Small Business Economics

, Volume 41, Issue 3, pp 701–715

Perceived credit constraints in the European Union

Authors

  • Erik Canton
    • Ecorys Nederland
  • Isabel Grilo
    • DG Economic and Financial AffairsEuropean Commission
  • Josefa Monteagudo
    • DG Economic and Financial AffairsEuropean Commission
    • Erasmus School of EconomicsErasmus University Rotterdam
    • Panteia/EIM Business and Policy Research
Article

DOI: 10.1007/s11187-012-9451-y

Cite this article as:
Canton, E., Grilo, I., Monteagudo, J. et al. Small Bus Econ (2013) 41: 701. doi:10.1007/s11187-012-9451-y

Abstract

The promotion and support of small and medium-sized enterprises (SMEs) is an essential component of policies designed to help improve Europe’s economic performance. A crucial issue is whether SMEs face difficulty obtaining bank loans. Using pre-crisis survey data from 2005 and 2006 for nearly 3,500 SMEs (firms with fewer than 250 employees) in the European Union (EU), we investigate the determinants of perceived bank loan accessibility at the firm level and at the country level. Based on hierarchical (multi-level) binomial logit regressions, our findings show that the youngest and smallest SMEs have the worst perception of access to bank loans. The SMEs in nations with concentrated banking sectors are more positive about loan accessibility. In addition, a high fraction of foreign-owned banks is associated with improved perception of loan accessibility in the EU 15 but not in the EU 10.

Keywords

Bank loansMoney supplySMEsCredit constraintsConcentration indexMulti-level

JEL Classifications

E44E51G15G21

1 Introduction

For many years, the European Union (EU) has identified the promotion of small and medium-sized enterprises (SMEs) as being of vital importance, not least because of the often-discussed relationship between entrepreneurship and economic prosperity (Carree and Thurik 2007, 2010; Van Praag and Versloot 2007). SME promotion is a significant component of the strategy for improving European competitiveness that was initiated in Lisbon at the European Council meeting in March 2000. This initiative was further strengthened in 2005 with the re-launch of the Lisbon Strategy and was brought to the fore by the Small Business Act in 2008, which is consistent with the objectives of the new Europe 2020 Strategy. In this context, the question of whether SMEs face (undue) difficulty in trying to obtain forms of external financing, for example bank loans, is pressing. Restricted access to bank loans can arise because of information asymmetry between the lender and the borrower if the banks cannot determine the quality of the projects undertaken by the borrowing SME. It is believed that SMEs may be especially sensitive to informational asymmetry, because of the greater perceived opacity of small and young firms.

The purpose of this paper is to shed more light on the factors that affect perception of access to bank loans of European SMEs with fewer than 250 employees, both at the firm level and at the country level. For this purpose, we use a 2005/2006 Eurobarometer dataset from the European Commission that includes detailed financing information for almost 3,500 SMEs in the European Union (EU 25). At the firm level, we are interested in whether perceived loan accessibility depends on firm age and firm size. The between-country variation in perception is linked to several characteristics of the banking sector. These characteristics include the degree of concentration of the banking system, the percentage of foreign-owned banks, and average bank size. Hierarchical multi-level regressions are performed to assess the influences of firm age, firm size, and the three banking sector characteristics on perceived loan accessibility.

Significant research has been performed to determine whether cash flow has a positive effect on firm growth, indicating credit constraints (Wagenvoort 2003a). We also note there has been much research (more theoretical than empirical) on under-investment, given the socially optimum levels (De Meza and Webb 1987; De Meza 2002). Denial rates for loan applications and their determinants have also received much attention (Levenson and Willard 2000). This paper adds to current knowledge of restricted access to finance by SMEs in at least two ways. First, this paper focuses on bank loans. This focus is important because banks are by far the most used financial institution when SMEs need financing (Petersen and Rajan 1994). Although some cross-country studies exist on the more general concept of financing obstacles to firm growth (Clarke et al. 2006; Beck et al. 2006, 2008), there have been no direct assessments of SMEs’ access to bank loans in both developed and less developed nations. Furthermore, this paper’s focus on bank loans is deeper because the country-specific variables are all related to the structure of the banking sector.

Second, there has not been much work on what affects perception of SMEs regarding the access to bank loans, although such information is vital if we want to encourage the creation and growth of SMEs. Concentrating on SME perception enables large-scale analysis because information on such perception is available for every firm in our sample. Hence, our sample also includes firms that require loans but do not apply for them because they fear they will be rejected. Such discouraged firms form a substantial part of all SMEs (Han et al. 2009) and should not be neglected in studies of credit accessibility (Chakravarty and Xiang 2009). Note that we use a representative sample of all SMEs in the EU. SMEs represent 99 % of all enterprises in the EU, and SMEs are responsible for 85 % of total employment growth (De Kok et al. 2011).

Our multi-level regression results at the firm level reveal that the youngest and smallest SMEs perceive it to be more difficult to access bank loans than do older and larger SMEs. At the country level, we find that SMEs in nations with concentrated banking sectors are, on average, more positive about loan accessibility than are SMEs in nations with less concentrated banking sectors. Furthermore, the existence of a large fraction of foreign-owned banks in a particular country is associated with more positive perception of loan accessibility, but only in the EU 15. It must be stressed that the survey data were collected well before the economic crisis. Thus, the results presented here describe the more structural elements of the relationship between perceived access to bank loans and the firm-level and country-level determinants studied in a “normal” economic situation.

This paper is structured as follows. Section 2 presents an overview of related literature and the main conjectures to be tested. In Sect. 3, the data are discussed. (The data are derived from a questionnaire that asked almost 3,500 SMEs in the European Union about their perception of access to bank loans and other firm-specific characteristics.) Our results and additional analysis are presented in Sects. 4 and 5, respectively. We conclude this article with discussion in Sect. 6.

