The influence of financial ratios on the ability of SMEs to acquire bank loans

11-01-2021 University of Groningen

The influence of financial ratios on the ability of

SMEs to acquire bank loans

In the Netherlands

Jeroen Reijntjes, S2939037

Master Thesis

Final Version

Master: Finance

2020-2021

Word Count: 9178

Supervisor: Dr. J.V. Tinang Nzesseu

Abstract

This paper investigates 350 SMEs in the Netherlands over a period of 5 years, to observe the influence of financial ratios on the ability of acquiring additional bank loans. The focus is on examining whether solvability ratios or liquidity ratios have the largest impact on the ability of SMEs to acquire additional bank loans. The two primary models used in this empirical research are a logit model and a time and entity fixed effects model. All results show a significant larger impact of solvability ratios than liquidity ratios. Furthermore, this paper can improve the decision making process of financial institutions when providing loans to firms that are affected by the economic effects of the COVID-19 pandemic. For this last section however, this paper provides an intermediate solution and further research upon this subject is recommended.

1

Table of Content

1. Introduction ... 2

2. Literature review and hypotheses ... 3

2.1 General ... 3

2.2 Conditions for Predictions ... 3

2.3 Financial ratios ... 4 2.4 Hypotheses ... 6 3. Methodology ... 6 3.1 Data ... 6 3.2 Regressions ... 9 3.3 Logit model ... 10

3.4 Likelihood ratio test ... 11

3.5 Data biases ... 11

3.6 Longitudinal data ... 12

4. Results ... 13

4.1 Determination of models to use ... 13

4.2 Complete logit model ... 14

4.3 Importance solvability and liquidity ... 16

4.4 Panel Model ... 18

5. Conclusion ... 19

References ... 22

2

1. Introduction

Financial ratios are a subject of extensive research over the last decades. Using financial ratios, a prediction can be established about the performance of companies with the main focus on bankruptcy. Moreover, comparison between companies can be made using financial ratios. Financial ratios are mathematical fractions between certain numerical values of a company’s financial statements. Most common are the solvability ratios which show the ability to repay long term debts (example: equity/liabilities) and the liquidity ratios which show the ability to repay short term debts (example: current liabilities/current assets). The two main research areas on financial ratios are on bankruptcy and the ability of comparing companies, which are in fact counter parts. On the one hand, when companies are performing poorly, research is focused on the ability of financial ratios to predict the terminal moment of these companies. On the other hand, when companies are performing well, research is focused on which company is performing the best using financial ratios to compare them.

This paper primarily focuses on the intermediate moment. Prior to the stage of bankruptcy, there is a stage where companies have trouble with receiving additional loans from banks. These additional loans could help companies that are at a turning point, to stay out of critical financial status or bankruptcy. In 2013, already 25% of the SMEs in the Netherlands communicated to be in some kind of financial difficulties (Het Financieele Dagblad, 2013). This paper examines the problem of financial difficulties for SMEs and analyses possible solutions. This group will probably increase extensively due to the COVID-19 pandemic. The reason for this is that, currently many firms are closed and are only able to survive because of governmental support. When the support of the government ends, those firms will face a difficult time readjusting to the market. Analysing the financial ratios in this area could be the key factor in determining the solution for this problem. Furthermore, it could also provide a basis for the current question for financial institutions and governments, which of the companies that are in trouble due to COVID-19 are worth saving. As previously mentioned, the two main groups of financial ratios are solvability ratios and liquidity ratios. This research investigates which of the two has the largest impact for SMEs in order to acquire additional loans. To elaborate on this point, this paper investigates which financial ratios are most important for receiving new bank loans and to what extend they influence this process. Extending to the rising problem of COVID-19, can there be made a separation point between companies which deserve aid by financial institutions and governments and which companies do not deserve aid using financial ratios.

3

additional bank loans and which of the two main groups of financial ratios has the largest impact.

This paper is structured as follows. Section 2 describes relevant literature, presents the hypotheses and the reasoning behind it. Section 3 shows the methodology of this research. Furthermore, in section 4 the results of the multiple regressions are shown. Finally, section 5 discusses the conclusions drawn from the results and further research opportunities and recommendations.

2. Literature review and hypotheses

Financial ratios have been extensively studied in the past decades. Edmonds et al. (2000) describes ratio analysis as “studying various relationships between different items reported in a set of financial statements”. Ratios can be created from values within the same financial statement or between different financial statements. For example, revenue/total assets is a financial ratio between different financial statements. The revenue value comes from the income statement, whereas total assets come from the balance sheet. In the income statement, all the income and expenditures of a company are presented and hence a company’s profit or loss. Furthermore, the balance sheet shows an overview of all the assets and the way they are financed. Objectives from financial ratios are to assess a firm by analysing how the firm has changed over time and it can also be used to compare a firm to other similar firms using a common set of financial ratios. Using these two principles, the main study focus areas have been on bankruptcy and performance. In this paper, there is proceeded on previous research determining which financial ratios are important to consider and what kind of research design shows the best results. Hereby, first examining the value of certain financial ratios and their power. Wilcox (1970), examines the predictive value of certain financial ratios and compares that to earlier research of Tinsley (1970) and Beaver (1966). In his research Wilcox (1970) additionally shows that financial ratios that are often used have fewer predictive values as less well-known ratios due to the ability of managers to guide ratios. The financial ratios that are considered, are shown in part 2.3 of the literature review. The notion of Wilcox that less well-known financial ratios have a better predictive power is also taken into account in this part together with other related literature.

2.2 Conditions for Predictions

4

measurement size, the necessity of the presence of a proper failure group is meant. Meaning that the success group and the failure group in this kind of research need to be of significant size in order for the data to be unbiased. The failure group would in this case be, the SMEs which are not able to acquire additional loans. In section 3.1 is the reasoning behind the sample size of 350 companies given. In the sample, the failure group is 32% of the total amount which is equals to 112 companies. This is a large enough amount to counter the problem of measurement size. Furthermore, the measurement timing can become a problem in this area. The problem of measurement timing is using data that can still fluctuate after the research. This could lead to false conclusions or predictions. Financial statements and therefore the financial ratios can change up until 2 years after their release in the Netherlands. Hence this paper will use the data from the 5 year period of 2014 to 2018 to avoid this problem. Finally, the relevance of industry comparisons is displayed in the papers mentioned. However, the companies evaluated in those studies are all listed firms and are therefore significantly larger than the SMEs researched in this paper. In the paper from Voulgaris et al. (2000) “On the evaluation of Greek industrial SME's performance via multicriteria analysis of financial ratios” the relevance of making a difference between industries with SMEs was debunked on the notion that it is an industry on itself. Additionally, the detailed information which can be found for listed firm is not the case for SMEs, where finding information about the management, niche, position or technical status is nearly impossible. SMEs are not obligated to provide as much information as listed firms. Furthermore, depending on their size, companies are obligated to provide more or less information to the chamber of commerce in the Netherlands. This leads to the assumption that all SMEs are part of the same industry. This is the same reasoning as is given in the paper from Voulgaris et al. (2000). Despite the exceptional predictive power from financial ratios which is shown by Ohlson(1980) on its own, Myšková et al. (2017) considered the combination of financial ratios and a linguistic analysis of annual reports. In this research, Myšková et al. finds a beneficial correlation between the precision of the annual reports and successful firms, especially in the cash flows and leverage ratios. This is another important factor that was considered while conducting this research.

