• No results found

COMPETITION IN THE BANKING SECTOR, FINANCIAL STABILITY AND GROWTH

N/A
N/A
Protected

Academic year: 2021

Share "COMPETITION IN THE BANKING SECTOR, FINANCIAL STABILITY AND GROWTH"

Copied!
66
0
0

Bezig met laden.... (Bekijk nu de volledige tekst)

Hele tekst

(1)

COMPETITION IN THE BANKING SECTOR,

FINANCIAL STABILITY AND GROWTH

Master Thesis

MSc. International Economics and Business

Groningen, August 2009

Author: Svilena Mihaylova (1822950, s.m.mihaylova@student.rug.nl)

Supervisor: dr. Robert Inklaar (r.c.inklaar@rug.nl)

(2)

Abstract

In this thesis we investigate the impact of bank competition on industry growth in the European Union (EU) and the way financial stability affects this relationship. Our empirical analysis, using data on 25 EU member states and 25 industries, over a time period from 1995 to 2004, shows that bank competition has an overall negative growth impact in EU-25. When we consider also the level of stability in the banking sector, we find that this negative effect holds only for countries with high levels of financial stability. Furthermore, we expand the study by exploring the impact of bank concentration on industry growth. We do not find evidence of a growth effect in the whole EU-25 but we do find a positive impact of a more concentrated banking sector on industry growth, when taking into account stability in this sector. In particular, we find that countries with more stable financial systems benefit from more bank concentration. The thesis confirms the lack of consensus in the banking literature on the important competition-stability-growth nexus and calls for future research on these issues.

(3)

Table of contents:

1. Introduction………...………4

2. Literature review…...………6

2.1. Bank competition, concentration and economic growth...6

2.2. Bank competition and financial stability...9

2.2.1 The “competition-fragility” nexus...9

2.2.2. The “competition-stability” nexus...11

2.3. Methods for measuring bank competition and concentration...11

3. Methodology, data sources and measures...16

3.1. Methodology...16

3.1.1. Basic model...16

3.1.2. Extended model...19

3.2. Data sources and measures...21

4. Results...30

4.1. Basic model...30

4.2. Extended model...37

5. Robustness tests...39

5.1. How are the new EU members and the incumbent EU countries affected? ...39

5.2. Is it the measure of dependence on external finance? ...46

6. Diagnostic checks………46

6.1. Are the estimates of Boone indicator reliable measures of bank competition? ...46

6.2. Outliers, heteroskedasticity, normality of residuals and multicollinearity………46

6.3. Model specification...48

7. Concluding remarks ...49

7.1. Conclusions...49

7.2. Limitations...51

7.3. Suggestions for future research...52

Reference ...54

(4)

1. Introduction

Considerable amount of recent theoretical and empirical studies have established that greater financial development stimulates economic growth and also that financial development is related to a country’s institutional characteristics. However, less is known about the effect of competition and market structure of the banking sector on economic growth. In the same time, as argued in Claessens and Laeven (2005), competition in banking is important for a number of reasons. As in other industries, the degree of competition in the financial sector can affect the efficiency of the production of services, the quality of products, and the degree of innovation in that sector. Furthermore, specific to the banking sector is the relationship between competition and stability that has been established in both theoretical and empirical research (Allen and Gale, 2004), as well as in the actual conduct of bank supervision. Last but not least, it has been shown both theoretically and empirically that the degree of competition in the banking sector can affect the access of firms to external financing. However, as cross-country research on the effects of competition in the banking system is still at an early stage and the results of existing studies are ambiguous, the directions of the above mentioned relationships remain unclear.

Furthermore, another problem in the literature is that the interpretation of the existing empirical studies using measures of regulations, market structure, and institutions is not always clear because some theoretical issues have not always been considered (Claessens and Laeven, 2005). Specifically, as the contestability literature has shown, the competitiveness of an industry cannot be measured by market structure indicators alone (such as the number of institutions, Herfindahl or other concentration indexes, or ownership structure, such as the degree of foreign or state ownership). Rather, in order to assess the degree of effective competition, one needs a structural model. Thus far, however, most works have not examined the degree of bank competition using a specific structural model. Therefore, as argued in Claessens and Laeven (2005) their findings about the impact of market structure on banking system performance, firm financing, and economic growth could reflect factors different from competition.

In this paper we explore the impact of bank competition on industry growth in EU-251 and the way financial stability affects this relationship. As outlined above, despite the ambiguity in the literature, the growth impact of bank competition, as well as the competition – stability nexus have been examined in previous works. However, the modifying effect of bank stability on the relationship between competition and growth has not yet been addressed

1

(5)

in previous studies. What is more, very few papers have examined the effect of bank competition in the context of the enlarged EU. Given that, as well the importance of these relationships in terms of policy implications, the issues discussed in this paper try to fill a gap in the existing literature. Moreover, in order to allow for comparisons with previous studies, in addition to bank competition, we also examine the growth impact of bank concentration, as well as how this effect changes at different levels of financial stability. Hence, our main research questions are:

• How does competition in the banking sector affect industry growth in EU-25 and how does this relationship change at different levels of financial stability?

• How does concentration in the banking sector affect industry growth in EU-25 and how does this relationship change at different levels of financial stability?

• Are the growth effects of bank competition and bank concentration heterogeneous across industries which differ in terms of their dependence on external finance?

Moreover, we also examine how the new and the old member states are affected by bank competition and bank concentration and whether these effects are different for different industries.

(6)

The outline of the paper is as follows: section 2 provides review of theoretical and empirical studies on the relationship between bank competition (concentration) and growth, as well as between competition (concentration) and stability in the banking sector. It also reviews the existing methods for measuring bank competition, as well as some studies which have tried to relate bank concentration and bank competition. Section 3 outlines the methodology, the data sources and the measures used. Section 4 presents our main estimation results and in section 5 some robustness tests are discussed. In Section 6 we outline the main diagnostic checks that we performed and in Section 7 we draw conclusions.

2. Literature review

In this section several related strands of literature are reviewed. First, we present different theories on bank competition, market power and economic growth, as well as a short review of the relevant empirical works. Second, we describe the opposing views about the relationship between competition (concentration) and stability in banking, as well as the empirical studies done in this area. Third, we discuss general theories on measuring bank competition and concentration and briefly review some of the empirical papers which have applied such measures and tried to relate competition in the banking sector with market structure.

2.1. Bank competition, concentration and economic growth

(7)

Thakor (1992) who show that competition in the banking sector leads to decrease in loan rates and increase in deposit rates. The negative impact of monopolistic banking market on economic growth is shown by Guzman (2000). He concludes that monopoly power in banking tends to depress capital accumulation via either credit rationing and/or excessive monitoring as relatively high loan rates induce entrepreneurs to undertake riskier projects. In a simple endogenous growth model Pagano (1993) shows that any departure from perfect competition in the credit market causes inefficiencies that harm firms’ access to credit, thus hindering growth. Hence, these models support the theory that market power in banking is detrimental to consumers and growth whereas bank competition fosters growth.

