DO FINANCIAL SECTOR STRUCTURE AND DEVELOPMENT MATTER FOR THE EFFECT OF BANK CAPITAL ON LENDING IN LARGE EU BANKS?

e magnitude of the eect of changes in bank capital on the extension of bank credit has been one of the most important questions of the crisis, due to the role that banks play in the economy. In the European Union (EU) context, with bank -oriented nancial systems, bank capital may be even more salient, as capital losses may result in reduced lending and therefore be a hindrance to real economy investment activity and thus economic growth. As implementation of more restrictive Basel III capital standards in the EU is accelerating, due to the formal acceptance of its rules into a directive and a regulation in 2013 and due to relatively scant evidence on the role of bank capital in lending activity in the EU, it seems vital to answer the question what the impact of capital ratios on the EU banks’ lending is. As previous cross-country studies suggest that the procyclicality of bank capital regulation is alleviated by more stringent bank regulations and supervision, we ask whether additional country specic factors inuence the size of the eect of the bank capital ratio on lending during contractions in the EU. We hypothesise that


Introduction
e magnitude of the e ect of changes in bank capital on the extension of bank credit has been one of the most important questions of the crisis, due to the role that banks play in the economy.In the European Union (EU) context, with bank -oriented nancial systems, bank capital may be even more salient, as capital losses may result in reduced lending and therefore be a hindrance to real economy investment activity and thus economic growth.As implementation of more restrictive Basel III capital standards in the EU is accelerating, due to the formal acceptance of its rules into a directive 1 and a regulation 2 in 2013 and due to relatively scant evidence on the role of bank capital in lending activity in the EU, it seems vital to answer the question what the impact of capital ratios on the EU banks' lending is.
As previous cross-country studies suggest that the procyclicality of bank capital regulation is alleviated by more stringent bank regulations and supervision, 3 we ask whether additional country speci c factors in uence the size of the e ect of the bank capital ratio on lending during contractions in the EU.We hypothesise that in countries with a nancial sector structure that is more capital markets oriented and with a more developed nancial sector the impact of capital ratio on loan supply in contractions is enhanced.
Previous papers concentrate 4 on the e ects of bank capital on lending, both in expansions and in recessions within a single country or in a limited sample of countries.ey do not take account of the role of the cross country di erences for the association between lending and capital ratios in contractions.Our study makes two contributions relative to the literature.First, we focus on the impact of nancial sector structure on the e ect of bank capital on loan growth in contractions.Second, we test the role of nancial sector development in the association between loan growth and the capital ratio in contractions.
To test our hypotheses we apply a two-step GMM robust estimator (Arrelano and  Bond, 1991, Blundell and Bond, 1998) for data spanning 1996-2011 on individual banks available in the Bankscope database.
e rest of the paper is organised as follows.Section 2 develops our hypotheses.We describe our sample and research design in Section 3. We discuss results in Section 4. Section 5 concludes our work.

Hypotheses
Banks facing external nancing frictions, such as the Myers and Majluf5 adverse-selection problem, cannot immediately restore equity capital declines, which may result in reduction of lending and thus have negative consequences for the economy. 6Some authors suggest that this reduction due to insu cient capital is stronger in recessionary than in expansionary periods. 7Our objective is to test if such an e ect exists in the case of EU countries, especially those which have more market-oriented economies or which have better developed nancial sector.e experience of the nancial crisis in 2007 and 2008 showed that countries in which the nancial sector plays an important role in the economy, are more prone to shocks stemming from nancial markets in generally, and from the banking sector in particular.We ask whether such shocks may be caused by (probably insu cient) capital ratios which banks keep in such economies.To resolve this problem we put forward two hypotheses: H1: In countries which are more capital market-oriented the association between loan growth and the capital ratio in contractionary periods is strengthened.H2: The higher the development of financial sector, the more important the bank capital for loan extension.