2.3 Financial ratios

z-5

score, then they are in immediate danger for a bankruptcy. The formula for the z-score from the paper of Altman (1968) equals to:

𝑍 = 0.012𝑋₁+ 0.014𝑋₃+ 0.033𝑋₃ + 0.006𝑋₄+ 0.999𝑋₅ (1)

Where, 𝑋₁ equals to working capital/total assets, 𝑋₂ equals to retained earnings/total assets, 𝑋₃ equals to earnings before interest and taxes/total assets, 𝑋₄ equals to market value of equity/book value of total debt, 𝑋₅ equals to sales/total assets and Z equals to the overall index. The ratios used in this formula consist every aspect of the financial statements of companies. Therefore, Altman’s research cannot distinguish between the importance of the several groups of financial ratios. To distinguish between the importance of solvability ratios and liquidity ratios, the research in this paper needs a significantly different approach from the research of Altman (1968). Furthermore, this paper does not focus on bankruptcy like the paper from Altman (1968). To equally propionate the weights of solvability ratios and liquidity ratios in this research, both the group of solvability ratios and liquidity ratios need to have the same amount of ratios.

Henceforth, this paper will focus on four financial ratios. Two ratios will be used to assess the solvability of a company and two ratios for the liquidity. According to Gibson (1987) there are two ratios for determining the liquidity significantly better than all the others. These ratios are the quick ratio and the current ratio which are calculated as follows:

𝑄𝑢𝑖𝑐𝑘 𝑟𝑎𝑡𝑖𝑜 =𝐶+𝑀𝑆+𝑅𝑒

𝐶𝐿 (2) and 𝐶𝑢𝑟𝑟𝑒𝑛𝑡 𝑟𝑎𝑡𝑖𝑜 = 𝐶𝐴 𝐶𝐿 (3)

where, C equals cash, MS are all marketable securities, Re are the receivables, CL are the current liabilities and CA are the current assets. These liquidity ratios have a rated significance of 7,1 and 6,3 respectively according to the research from Gibson (1987). These two ratios are used in this paper to assess the liquidity of the companies. To equally measure the solvability, two ratios out of the same research are chosen with a combined equal significance (13,4) to avoid any bias to one of the two measurements. In the paper from Gibson (1987), the solvability ratios of liability based and asset based have also a combined significance of 13,4 (7 and 6,4 respectively) and are calculated as follows:

𝐿𝑖𝑎𝑏𝑖𝑙𝑖𝑡𝑦 𝑏𝑎𝑠𝑒𝑑 =𝑇𝐸

𝑇𝐿 (4) and 𝐴𝑠𝑠𝑒𝑡 𝑏𝑎𝑠𝑒𝑑 = 𝑇𝐸 𝑇𝐴 (5)

6

besides the already mentioned ratios, also the ratios of cash to current liabilities, equity to long-term liabilities and equity to fixed assets discussed. This research showed as well that the best performing liquidity ratios are the current and quick ratio. The cash to current liabilities ratio is performing well at a short term base, however it is not performing well for forecasting over a longer time period. Furthermore with the solvency ratios, the liability based solvency ratio, which was also assessed by Gibson (1987), was the only solvency ratio which showed any significant results. In addition, the research of Chen (1981) and Höbarth (2006) also only consider the solvability and liquidity ratios that are used in this research.

2.4 Hypotheses

As was mentioned so far, all the related papers show the power and importance of financial ratios and their capabilities. However, all these papers look at the two outer states of a company’s performance. Moreover, they make no difference in the importance of specific financial ratios. Whereas in this paper, the expectancy is that there will be a larger influence from solvability ratios than from liquidity ratios. This assumption rises from the idea that the liquidity can be fixed with securing additional bank loans and should therefore be a less dependent factor in securing additional bank loans. Furthermore, a company’s solvability ratios decrease while attracting additional loans. Finally, the definition for solvability ratios is the ability of paying long term debt which is the essence for a bank. Hence, this would suggest that a company is more dependent on its solvability ratio to receive additional bank loans than its liquidity ratios.

Therefore, the hypotheses this paper will investigate are:

H0: Solvability ratios of asset based and liability based have a larger impact than the liquidity ratios of current and quick ratio, on the ability of acquiring additional bank loans for SMEs in financial difficulties.

H1: Solvability ratios of asset based and liability based do not have a larger impact than the liquidity ratios of current and quick ratio, on the ability of acquiring additional bank loans for SMEs in financial difficulties.

3. Methodology

7

and 10 are excluded due to the fundamental difference in financing those firms. Generally with sole proprietorships or micro-firms, the firms consist out of pure equity financing. The rationality behind this phenomenon is the low start-up cost which accompanies this kind of businesses. Often these people start working from home and do not have any significant investment prior to the start of their business. Therefore the research group in this paper will focus on companies with the amount of employees between 11 and 250. As previously mentioned, in 2018 this amount was 22.422 firms (Orbis, cbs.nl), therefore a sample size of 350 firms has been chosen for this research. In the appendix a list of all the companies can be found. The sample size is based on research from the University of California, San Francisco (UCSF) and the papers from Simel et al. (1991) and Dziak et al. (2014). In the next section of this paper (regressions), is shown that a logistic regression is needed in order to model the data. The UCSF created a simplistic calculator to find the required sample size for logistic regressions with the following formulas.

𝑇𝑜𝑡𝑎𝑙(𝑛) = 𝑁₁(𝑟𝑎𝑤) + 𝑁₀(𝑟𝑎𝑤) (6) 𝑁0(𝑟𝑎𝑤) = ( 𝑞0 𝑞1) ∗ 𝑁1(𝑟𝑎𝑤) (7) 𝑡_𝑎 = 1 − (∝/2) (8) 𝐷𝐹 = 𝑁₁(𝑟𝑎𝑤) + 𝑁0(𝑟𝑎𝑤) − 2 (9) 𝑁𝐶𝑃 = 𝐿𝑁(𝑂𝑅)/√_𝑁 1 1(𝑟𝑎𝑤)+ 1 𝑁0(𝑟𝑎𝑤) (10)

Where 𝑞₀ equals 0.32, 𝑞₁equals 0.68 which are the proportions of the fail and succeed groups in the data. The α equals 0.01 and represents the probability of rejecting the null hypothesis (type 1 error rate) and a β of the probability of falling to reject the null hypothesis of 0.2 (type 2 error rate). Finally an odd ratio of 1.5 is used which is a measure of association between an exposure and an outcome. The odd ratio represents the odds that an outcome will occur given a particular exposure, compared to the odds of the outcome occurring in the absence of that exposure. These variables correspond to similar values used in the research of Simel et al. (1991) and Dziak et al. (2014). The 𝑁₁(𝑟𝑎𝑤)is found by setting the equations above equal to each other to find a 𝑡𝑎 with a probability from the non-central t distribution conditional to the

8

over the Netherlands. The amounts given in table 1 below are used to pick a random selection out of each part of the Netherlands.

Table 1: SMEs Companies with 11-250 Employees. Column 1 shows the area of the population. Column 2 shows the total population of the entire Netherlands and their provinces. Column 3 shows the amount used in the sample. Column 4 shows the percentages of amount per province to the total amount, with the exception (*) in row two, where it is the percentage between the total sample and the total population.

Total Sample Percentage

Netherlands 22422 350 2%(*) Drenthe 459 7 2% Flevoland 476 7 2% Friesland 678 11 3% Gelderland 2812 44 13% Groningen 517 8 2% Limburg 1369 21 6% Noord-Brabant 3902 61 17% Noord-Holland 3533 55 16% Overijssel 1578 25 7% Utrecht 1910 30 19% Zeeland 494 8 2% Zuid-Holland 4694 73 21%

From these 350 firms, the yearly reports of all financial data were requested and the credit ratings of banks in order to assess their ability of acquiring additional bank loans during the researched years. The data was collected over the years 2014-2018 which makes a total of 1750 data points. Within those 1750 data points the following data is outlined: Current Ratio, Quick Ratio, Solvability Ratio Asset Based, Solvability Ratio Liability Based, Operating Revenue, Net Income and Number of Employees. Table 2 shows a summary of the descriptive statistics.

Table 2: Descriptive Statistics of all variables. Column 1 shows the variables, which are explained below the table. Column 2 shows the total amount of observations, column 3 shows the mean of the variable, column 4 shows the standard deviation and column 5 and 6 showing the minimum and maximum value of the variable.