2 The factors that determine perceived loan accessibility

Numerous studies have attempted to find direct evidence of financing constraints related to bank loans. For example, studies have investigated the funding granted relative to the funding requested (Freel 2007, Parker and Van Praag 2006) or have focussed on the actual denial of bank loan requests (Levenson and Willard 2000; Storey 2004; Kim 2006). This paper focuses on SMEs’ perception of the accessibility of bank loans instead of analysing objective information regarding the success of loan applications. By investigating perception of loan accessibility, we also consider so-called discouraged borrowers. Discouraged borrowers require loans but do not apply for them because they fear that they will be rejected. Such borrowers are rarely investigated, but they comprise a non-negligible pool of firms (Levenson and Willard 2000) and should not be omitted from studies of credit accessibility (Chakravarty and Xiang 2009). This section will draw on studies on the concept of borrower discouragement to introduce the firm-level and country-level factors that determine SMEs’ perception of loan accessibility.

2.1 Firm-level determinants: firm size and firm age

In their theoretical contribution, Kon and Storey (2003) postulate that the extent of borrower discouragement depends on bank screening errors, application costs for firms, and the difference between the bank’s interest rate and that of other lenders. Importantly, the extent to which banks are informed about borrowers’ projects and prospects—or the (in)ability of banks to distinguish between the relative quality of potential borrowers—is an important factor in the prevalence of discouraged borrowers. More specifically, Kon and Storey (2003, p. 47) conclude, “Discouragement is therefore at maximum where there is some, but not perfect, information.” In addition, Han et al. (2009, p. 416) state that “Imperfect information lies at the heart of the concept of discouraged borrowers (…).” More generally, asymmetric information has been a prominent factor in theoretical and empirical work on financing constraints, which does not necessarily consider bank loans (Stiglitz and Weiss 1981; De Meza and Webb 1987).

This information asymmetry may arise if the firm, as an “insider” (Williamson 1975), is better informed than the “outsiders” (the market). It is commonly believed that this effect is accentuated with smaller firms, partially because larger firms have better accounting records and are obliged to obey the strict regulations they must follow to be publicly listed. For example, firms with certified audited financial statements are generally larger and more transparent. Also, smaller firms can be more reluctant to be fully open about their ownership structures and strategic objectives. Thus, larger firms are more transparent than smaller ones. Furthermore, monitoring costs, which tend to be fixed, affect smaller-scale projects more drastically (Beck et al. 2006).

A firm’s reputation or track record can also be important in reducing the extent of information asymmetry. Younger firms have a limited credit history (Gertler 1988; Devereux and Schiantarelli 1990; Wagenvoort 2003a, b; Beck et al. 2006; Cabral and Mata 2003).1 Their restricted credit history makes it difficult for banks to predict the future probability of loan repayment.

Although firm size and firm age have traditionally been seen as important determinants of asymmetric information (Gertler 1988), this relationship has rarely been tested (Hyytinen and Pajarinen 2008), and the opposite hypothesis may also be valid. For example, smaller firms have simpler informational systems than larger firms, which could reduce the opaqueness of small firms. In addition, the Enron and Parmalat scandals have clearly shown that larger companies can also be opaque with regard to information. Hyytinen and Pajarinen (2008) further investigate the conventional wisdom that small and young firms are informationally opaque by comparing data from two credit information companies and examining their disagreements about the creditworthiness of firms. The researchers find that disagreements regarding ratings are inversely related to firm age, but they do not obtain a robust relationship between such disagreements and firm size. These authors conclude that firm age, rather than firm size, is the main proxy for information asymmetry (measured as total assets, total number of employees, or total net sales in their study).

On the basis of these insights, we empirically investigate the importance of firm age and firm size as determinants of perceived loan accessibility. We suggest that younger and smaller SMEs are likely to be more negative about loan access than are older and larger SMEs.

2.2 Country-level determinants: banking sector structure

Many existing studies of access to credit have a single-country focus. We use an international dataset that includes information on 25 EU Member States. This breadth enables us to investigate whether cross-country variations in perceived access to loans can be explained by particular banking sector characteristics when general economic conditions are controlled for. This approach can thus reveal which aspects of the banking system most improve perceived access to bank loans. Specifically, we investigate the importance of the following features of the banking sector.

First, a concentrated banking system enables lenders to use stricter credit conditions. Because banks exert more power in a concentrated banking sector, SMEs may be more negative about potential loan access in this type of climate. Indeed, Kon and Storey (2003) argue that borrower discouragement can be greater when there are fewer alternative sources of funding. There is empirical evidence of greater financing obstacles in such countries (Clarke et al. 2006).2 This study tests whether concentrated banking systems in the EU are associated with negative perception of loan accessibility. On the basis of US data, however, Han et al. (2009) show that a more concentrated banking market is associated with less borrower discouragement. This result is somewhat surprising, though other studies have indeed shown that a more concentrated banking system—as indicated by economies of scale and scope—could be more efficient (European Central Bank 2005).

In tandem with the transition process from central planning to market economies that many Eastern European countries have undergone, foreign ownership of banks has started to increase. Thus, we investigate whether the ownership structure of banks can explain perceived loan accessibility. That is, do the SMEs in countries with more foreign-owned banks believe that access to loans is more limited than do SMEs in countries with more domestically owned banks? There is reason to believe that this could be so: according to Berger et al. (2001, p. 2134), “(…) a foreign-owned bank may be headquartered in a very different market environment, with a different language, culture, supervisory/regulatory structure, and so forth.” However, some studies also note the efficiency gains in the banking systems with more foreign-owned banks, at least in some countries (Claessens et al. 2001), which could improve firms’ perception of loan accessibility (The World Bank 2008, Chapter 2).