However, there are other theoretical studies that show an ambiguous relationship between banking system competition, firm access to external financing and, in turn, firm growth. Capital accumulation depends not only on the volume of credit supply but also on the efficiency of its allocation and there are theoretical studies arguing that market power need not necessarily have a negative effect on allocative efficiency. Such theoretical approaches are those which focus on relationship lending and screening. Regarding the first one, it is argued that more competitive banking systems may provide less financing to firms because they have fewer incentives to invest in close relationships with firms (Rajan 1992; Petersen and Rajan 1995). As a result, especially those firms and sectors that rely heavily on external finance should grow more slowly in systems that are more competitive. Conversely, a bank with market power is more eager to engage in relationship lending, with the result that the credit supply for young firms is higher (and the cost of these funds is lower) than in a more competitive banking system. Hence, credit access and industry growth can improve in banking markets that are characterized by more market power.

(8)

The foregoing review of the two opposing theories shows that the impact of bank competition on economic growth is not clear and straightforward despite the considerable amount of theoretical literature on the topic.

Empirical literature on banking market power, competition and economic growth also produces ambiguous results. It usually uses the number of banks or the degree of concentration in the banking sector as a proxy for market power. The majority of the early studies use U.S. data to explore the relationship between bank profitability and concentration. However, a major shortcoming of most of these studies is that they do not take into consideration differences in productive efficiency. Those studies which have tried to control for differences in efficiency, produce mixed results (Berger, 1995). Recent empirical literature uses panel data to explore the impact of bank concentration on economic growth. In a cross-country, cross-industry study Cetorelli and Gambera (2001) test the average effect of bank concentration on industrial growth. They find that market power has an overall depressing effect on growth, but that the effect is heterogeneous across industries. Sectors where young firms are more dependent on external finance grow faster in countries that have a more concentrated banking sector (which supports the concept of relationship lending). Using the same data and similar methodology, Deidda and Fattouh (2002) find that banking concentration is negatively related to per capita growth and industrial growth only in low-income countries, while in high-low-income countries there is no significant relationship. Fritzer (2004) also finds that concentration of banks has a detrimental impact on growth, with this impact depending on the country’s initial stage of economic development and being lower for more developed countries. Similarly, Beck, Demirgüç-Kunt, and Maksimovic (2003) use a dataset of developed and developing countries to examine the effects of concentration on credit availability while controlling for regulatory policies such as entry, ownership structure, and restrictions on bank activities. They find that firms face higher financing obstacles in concentrated banking markets. However, the effect is insignificant for countries that have a well-developed financial system. Another study that supports the view about the negative growth impact of concentration is Cetorelli and Strahan (2006). They examine small firms in the United States and find a positive effect of lower bank concentration on firm growth.

(9)

that in an earlier study by the same authors (Claessens and Laeven , 2003), they find that the effects of bank competition on access to financing (and growth) can depend on the level of development of the financial system. Specifically, in countries with less developed financial systems financially dependent industries grow faster when the financial system is less competitive while in more developed financial systems, more competition is associated with higher growth. Another study (Koivu, 2002), which is one of the few works that focus on transition economies, also finds support for the beneficial effect of competition in banking. The study shows that lower interest rate spreads (i.e. higher competition) lead to faster GDP growth in transition countries in Eastern Europe in the 1990s.

The discussed empirical studies above show that although a positive growth impact of bank competition and negative effect of concentration are the most common results, these relationships are not always straightforward. First, the empirical literature that links concentration to higher profits is not convincing. Second, it was shown above that the relationship between competition and economic growth is also ambiguous in the sense that it depends on whether differences in industries’ or countries’ characteristics are taken into account.

2.2. Bank competition and financial stability

Similar to the above-discussed studies on the growth impact of bank competition, the literature on bank competition, concentration and stability is even more ambiguous. Two opposing lines of literature are reviewed below as well as the supporting empirical works.

2.2.1 The “Competition-fragility” nexus

(10)

less risky policies that would result in fewer bank bankruptcies and more stability in the banking sector as a whole (Jiménez, Lopez, Saurina, 2007). Conversely, too much competition could decrease the franchise values of banks, which might encourage banks’ shareholders to take more risk in order to be able to obtain their former profits. This in turn, could threaten the solvency of the banks and, at an aggregate level, impede the stability of the banking system as a whole. Studies of Marcus (1984), Furlong and Keeley (1989), Besanko and Thakor (1993), Suárez (1994), Matutes and Vives (2000), Hellman, Murdock, and Stiglitz (2000) and Repullo (2004) support the “franchise value” paradigm and contain models showing a trade-off between competition and financial stability.

In addition to the charter-value arguments, the screening theories described earlier also imply that market power in banking leads to higher financial stability because it improves the quality of banks’ assets. For example, Shaffer (1998) develops a model which demonstrates that a higher number of banks may cause the quality of a bank’s lending portfolio to deteriorate because of the greater number of low-quality loans. Another study (Cordella and Yeyati, 2002) investigates the effect of competition on banks’ incentive to monitor, which determines the riskiness of their portfolio. Consistent with previous studies, Cordella and Yeyati show that competition results in a lower investment in monitoring by the banks. However, they also argue that policies that promote transparent disclosure of the riskiness of bank portfolios and/or risk-based deposit insurance can provide incentives to screen even in a competitive environment. Hence, the relationship between bank competition/market power and the incentives for screening which are tightly related to the riskiness of banks’ portfolios is not clear-cut in the literature.

(11)

concentration and competition increase stability. The authors attempt to reconcile this result by testing if concentration is a proxy for better diversification or easier monitoring by supervisors, but the outcomes are mixed.

2.2.2. The “Competition-stability” nexus

The above discussion on bank competition, market power and financial stability implies that the traditional view about the importance of market power for financial stability is widely supported both theoretically and empirically. However, the existence of opposing theories such as the market power-fragility hypothesis makes the literature on the relationship between competition, risk and stability far from conclusive. Boyd and De Nicoló (2005) propose an alternative view that market concentration could impact bank stability in different ways. Specifically, concentration in the loan market could lead to higher lending rates that raise borrowers’ debt loads and default probabilities as well as their incentive to take more risk. The authors argue that, apart from the deposit channel, there is also a loan market channel that could eliminate the trade-off between competition and financial stability. In fact, they argue for a new paradigm for banking supervision: more competition decreases credit risk and enhances financial stability.