Variable Observations Mean Std. Dev. Min Max

PosLoan 1,750 .6765714 .4679184 0 1

9

SolRatAs 1,697 40.70004 26.73037 -89.016 99.992 SolRatLi 1,081 45.10361 27.73209 .052 99.914

NumEmp 1,712 93.79439 59.1914 11 250

PosLoan=Possibilty for additional loan, OperRev=Operating Revenue, CurRat=Current Ratio, LiqRat=Quick Ratio, SolRatAs=Solvability Ratio Asset Based, SolRatLi=Solvability Ratio Liabilities Based and NumEmp=Number of Employees

3.2 Regressions

In order to establish the best way to test the hypotheses, Muller et al. (2009) analysed several different ways of testing the predictions techniques of financial ratios and compared their abilities. Muller et al. (2009) continued on the work of Aziz and Dar (2006), who analysed 46 articles which covered 89 empirical studies on prediction of bankruptcy. The three differentiations made by Aziz and Dar are: statistical models, artificial intelligence expects systems (AIES) models and theoretical models. Muller et al. specified these models into four analysing techniques which were compared against each other. The four analysing techniques considered by Muller et al. are the following: Multiple discriminant analysis (MDA), Recursive partitioning (RP), Logit analysis (LA) and Neural networks (NN). Muller et al. showed that LA and NN had the best predictive capabilities, however they had likewise the highest chance of making a type 1 or type 2 statistical error. Where, a type 1 error rejects the null hypothesis which is in fact true (false positive) and a type 2 error fails to reject the null hypothesis which is in fact false (false negative).

The following paragraph elaborates on the problem of what modelling technique to use for this research. The question of whether a SME gets an additional bank loan is a question with only two possible outcomes. In cases where there are only two possibilities, dummy variables can be used, since they can either get the values of 0 and 1. In this case, if the SME gets a bank loan the variable would be 1 and 0 otherwise. However, a dummy variable in this case, is the dependent variable where it is often only used as explanatory variable. According to Brooks (2019), models with a dummy in the dependent variable can only be described by three models. The most straightforward model here is the linear probability model. Yet, the linear probability model assumes the data to be linear. The problem with financial data is that it is often non-linear. Therefore, the linear probability model is also the most flawed way of dealing with a binary dependent variable in this case. The linear probability model has two main problems: probability outcomes which are outside the barriers of 0-1 and the data showing signs of heteroscedasticity.

10

transforming the function in a way that the fitted values of the regression stay between the boundaries of 0-1. Secondly, the heteroscedasticity problem can be solved by using robust standard errors. The logit model is traditionally the preferred method according to Stock and Watson (2006), since it does not need an integral to calculate the correct value which makes finding the true value easier. However, with the current technology, this is not a valid reason anymore. Nevertheless, the logit model has become the more accepted standard in financial research in previous years. So for this paper, a logit analysis is one of the primary methods for analysing the given hypotheses. This is due to the reasoning above which was covered by Brooks (2019) and the results of the previous paper mentioned by Muller et al. (2009).

Therefore, to find the necessary results to answer the hypotheses, multiple regressions will be performed as will be explained in the next section. Moreover, in this paper there is a logit model to compare and analyse the importance of solvability ratios for receiving additional bank loans. Furthermore, the suitability of the logit model will also be tested against the alternative of longitudinal models.

3.3 Logit model

A logit model transforms the linear probability model into a regression model where the fitted values are between the boundaries of 0-1. Therefore the visually straight-line of the linear model will be shaped into a S-shaped model. The formula for the logit model is a logistic function of F with, z, being any random variable:

𝐹(𝑍_𝑖) = 𝑒𝑧𝑖

1+𝑒𝑧𝑖= 1

1+𝑒−𝑧𝑖 (11)

With, e, being the exponential in the logit approach. Whereas, the function F is actually a cumulative logistic distribution. Therefore the logistic model can be estimated by:

𝑃 = 1

1+𝑒−(𝛽1+𝛽2𝑥2𝑖+⋯+𝛽𝑘𝑥𝑘𝑖+𝜇𝑖) (12)

11

(ceteris paribus). This is however not the case when using a logit model with maximum likelihood. With a logit model the results shown are in log odds, which is the natural log of the odds. Log odds are extremely hard to interpret without further aid. Using Stata, the estimates can be changed from log odds to odds. Where, odds can be interpreted by the probability of success (event occurring) divided by the probability of failure (event not occurring). To further ease the way of interpreting, the estimates of the average marginal effect can be taken to change the results form odds to probabilities. Where, probabilities are the number of times the event occurs divided by the number of times the event could occur and are hence the easiest way of interpreting.

3.4 Likelihood ratio test

As previously mentioned, multiple regressions are needed to determine if and how much financial ratios affect the ability of SMEs to acquire new bank loans. First of all a complete regression is needed with both the solvability and liquidity ratios. Doing so set a basis for further testing and helps determining whether the ratios have any statistically significant effect on the ability of SMEs to acquire new bank loans. Secondly, two reduced models are created with one only containing solvability ratios and one only containing liquidity ratios. These two reduced models will both be compared with the complete model using a likelihood ratio test. The formula used for the joint hypothesis test for a likelihood ratio test equals:

𝐿𝑅 = 𝑇{𝑙𝑜𝑔 (Ʃ̂𝑟 ) − log(Ʃ̂𝑐 )} (13)

Where, Ʃ̂𝑟 is the determinant of the variance–covariance matrix of the residuals for the reduced

model, Ʃ̂𝑐 is the same for the complete model and T is the total number of observations. The

test statistic has a chi square distribution with the total number of restrictions as the degrees of freedom. These regressions can determine the importance of financial ratios and whether solvability ratios have a larger impact than liquidity ratios.

3.5 Data biases

12

are: operational revenue, net income and amount of employees. Moreover, when considering the data as a panel set, a broader perspective is shown as well as several different ways of testing the data. These different kinds of testing can address and counter endogenous problems like endogeneity. The specific use of longitudinal data and its possibilities will be further explained in the next section.

3.6 Longitudinal data

So far, the data has been treated as cross-sectional data. However, the data contains observations of multiple entities over a time span of several years. Therefore, the data can also be treated as longitudinal or panel data. When treading the data as longitudinal data, changes over time or between firms can be observed instead of purely looking at the total influence of the observations. This can answer more complex problems, as it gives a broader view on the situation and is able to find potential endogeneity problems. The econometric setup for panel data equals to:

𝑌𝑖𝑡 = 𝛼 + 𝛽𝑥𝑖𝑡+ 𝜇𝑖𝑡 (14)

13

4. Results

4.1 Determination of models to use

In this section the results of the multiple regressions are shown and interpreted. First of all, the significance of the appropriate model is illustrated. As mentioned in the methodology, the easiest model for a dummy variable in the dependent variable is a linear probability model (LPM). However, the two main problems that accompany the LPM can lead to inaccurate results and these results can diverge significantly from more proper models. In graph 1 and 2 below, the fitted values are shown for the results using a LPM and a logit model. In these results the problem of heteroscedasticity has already been taken into account by using robust standard errors, nonetheless the boundaries of 0 and 1 are clearly exceeded by the LPM. This shows that the LPM is not an appropriate model to describe the data.

Graph 1: Fitted Values of Linear Probability Model Graph 2: Fitted Values of Logit Model

When using an inappropriate model the consequences to the further interpretation of the coefficients can have large discrepancies. In table 3 the coefficients and their significance of the LPM and the logit model are described next to each other. This table shows that the asset based solvability ratio is in the LPM not significant, whereas in the logit model it is significant at the 1% significance level. The difference between those results displays the impact of choosing a certain model. The coefficients of the two models are further explained in the appendix in table 4 and 5.

Table 3: Column 1 shows all the variables and is explained below the table. Column 2 shows the first model, which is a linear probability model with robust standard errors. Column 3 shows the second model, which is a logit model with odd ratios and robust standard errors. Where the dependent variable in both models is the possibility of additional loans.