Another useful analysis of the organisation of firms, information production, and the allocation of capital is that of Stein (2002). Stein’s analysis is motivated by the concern that consolidation in the banking industry will lead to less lending to small businesses. Lending to small businesses often heavily relies on “soft” information: for instance, on the personality and competences of the owner of the SME. Small banks are argued to be more able to handle soft information. Cross-country evidence of access to financing relative to bank size is scarce. Our supposition is that the SMEs in countries with smaller banks are generally more positive regarding access to bank loans than are SMEs in countries with larger banks.

Finally, we control for economic development using GDP per capita. GDP per capita can partially reflect institutional quality (Clarke et al. 2006). Empirical evidence suggests that financing obstacles (although they have been differently measured in different studies) are lower in countries with higher levels of GDP per capita (Beck et al. 2006; Clarke et al. 2006).

3 Data and methodology

3.1 Dataset

Data are used from two identical Flash Eurobarometer Surveys on Access to Finance, of which one (no. 174) was conducted in the 15 old Member States of the European Union (EU) and one (no. 184) was conducted in the 10 newer Member States.3 The surveys were conducted on behalf of the DG Enterprise and Industry of the European Commission. Together, the datasets cover 4,583 firms in the 25 EU countries, of which 3,047 firms (66 %) belong to the EU 15 and the remaining 1,536 firms (34 %) belong to the EU 10. Randomized telephone interviews were conducted by The Gallup Organization in September 2005 in the EU 15 and in April and May 2006 in the EU 10. Each company that was not in the agriculture, public administration, or non-profit sectors and that employed from 1 to 249 persons was eligible to participate. The person interviewed at each firm was a top-level executive, i.e., s/he worked in general management, as a financial director/manager, or as a chief accountant. The target sample sizes range from 100 to 300 SMEs for each country.

The combined dataset enables an appropriate in-depth study of the SMEs’ perception of the credit market and the country-level differences. The dataset enables us to look closely at which firm or banking sector characteristics could cause variability in perception of loan accessibility. Because the dataset only includes SMEs, we cannot make any inferences regarding perception of bank loan accessibility among large companies with at least 250 employees.

3.2 Measurement

3.2.1 Dependent variable

To determine perception of loan accessibility among SMEs, we use the following question from both Flash Eurobarometer surveys: “Would you say that today, access to loans granted by banks is very easy, fairly easy, fairly difficult, or very difficult?”

We use a dependent variable that takes a value of 1 if the answer is “very difficult” or “fairly difficult” and takes a value of 0 if the answer is “very easy” or “fairly easy”.

Table 1, panel A, shows the distribution of the responses across countries. The calculations are based on the same set of firms that will be included in the hierarchical regression models in the remainder of this paper. Listwise deletion of missing values for the firm-level variables yields a sample of 3,289 firms. All of the descriptive statistics have been weighted. The weights denote the inverse of the probability that a firm is included in the sample and have been constructed on the basis of the true distribution of the SMEs in each country according to firm size and sector of activity.
Table 1

Perceived difficulty of loan accessibility among SMEs in the EU

Country

Panel A: all SMEs

Panel B: by firm age

Panel C: by firm size

<10 Years

≥10 Years

1–9 (Micro)

10–49 (Small)

50–249 (Medium)

Observations

% Easy

% Difficult

% Difficult

% Difficult

% Difficult

% Difficult

% Difficult

Austria

121

0.56

0.44

0.44

0.44

0.45

0.40

0.40

Belgium

146

0.55

0.45

0.42

0.45

0.50

0.20

0.25

Denmark

138

0.90

0.10

0.22

0.06

0.09

0.17

0.10

Finland

79

0.96

0.04

0.09

0.03

0.04

0.05

0.09

France

270

0.62

0.38

0.40

0.38

0.39

0.26

0.45

Germany

242

0.14

0.86

0.76

0.87

0.86

0.86

0.64

Greece

84

0.72

0.28

0.28

0.28

0.29

0.25

0.20

Ireland

53

0.84

0.16

0.06

0.20

0.17

0.17

0.09

Italy

236

0.42

0.58

0.67

0.52

0.58

0.59

0.56

Luxembourg

64

0.40

0.60

0.72

0.50

0.64

0.50

0.38

Netherlands

137

0.45

0.55

0.63

0.43

0.58

0.44

0.35

Portugal

59

0.63

0.37

0.50

0.15

0.38

0.20

0.31

Spain

201

0.72

0.28

0.36

0.20

0.28

0.30

0.21

Sweden

187

0.60

0.40

0.43

0.22

0.42

0.34

0.15

United Kingdom

209

0.79

0.21

0.24

0.21

0.21

0.24

0.12

EU 15

2,226

0.59

0.41

0.46

0.39

0.42

0.38

0.29

Cyprus

79

0.76

0.24

0.24

0.25

0.25

0.16

0.22

Czech Republic

155

0.53

0.47

0.44

0.50

0.48

0.42

0.37

Estonia

73

0.79

0.21

0.23

0.18

0.32

0.09

0.08

Hungary

153

0.43

0.57

0.58

0.54

0.57

0.58

0.39

Latvia

67

0.77

0.23

0.27

0.15

0.30

0.09

0.19

Lithuania

71

0.66

0.34

0.27

0.48

0.37

0.30

0.29

Malta

58

0.57

0.43

0.25

0.47

0.37

0.53

0.22

Poland

228

0.49

0.51

0.58

0.47

0.51

0.54

0.56

Slovakia

84

0.56

0.44

0.39

0.54

0.46

0.41

0.19

Slovenia

95

0.67

0.33

0.34

0.32

0.37

0.14

0.19

EU 10

1,063

0.58

0.42

0.41

0.42

0.45

0.34

0.26

EU 25

3,289

0.58

0.42

0.44

0.40

0.43

0.37

0.28

Source Flash Eurobarometer survey on access to finance (no. 174 and no. 184). “% Easy” represents the weighted percentage of firms responding “very easy” or “fairly easy” to the question “Would you say that today, access to loans granted by banks is very easy, fairly easy, fairly difficult or very difficult?”. % Difficult represents the weighted percentage of firms responding “very difficult” or “fairly difficult”