The opposing theories discussed above explicitly show that the relationship between competition, risk and stability in the banking sector is not clear but complex and multifaceted involving more than a simple trade-off. Among all the theoretical and empirical studies on bank competition, market structure, financial stability and economic growth, the most widespread view is that increased competition in the banking sector leads to excessive risk, thus to higher fragility but in the same time spurs economic growth. Conversely, concentration in the banking industry is related to decreased risk, higher stability but also lower efficiency and a hampering effect on growth. However, as noted above the literature on this topic is by no means conclusive and therefore needs further research.

2.3. Methods for measuring bank competition and concentration

(12)

competition can be measured by concentration indices such as the k bank ratio (sum of the market shares of the three (five) largest banks), or by the Herfindahl index. With regard to the first one, simplicity and limited data requirements make the k bank concentration ratio one of the most widely used measures of concentration in the empirical literature. The Herfindahl -Hirschman Index is the most common summary measure of concentration in the theoretical literature. It is the sum of the squares of the market shares of the individual market participants. It can range from 0 to 10 000, moving from a huge number of very small firms to monopoly with a single producer. The idea behind these concentration indices is that banks with larger market shares may have more market power and use that. However, as outlined in Northcott (2004), there are several problems with the SCP approach. First, accounting data on profits may not provide a precise measure of economic profits and market power. Next, in order to measure a structural variable such as concentration, one must define the relevant product and geographical markets which is quite difficult. Another important issue is the existence of a competing hypothesis that predicts the same positive relationship between concentration and profits. According to the efficient structure hypothesis, firms with greater productive efficiency have lower costs and therefore higher profits. These firms tend to perform better and eventually gain market share, which can result in concentration. Hence, concentration reflects more efficient banks, not necessarily an increase in market power. The empirical results that relate concentration with profits are inconclusive.

The new approaches for measuring bank competition are based on non-structural tests and thus try to overcome the problems with the traditional IO approach. The new empirical IO theory assesses banks' behavior directly. Moreover, it enables us to consider the actual conduct of banks by taking contestability (the degree of competitive behavior in the market) into account. The contestability literature focuses on the competitive behavior of banks, rather than on concentration or the number of banks. As described in this literature, competitive outcomes can occur in very concentrated banking markets, and there might be collusion even when there is a high number of firms. Two commonly used methods to empirically measure the degree of competitive behavior in the market have been introduced by Panzar and Rosse (Panzar and Rosse, 1987) and Bresnahan and Lau (Bresnahan, 1982; Lau, 1982). Each technique tries to assess the competitive behavior of banks ignoring the impact of bank concentration.

(13)

input prices will increase marginal costs, reduce equilibrium output, and (as a result) reduce total revenues. The H-statistic can be interpreted as follows: H<0 indicates a monopoly; H=1 indicates perfect competition; and 0<H<1 indicates monopolistic competition. The advantage of the Rosse-Panzar model is that it uses bank-level data and allows for bank-specific differences in production function (Claessens and Laeven, 2003). However, there are two main drawbacks of this method (Boone et al., 2007). First, if H ≤ 0, we actually do not learn anything, except that the sector is not in a long run competitive equilibrium. This is because a negative sum of elasticities could reflect both monopoly and oligopoly. Second, calculation of H-statistic requires information on factor prices which is usually hard to obtain. These difficulties could probably explain why only a limited number of studies has applied Rosse-Panzar model in the banking industry.

In the Bresnahan and Lau model, profit-maximizing firms set marginal cost equal to their perceived marginal revenue to determine a product’s price and the quantity they will supply. Under perfect competition, the perceived marginal revenue equals the demand price. Under monopoly, however, the perceived marginal revenue is not equal to industry demand. One advantage of the model is that it allows for an easy to obtain test statistic (λ) which calculates firms’ deviations from marginal cost (competitive) pricing. If λ=0, firms behave in a perfectly competitive manner. If λ=1, firms price according to the industry’s marginal revenue curve, which is consistent with monopoly. Values of λ between 1 and 0 reflect varying degrees of imperfect competition. An empirical advantage of the Bresnahan and Lau model is that only industry aggregate data is needed for the estimation, although using firm-specific data is also possible. However, empirical studies applying this model are rather scarce.

Other approaches relate market power (respectively competition) to profit in the sense that very large profits may be indicative of lack of competition. A widely used measure of market power is the Lerner index which shows the degree to which a firm can increase their marginal price beyond their marginal cost. The Lerner index is calculated as the difference between the price of output (loans) and marginal cost divided by price. The Lerner index is an inverse measure of competition, i.e., a higher Lerner index means lower competition. Another measure is the price-cost margin which is often used as an empirical approximation of the theoretical Lerner index. It is equal to the output price minus the marginal costs, divided by the output price.

(14)

explained in Boone et al. (2007), there are two ways in which competition can be increased. First, competition becomes more intense as the number of firms in a market increases due to a fall in entry costs. Second, competition increases as firms’ behavior becomes more aggressive (for example due to changes in consumer preferences). Using simulations Boone et al. show that the competition measures price cost margins and Herfindahl index are consistent in the former case but not in the latter. In other words, more intense competition (lower concentration) due to more entry in the market is correctly captured by a concentration measure like the Herfindahl index. The problem with concentration measures as indicators of competition is that a switch to more aggressive behavior by firms crowds inefficient firms out of the market. This increases concentration, but as stressed in Boone et al. (2007) should not be interpreted as a fall in competition. Moreover, an increase in competition tends to raise the market shares of efficient banks at the expense of inefficient banks. Such a reallocation effect also tends to raise the Herfindahl index, contrary to common expectations. The same problem concerns price cost margins as a measure of competition. Generally, more intense competition reduces the price cost margins of all banks. But since more efficient banks may have a higher price cost margins (gaining larger profits due to their higher efficiency), the increase of their market share may raise the industry’s average price cost margins, contrary to what is expected. Hence, as argued in Boone et al. (2003), price cost margins is not a reliable measure of competition because it tends to misrepresent the development of competition over time in markets with few firms and high concentration.

Outlining all the drawbacks of the commonly used competition measures, Boone et al. (2007) suggest a better measure of competition called profit elasticity. It shows the elasticity of a firm’s profits with respect to its marginal costs. A higher value of this profit elasticity indicates more intense competition. The underlying intuition is as follows: in all markets, an increase in marginal costs leads to a reduction of firm’s profits. However, in case of a competitive market, the same percentage increase in marginal costs will lead to a greater fall in profits. The reason, as explained by Boone et al., is that in a more competitive market, firms are punished more harshly (in terms of reduction in profits) for their inefficiency. The author demonstrates that this measure is more informative and reliable than standard competition measures like Herfindahl and price cost margins because it picks up all forms of changes in competition correctly. It is given in the following equation:

ln(πk) = α + β ln(ck)

(15)

negative relationship between profits and marginal costs (profits of banks increase when their marginal costs fall). The stronger competition is, the stronger this effect will be, and the larger, in absolute terms, this (negative) value of β.

