible for Macroeconomic studies and failures. Economix study day, Paris Ouest Nanterre University. 2012.
 Lin T. A cross model study of corporate financial distress prediction in Taiwan: Multiple discriminant analysis, logit, probit and neural networks models. Neurocomputing. 2009; 72: 35073516.
 Malecot JF. Theoretical analyzes of business failures: A review of the literature. Financial Economics Review. 1991; 205227.
 Marco L. The rise of bankruptcies in France: 19th and 20th century, Editions Le Harmattan. Collection Economic Logic. 1989; 191.
 Marco L. Bankruptcies and economic crises in France in the nineteenth century, Annals Economies, Societies, Civilizations, 44th year. 1989; 2: 355378.
 Martinet AC. Strategic diagnosis, Vuibert.
 Matoussi H, Mouelhi R, Salah S. The prediction of bankruptcy of Tunisian companies by the logistic regression. 20th Congress of AFC, May 1999, France. 2011.
 Mestiri D, Hamdi M. The prediction of the financial distress of Tunisian firms by the random effects logistic regression model. 2015.
 Mossman CE, Bell GG, Swartz LM, Turtle H. An empirical comparison of bankruptcy model. Financial Rev. 1998; 33: 3554.
 Mrabet R, Tazi M. The causes of business failure: an explanation essay. Manag Society Rev. 1999.
 Newton GW. Bankruptcy and Insolvency Accounting. John Wiley and Sons, 3rd Edition, New York. 2005.
 Ooghe H, Wymeersch VC. Predictive models of bankruptcy. J University of Liege. 1986; 3: 183195.
 Ooghe H, Waeyaert N. Oorzaken van faling en falingspaden Literatuur Overzicht en conceptueel verklarings model. Economisch en Sociaal Tijdschrift. 2004; 57: 367393.
 Pompe PM, Bilderbeek J. The prediction of bankruptcy of small and medium sized industrial firms. J Buss Ventur. 2005; 20: 847868.
 Porter ME. Competitive Strategy: Techniques for analyzing industries and competitors. Free Press, New York. 1980; 396.
 Refait C. Bankruptcy prediction based on the financial analysis of the company: an inventory. Economy and Forecasting. 2004; 162: 129147.
 Sheppard JP. Strategy and bankruptcy: an exploration into organizational death. J Manag. 1994; 20: 795833.
 Stafford J, Bodson P. Multivariate analysis with SPSS. University of Quebec Press. 2006.
 Taffler RJ. Forecasting company failure in the UK using discriminant analysis and financial ratio data. J Royal Statistical Society. Seriesa. 1982.
 Thornhill S, Amit R. Understanding failure, organizational mortality and the resourcebased approach. Analytical Studies. 2003.
 Tirapat S, Nittayagasetwat A. An investigation of Thai listed firm’s financial distress using macro and micro variables. Multinational Finance J. 1999; 3: 103125.