The distinction between the two age categories (<10 years vs. ≥10 years) ensures enough observations for each country. However, for two countries, there are <10 “older” firms (i.e., Ireland and Finland); therefore, the threshold for these countries is set at 20 years

Table 1, panel A, reveals that views about the difficulty of access to loans are mixed: 42 % of all SMEs believe that obtaining bank loans is difficult, whereas 58 % perceive it to be easy. Average perception in the EU 15 and the EU 10 areas is similar. However, the data show significant variation across countries: 96 % of all Finnish firms believe that access to loans is easy whereas 86 % of the German SMEs find it difficult. An ANOVA test of the weighted averages reveals sufficient between-country variation, which justifies the use of hierarchical regressions (F value = 22.94; corresponding p value < 0.001). Specifically, hierarchical binomial logit regressions will be used to determine the sources of this cross-country variation.

Furthermore, Table 1 provides some indication of the variation in the SME perception based on firm age and firm size. More specifically, panel B distinguishes between young firms (those that have been in existence for less than 10 years) and established firms (those that have been in existence for at least 10 years) and shows that for most countries, younger firms are indeed more negative about loan accessibility than established firms. However, in some EU 10 countries, the relationship is less straightforward. Panel C shows the perception across three firm size classes: micro firms (1–9 employees), small firms (10–49 employees), and medium-sized firms (50–249 employees). The averages reveal a clear pattern: perception of access to bank loans becomes more positive as firm size increases. The exact relationships between firm age and firm size on the one hand and perceived loan accessibility on the other hand will be determined in a multivariate and multi-level context in the remainder of this paper.

3.2.2 Subjective versus objective measurement

It should be noted that we examine the perceived difficulty of access to bank loans and do not have data for more objective measures of the difficulty of obtaining bank loans. A self-assessment criterion based on surveys among firms is used in other studies also.4 One advantage of this type of approach is that it enables us to include a large group of firms that would have been otherwise omitted, including discouraged borrowers. Furthermore, the perception of a wide range of SMEs can be more relevant than objective criteria if our objective is to understand firm behaviour. This analysis provides insight into how the initiatives intended to improve access to finance could not only benefit the firms that have (un)successfully applied for loans in the past but could also assist firms that have not applied for loans. We expect at least a moderately positive relationship between subjective and objective indicators of access to credit (see also Egeln et al. 1997). To find support for this claim, we use the Global Competitiveness Report 2005–2006 by the World Economic Forum, which contains information on the difficulty of accessing bank loans in many countries. This measure of the difficulty of accessing bank loans is based on experts’ responses to the following question: “How easy is it to obtain a bank loan in your country with only a good business plan and no collateral?” Each expert rates this statement on a seven-point Likert-scale (ranging from 1 = “impossible” to 7 = “very easy”) such that each country receives a value between 1 and 7 (to 1 decimal place) that indicates the difficulty of accessing bank loans. It turns out that the Pearson correlation coefficient of the weighted country averages for our perception variable and the country scores from the Global Competitiveness Report equals 0.41 (significant at the 0.05 level; 25 observations), which indeed indicates a modest positive relationship between the two measures.

3.2.3 Firm age and firm size

We construct a series of dummy variables to reflect firm age: age < 10 if the firm has existed for fewer than 10 years; age 10–20 for a firm aged between 10 and 20 years; age 20–30 for an age between 20 and 30 years; and age > 30 if the firm has been in existence for more than 30 years. Age > 30 is used as the reference category in our analyses. We do not include a continuous specification for age because such a continuous measure is only available for the firms in the EU 15.

Again, a set of dummies is used to measure firm size. Three size classes are available that indicate the total number of employees: micro firms have 1–9 employees, small firms have 10–49 employees, and medium-sized firms have 50–249 employees. The group of 50–249 employees is used as the reference category in the remaining analyses.

3.2.4 Control variables at the firm level

Dynamic firms, both those with a growing number of employees and those with improved performance, are expected to be more optimistic in their perception of loan accessibility. This is because such firms may actually not feel constrained by their lack of external financing given that they can use their increasing cash flows for their financing purposes or reinvest their profits. However, it could also be argued that growing firms require greater financial inflows (Freel 2007) and that they may therefore be more negative regarding loan availability. The growth in the number of employees is captured by the variable employment. This variable takes a value of 2 if a firm’s number of employees has increased since the previous year, a value of 1 if it has remained the same, and a value of 0 if it has decreased. A firm’s dynamics in terms of its cash flow and investments are also taken into account. The variables cash flow and investments take a value of 2 when the firm’s situation has improved in this regard since the previous year, a value of 1 if there has been no change, and a value of 0 if cash flow or investments have worsened.

Because perception of access to bank loans clearly depends on earlier experiences, we include previous loan, which takes a value of 1 when a firm has made use of a short-term or a long-term loan in the past and 0 otherwise.

Evidence has been found that foreign-owned firms have easier access to external financing than do nationally owned firms (Schiantarelli and Sembenelli 2000; Harrison and McMillan 2003; Beck et al. 2006). Hence, the ownership of the firm is controlled for by using the following variables. Family takes a value of 1 if the firm is partly or exclusively family-owned and a value of 0 otherwise. Domestic equals 1 if the firm is exclusively owned by a national company and 0 otherwise. Foreign takes a value of 1 if the firm is partly or entirely owned by an international company and a value of 0 otherwise. Other ownership structures are assembled in the variable other. In our regressions, domestic is used as the reference category.