Since Boone indicator is a new measure, it has been applied in very few studies and the question, whether it is able to correctly measure competition in practice, is an unanswered question yet. However, its advantages have been outlined in several studies. The work of Van Leuvensteijn and Bikker (2007) which is the first that applies this method in the banking industry, measures competition in several EU countries, USA and Japan. They even improve Boone indicator because instead of using average variable costs as an approximation of marginal costs (as in Boone et al.), they calculate marginal costs. Furthermore, instead of profits, they use market shares of banks which have the advantage of being only positive, contrary to profits (losses). The main innovation of their approach, as argued in the study, is that it allows measurement of competition not only for the entire banking market, but also for separate product markets, such as the loan market, and for single types of banks, such as commercial, savings and cooperative banks. Another study (Schaeck and Cihak, 2008) which uses Boone indicator, examines the effect of competition on bank soundness. Their main finding (competition, measured by Boone indicator, is positively related to bank soundness), contrasts with the prevailing view both in the literature and in policymaking. The authors attribute this finding partially to the new methodology. However, a study by Van Leuvensteijn (2008) provides empirical evidence that the Boone-indicator is an appropriate measure of competition. More specifically, he demonstrates empirically that Boone-indicator is better able to identify the different regimes of competition than Lerner index. Given this, as well as the shortcomings of the other measures of competition discusses above, Boone indicator is considered as one of the recent most innovative methods of measuring competition in banking. However, like all other model-based measures, the Boone indicator has several shortcomings. First, as discussed in Bikker and Van Leuvensteijn (2007) the Boone indicator assumes that banks generally pass at least part of their efficiency gains to their clients. Second, like many other measures, it ignores differences in bank product quality and design, as well as the attractiveness of innovations. Finally, compared to direct measures of competition, Boone indicator may suffer from the disadvantage of being an estimate and is therefore surrounded by a degree of uncertainty.

(16)

Laeven (2003) build on this work by attempting to link the competitiveness of a country’s banking sector to structural and regulatory indicators of the financial system. They use panel data (1994–2001) to construct H-statistics for 50 countries. Claessens and Laeven find that contestability is positively related to foreign bank presence, less-severe entry restrictions, and few activity restrictions. In all specifications, contestability is positively related to concentration and negatively related to the number of banks. This suggests that measures of markets structures do not translate in measures of effective competition and it is absolutely possible that more concentrated banking systems are more competitive (which was actually found in Claessens and Laeven, 2003).

To sum up, the ambiguous outcomes of the concentration approach and the implications of the emerging contestability literature suggest that although it may seem counterintuitive, concentration and competition can exist together in the banking market. Therefore, this should be taken into account when interpreting the relation between different measures of bank competition and concentration.

3. Methodology, data sources and measures

3.1. Methodology

The empirical analysis on the growth effects of bank competition (and bank concentration) and the way financial stability modifies this relationship is based on two main model specifications – a basic and an extended one and they are both specified as interaction models.

3.1.1. Basic model

The basic model explores how stability in the banking sector modifies the effect of bank competition on economic growth at large. In other words this model enables us to identify an economy-wide effect of stability on the relationship between competition and economic growth, regardless of industry-specific characteristics.

The basic growth equation is:

(1) gYij= α*SHij + β*COMPj + γ*STABj + δ*COMPj*STABj + Industry dummies(i) + Country controls(j) + εij

(17)

and competition (COMPj*STABj); i is the industry index, j – the country index, εij is the error term.

In the specification of the model in (1), we build on one of the prominent studies in the area - Cetorelli and Gambera (2001) which explores the impact of banking market structure on industrial growth and uses a very similar model. Specifically, they find that bank concentration has an overall depressing effect on industry growth. As in Cetorelli and Gambera (2001) and most of the other similar cross-country growth studies, the measures of industry growth, competition and stability are averaged over the period under study (1995-2004). The industry’s share in value added which is included as a control variable (as in Rajan and Zingales, 1998; Cetorelli and Gambera, 2001; Inklaar and Koetter, 2008), is calculated in the beginning of the period (1995) and captures an industry-specific convergence effect: sectors that have already grown substantially in the past are unlikely to continue to grow at a high rate in the future. Therefore, α is expected to have a negative sign. In order to correct for industry-specific effects we include industry dummies. We also follow Cetorelli and Gambera (2001) and include country controls in order to reduce the probability of model misspecification due to the omission of important variables. The country controls that we use are commonly used in cross-country growth studies (also used in Cetorelli and Gambera, 2001) and they include bank development in 1995, the logarithm of per-capita income in 1995 and stock market capitalization over GDP in 1995. The measure of bank development is the commonly used ratio between private credit and GDP and the sign of the coefficient is expected to be positive. Per-capita GDP captures convergence effect of the economy as a whole and we expect that it will have a negative sign. Stock market capitalization controls for the importance of alternative sources of external finance and is expected to have positive impact. In the study of Cetorelli and Gambera (2001) the control variables were found to have exactly these signs. However, similar to Rajan and Zingales (1998), Cetorelli and Gambera use two more control variables – accounting standards and the level of human capital. However, due to data restrictions, we use only the first three country controls.

(18)

variable. Such a setting is also well-grounded in theory and practice. It has been shown that financial stability is important for economic growth. As for competition, given the opposing “competition-fragility” and “competition-stability” theories, its modifying effect on this relationship is not clear and is therefore, an interesting question to look at. However, as seen, specifying a model like that changes thoroughly the research question because we are no longer investigating the growth effect of bank competition but are primarily interested in the effect of financial stability. That is why, using an interaction model like (1) is more appropriate in terms of finding answers for our research questions. More specifically, it implies that the effect of a change in bank competition on economic growth depends on the level of financial stability:

dYij/dCOMPj = β + δ *STABj

Because of the existence of opposing theories on the relationship between competition, economic growth and stability, all the coefficients in the growth equation are a priori ambiguous. Our predictions about the signs of the coefficients would differ depending on the strand of theories we address.

According to the widely acknowledged view about the trade-off between economic efficiency and stability in the banking system, a competitive banking system is more efficient in the allocation of capital and therefore fosters growth. In the same time, as the “franchise value” paradigm states, competition also erodes the market power and results in excessive risk-taking, meaning that there exists a trade-off between competition and financial stability.

Hence, based on this strand of theories, it could be expected that the positive growth impact of bank competition is likely to decrease when the level of financial stability is very high.

(19)

The positive effect of competition on industry growth is likely to decrease when there is less stability in the banking sector.

Although many recent studies find support of a positive effect of bank competition on industry growth, the literature seems to be ambiguous about this relationship. There are some theories suggesting that competition has a negative impact on industry growth because banks have fewer incentives to invest in relationship lending.