(1) (2)

VARIABLES LPM Logit OR CurRat -0.033 -0.204

14 (0.003) (0.028) SolRatAs 0.000 -0.101*** (0.006) (0.039) SolRatLi 0.008*** 0.100*** (0.003) (0.021) OperRev 0.000 0.000 (0.000) (0.000) NetIncome 0.000*** 0.000** (0.000) (0.000) NumEmp -0.000 -0.001 (0.000) (0.001) Constant 0.254*** -0.709** (0.058) (0.313) Observations 894 894 R-squared 0.244

Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

CurRat=Current Ratio, LiqRat=Quick Ratio, SolRatAs=Solvability Ratio Asset Based, SolRatLi=Solvability Ratio Liabilities Based, OperRev=Operating Revenue, NumEmp=Number of Employees and p= the significance level

4.2 Complete logit model

15

lead to an increase of 36.1% higher change of receiving an additional loan (ceteris paribus). For the quick ratio, the mean is 1.8, which would lead to an increase of 2% of the probability in the model (ceteris paribus). Finally, the mean of net income equals to 2,859,056, this would lead to an increase of 3% within the model (ceteris paribus). This standardization shows the large impact of the solvability ratio and this variable is therefore also the only variable which is significant at the 1% level.

Table 6: logit model with marginal effect and the dependent variable being possibility of additional loans. Where column 1 shows all the variable, column 2 the margins of the coefficients, column 3 the standard errors, column 4 the z-score of the statistical test, column 5 shows the p-values associated with the z-scores and column 6 and 7 show the outer boundaries of the 95% confidence interval.

Delta-method

dy/dx Std. Err. z P>|z| [95% Conf. Interval] CurRat -.032612 .0220819 -1.48 0.140 -.0758916 .0106677 LiqRat .0112122 .0049072 2.28 0.022 .0015943 .02083 SolRatLi .0081563 .0004343 18.78 0.000 .007305 .0090076 OperRev 2.01e-10 1.53e-10 1.31 0.190 -9.94e-11 5.02e-10 NetIncome 1.04e-08 5.20e-09 2.00 0.045 2.16e-10 2.06e-08 NumEmp -.0001425 .0002385 -0.60 0.550 -.0006099 .000325 CurRat=Current Ratio, LiqRat=Quick Ratio, SolRatLi=Solvability Ratio Liabilities Based, OperRev=Operating Revenue, NumEmp=Number of Employees

With the use of this logit model, predictions can be made about the ability of the SMEs to receive additional loans as is shown in table 8 below. The mean of the dependent dummy variable (possibility for a loan) equals to 0.68 as can be seen in the descriptive statistics in table 1. So in the prediction model, all predicted values classified as 0.68 or higher are put into the category of able to get additional loans and everything below in the category unable to get additional loans. The model allocates all data points a + if according to the model they are able to get an additional loan and a – if they are according to the model unable to get an additional loan. All data points in column D are effectively able to get additional loans and all data points in column ~D are all unable to get additional loans. So, the model correctly predicted 61% of the firms in category D, the companies which are able to get a loan in truth (sensitivity). Furthermore, the model correctly classified 86% of the firms in category ~D, which are unable to get a loan (specificity). Combining these results to their weights, show a total of 70% being correctly predicted. Furthermore, the model correctly allocates 88% in the + category and 57% in the – category.

Table 8: Predictive Logistic Model for the dependent variable possibility for additional loans. The classification in column 1 is based on the predictions of the model. The values D and ~D in column 2 and 3 from true are the actual values according to the data. Column 4 and row 5 show all the totals.

Predictive model True

Classified D ~D Total

+ 335 47 382

- 218 294 512

16

Classified + if predicted (D)>=0.68 True D defined as Possibility Loan=1 and ~D=0

Sensitivity Predicted(+/D) 60,58%

Specificity Predicted(-/~D) 86,22% Positive predictive value Predicted(D/+) 87,70% Negative predictive value Predicted(~D/-) 57,42% False + rate for true ~D Predicted(+/~D) 13,78% False - rate for true D Predicted(-/D) 39,42% False + rate for classified + Predicted(~D/+) 12,30% False - rate for classified - Predicted(D/-) 42.58%

Predicted values correctly classified 70,36%

Column 1 of the second part of the table shows the name of the prediction. Column 2 shows the combination in the table of the prediction and the actual value. Column 3 shows the percentages of the combination of column 2 to the according total.

4.3 Importance solvability and liquidity

To find the significance of the solvability and liquidity, several comparison regressions are considered. The first two comparison models show the full model against one with either missing the solvability ratios or the liquidity ratios. This can be seen in table 9 and 10 below. The difference between the full logit model in table 9 and the logit in table 3 is the absence of robust standard errors with table 9. In general the robust standard errors are preferred, however while using robust standard errors additional testing is not possible. Therefore, to perform the comparison likelihood ratio test the absence of robust standard errors is necessary. In the reduced model of table 9 the solvability ratios which are both significant at the 1% significance level in the full model are removed. Due to this change, the significance level of the current ratio changes from being significant at the 10% level to the 1% level. Furthermore, below the table the likelihood ratio test between the full model and reduced model shows that the reduced model, at the 1% level, significantly changed due to the removal of the two variables. Therefore, the full model is statistically better and the solvability ratios cannot be removed from the model. On the other hand, table 10 shows the full model compared against a reduced model without the liquidity ratios. In this model, only the significance level of the constant changes and the likelihood ratio test shows no significant result. This means that the liquidity ratio can be removed from the model without serious consequences. Hence, the solvability ratios are clearly of more importance than the liquidity ratios. In table 11, 12 and 13 in the appendix are 3 other comparing models of the key variables.

Table 9: Comparison model 1, full model vs liquidity ratio model and likelihood ratio test with the possibility for additional loans as the dependent variable and both in odd ratios.

(1) (2)

17 (0.106) (0.117) LiqRat 0.058 0.006 (0.055) (0.038) SolRatAs -0.101*** (0.038) SolRatLi 0.100*** (0.021) OperRev 0.000 -0.000 (0.000) (0.000) NetIncome 0.000** 0.000*** (0.000) (0.000) NumEmp -0.001 0.001 (0.001) (0.001) Constant -0.709** -0.165 (0.309) (0.208) Observations 894 894

Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

CurRat=Current Ratio, LiqRat=Quick Ratio, SolRatAs=Solvability Ratio Asset Based, SolRatLi=Solvability Ratio Liabilities Based, OperRev=Operating Revenue, NumEmp=Number of Employees and p= the significance level

Likelihood-ratio test LR chi2(2) = 214.19 (Assumption: reduced_model nested in full_model) Prob > chi2 = 0.0000

Table 10: Comparison model 2, full model vs solvability ratio model and likelihood ratio test with the possibility for additional loans as the dependent variable and both in odd ratios.

(1) (2)

VARIABLES Full Model Sol Ratio CurRat -0.204* (0.106) LiqRat 0.058 (0.055) SolRatAs -0.101*** -0.100*** (0.038) (0.038) SolRatLi 0.100*** 0.097*** (0.021) (0.021) OperRev 0.000 0.000 (0.000) (0.000) NetIncome 0.000** 0.000** (0.000) (0.000) NumEmp -0.001 -0.001 (0.001) (0.001) Constant -0.709** -0.853*** (0.309) (0.286) Observations 894 894

18

CurRat=Current Ratio, LiqRat=Quick Ratio, SolRatAs=Solvability Ratio Asset Based, SolRatLi=Solvability Ratio Liabilities Based, OperRev=Operating Revenue, NumEmp=Number of Employees and p= the significance level

Likelihood-ratio test LR chi2(2) = 4.26 (Assumption: reduced_model nested in full_model) Prob > chi2 = 0.1190

4.4 Panel Model

19

Table 14: Coefficients of Panel Data for the variables shown in column 1, with pooled ordinary least squares regression with robust standard errors in column 2, entity fixed effect in column 3, time and entity fixed effect in column 4 and random effects in column 5. With all models having possibility for additional loans as the dependent variable.