Because loan demand can vary across sectors, we control for sector-specific characteristics by using sector dummies. We distinguish between eight sectors: construction or civil engineering, financial services, production and manufacturing of goods, extraction or production of raw materials, trade and distribution, transport, other services to businesses, and other services to consumers. Construction or civil engineering is used as the reference category in the regression analyses.

3.2.5 Country-level variables

GDP per capita figures for the year 2004 have been retrieved from Eurostat. We use a logarithmic transformation of this variable, and GDP per capita is expressed in Euros (and in terms of purchasing power parity).

To measure the degree of concentration in the banking sector, the Herfindahl–Hirschman index for 2004 is used. Concentration is measured as the sum of the squares of the market shares of all credit institutions according to total assets. The corresponding 2004 data can be found in European Central Bank (2005). The numbers range from 178 for Germany to 3,887 for Estonia; the theoretical maximum value is 10,000. The higher the number, the more concentrated the market is. Values higher than 1,800 usually indicate a concentrated banking industry. This threshold for distinguishing between high and low-concentration banking markets has been used for the US bank merger guidelines (Federal Reserve Bank 1998; Han et al. 2009).

To reflect the percentage of foreign-owned banks, we use the share of the assets of foreign credit institutions as a percentage of the total assets of domestic credit institutions. The data from 2004 can be found in Allen et al. (2006). Higher numbers indicate greater foreign ownership in the banking sector. Percentages are high in numerous EU 10 countries, including the Czech Republic, Estonia, Lithuania, and Slovakia, and in Luxembourg (all over 90 %). Many EU 15 countries such as Germany and Italy have low values (below 10 %).

The average bank size for the year 2004 is defined as the total assets of all credit institutions in a country divided by the number of credit institutions in that country, as a percentage of the country’s GDP (calculations based on European Central Bank 2005). Higher numbers indicate larger banks in terms of their average asset value (scaled by the country’s GDP).

Note that the relatively small number of countries included in this study imposes constraints on the maximum number of variables that can be taken into account at the country level.

3.3 Analysis

Hierarchical (multi-level) binomial logit regressions are performed to simultaneously assess the influence of firm-level and country-level variables on the dependent binomial variable, i.e., perceived loan accessibility. Hierarchical modelling takes into account the hierarchical structure of the dataset (Peterson et al. 2012), i.e., the 3,289 firms are nested within the 25 countries. Random intercepts are incorporated into our model to explain the between-country variation in the perceived difficulty of obtaining bank loans while controlling for the relevant firm-level variables (Peterson et al. 2012). The model contains country-specific intercepts that depend on country-specific variables and a normally distributed disturbance term (Block et al., 2012).

HLM 7.0 software is used for the multi-level analyses (Raudenbush et al. 2011). Standard errors robust to heteroskedastic disturbance terms are used (more technical details are given by Raudenbush and Bryk 2002, Chapter 9). The PQL (penalised quasi-likelihood) method is used to obtain coefficient estimates at the firm level and at the country level (Raudenbush and Bryk 2002, Chapter 14). The variables have not been centred around the overall means or the country means. Finally, inferences are based on population average models rather than on unit-specific models. Raudenbush and Bryk (2002, p. 304) conclude that “population-average inferences are based on fewer assumptions and will be quite robust to erroneous assumptions about the random effects in the model.”

The coefficients of the firm-level variables can be interpreted as they are in ordinary binomial logit regressions (Block et al. 2012). A significant positive (negative) coefficient implies an increase (a decrease) in the probability of finding access to bank loans difficult as the value of a particular variable increases. Regarding the country-level variables, a significant positive (negative) coefficient means that higher values for the particular variable are associated with a tendency of firms to perceive obtaining loans as difficult (easy) in such countries.

4 Results

A correlation matrix for the firm and country-level variables is provided in Table 2. The low between-variable correlations generate no serious concerns with regard to multicollinearity. Table 2 also reports the weighted sample averages for the entire sample, the EU 15 sample, and the EU 10 sample.
Table 2

Correlation matrix firm-level and country-level variables

https://static-content.springer.com/image/art%3A10.1007%2Fs11187-012-9451-y/MediaObjects/11187_2012_9451_Tab2_HTML.gif
The estimated coefficients of our hierarchical binomial logit regression are listed in Table 3. Model 1 includes the firm-level variables only while allowing for a random intercept at the country level.
Table 3

Results of hierarchical binomial logit regressions

 

Model 1

Model 2

Model 3

Model 4

Model 5

Coeff.

SE

Coeff.

SE

Coeff.

SE

Coeff.

SE

Coeff.

SE

Firm level

 Age <10

0.35***

(0.10)

0.36***

(0.11)

0.37***

(0.12)

0.34***

(0.11)

0.37**

(0.16)

 Age 10–20

0.09

(0.10)

0.10

(0.11)

0.10

(0.12)

0.07

(0.12)

0.00

(0.16)

 Age 20–30

−0.30*

(0.16)

−0.31**

(0.15)

−0.32**

(0.16)

−0.33**

(0.16)

−0.44**

(0.18)

 Employees 1–9

0.28**

(0.12)

0.29**

(0.12)

0.30**

(0.12)

  

0.34**

(0.14)

 Employees 10–49

0.15

(0.11)

0.15

(0.11)

0.15

(0.12)

  

0.14

(0.11)

 Turnover <500 k

      

0.43***

(0.12)

  

 Turnover 500–2,500 k

      

0.38***

(0.12)

  

 Turnover 2,500–5,000 k

      

0.14

(0.16)

  

 Dynamics: employ.