Hence, a negative impact of bank competition on growth, depending on the level of financial stability, could be also expected.

3.1.2.. Extended model

In order to correct for country and industry specific effects we use an approach similar to the one used by Rajan and Zingales (1998) who take into account within-country differences between industries using an interaction between a country and an industry characteristic. In particular, the country characteristic they use is financial development and the industry characteristic is industries’ dependence on external finance. It is argued that this approach has a significant advantage over previous cross-country studies and is less subject to criticism about an omitted variable bias. What Rajan and Zingales show is that industries that are more dependent on external financing have relatively higher growth rates in countries that have more developed financial markets. An important element of this approach is that the authors use data on U.S. firms as proxies at the industry level to construct a measure of the typical external dependence for a particular industrial sector. Basically, their model assumes that the external financing needs across industries are similar across countries and also that the USA is a good proxy for the demand for external financing in other countries.

(20)

The extended model specification can provide us with better understanding of the modifying effect of financial stability on the relationship between competition and growth because it enables us to decompose the economy-wide effect of financial stability illustrated by the basic model, in a sector-specific level. The extended model is as follows:

(2) gYij= α1*SHij + β1*DEPij*COMPj + γ1* DEPij *STABj + δ1* DEPij*COMPj*STABj +

ηi + ηj + εij

where DEPij indicates external financial dependence of industry i in country j. The others are the same as those described in the basic model. The measures of industry growth, financial dependence, competition and stability are all averaged over the period under study (1995-2004) and country and industry dummies are also included.

In this extended model we include interactions between external financial dependence of industry j in country i and each of the explanatory variables. Thus we can explore whether the growth impact of bank competition and concentration, as well as the modifying effect of financial stability are heterogeneous across industries. More specifically, the first interaction term shows if industrial sectors that typically use more external financing grow faster (slower) in countries with less (more) bank competition. The second one illustrates whether industries that are more dependent on external financing grow faster or slower in countries with more stable (less) banking systems. With the third interaction term we can test if the growth effects of bank competition (and concentration) and the modifying effect of financial stability on these relationships differ across industries. Taking into consideration the above-discussed ambiguity of the theory on bank competition, stability and growth, the signs of the coefficients in this equation are a priori ambiguous.

In order to allow for comparisons with other studies (Cetorelli and Gambera, 2001 in particular), we also estimate the above equations using bank concentration instead of the competition measure. More specifically, we estimate an interaction model, similar to (1):

(3) gYij= α2*SHij + β2*CONCj + γ2*STABj + δ2*CONCj*STABj + Industry dummies(i) + Country controls(j) + εij

(21)

To check if the growth impact of bank concentration and the modifying effect of stability are different across industries that are more or less dependent on external finance, we also estimate an extended model, similar to equation (2):

(4) gYij= α3*SHij + β3*DEPij*CONCj + γ3* DEPij *STABj + δ3* DEPij*CONCj*STABj +

ηi + ηj + εji

As the literature on banking market structure and growth is mixed, the expected outcomes of the regressions are also a priori ambiguous. In particular, according to the widely acknowledged view about the trade-off between economic efficiency and stability in the banking system, market power (associated with a concentrated banking sector) tends to exert a negative effect on growth. Based on that, as well as on the supporting evidence of many empirical studies, we expect a negative sign of β2 - the coefficient of bank concentration in the interaction model (3). However, there are recent strands of literature such as the theories about relationship lending and screening which suggest a positive impact of bank concentration on industry growth. According to them, a positive coefficient on bank concentration can be also expected. How the growth impact of bank concentration changes depending on the level of stability in the banking sector, is a question that also cannot be given straightforward answer. First, as discussed earlier, the literature on the market power – stability nexus is ambiguous and second, the problem setting that we use (linking concentration, stability and growth), has not been analyzed before. As for the extended model in (4), the ambiguity of the theories and empirical works also prevents us from making strong predictions. One of the few studies that explore the growth impact of bank concentration taking into account industries’ differences, is Cetorelli and Gambera (2001) who find that although bank concentration has a negative effect on growth as whole, industries that are more dependent on external finance, enjoy a beneficial effect from a concentrated banking sector. Hence, based on that, we expect β3 in equation (4) to be positive. As for the modifying effect of financial stability this model, as in model (3) it is a priori ambiguous.

3.2. Data sources and measures

(22)

Industrial growth

The empirical analysis is based on an augmented version of Inklaar and Koetter (2008) dataset. The sample includes 25 countries (those which were EU members before 2007) and for each of them, 25 industries. The source for this dataset is EU KLEMS database (March 2007 version). It contains detailed industry-level data on outputs, inputs and productivity, starting in 1970 and covering 25 EU countries as well as important non-EU countries like the US. All data are on the same industry classification to make international comparisons easier. The 25 industries included in the study cover the non-financial market economy at the one to two-digit level of industry detail and omit the financial sector, government, education, health and real estate. The financial sector is omitted because the aim is to explore the impact of competition and stability in the banking sector on the growth in other industries. Government, education and health are excluded because output measurement in those industries is quite problematic. A list of all industries can be found in Appendix Table 6.

The measure of industry growth that we use, namely growth of value added in each industry in each county for the period 1995-2004, is taken directly from this dataset. As in most of the studies analyzing the impact of different aspects of the banking sector on economic growth, we use period-average data on growth of industrial value added in each country.

Financial stability

To measure stability in the banking sector we use the Z-index which has been widely used in recent studies (e.g., Mercieca, Schaeck, and Wolfe, 2007; Schaeck and Cihak, 2008; Demirgüç-Kunt, Detragiache, Tressel, 2006). It is given by the formula:

Zk = (ROAk +E/TAk) / σROAk

where ROAk is the average return on assets for bank k, E/TA represents the

period-average equity to total assets ratio for bank k, and σROAk is the standard deviation of return on assets over the period under study. The Z-index increases with higher profitability and capitalization levels, and decreases with unstable earnings reflected by a higher standard deviation of return on assets. It inversely signals the bank’s probability of failure and is an indicator of financial stability at the firm level (a larger value reflects a higher bank stability and less overall bank risk).

(23)

theoretically grounded. More specifically, it can be shown that the Z index is an inverse proxy for financial institution’s insolvency, i.e. the probability that the value of its assets becomes lower than the value of its debt. Second, the Z index has an important practical advantage because it can be calculated in an easy and transparent fashion for all the banks in the sample since only accounting information is needed. Third, empirical studies confirm that the Z-score is indeed a useful measure of bank stability. For example, Cihak (2008), using a sample of 29 countries, including 12 with systemic banking crises, finds that the latter are characterized by significantly lower Z-scores than other banks. A study on the effect of competition on bank efficiency and stability (Schaeck, Čihák, 2008) also finds that the mean Z-index in failed banks is much smaller than the Z-scores in the rest of the sample.