(1) (2) (3) (4)

VARIABLES Pool Fixed Time Random

CurRat -0.033 -0.039* -0.035 -0.038** (0.022) (0.023) (0.023) (0.018) LiqRat 0.011*** 0.001 0.001 0.011 (0.003) (0.008) (0.008) (0.007) SolRatAs 0.000 0.043*** 0.042*** 0.008 (0.006) (0.009) (0.009) (0.006) SolRatLi 0.008*** -0.003 -0.003 0.007** (0.003) (0.004) (0.004) (0.003) OperRev 0.000 0.000 -0.000 0.000 (0.000) (0.000) (0.000) (0.000) NetIncome 0.000*** 0.000*** 0.000*** 0.000*** (0.000) (0.000) (0.000) (0.000) NumEmp -0.000 0.001 0.000 -0.000 (0.000) (0.001) (0.001) (0.000) year2014 -0.123*** (0.042) year2015 -0.134*** (0.041) year2016 -0.130*** (0.041) year2017 -0.046 (0.038) Constant 0.254*** -0.531*** -0.364*** 0.114* (0.058) (0.115) (0.123) (0.066) Observations 894 894 894 894 R-squared 0.244 0.346 0.362 Number of Comp 267 267 267

Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

(RFE) F test that all u_i=0: F(266, 616) = 2.68 Prob > F = 0.0000

CurRat=Current Ratio, LiqRat=Quick Ratio, SolRatAs=Solvability Ratio Asset Based, SolRatLi=Solvability Ratio Liabilities Based, OperRev=Operating Revenue, NumEmp=Number of Employees, with year2014, year2015, year2016 and year 2017 as the year dummy variables, p= the significance level and RFE=Redundant Fixed Effects

5. Conclusion

20

loans. This research examines 350 firms over a time span of 5 years to check the null hypothesis:

The solvability ratios of asset based and liability based have a larger impact than the liquidity ratios of current ratio and quick ratio, on the ability of acquiring additional bank loans for SMEs in financial difficulties.

In all of the tests performed, solvability ratios show to have the largest impact and thus, confirming the null hypotheses. The comparison models of table 9 and 10 of the logit model show that the model is dependent on the solvability ratios using a likelihood ratio test at a 1% significance level. On the other hand, the removal of the liquidity ratio out of the model gives no statistical significant result. Furthermore, the model correctly classifies the ability for SMEs to acquire additional bank loans for 70% of the times. In the prediction model the specificity equals to 86% and the sensitivity equals to 61%, which means that the model is better in classifying the cases when companies cannot receive additional bank loans than in the cases when they can. Furthermore, the time and entity fixed effects model shows a significance of the asset based liability ratio at a 1% level. Yet, none of the liquidity ratios show a statistical significant result. The time and entity fixed effect model is in this paper considered as the best model for describing the data, due to the ability of the model to counter the problems of heteroscedasticity and endogeneity. Other models used to describe the data are the linear probability model, pooled ordinary least squares, entity fixed effects models and random affects models.

Not being able to be reject the null hypothesis, confirms to a certain extent the chain of thought behind the hypotheses. The reasoning behind the hypotheses was that by acquiring additional bank loans, any liquidity problems could be solved. Therefore, making liquidity ratios obsolete for assessing the ability for firms to get additional bank loans and making solvability ratios the key factor in predicting that ability.

21

22

References

Altman, E. I. (1968). Financial ratios, discriminant analysis and the prediction of corporate bankruptcy. The journal of finance, 23(4), 589-609.

Altman, E. I., Iwanicz‐Drozdowska, M., Laitinen, E. K., & Suvas, A. (2017). Financial distress prediction in an international context: A review and empirical analysis of Altman's Z‐ score model. Journal of International Financial Management & Accounting, 28(2), 131-171. Aziz, M. A., & Dar, H. A. (2006). Predicting corporate bankruptcy: where we stand?. Corporate Governance: The international journal of business in society.

Barnes, P. (1987). The analysis and use of financial ratios: A review article. Journal of Business Finance & Accounting, 14(4), 449-461.

Baum, C. F., & Christopher, F. (2006). An introduction to modern econometrics using Stata. Stata press.

Beaver, W. H. (1966). Financial ratios as predictors of failure. Journal of accounting

research, 71-111.

Berk, J., DeMarzo, P., Harford, J., & Stangeland, D. A. (2012). Fundamentals of Corporate

Finance, Canadian Edition with MyFinanceLab. Pearson Education Canada.

Brooks, C. (2019). Introductory econometrics for finance. Cambridge university press.

Chen, K. H., & Shimerda, T. A. (1981). An empirical analysis of useful financial ratios. Financial management, 51-60.

Dziak, J. J., Lanza, S. T., & Tan, X. (2014). Effect size, statistical power, and sample size requirements for the bootstrap likelihood ratio test in latent class analysis. Structural equation modeling: a multidisciplinary journal, 21(4), 534-552.

Edmonds, T. P., Edmonds, C. D., Tsay, B. Y., & Olds, P. R. (2000). Fundamental managerial

accounting concepts. Irwin/McGraw-Hill.

Elam, R. (1975). The effect of lease data on the predictive ability of financial ratios. The Accounting Review, 50(1), 25-43.

Gibson, C. (1987). How chartered financial analysts view financial ratios. Financial Analysts Journal, 43(3), 74-76.

23

Lev, B., & Sunder, S. (1979). Methodological issues in the use of financial ratios. Journal of

Accounting and Economics, 1(3), 187-210.

Monea, M. (2009). Financial ratios–reveal how a business is doing?. Annals of the University

of Petroşani, Economics, 9(2), 137-145.

Moore, D. S., McCabe, G. P., Alwan, L. C., Craig, B. A., & Duckworth, W. M. (2016). The practice of statistics for business and economics. WH Freeman.

Muller, G. H., Steyn-Bruwer, B. W., & Hamman, W. D. (2009). Predicting financial distress of companies listed on the jse - a comparison of techniques. South African Journal of

Business Management, 40(1), 21–32.

Myšková, R., & Hájek, P. (2017). Comprehensive assessment of firm financial performance using financial ratios and linguistic analysis of annual reports. Journal of International

Studies, volume 10, issue: 4.

Ohlson, J. A. (1980). Financial ratios and the probabilistic prediction of bankruptcy. Journal

of accounting research, 109-131.

Palepu, K. G., Healy, P. M., & Peek, E. (2013). Business analysis and valuation: IFRS

edition. Cengage learning.

Simel, D. L., Samsa, G. P., & Matchar, D. B. (1991). Likelihood ratios with confidence: sample size estimation for diagnostic test studies. Journal of clinical epidemiology, 44(8), 763-770.

Stock, J. H., & Watson, M. W. (1988). Testing for common trends. Journal of the American

statistical Association, 83(404), 1097-1107.

Stock, J. H., & Watson, M. W. (2006). Forecasting with many predictors. Handbook of

economic forecasting, 1, 515-554.

Tinsley, P. A. (1970). Capital structure, precautionary balances, and valuation of the firm: The problem of financial risk. Journal of Financial and Quantitative Analysis, 33-62.

Voulgaris, F., Doumpos, M., & Zopounidis, C. (2000). On the evaluation of Greek industrial SME's performance via multicriteria analysis of financial ratios. Small business

economics, 15(2), 127-136.