−0.17***

(0.05)

−0.18***

(0.05)

−0.19***

(0.06)

−0.17***

(0.06)

−0.25***

(0.06)

 Dynamics: cash flow

−0.15***

(0.05)

−0.16***

(0.04)

−0.16***

(0.06)

−0.15***

(0.06)

−0.15***

(0.05)

 Dynamics: investments

−0.11*

(0.06)

−0.11*

(0.06)

−0.12*

(0.06)

−0.12*

(0.06)

−0.13*

(0.08)

 Previous loan

0.00

(0.09)

0.00

(0.09)

0.00

(0.08)

0.00

(0.08)

  

 Ownership: family

0.13

(0.08)

0.12

(0.08)

0.13

(0.14)

0.08

(0.14)

0.14

(0.13)

 Ownership: foreign

0.01

(0.14)

0.00

(0.15)

−0.00

(0.21)

0.03

(0.21)

0.30*

(0.16)

 Ownership: other

−0.04

(0.14)

−0.04

(0.14)

−0.04

(0.19)

−0.08

(0.19)

−0.12

(0.20)

 Sector: finance

−0.15

(0.26)

−0.16

(0.27)

−0.16

(0.29)

−0.17

(0.28)

−0.24

(0.27)

 Sector: production

0.05

(0.15)

0.05

(0.15)

0.05

(0.14)

0.05

(0.14)

0.09

(0.14)

 Sector: raw materials

−0.22

(0.26)

−0.22

(0.26)

−0.22

(0.33)

−0.19

(0.33)

−0.11

(0.41)

 Sector: service bus.

0.08

(0.15)

0.08

(0.15)

0.08

(0.15)

0.09

(0.15)

0.09

(0.18)

 Sector: service consumers

0.11

(0.15)

0.10

(0.15)

0.11

(0.15)

0.10

(0.15)

0.23

(0.18)

 Sector: trade

0.12

(0.14)

0.12

(0.14)

0.12

(0.13)

0.17

(0.13)

0.17

(0.16)

 Sector: transport

0.09

(0.17)

0.09

(0.17)

0.10

(0.21)

0.11

(0.21)

0.26

(0.21)

Country-level

 Log (GDP per capita)

  

−0.16

(0.38)

1.07

(0.83)

−0.10

(0.49)

−0.22

(0.40)

 Concentration/1,000

  

−0.48**

(0.20)

−0.57**

(0.21)

−0.48**

(0.23)

−0.58***

(0.20)

 % Foreign owned/100

  

−0.09

(0.40)

1.77

(1.07)

−0.12

(0.61)

−0.18

(0.40)

 Bank size (% of GDP)

  

0.02*

(0.01)

0.04

(0.03)

0.02

(0.03)

0.02*

(0.01)

 EU 15 × % foreign owned/100

    

−3.97**

(1.66)

    

 Obs. firm level

3,289

 

3,289

 

3,289

 

3,289

 

2,093

 

 Obs. country level

25

 

25

 

25

 

25

 

25

 

 Max. value log likelihood

−4,568

 

−4,586

 

−4,598

 

−4,589

 

−2,933

 

Dependent variable takes a value of 1 (access to loans is perceived as difficult) or 0 (access to loans is perceived as easy). Reference categories: Age >30; Employees 50–249 (medium); Turnover >500 k; Ownership: Domestic; Sector: construction or civil engineering. Robust SE are displayed between parentheses (robust standard errors could not be computed for Models 3 and 4; regular SE are computed instead). The estimates of the intercepts are not reported, but are available upon request. Model 3 contains the EU 15 dummy variable; its estimates of the coefficient and SE are also available upon request. Convergence is achieved after seven (macro) iterations at most for each model. The iteration process stops when the largest parameter estimate change is <0.000001

* Significant at 0.10; ** significant at 0.05; *** significant at 0.01 (two-tailed tests)

4.1 Firm age and firm size

Model 1 in Table 3 reveals that the youngest SMEs (those that have been in existence for up to 10 years; p value < 0.01) have the worst perception of loan accessibility. This finding implies that more time enables firms to develop a track record that relaxes such perception. Note that the relationship between firm age and perceived loan access is non-monotonic: the firms that are 20–30 years old have the best perception, even compared with those that are more than 30 years old (p < 0.10).

Firm size also affects perceived loan access. The smallest firms (1–9 employees, p < 0.05) have the worst perception. There are no significant differences in perception for small (10–49 employees) versus medium-sized (50–249 employees) firms (p > 0.10).

4.2 Control variables at the firm level

Significant and consistent results are found for our measures of firm dynamics. That is, firms that are growing—in terms of their number of employees (p < 0.01), cash flow (p < 0.01), or investments (p < 0.10)—are more positive about access to bank loans than are firms that are not growing. The other control variables are insignificant (p > 0.10 for all variables). That is, the firms that have received loans in the past are no different in terms of their perception from the firms that have not received loans. In addition, the ownership structure of a firm is not significantly related to its perception of loan accessibility. Last, a firm’s sector of economic activity does not matter in this regard.

4.3 Country-level variables

Model 2 in Table 3 adds the country-level variables. We find that the extent of concentration in the banking sector has a strong negative influence (p < 0.05). Contrary to our expectations, concentration improves perceived access to bank loans. Furthermore, the percentage of foreign-owned banks is not an important factor in explaining the variation in perception across countries (p > 0.10). Finally, the average bank size is positively associated with the perceived difficulty of obtaining bank loans (p < 0.10). This finding implies that countries with larger banks are, on average, characterised by unfavourable SME perception of loan accessibility. Our control variable, GDP per capita, is not significantly related to the cross-country variation in perceived loan accessibility (p > 0.10).