Data
We use pooled cross-section and time series data of individual banks' balance sheet items and pro t and loss accounts from 27 EU countries and country-speci c macroeconomic indicators for these countries, over a period from 1996 to 2011.
e balance sheet and pro t and loss account data are taken from unconsolidated nancials available in the Bankscope database, whereas the macroeconomic data were accessed from the EUROSTAT and the IMF web pages.We exclude from our sample outlier banks by eliminating the extreme bank-speci c observations when a given variable adopts extreme values.Since most of these institutions are located in Ireland, the number of countries included in the nal sample drops to 26.Based on this selection strategy, the number of banks included in our sample is 2523 (27359 observations and 26 countries).
To explore the relation between the sensitivity of loan growth to the capital ratio in contractions and the nancial structure (FINSTR) we apply the aggregated indicator constructed by Beck and Levine 8 which is the rst principal component of two variables that measure the comparative activity and size of markets and banks.Each of the underlying variables is constructed in such a way that higher values indicate more market-based nancial systems.e rst variable equals the log of the ratio of value traded (equal to the value of stock transactions as a share of national output) to bank credit (which equals the claims of the banking sector on the private sector as a share of GDP).e second variable equals the log of the ratio of market capitalisation to bank credit.Following Beck and Levine9 we de ne market capitalisation as the value of listed shares divided by GDP, and it is our measure of the size of stock markets relative to the economy.We use data for FINSTR averaged over the period of 1996-2010.We take all values, i.e. value traded, bank credit and market capitalisation from Beck et al. 10 database updated for the current data.e computed principal component ranges from -2.41 (in Slovenia) to 2.45 (in Sweden) and the higher its value, the more important the capital market in the economy.
We follow Beck and Levine11 and use Finance-Aggregate (FINDEV), which equals the rst principal component of the two underlying measures of nancial development.e rst underlying measure is a measure of the overall activity of the nancial intermediaries and markets.It equals the log of the product of private credit (the value of credits extended by nancial intermediaries, both bank and nonbank intermediaries, to the private sector divided by GDP) and value traded (the value of total shares traded on the stock market exchange divided by GDP).e second underlying measure of nancial development is a measure of the overall size of the nancial sector and equals the log of the sum of private credit and market capitalisation.We aggregate data over the period of 1996-2010.We take all the values, i.e. value traded, bank credit and market capitalisation from Beck et al. 12 database updated for the current data.e values of the rst principal component range between -2.55 (in Romania) and 2.03 (in the United Kingdom), with higher values suggesting a more developed nancial sector.
In our study we focus on the largest banks operating in each of the EU countries.We de ne the largest banks as 30% of banks with the largest assets in a given country. 13 empirical models that addressed the question of whether a bank capital induced credit crunch was hindering the recovery were developed in the early-and mid-1990 s in the US.We follow contemporary adoptions of those models,14 and our basic model reads as follows:  Annual change in unemployment rate is our measure of demand for loans.e unemployment rate is included because it not only re ects the business cycle but also longer term and structural imbalances in economies.We hypothesise that microprudential behaviour of banks is re ected by a positive correlation with unemployment.One can also expect banks operating in countries with lower unemployment to meet a higher credit demand as the income may be considered to be more stable. 15lements Country j j=11 27 and 11 are a set of country and time dummy variables, ϑ are unobservable bank speci c e ects that are not constant over time but vary across banks.Finally, ε is a white-noise error term.
We predict a negative coe cient on Contraction, if loan supply declines during contractions for reasons other than capital and liquidity constraints. 16f external nancing is not frictionless, and banks are concerned that they might violate regulatory capital requirements, then the coe cient on CAP is expected to be positive, i.e. banks with higher capital ratio will extend more loans.
To test the impact of nancial sector structure and nancial sector development, in particular during contraction periods, we include in equation (1) the indices which measure nancial sector structure and development as well as interaction terms between each of those indices and ContractionxCAP.We run separate regression for each of these indices and interaction terms between them and ContractionxCAP.
To identify contractionary periods we refer to the dataset available in a study of Olszak et al., 17 which was prepared following the approach of Lenart and Pipień. 18ur econometric model involves explanatory variables that may not be exogenous.is means that variables are correlated with error terms, both current and lagged.
One may also observe heteroskedasticity e ects and autocorrelation within individuals.erefore, we apply an approach that involves instrumental variables.In order to limit the possible estimation bias, we consider the system of generalised method of moments (GMM). 19is method has a proven track record and seems to be the best approach to address three relevant econometric issues that are inherent to our analysis: (1) the presence of unobserved bank speci c e ects, which is eliminated by taking rst di erences of all variables; (2) the inclusion of lags of the dependent variable needed to capture the dynamic nature of loan growth, which brings about the autoregressive nature of the data regarding the behaviour of lending; and (3) the likely endogeneity of the explanatory variables, mentioned above.
We control for the potential endogeneity of CAP, LIQGAP, DEPBANKS, ∆CAP and QLP in the two step system GMM estimation procedure by the inclusion of up to four lags of explanatory variables as instruments.e UNEMPL, as well as country and time dummy variables are the only variables considered exogenous.As the consistency of the GMM estimator depends on the validity of the instruments, we consider two speci cation tests.e rst is the test verifying the hypothesis of absence of second-order serial correlation in the rst di erence residuals (AR(2)) and the absence of rst-order serial correlation in the di erentiated residuals (AR(1)).In particular, it is important that in the models applied there is no second-order serial correlation in error terms.e second test which we apply is the Hansen's J statistic for overidentifying restrictions, which tests the overall validity of instrument tests.When interpreting the p-values of Hansen's J statistics we follow Roodman's warning20 that the Hansen test should not be relied upon too faithfully, as it is prone to instrument proliferation.