24

Appendix

Table 4: complete linear probability model, with column 1 as the variables, column 2 as the coefficients, column 3 as the standard errors, column 4 as the t-statistic for the statistical test, with column 5 as the p-value connected to the t-statistic in column 4 and column 6 and 7 are the boundaries of the 95% confidence interval according to the t-test

Robust

PosLoan Coef. Std. Err. t P>|t| [95% Conf. Interval] CurRat -.0326965 .0220269 -1.48 0.138 -.0759274 .0105345 LiqRat .0112203 .0028569 3.93 0.000 .0056132 .0168275 SolRatAs .00039 .005564 0.07 0.944 -.0105302 .0113102 SolRatLi .008446 .00264 3.20 0.001 .0032645 .0136274 OperRev 1.73e-10 1.63e-10 1.06 0.290 -1.47e-10 4.93e-10 NetIncome 5.49e-09 1.20e-09 4.57 0.000 3.13e-09 7.85e-09 NumEmp -.0001454 .0002379 -0.61 0.541 -.0006123 .0003215 _cons .2540703 .0576216 4.41 0.000 .1409796 .3671611 PosLoan= Possibilty for additional loan, CurRat=Current Ratio, LiqRat=Quick Ratio, SolRatAs=Solvability Ratio Asset Based, SolRatLi=Solvability Ratio Liabilities Based, OperRev=Operating Revenue, NumEmp=Number of Employees

Table 5: logit model with odd ratios, with column 1 as the variables, column 2 as the coefficients, column 3 as the standard errors, column 4 as the t-statistic for the statistical test, with column 5 as the p-value connected to the t-statistic in column 4 and column 6 and 7 are the boundaries of the 95% confidence interval according to the t-test

Robust

PosLoan Odds Ratio Std. Err. z P>|z| [95% Conf. Interval] CurRat .8154479 .1159584 -1.43 0.151 .6170972 1.077554 LiqRat 1.059424 .0294847 2.07 0.038 1.003183 1.118818 SolRatAs .9041339 .0350706 -2.60 0.009 .8379447 .9755514 SolRatLi 1.10489 .0235882 4.67 0.000 1.059611 1.152103 OperRev 1 8.52e-10 1.11 0.266 1 1 NetIncome 1 3.30e-08 2.06 0.039 1 1 NumEmp .9991901 .0013614 -0.59 0.552 .9965254 1.001862 _cons .4919781 .1539686 -2.27 0.023 .2664129 .908524 PosLoan= Possibilty for additional loan, CurRat=Current Ratio, LiqRat=Quick Ratio, SolRatAs=Solvability Ratio Asset Based, SolRatLi=Solvability Ratio Liabilities Based, OperRev=Operating Revenue, NumEmp=Number of Employees

Table 7: correlation diagram between variables

CurRat LiqRat SolRatLi

CurRat 1,0000

LiqRat 0,4002 1,0000

SolRatLi 0,3056 0,0409 1,0000

CurRat=Current Ratio, LiqRat=Quick Ratio, SolRatLi=Solvability Ratio Liabilities

Table 11: Key Variables Comparison 1, with liquidity ratios and solvability ratios vs liquidity ratios. With possibility for additional loans as the dependent variable.

(1) (2)

25 (0.042) (0.034) SolRatAs -0.036 (0.032) SolRatLi 0.067*** (0.017) Constant -1.292*** -0.346** (0.253) (0.167) Observations 1,078 1,078

Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

CurRat=Current Ratio, LiqRat=Quick Ratio, SolRatAs=Solvability Ratio Asset Based, SolRatLi=Solvability Ratio Liabilities Based and p= the significance level

Likelihood-ratio test LR chi2(2) = 286.53 (Assumption: reduced_model nested in full_model) Prob > chi2 = 0.0000

Table 12: Key Variables Comparison 2, with liquidity ratios and solvability ratios vs solvability ratios. With possibility for additional loans as the dependent variable.

(1) (2)

VARIABLES Full Reduced CurRat -0.151 (0.099) LiqRat 0.052 (0.042) SolRatAs -0.036 -0.038 (0.032) (0.032) SolRatLi 0.067*** 0.066*** (0.017) (0.017) Constant -1.292*** -1.378*** (0.253) (0.232) Observations 1,078 1,078

Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

CurRat=Current Ratio, LiqRat=Quick Ratio, SolRatAs=Solvability Ratio Asset Based, SolRatLi=Solvability Ratio Liabilities Based and p= the significance level

Likelihood-ratio test LR chi2(2) = 3.16 (Assumption: reduced_model nested in full_model) Prob > chi2 = 0.2060

Table 13: Comparison model 3, with total model without liquidity ratios vs total model without solvency ratios. With possibility for additional loans as the dependent variable.

(1) (2)

26 NetIncome 0.000** 0.000*** (0.000) (0.000) NumEmp -0.001 0.001 (0.001) (0.001) CurRat 0.303*** (0.117) LiqRat 0.006 (0.038) Constant -0.853*** -0.165 (0.286) (0.208) Observations 894 894

Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

CurRat=Current Ratio, LiqRat=Quick Ratio, SolRatAs=Solvability Ratio Asset Based, SolRatLi=Solvability Ratio Liabilities Based, OperRev=Operating Revenue, NumEmp=Number of Employees and p= the significance level

Table 15: Time dummies redundancy test for the dummies of table 14. test year2014=year2015=year2016=year2017==0 ( 1) year2014 - year2015 = 0 ( 2) year2014 - year2016 = 0 ( 3) year2014 - year2017 = 0 ( 4) year2014 = 0 F( 4, 616) = 4.01 Prob > F = 0.0032

Table 16: Hausman fixed random test with possibility for additional loans as the dependent variable.

Variables (b) Fixed (B) random (b-B) Difference sqrt(diag(V_b-V_B)) S.E. CurRat -.0389389 -.0377903 -.0011486 .0141594

LiqRat .0010735 .0105067 -.0094332 .0048843

SolRatAs .04287 .0079553 .0349147 .0058019

27

NetIncome 6.22e-09 6.00e-09 2.14e-10 1.16e-09

NumEmp .0009785 -.0000805 .001059 .0006218 CurRat=Current Ratio, LiqRat=Quick Ratio, SolRatAs=Solvability Ratio Asset Based, SolRatLi=Solvability Ratio Liabilities Based, OperRev=Operating Revenue, NumEmp=Number of Employees and p= the significance level

b = consistent under Ho and Ha; obtained from xtreg

B = inconsistent under Ha, efficient under Ho; obtained from xtreg Test: Ho: difference in coefficients not systematic

chi2(5) = (b-B)'[(V_b-V_B)^(-1)](b-B) = 89.42 Prob>chi2 = 0.0000 List of companies: 1. 3 PET HOLDING B.V. 2. A WARE WORKUM B.V. 3. ABB HOLDINGS B.V.

4. ABBOTT VASCULAR NETHERLANDS B.V. 5. AC ANALYTICAL CONTROLS B.V.

6. ACT COMMODITIES GROUP B.V.

7. ACTIVISION BLIZZARD INTERNATIONAL B.V. 8. ADOMEX INTERNATIONAL B.V.

9. AFA DISPENSING GROUP B.V.

10. AKAMAI TECHNOLOGIES NETHERLANDS B.V. 11. ALBEL HOLDING B.V.

12. ALDI HOLDING B.V. 13. ALIGN TECHNOLOGY, B.V.

14. ALL OPTIONS INTERNATIONAL HOLDING B.V. 15. ALPINVEST PARTNERS B.V.

16. ALSO INTERNATIONAL B.V.

17. AMF BAKERY SYSTEMS EUROPE B.V. 18. ANCOFERWALDRAM STEELPLATES B.V. 19. ANDRE RIEU PRODUCTIONS B.V.

20. ANNA VAN TOOR HOLDING B.V. 21. ANSALDO THOMASSEN B.V. 22. ANTALIS B.V.

23. ANTHURA B.V. 24. AO&G HOLDING B.V. 25. AP B.V.

26. APL LOGISTICS EUROPE B.V. 27. AS 24 NEDERLAND B.V. 28. ATOTECH NEDERLAND B.V. 29. ATRIUM GROUP SERVICES B.V. 30. AUTIMAAT B.V.