We further investigate the country dimension by investigating the differences between the EU 15 and the EU 10. Interaction terms between an EU 15 dummy variable that takes a value of 1 for an EU 15 Member State and 0 otherwise and each country-level variable are consecutively added to Model 2. It seems that the percentage of foreign-owned banks is the only country-level variable with different influences between the EU 15 and the EU 10. Model 3 in Table 3 shows the relevant results, adding the interaction term EU15 × % Foreign owned/100 to Model 2. In Model 3, the coefficient of the variable % Foreign owned/100 indicates the influence for the EU 10 (p > 0.10). An additional test of the linear combination of this coefficient and the coefficient of the interaction term shows that the percentage of foreign-owned banks has a significant negative influence for the EU 15 (p < 0.05).

5 Additional analyses

Our first robustness check involves using a different measure of firm size, which is annual turnover instead of the number of employees. Four dummy variables are created: turnover <500 k if a firm’s turnover in the previous fiscal year was less than €500,000, turnover 500–2,500 k for a turnover between €500,000 and €2.5 million, turnover 2,500–5,000 k for a turnover between €2.5 million and €5 million, and turnover >5,000 k when the previous year’s turnover exceeded €5 million. Turnover >5,000 k is used as the reference category. The estimation results presented in Model 4 replicate our findings regarding firm size: the smallest firms—those with up to an annual turnover of €2.5 million—have the worst perception (p values < 0.01). These firms represent 88 % of all SMEs in our sample.

Firms that have received loans in the past base their assessment of loan availability on different information from firms that have not received loans. We rerun our analysis of Model 2 using the subset of firms that have received bank loans in the past, i.e., SMEs for which the variable previous loan equals 1. The results are displayed in Model 5. The findings are qualitatively similar for Model 2 and Model 5. Again, the youngest (p < 0.05) and smallest (p < 0.05) firms have the worst perception even though these firms apply for substantially smaller amounts than the older and larger firms. This finding is illustrated in Table 4. Table 4 shows the most recent loan amounts for which firms have applied. Panel A of Table 4 shows that, on average, 36 % of all firms that applied for a loan in the past made a recent request of less than 25,000 Euros; 37 % requested a loan of between 25,000 and 100,000 Euros, and 26 % requested more than 100,000 Euros. However, variation in these figures with reference to firm size is substantial. Whereas 43 % of the micro firms requested less than 25,000 Euros, the corresponding figure is only 15 % for small firms, and it is only 6 % for medium-sized firms. Additionally, the percentages increase from 19 to 48 % and 75 % for the requests that are larger than 100,000 Euros for micro firms, small firms, and medium-sized firms, respectively (all weighted percentages). This pattern can also be observed for the separate EU 15 (Panel B) and EU 10 (Panel C) samples.
Table 4

Size of most recent loan, by firm age

 

% of SMEs that made use of a loan in the past (% for which previous loan = 1)

Distribution of responses to the question “What was the approximate amount of the last loan which your company applied for?a

<25,000 €

25,000–100,000 €

>100,000 €

Panel A

 EU 25: all SMEs

0.60

0.36

0.37

0.26

 EU 25: micro firms (employees 1–9)

0.58

0.43

0.38

0.19

 EU 25: small firms (employees 10–49)

0.69

0.15

0.37

0.48

 EU 25: medium firms (employees 50–249)

0.72

0.06

0.19

0.75

Panel B

 EU 15: all SMEs

0.64

0.34

0.39

0.28

 EU 15: micro firms (employees 1–9)

0.62

0.38

0.41

0.21

 EU 15: small firms (employees 10–49)

0.73

0.15

0.35

0.50

 EU 15: medium firms (employees 50–249)

0.75

0.06

0.16

0.79

Panel C

 EU 10: all SMEs

0.52

0.44

0.32

0.24

 EU 10: micro firms (employees 1–9)

0.48

0.57

0.30

0.13

 EU 10: small firms (employees 10–49)

0.62

0.14

0.42

0.44

 EU 10: medium firms (employees 50–249)

0.70

0.06

0.23

0.71

Source Flash Eurobarometer survey on access to finance (no. 174 and no. 184)

aNote that the percentages in the three columns add to 1.00 (= 100 %) in each row

Regarding the country-level variables, the results are also qualitatively similar for Model 2 and Model 5. The relationship between perceived loan accessibility and the degree of concentration is even stronger in Model 5 than in Model 2 (p < 0.01).

At the country level, we assess the robustness of our findings to the degree of concentration. An alternative measure of concentration replaces the Herfindahl index in Model 2 with the sum of the shares of total assets of the five largest credit institutions, known as the five-firm concentration ratio (source: European Central Bank 2005). Again, the highest value is found for Estonia (98.0 %), whereas Germany ranks lowest in terms of the degree of concentration (6.3 %). This variable also has a significant negative coefficient (p < 0.10). The exact estimation results are available from the authors upon request.

6 Discussion and conclusion

The determinants of perceived bank loan accessibility among SMEs are of central importance in this paper, both at the firm level and at the country level. We use a firm-level dataset that contains information for nearly 3,500 SMEs in 25 European Union Member States. At the firm level, our multi-level results indicate that the youngest and smallest SMEs perceive it to be more difficult to obtain bank loans than do the older and larger SMEs. For firms, growing older establishes a track record that reduces the information asymmetry between them and banks. Our results suggest that investing time and energy in “relationship banking” could reduce this information asymmetry between the lender and the borrower and could thus improve perceived access to bank loans (Petersen and Rajan 1994, 1995; Berger and Udell 1995; Harhoff and Körting 1998; Wagenvoort 2003b). Interestingly, our results do not imply an entirely monotonic relationship between firm age and perceived loan accessibility. Indeed, firm age relaxes perception of loan accessibility, but only until a firm age of 30 years. The oldest SMEs (those that are more than 30 years old) are more pessimistic about access to bank loans than are SMEs that have been in existence for between 20 and 30 years. Hence, relationship banking may be the most useful in the earliest phase of a firm’s life cycle, but these benefits may be eliminated as firms age. One explanation could be that the oldest firms are not as dependent on banks as the younger firms are because of their stronger internal financial situation (Petersen and Rajan 1994), which diminishes the strength of the older SMEs’ firm–bank relationships. Furthermore, the loan amounts requested by the oldest firms are relatively high (observation on the basis of our dataset), and the probability of denial could also increase as a result.