The bank-specific data needed for the calculation of the Z-index are derived from the Bankscope database. The sample includes 8741 banks from the 25 EU member states and covers the period 1995-2004. We use period-average data to compute period-average Z-scores for each bank in the sample and then we average them across banks in order to get a measure of stability for each country. The values of Z index for each country in the sample are given in Table 1 below.

(24)

As seen from Table 1, Z index varies significantly across countries - the countries with higher Z-index, indicating a higher level of bank stability are mainly the older member states, whereas those countries which became members of the EU in 2004 have lower values of this indicator, i.e. they are characterized by less stability in the banking sector (especially Cyprus, Hungary, Latvia). The mean value of Z index is 50.09585. 17 countries from EU-25 have values of Z index lower than the mean. It is noteworthy that this refers to all the 10 new EU members. The other 8 countries, all of them incumbent member states (Sweden, Belgium, Germany, The Netherlands, Spain, Italy, Austria, UK) have values of Z index above the mean.

Bank competition

For measuring bank competition we chose an innovative measure introduced by Boone et al. (2007) which overcomes the drawbacks of previous measures used in the literature (the Boone indicator as well as the other main measures of competition were described in detail in the literature review). As explained earlier, Boone indicator shows the elasticity of a firm’s profits with respect to its cost level. A higher value of this profit elasticity signals more intense competition. Boone et al. (2007) argue that this measure is more informative and reliable than the standard competition measures like Herfindahl and price cost margins because it picks up all forms of changes in competition correctly.

Based on that, and taking into account all the above discussed drawbacks of the other competition measures, in our analysis we choose the Boone indicator as a measure of competition. It is given in the following equation:

ln(πk) = α + β ln(ck)

The slope β indicates the percentage fall in profits due to a one percent increase in marginal costs. We refer to the β parameter as the Boone indicator. The slope β is expected to be negative because of the negative relationship between profits and marginal costs. The stronger competition is, the stronger this effect will be, and the larger, in absolute terms, this (negative) value of β. As in all empirical studies using Boone indicator, we specify the above equation in log-log terms. As argued in Van Leuvensteijn, Bikker (2007) choosing this specification helps in dealing with heteroskedasticty and also in interpreting the slope β as it implies that β is elasticity.

(25)

database) doesn’t provide comprehensive information about all variable costs but mostly about overheads. As overheads are part of variable costs (i.e. indirect variable costs) we use them as an approximation of variable costs. Furthermore, as in Boone et al. we do not have data on gross output and therefore we calculate the average variable costs by dividing overheads by revenue. To calculate revenue we sum net interest revenue and commission income as they represent the biggest part of total bank revenue. As for profits, we use profit before taxes directly from the Bankscope Database. As in Boone et al. (2007) in order to obtain reliable bank-level data we remove observations of banks with negative profits and negative average variable costs. This reduces the sample to 8449 banks. The time period covered is 1995-2004.

We obtained the Boone indicator for bank competition for each country in the sample by estimating a panel data model for each country. As in Schaeck and Cihak (2008) in order to prevent our results from being driven by extreme values, we dropped observations for which the respective variables lie in the 1st or 99th percentile of the distribution. The number of observations for each country, as well as the mean, minimum and maximum values of profits and costs for the banks in each country is given in Appendix Table 3. As seen from the Appendix Table 3, Germany, Italy, France and UK dominate the sample with, respectively, 15299, 5690, 4181 and 2479 observations. The countries with the smallest number of observations are Malta and Estonia (80 and 96). The countries where the banks have the lowest mean profit are Latvia and Lithuania while the ones with the highest mean profit are in the UK and the Netherlands. The countries, where the mean average variable costs of the banks are the lowest, are Luxemburg and Austria whereas the ones, where banks have the highest mean costs, are UK and Lithuania.

In order to obtain the Boone indicator for competition, we estimate the following equation for each country:

ln(Πkt) = α + β ln(Ckt) + εkt

(26)

allows estimation in the presence of autocorrelation within panels and cross-sectional correlation and heteroskedasticity across panels. For the above-mentioned four countries we corrected only for heteroskedasticity by obtaining robust standard errors. The values of Boone indicator for all countries in the sample are given in Table 2.

Table 2, Boone indicator for EU countries for the period 1995-2004 Country Boone indicator

(27)

(0.049) Sweden -0.860*** (0.066) Slovenia -0.196 (0.335) UK -0.539*** (0.029)

* denotes a coefficient significantly different from zero at the 10%-level, ** at the 5%-level and *** at the 1%-level

All coefficients are negative as predicted, indicating a negative relationship between profits and costs. We obtained statistically significant coefficients for all countries but three – Malta, Slovenia and Slovakia. The insignificant results could be attributed to the smaller number of observations for these countries. Therefore, Boone indicators for these countries were not included in the final data set. The Boone indicator is the lowest (in absolute terms) for France, Portugal and Greece, suggesting that in these countries competition among banks is the lowest. The Boone indicator is the highest (in absolute terms) in Estonia, Germany and Italy, implying that in these countries bank competition is the strongest. The mean value of Boone indicator is -0.6937555. Nine countries (Lithuania, Spain, Denmark, Sweden, Austria, Latvia, Italy, Germany, and Estonia) have Boone indicators less than that, meaning that in these countries competition is higher than the mean for the countries in the sample as a whole.

As explained above, the Boone indicator is decreasing in the degree of competition, meaning that if competition increases industrial growth, the relation between the measure of industrial growth and Boone indicator can be expected to be negative. Based on that and also taking into account the specification of the model as interaction model, for easier interpretation of results, in the dataset we include the absolute values of Boone indicator for the countries in the sample.

Bank concentration

(28)

Table 3: Bank concentration in EU countries for the period 1995-2004 Country 3-bank concentration ratio Austria Belgium Cyprus Czech Germany Denmark Estonia Spain Finland France UK Greece Hungary Ireland Italy Lithuania Luxembourg Latvia Malta Netherlands Poland Portugal Sweden Slovenia Slovakia 0.235 0.240 0.408 0.478 0.151 0.415 0.635 0.315 0.673 0.208 0.215 0.388 0.444 0.375 0.194 0.656 0.204 0.482 1.053 0.535 0.420 0.321 0.401 0.507 0.610

As seen from the table, the counties with the least concentrated markets are Germany, Italy, Luxemburg, France, UK, whereas Malta, Estonia, Lithuania and Finland have the most concentrated banking industries.

Financial Dependence

(29)

of the UK as a benchmark is motivated by the fact that it is one of the countries with the highest levels of financial development. Moreover, the large number of UK firms in the sample ensures that the financial dependence measures will not be influenced by industries with few firm observations. The measure of financial dependence is also averaged over the period under study, namely 1995-2004.