31. AUTO HAAIMA B.V.

32. AUTOBEDRIJF NOTEBOOM ROTTERDAM B.V. 33. AUTOLAND VAN DEN BRUG B.V.

28

37. B.V. NEWOMIJ-GROEP

38. B.V. SCHEEPSWERF MAASKANT 39. BAN BOUW HOLDING B.V. 40. BASF AGRO B.V. 41. BAY HOLLAND B.V. 42. BAYARDS HOLDING B.V. 43. BEGIJNENHOF B.V. 44. BEHOLD WAALWIJK B.V. 45. BELEGGINGSMAATSCHAPPIJ "ECA" B.V. 46. BENTLEY SYSTEMS NETHERLANDS B.V. 47. BERCO CAR CARPETS B.V.

48. BIOGEN NETHERLANDS B.V. 49. BLUE SKY GROUP HOLDING B.V. 50. BLUEKENS TRUCK EN BUS B.V. 51. BOL HOLDING B.V.

52. BOLIDT INTERNATIONAL HOLDING B.V. 53. BORCHWERF VASTGOED B.V.

54. BOREALIS PLASTOMERS B.V. 55. BOUWBEDRIJF JOOS HOEX B.V. 56. BOUWERS MET VISIE B.V. 57. BRAINNET B.V.

58. BRASKEM NETHERLANDS B.V. 59. BRB INTERNATIONAL B.V. 60. BURANDO HOLDING B.V.

61. BURGHOUWT BOUWBESLAG B.V. 62. BURO VAN ROOSMALEN GGZ B.V. 63. CAE CENTER AMSTERDAM B.V. 64. CALDERYS THE NETHERLANDS B.V. 65. CALVIN KLEIN EUROPE B.V.

66. CAREYN MAATSCHAPPELIJKE DIENSTVERLENING B.V. 67. CARU CONTAINERS B.V.

68. CBS INTERNATIONAL (NETHERLANDS) B.V. 69. CENTRALPOINT AMSTELVEEN B.V.

70. CHECKPOINT APPAREL LABELING B.V. 71. CHEP PALLECON SOLUTIONS B.V. 72. CISCO SYSTEMS MANAGEMENT B.V. 73. CLARIOS NETHERLANDS B.V. 74. COATEX NETHERLANDS B.V. 75. COMPAREX NEDERLAND B.V. 76. CONTAINERSHIPS ROTTERDAM B.V. 77. COXGEELEN B.V. 78. CPH CHEMICALS B.V. 79. CROCS EUROPE B.V.

80. CYCLING SPORTS GROUP EUROPE B.V. 81. D.P. SUPPLY B.V.

82. DAIRY TRADING INTERNATIONAL B.V. 83. DANISH CROWN FOODS HAARLEM B.V. 84. DAS LEGAL FINANCE B.V.

85. DE WAAL AUTOGROEP B.V.

86. DEKRA CLAIMS SERVICES NETHERLANDS B.V. 87. DEVELING BEHEER B.V.

29

90. DOCS INSOURCING B.V. 91. DOEHLER ROGGEL B.V. 92. DRAKA POLYMER FILMS B.V. 93. DSV AIR & SEA B.V.

94. DUNI BENELUX B.V. 95. EASTMAN CHEMICAL B.V. 96. EBN B.V.

97. ECCO LEATHER B.V.

98. ECEM EUROPEAN CHEMICAL MARKETING B.V. 99. EDELMAN B.V.

100. EISAI B.V.

101. ELLIS ENTERPRISES B.V. 102. EMESA NEDERLAND B.V. 103. EUROFINS LCPL B.V.

104. EVONIK INTERNATIONAL HOLDING B.V. 105. EXCLUTON B.V.

106. FABRICOM B.V.

107. FAMILIE TER MATEN BEHEER B.V. 108. FAURECIA AUTOMOTIVE SEATING B.V. 109. FEI EUROPE B.V.

110. FONDSENBEHEER NEDERLAND B.V. 111. FOREVER DIRECT E.U. B.V.

112. FOREVER DIRECT E.U. HOLDING B.V. 113. FRIEDE & GOLDMAN MARKETING B.V. 114. FUNDA B.V.

115. FUNDA REAL ESTATE B.V. 116. FUTURE PIPE INDUSTRIES B.V. 117. GARTNER NEDERLAND B.V. 118. GASTERRA B.V.

119. GEAS ENERGIEWACHT B.V. 120. GEBERIT B.V.

121. GEBR. BURGHOUWT PARTICIPATIE B.V. 122. GEDORE TECHNAG B.V.

123. GEERT-JAN DE KOK HOLDING B.V. 124. GENMAB B.V.

125. GEODIS SCO NETHERLANDS B.V.

126. GERMANISCHER LLOYD NETHERLANDS B.V. 127. GIBARI B.V.

128. GILEAD SCIENCES NETHERLANDS B.V. 129. GOEDEGEBUUR VLEES ROTTERDAM B.V. 130. GOODYEAR EUROPE B.V.

131. GREENSTREAM B.V. 132. GROENLEVEN B.V. 133. GROUPM B.V.

134. GS STAALWERKEN B.V. 135. GULF HOTELS HOLDINGS B.V. 136. GVT TRANSPORT & LOGISTICS B.V. 137. HABOMIJ I. B.V.

30

143. HEMPEL (THE NETHERLANDS) B.V. 144. HENRY BATH B.V.

145. HEROS SLUISKIL B.V.

146. HEUVER BANDEN GROOTHANDEL B.V. 147. HIROSE ELECTRIC EUROPE B.V.

148. HITACHI CAPITAL MOBILITY HOLDING NETHERLANDS B.V. 149. HMSHOST INTERNATIONAL B.V.

150. HOCHWALD FOODS NEDERLAND B.V. 151. HOLDING DAELMANS II B.V.

152. HOLLANDIA INFRA B.V.

153. HOLLANDIA STRUCTURES B.V. 154. HOLLANDIA SYSTEMS B.V.

155. HOTEL MAATSCHAPPIJ OUD-AMSTERDAM B.V. 156. IBSOR B.V.

157. IDDINK LEARNING MATERIALS B.V. 158. IGEPA NEDERLAND HOLDING B.V. 159. IGLO NEDERLAND B.V.

160. IGT-EUROPE B.V. 161. IMC TRADING B.V.

162. INDUSTRIE- EN HANDELSONDERNEMING VREUGDENHIL B.V. 163. INFRADATA GROUP B.V.

164. INGMA HOLDING B.V. 165. INTECSEA B.V.

166. INTER METALS TRADING B.V. 167. INTERFOOD B.V.

168. INTERNATIONAL FURAN CHEMICALS B.V. 169. INTER-SPRINT BANDEN B.V.

170. INTERSYSTEMS B.V. 171. IVECO NEDERLAND B.V.

172. J.A. TER MATEN PLUIMVEEBEDRIJF B.V.

173. J.M. DE JONG VERENIGDE WERKPLAATSEN BEHEER B.V. 174. J.P. BEEMSTERBOER FOOD TRADERS B.V.

175. JAN SNEL BOUW B.V. 176. JAN ZANDBERGEN B.V. 177. JOIN B.V.

178. JONGEN PROJECTONTWIKKELING B.V. 179. JORRITSMA BOUW B.V.

180. JUNIPER NETWORKS INTERNATIONAL B.V. 181. K.B.K. VASTGOEDONDERHOUD B.V. 182. K2M HOLDING B.V.

183. KALLISTE WONINGBOUWONTWIKKELING B.V. 184. KASNED B.V.

185. KINDERRIJK BUITENSCHOOLSE OPVANG B.V. 186. KIVO PLASTIC VERPAKKINGEN B.V.

187. KLAAS PUUL B.V. 188. KMG KASHAGAN B.V.

31 196. LEGRAND NEDERLAND B.V. 197. LG CNS EUROPE B.V. 198. LINDORFF NETHERLANDS B.V. 199. LOPAREX B.V. 200. LOSUR B.V.