At least one other result at the firm level may indicate the importance of relationship banking and thus of asymmetric information in relaxing firms’ perception of loan accessibility. The firm-level results consistently indicate that growing SMEs are more positive regarding access to bank loans than are non-growing SMEs. This result may be explained by lower information asymmetry between growing firms and banks because banks “may be more willing to invest in developing a closer working relationship with growing businesses” (Binks and Ennew 1996, p. 20). In addition, the growing firms could be more open about their internal structures because of their greater reliance on banks in planning for future investments. These results are interesting and further develop the research that finds negative (or non-significant) relationships between firm growth and measures of loan accessibility (Freel 2007).

There are considerable cross-country differences in terms of perceived loan accessibility. This between-country variation can be largely explained by the degree of concentration in the banking sector. Specifically, we find that a concentrated banking sector is associated with more favourable perception of access to bank loans. Our results add to the current literature that addresses the implications of the increased market power of banks (Turk-Ariss 2010). A similar relationship between access to finance and the degree of concentration is proposed in Han et al. (2009) and European Central Bank (2005). Our results suggest that the banks in concentrated markets are more willing to invest in relationships with SMEs than are banks in less concentrated markets, which positively affects SMEs’ perception of loan accessibility. Clearly, our conclusions contradict the argument that greater concentration generates less competition between banks, which would then enable lenders to make credit conditions more strict (empirical evidence on this subject is given by Clarke et al. 2006). Similarly, other literature has supposed that increased competition will generate improved access to external finance because it forces the incumbent banks to work more efficiently at providing financial services (The World Bank 2008, Chapter 4).

The liberalisation of the financial markets in the EU 10 has led to an increased presence of foreign banks. The percentage of foreign-owned banks does not explain the cross-country variation in perception of loan accessibility in the EU 10 and in the EU 25 as a whole. Clearly, the suggested positive effects of foreign banks offset some of the proposed negative consequences. Furthermore, it is possible that the effect of foreign banks is heterogeneous, i.e., some types of firm benefit more from foreign banks than do other types (Lin 2011). However, this issue was not investigated in this paper. Interestingly, in the EU 15, the presence of more foreign-owned banks is associated with improved perception of access to bank loans. There is, indeed, some evidence that financing obstacles decrease in the presence of foreign banks (Clarke et al. 2006, The World Bank 2008, Chapter 2), but these analyses typically include larger firms and are restricted to developing or transition economies.

In common with many other studies, this one addresses established firms only. It might be possible, for example, that restricted access to bank loans prevents the successful start up of a business (see, for instance, Aghion et al. 2007). Thus, it is possible that this study underestimates the impacts of firm age and firm size on the perceived difficulty of accessing bank loans, because potential entrepreneurs who cannot obtain sufficient credit to start a business are not included in the sample. Another form of potential selection bias is associated with the possibility that older firms are generally “higher quality” firms because lower quality firms are less likely to survive. In other words, when young firms perceive the access to credit to be limited, this difficulty may arise as a result of banks’ negative risk assessments of these firms rather than because of information problems.

Further research should develop better proxies for asymmetric information in attempting to analyse perception of access to bank loans or the determinants of borrower discouragement. For example, many empirical studies show that collateral provides an incentive and a means for good lenders to identify themselves (Besanko and Thakor 1987; Chan and Kanatas 1985; Bruns and Fletcher 2008; Kon and Storey 2003). This finding suggests that an increased ability to supply collateral should lead to improved perception of loan accessibility or to less borrower discouragement. Furthermore, our analysis focuses primarily on supply-side constraints because the factors that are investigated are not associated with the viability of the projects themselves. Future research could more specifically explore the demand side constraints related to information and advice that may exist when firms do not obtain optimum amounts of credit because of their lack of knowledge, their inadequate presentation of their proposals, or their poor management (Cressy 2002, 2003; De Meza and Webb 1987; De Meza 2002).

Footnotes
1

Alternatively, several authors use firm size as an indicator of the borrower’s reputation. There is, however, little evidence that small firms are riskier than large companies (St. Pierre and Bahri 2011). In fact, Behr and Güttler (2007) find that the effect of firm size on default risk weakens when other control variables are included in the analysis.

 
2

However, in Clarke et al. (2006), the managers of SMEs and larger firms in developing and transition countries are not directly asked about their perception of loan accessibility but are rather asked about how problematic access to long-term loans is for their firm’s operations and growth.

 
3

Two Member States of the European Union are not represented in the present dataset: Bulgaria and Romania joined the European Union in 2007.

 
4

Examples include Egeln et al. (1997) and Beck et al. (2006), although the firms are less directly asked about loan accessibility than in this study.

 

Acknowledgments

At the time of writing this paper, Erik Canton was employed at the European Commission (DG Enterprise and Industry). The views expressed here are those of the authors and should not be attributed to Ecorys or the European Commission. The authors would like to thank Jozef Konings, Jacques Mairesse, Roy Thurik, seminar participants at the European Commission and Erasmus University Rotterdam, and two anonymous reviewers for useful comments. This study benefited from a grant by the “Van Cappellen Stichting”.

Copyright information

© Springer Science+Business Media New York 2012