(30)

Table 4: Correlation matrix Variables Growth Share in

value added Boone indicator 3-bank ratio Z index Fin. depen dence Stock market /GDP Priv. credit /GDP Log GDP Growth 1 Share in value added 0.024 1 Boone Indicator 0.062 0.010 1 3-bank Ratio 0.092* 0.023 0.039 1 Z index -0.119* -0.0072 0.183* -0.376* 1 Fin. dependence -0.066 0.082* -0.000 -0.000 -0.000 1 Stock market /GDP 0.023 -0.055 -0.065 -0.325* 0.205* -0.000 1 Priv. credit/ GDP -0.194* -0.019 -0.190* -0.174* 0.278* 0.000 0.452* 1 Log GDP -0.212* -0.039 -0.159* -0.500* 0.501* 0.000 0.684* 0.616* 1

* indicates that the correlation is different from zero at the 5 percent significant level.

4. Results

4.1. Basic model

Given the data on industrial growth, bank competition, concentration and financial stability, we turn next to the estimation of the basic model. As mentioned in the methodology section, in order to allow for comparisons with previous studies, we also estimate the model using bank concentration instead of bank competition. Results are reported in Table 5 below.

(31)

distribution. (Detection of outliers, as well as other diagnostic checks will be discussed in Section 6).

Table 5: Industry value added growth, bank competition, concentration and stability in the banking sector

Notes: Regressions are estimated using robust regression technique. Standard errors are in parentheses * denotes a coefficient significantly different from zero at the 10%-level, ** at the 5%-level and *** at the 1%-level

In the first column we estimate the average effect of bank competition on industry growth in an unconditional model, i.e. is it is not included in an interaction term. We find a negative significant effect of bank competition on industrial growth. This result seems to contrast with the widely acknowledged view about the positive link between bank competition and efficiency, hence growth. As outlined in the literature review, many theoretical, as well as empirical works state and find that bank competition maximizes welfare by providing the greatest quantity of credit at the lowest price, thus benefiting especially financially dependent firms. However, although most recent studies find a positive relationship between competition and growth, there are studies that suggest an ambiguous Dependent variable:

growth of real value added

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

Industry share in value added -0.098 (0.076) -0.191*** (0.070) -0.098 (0.075) -0.102 (0.075) Bank development -0.005 (0.003) -0.008** (0.004) -0.009*** (0.004) -0.009*** (0.004) Log of per-capita GDP -0.011** (0.005) -0.016*** (0.004) -0.023*** (0.005) -0.020*** (0.005) Boone indicator -0.013*** (0.005) 0.027*** (0.007) Stock market capitalization/GDP 0.007 (0.004) 0.015*** (0.004) 0.011** (0.004) Z index 0.0003*** (0.0001) -0.0002** (0.0001) Boone indicator *Zindex -0.0005*** (0.0001) Bank concentration 0.002 (0.006) -0.030* (0.016) Bank concentration* Z index 0.0006** (0.0003)

Industry dummies Yes Yes Yes Yes

Country dummies No No No No

(32)

relationship between the degree of banking system competition and firm access to external financing and, in turn, firm growth. According to this strand of literature (Rajan 1992; Petersen and Rajan (1995), Cetorelli and Peretto (2000) more competitive banking systems may channel less financing to firms because they have less of an incentive to invest in close relationships with firms. As a result, especially those sectors that are heavily reliant on external financing grow more slowly in systems that are more competitive. An empirical study of Claessens and Laeven (2003) confirms this theory – they find that their competitiveness measure (H-statistic) is negatively associated with countries’ growth, suggesting that less competitive banking systems are better at providing financing to financially dependent firms. However, they also find that that the effects of bank competition on growth vary with the level of countries’ financial sector development and only in countries with less developed financial systems do financially dependent industries grow faster if the financial system is less competitive while in more developed financial systems they grow faster when the financial system is more competitive.

In column (2) we estimate the interaction model specified in equation (1). However, stock capitalization was not included as a control variable in this estimation because diagnostic checks (specification link test for single-equation models and regression specification error test for omitted variables) showed that the model is better specified this way. All variables are statistically significant. The industry’s share in value added and per-capita GDP have negative signs as predicted. Contrary to Cetorelli and Gambera (2001) and similar studies, we find a negative effect of bank development on growth. This may be due to the different sample of countries that we use, as well as the different time period covered (we use the ratio between private credit and GDP in the beginning of the sample period, namely 1995). As for the three main independent variables (Boone indicator for competition, Z index measuring financial stability and the interaction term between both), all of them are statistically significant, the first two being positive and the third one – negative.

(33)

independent variable in an interaction model (competition in our case) shows the effect of this variable on the dependent variable (growth) when the modifying variable (stability) is zero. However, since there are no cases in which Z index is zero, the results from the Table 7 (column 2) are not that informative and to illustrate the marginal effect of competition (column 2) across the range of values of the modifying variable (stability) we use Figure 1. It has also been suggested by Brambor, Clark and Golder as a better way of interpreting interaction models.

Figure 1: Marginal effect of competition on industry growth in EU-25

-. 1 -. 0 5 0 .0 5 “M a rg in a l E ff e c t o f C o m p e ti ti o n ” 0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 Stability

“Marginal Effect of Competition” “95% Confidence Interval”

(34)

the incumbent EU members, all other countries, which experience positive growth impact of bank competition, are new member states (which are also characterized by lower financial stability). This outcome could be explained with the fact that in a more risky environment banks do not tend to perform too much credit rationing or excessive monitoring and hence, they provide larger amount of credit to more borrowers, thus fostering industry growth. The positive growth impact of bank competition in the new member states could be also related to the fact that one of the major factors that shape bank competition in transition economies is the intense entry of foreign-owned banks. It is argued that the latter tends to raise competition and also brings positive spill-over effects, such as new financial services, management innovations, better bank supervision (Lensink and Hermes, 2004). There are very few studies which examine these relationships in the context of the new member states or the enlarged EU. In a study on Eastern Europe Koivu (2002) finds that lower interest rate spreads (more intense competition) lead to faster GDP growth in transition countries in the 1990s. Botric and Slijepcevic (2008) perform a similar analysis on southeast European countries and confirm the positive effect of lower interest rate spreads on growth. The only recent study on the enlarged EU (Pellenyi and Borko, 2009) also finds evidence of a beneficial effect of bank competition on firm entry and economic performance. Hence, our finding of a positive growth impact of bank competition in some of the new member states falls in line with this strand of empirical studies.