201. LUCHTHAVEN HOTEL BELEGGINGSMAATSCHAPPIJ B.V. 202. LUKINTER FINANCE B.V.

203. M. DE KONING VASTGOED B.V. 204. MAARS HOLDING B.V.

205. MAASTRICHT AACHEN AIRPORT BEHEER & INFRASTRUCTUUR B.V. 206. MACSTEEL INTERNATIONAL TRADING B.V.

207. MARRIOTT INTERNATIONAL MANAGEMENT COMPANY B.V. 208. MARS FOOD EUROPE B.V.

209. MARTIJN TRADING HOLDING B.V. 210. MEDA PHARMA B.V.

211. MEHOCO B.V.

212. MERCK SHARP & DOHME INTERNATIONAL SERVICES B.V. 213. MICHEL OPREY & BEISTERVELD NATUURSTEEN B.V. 214. MICROSOFT DATACENTER NETHERLANDS B.V. 215. MK2 GROEP B.V.

216. MODEHUIZEN CLAUDIA-STRATER B.V. 217. MONIER B.V.

218. MUNDIPHARMA DC B.V.

219. NAGRON NATIONAAL GRONDBEZIT B.V. 220. NALCO NETHERLANDS B.V.

221. NEOMET BEHEER B.V.

222. NETAPP HOLDING & MANUFACTURING B.V. 223. NEXANS NEDERLAND B.V.

224. NIKE TRADING COMPANY B.V. 225. NIKON INSTRUMENTS EUROPE B.V. 226. NNZ BEHEER B.V.

227. NOVO NORDISK B.V.

228. OBESITAS NEDERLAND B.V.

229. OERLEMANS FOODS WAALWIJK B.V. 230. OI EUROPEAN GROUP B.V.

231. OILTANKING TERNEUZEN B.V. 232. OLYMPIC FRUIT B.V.

233. OOSTERHOF HOLMAN INFRA B.V. 234. OPEL NEDERLAND B.V.

235. ORANGE CYBERDEFENSE NETHERLANDS B.V. 236. ORMCO B.V.

237. ORTHOPEDISCH CENTRUM OOST-NEDERLAND B.V. 238. P.M.S. BEHEER B.V.

239. P.W. KUSTER BEHEER B.V. 240. PACTON TRAILERS B.V. 241. PAK GROUP B.V.

242. PALING- EN ZALMFILEERDERIJ J. FOPPEN JZN. B.V. 243. PAPYRUS GROEP NEDERLAND B.V.

32

249. PRIME OIL & GAS B.V. 250. PVG HOLDING B.V.

251. QUADIENT TECHNOLOGIES NETHERLANDS B.V. 252. QUAKER CHEMICAL B.V.

253. RAINBOW INTERNATIONAL B.V. 254. RECKITT BENCKISER (ENA) B.V. 255. RECYCLING KOMBINATIE REKO B.V. 256. RICHTORIA B.V.

257. RICOH EUROPE (NETHERLANDS) B.V. 258. RICOH EUROPE SCM B.V.

259. ROCKWELL AUTOMATION B.V. 260. ROLAND B.V.

261. ROLLDOCK B.V.

262. ROTTERDAM WORLD GATEWAY B.V. 263. S.A.B. PROFIEL B.V.

264. S.I.T. CONTROLS B.V.

265. SAFANDARLEY HOLDING B.V. 266. SCHAAP EN CITROEN B.V.

267. SCHOLTENS HOLDING GROEP B.V. 268. SCHUR FLEXIBLES BENELUX B.V. 269. SEA-OIL HOLDING B.V.

270. SHELL CHEMICALS EUROPE B.V.

271. SHELL KAZAKHSTAN DEVELOPMENT B.V. 272. SHELL LUBRICANTS SUPPLY COMPANY B.V. 273. SHELL TRADING ROTTERDAM B.V.

274. SHELL TRADING RUSSIA B.V. 275. SHIPYARD DE HOOP B.V. 276. SIKA NEDERLAND B.V.

277. SKECHERS USA BENELUX B.V. 278. SLIM MET ENERGIE B.V. 279. SLTN IT PRODUCTS B.V. 280. SMEETS BOUW B.V. 281. SMIT'S BOUWBEDRIJF B.V. 282. SNACK CONNECTION B.V.

283. SOLIDUS SOLUTIONS ZUTPHEN B.V. 284. SOLINA NETHERLANDS HOLDING B.V. 285. SQUARE DRANKEN NEDERLAND B.V. 286. STAFFING ENTERPRISES B.V.

287. STAPPERT NOXON B.V.

288. STIGTERSTAAL HOLDING B.V. 289. STONE FASHION GROUP B.V. 290. STRUPLAST (HOLDING) B.V.

291. STRYKER EMEA SUPPLY CHAIN SERVICES B.V. 292. STUURGROEP HOLLAND B.V.

293. SUNRISE MEDICAL B.V. 294. SUPPLY POINT CUIJK B.V. 295. SURFMARKET B.V. 296. SURFNET B.V.

297. SWEDISH MATCH LIGHTERS B.V. 298. SYNRES B.V.

299. SYNTEGON PACKAGING TECHNOLOGY B.V. 300. SYSTEMAIR B.V.

33

302. TASS INTERNATIONAL SAFETY CENTER B.V. 303. TECHNOGYM BENELUX B.V.

304. TELEDYNE DALSA B.V. 305. TERBERG MACHINES B.V.

306. TESLIN CAPITAL MANAGEMENT B.V. 307. THEOBROMA B.V.

308. THERMEN HOLIDAY HOLDING B.V. 309. THERMO KING TRANSPORTKOELING B.V.

310. THOMSON MULTIMEDIA DISTRIBUTION (NETHERLANDS) B.V. 311. THUISZORG INIS B.V.

312. TIMZO BEHEER B.V.

313. TIP TRAILER SERVICES MANAGEMENT B.V. 314. TOINE BROCK CONSTRUCTIE\MECHANISATIE B.V. 315. TOLL GLOBAL FORWARDING (NETHERLANDS) B.V. 316. TOLSMA TECHNIEK EMMELOORD B.V.

317. TREDEGAR FILM PRODUCTS B.V. 318. TREFOIL TRADING B.V.

319. TRIFLEET LEASING (THE NETHERLANDS) B.V. 320. TRIFLEET LEASING HOLDING B.V.

321. TRIMBLE EUROPE B.V.

322. TROMP MEDICAL ENGINEERING B.V. 323. UNDER ARMOUR EUROPE B.V. 324. UNIVAR SOLUTIONS B.V.

325. V. HOUT KABELRECYCLING B.V. 326. VALK WELDING B.V.

327. VALLEI AUTO GROEP HOLDING B.V. 328. VAN DE REYT MESTSTOFFEN B.V. 329. VAN DEN BOSCH TRANSPORTEN B.V.

330. VAN DEN HEUVEL AANNEMINGSBEDRIJF B.V. 331. VAN DIJK GROEP WOERDEN B.V.

332. VAN DIJK HOLDING B.V.

333. VAN GALEN SERVICE EN ONDERHOUD ROTTERDAM B.V. 334. VAN HOLLANT HEILOO B.V.

335. VAN MIERT PROCURATIE BREUKELEN B.V. 336. VAN SILLEVOLDT RIJST B.V.

337. VAN SPIJK B.V.

338. VANDEBRON ENERGIE B.V.

339. VARO ENERGY NETHERLANDS B.V. 340. VASTGOED BEHEER PG B.V.

341. VDB GROEP B.V.

342. VEKOMA RIDES MANUFACTURING B.V. 343. VHP SECURITY PAPER B.V.

344. VVM BEHEER B.V.

345. VWR INTERNATIONAL B.V. 346. W. HERMS & ZN. B.V.

347. WARNER BROS. ENTERTAINMENT NEDERLAND B.V.

348. WILLEMSEN INTERIEURBOUW & SCHEEPSBETIMMERING B.V. 349. YARDEN UITVAARTZORG B.V.