Empirical results
Results using observations for the whole period of 1996-2011 are shown in Table 1.
To provide an overview of speci cation, we rst estimate equation (1) using a pooled, time series regression including all large banks in all EU countries for the whole period.e results give also some evidence in favour of capital crunch hypothesis for large banks, because the association between loan growth rate and capital ratio is positive and statistically signi cant.
e results reported in Table 1 show that loan growth is generally lower in countries with more stock market-oriented nancial sector, as the regression coe cient on FINSTR is negative -but not statistically signi cant.On the other hand, the impact of nancial development (FINDEV) on loan growth is stronger, around -1,1, but only marginally signi cant.
As for the role of nancial structure in the e ects of capital ratio on lending during contractions, we nd that the relationship between lending and interaction term between FINSTR and ContractioxCAP is positive and statistically signi cant.Similarly, the association between loan growth and ContractioxCAPxFINDEV is positive.is gives empirical support to the two hypotheses put forward in our paper.
erefore, we conclude that both measures of the characteristics of nancial sector seem to increase the procyclical e ects of the bank capital ratio on loan growth.
With respect to other variables, we nd that liquidity stemming from stable nancing (measured with LIQGAP) plays some role in the case of large banks.Better access to retail interbank nancing does not a ect lending capacity of our sample of banks.Increases in capital ratios, as expected, lead to decreased loan growth in all types of banks.Relatively poor quality of loans, as measured by loan loss provisions over average loans (QLP), tends to be associated with slower loan growth rates.Size also matters for the lending capacity of banks.On average, banks with larger assets extend more new loans, as the regression coe cient of size is positive and statistically signi cant.Such a result supports the view that big banks should be less prone to adjusting their credit portfolio in the event of external shocks (such as monetary policy changes or crises).We also nd that loan growth is higher when unemployment rate is higher.is supports the notion that in the case of large banks, supply factors are more important for loan growth than demand e ects.Source: Authors' calculations.
e model is given by equation (1).Results are obtained for banks included in the category of the biggest banks, i.e. banks which belong to the 30% banks with the largest assets.e symbols have the following meaning: ∆loan -annual loan growth rate; Contraction -dummy equal to one in contractions and 0 otherwise; CAP -capital ratio, i.e. equity capital to total assets; ContractionxCAP -interaction between contraction and capital ratio (CAP); ∆CAP -annual change in capital ratio; DEPBANKS -deposits from banks to total assets; LIQGAP -loans less total customer deposits less deposits from banks divided by loans; size -logarithm of total assets; QLP -loan loss provisions divided by average loans; ∆UNEMPL -change in annual unemployment rate.FINSTR is nancial sector structure index.FINDEV is the index measuring the nancial sector development.Coe cients for the country and time dummies are not reported.e models have been estimated using the GMM estimator with robust standard errors.e p-val denotes signi cance levels.T-statistics are given in brackets.Data range: 1996-2011.

Conclusions
In this paper we investigate the extent to which country speci c environment proxied with nancial sector structure and development a ect the relationship between loan growth rate and the capital ratio of largest EU banks during contractionary periods.
e results of our study show that the role of capital ratio for loan growth is stronger than previous literature has found for other countries, in particular for the U. S. In the full sample of large banks the role of capital ratio on loan growth in contractions is relatively weak.However, if we take into account the di erences in nancial sector structure and development between EU countries, we nd that the e ect of capital ratio on lending is positive and statistically signi cant.erefore, our results suggest that capital ratios are an important determinant of lending in the large EU banks in those countries which have a nancial sector with greater dominance of stock markets or with a better developed nancial sector.
Our results support the view that economies with both better developed nancial sector and more capital market-oriented nancial sector are more prone to procyclical impact of capital standards, as the impact of capital ratio on lending is strengthened in those countries which exhibit greater nancial sector development and greater reliance to stock market nancing.

Do Financial Sector Structure and Development Matter for the E ect of Bank Capital on Lending in Large EU Banks?
e paper aims at nding out what is the impact of bank capital ratios on loan supply in the EU and what factors explain the potential diversity of this impact.Applying the Blundell and Bond (1998) two step GMM estimator, we nd that in the full sample of large banks the role of capital ratio on loan growth in contractions is relatively weak.However, if we take into account the di erences in nancial sector structure and development between EU countries, we nd that the e ect of capital ratio on lending is positive and statistically signi cant.erefore, our results suggest that capital ratios are an important determinant of lending in the large EU banks in those countries where nancial sector is more dominated by stock markets of is better developed.us, our results provide support for the view that more nancially developed economies are prone to greater procyclical impact of capital ratios on lending of banks.
Keywords: loan supply, capital ratio, nancial sector structure, nancial sector development, procyclicality Est-ce que la structure du secteur nancier et son développement sontils importants pour l'e et de capital de la banque sur les prêts dans les grandes banques de l'UE?

Figure
Figure 1.Indices of financial sector structure (FINSTR) and financial sector development (FINDEV)