(35)

grow faster when the financial system is less competitive while in more developed financial systems, more competition is associated with higher growth. However, our finding of a negative growth impact of bank competition in the above-mentioned four incumbent member states, contradicts this evidence (these countries have advanced and developed financial systems and one would expect, as shown in Claessens and Laeven (2003), that more competition in the banking sector would be beneficial for industry growth). The reason for this difference might be that we do not examine the growth effect of bank competition alone but we take into account also the level of financial stability. As we are not aware of studies which explore the effect of bank competition in this way, we cannot make a direct comparison. Second, the differences in results might be due to the different sample of countries, the time period under study, as well as the estimation technique used (many studies in this field, apply instrumental variables while we use robust regression technique). Last but not least, as outlined in the literature review, the chosen measure of bank competition (Boone indicator) suffers from the drawback of being an estimate and this could adversely affect the final results.

As mentioned above, in order to allow for a direct comparison with the study of Cetorelli and Gambera (2001), we also estimate the effect of bank concentration on industry growth. As in their study we use the three-bank ratio as a measure of bank concentration which is calculated as the sum of market shares (measured in total assets) of the three largest banks in each country. The results from the regression are shown in the last two columns of Table 5. In column (3) we examine the average effect of bank concentration on industry growth without including it in an interaction term whereas in column (4) we add Z index as a measure of stability and an interaction between the measure of concentration and stability. However, in contrast to Cetorelli and Gambera (2001) we find no effect of bank concentration on industry growth (the coefficient of bank concentration in column (3) is insignificant). The differences may be due to the different sample of countries, the different time span (the time period they study is 1980-1990), the estimation techniques used (Cetorelli and Gambera, 2001 also use instrumental variables and have two more control variables – accounting standards index and human capital). However, the lack of effect of bank market structure on growth of value added is in line with the findings of another prominent study – Claessens and Laeven (2005). They use the same specification as Cetorelli and Gambera (2001) but find no evidence that market structure—that is, concentration in the banking system—helps explain industrial sector growth.

(36)

(negative for bank concentration and Z index and positive for the interaction term). In order to interpret these results, we illustrate them with the graph on Figure 2.

Figure 2: Marginal effect of bank concentration on industry growth.

-. 0 5 0 .0 5 .1 .1 5 “M a rg in a l E ff e c t o f C o n c e n tr a ti o n ” 0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 Stability

“Marginal Effect of Concentration” “90% Confidence Interval”

(37)

link between the growth of industries which are more dependent on external finance and the concentration of the banking sector.

4.2. Extended model

In the extended model we estimate the impact of bank competition on industry growth and how stability modifies this relationship taking into account differences between industries. We do so by including dependence on external finance which interacts with the independent variables in the model. As mentioned in the methodology part, this approach has been suggested by Rajan and Zingales (1998) and has been used in a number of studies. As in Inklaar and Koetter (2008) the measure for financial dependence is based on UK firms as the benchmark country and the growth observations for the UK are excluded in order to avoid endogeneity. The results of the regressions based on this specification are shown in Table 6.

Table 6: Industry value added growth, financial dependence, competition, concentration and stability in the banking sector in EU-25

Notes: Regressions are estimated using robust regression technique. Standard errors in parentheses * denotes a coefficient significantly different from zero at the 10%-level, ** at the 5%-level and *** at the 1%-level

Dependent variable: growth of real value added

(1) (2) (3)

Industry share in value added -0.140** (0.058) -0.183*** (0.065) -0.175*** (0.067) Financial dependence * Boone indicator -0.271*** (0.076) Financial dependence * Z index -0.002* (0.001) -0.001 (0.002) Financial dependence *

Zindex * Boone indicator

0.004** (0.002) Financial dependence * Bank

development

0.144*** (0.041) Financial dependence * Bank

concentration -0.098 (0.082) -0.340 (0.206) Financial dependence*Bank concentration*Zindex 0.005 (0.004)

Industry dummies Yes Yes Yes

Country dummies Yes Yes Yes

(38)

In the first column we show the results from the estimation of the extended model using Boone indicator as a measure of competition and in the third column we estimate the same equation but instead of Boone indicator, we use 3-bank concentration ratio. The extended model in column (1) gives significant coefficients for all variables. The negative coefficient of the interaction term between financial dependence and Boone indicator implies that industrial sectors that typically use more external financing, grow faster in countries with less competition in the banking sector. This finding falls in line with theoretical approaches and empirical works which support a negative relationship between bank competition and growth of industries that depend heavily on external financing (Rajan 1992; Petersen and Rajan 1995; Claessens and Laeven (2003). Furthermore, it should be noted that the positive sign of the interaction term between the three variables suggests that the reductive impact of bank competition on growth of industries, that are more in need of external finance, tends to decrease as stability in the banking sector improves. This could be explained with the fact that in a more stable environment banks are likely to do more screening and monitoring of their borrowers as well as more relationship lending, which results in more credit supply for industries that rely heavily on external finance. Because of that, the negative growth effect of bank competition becomes less pronounced.

In column (2), to allow for a comparison with Cetorelli and Gambera (2001), we estimate the impact of bank concentration using the same specification, i.e. including industry share in value added, interaction between financial dependence and bank concentration and between financial dependence and bank development. Control variables are also included. As in their study, we find significant negative coefficient on industry share in value added as well as a significant positive coefficient on the interaction between financial dependence and bank development. The latter implies that industries that are more reliant on external finance grow faster in countries with more developed bank systems. However, we find the coefficient on the interaction between financial dependence and bank concentration to be insignificant whereas Cetorelli and Gambera (2001) find a significant positive coefficient, indicating that bank market power, by facilitating the formation of lending relationships, enhances the growth of those sectors that are more dependent on external finance. Apart from the explanations given earlier, another reason for the different results might be that Cetorelli and Gambera (2001) focus on the external financing needs of young firms (less than ten years old), whereas our analysis covers all firms.

Referenties

GERELATEERDE DOCUMENTEN

J aar lij ks wordt gemest met een comp ost g emaakt van stalmest vermen gd m et oogstafval en ander or gani sch m ateriaal dat bij het beheer vrijk omt.. Het

The next step in the methodology, after having found evidence of long-run cointegration relationships between stock market development, economic growth and investment, is the

In other words, when using Boone indicator, which accounts for changes in competition more comprehensively (Schaek and Čihak, 2008b), results generally suggest that

In this paper I used data on a sample of ten of the largest oil exporting countries to illustrate the possible effects, positive or negative, of the financial-structure

The results show that all financial indicators have a positive significant effect on capacity and a negative significant effect on the share of capacity, except

Figure 2 presents the regression results for the interaction effect of the total sum of market share of cooperative and savings banks and the commercial bank

where R Cit represents the natural log of the actual yearly excess stock return of bank i in period t, Cλi represents the risk premium awarded for exposure to the factor

The instruments for the Boone indicator and bank size in regression (7) and (8) are: Financial freedom, Activity restrictions, the percent of foreign owned banks,