Download bank capital and lending behavior: empirical evidence for italy and more Lecture notes Finance in PDF only on Docsity!
BANK CAPITAL AND LENDING BEHAVIOR:
EMPIRICAL EVIDENCE FOR ITALY
by Leonardo Gambacorta *^ and Paolo Emilio Mistrulli *
This version: February 14, 2003 Abstract This paper investigates the existence of cross-sectional differences in the response of lending to monetary policy and GDP shocks due to a different degree of bank capitalization. The effects on lending of shocks to bank capital, that are caused by a specific (higher than 8 per cent) solvency ratio for highly risky banks, are also analyzed. The paper adds to the existing literature in three ways. First, it considers a measure of capitalization (the excess capital) that is better able to control for the riskiness of bank’s portfolio than the well-known capital-to-asset ratio. Second, it disentangles the effects of the “bank lending channel” from those of the “bank capital channel” in the case of a monetary shock; it also provides an explanation for asymmetric effects of GDP shocks on lending based on the link between bank capital and risk-aversion. Third, it uses a unique dataset of quarterly data for Italian banks over the period 1992-2001; the full coverage of banks and the long sample period helps to overcome some distributional bias detected for other public available dataset. The results indicate that well-capitalized banks can better shield their lending from monetary policy shocks as they have, consistently with the “bank lending channel” hypothesis, an easier access to non-deposit fund raising. A “bank capital channel” is also detected, with higher effects for cooperative banks that suffer a higher maturity mismatching. Capitalization also influences the way banks react to GDP shocks. Again, the credit supply of well-capitalized banks is less pro-cyclical. The introduction of a specific solvency ratio for highly risky banks determines an overall reduction in lending. JEL classification: E44, E51, E Keywords: Basle standards, monetary transmission mechanisms, bank lending, bank capital.
Index
- Introduction.......................................................................................................................... 2
- Bank capital and the business cycle..................................................................................... 5
- Some stylized facts on bank capital................................................................................... 10
- The econometric model and the data ................................................................................. 14
- The results.......................................................................................................................... 18 5.1 Robustness checks ...................................................................................................... 21
- Conclusions........................................................................................................................ 22 Appendix 1 – A simple theoretical model.............................................................................. 25 Appendix 2 - Description of the database .............................................................................. 33 References .............................................................................................................................. 35
- (^) Banca d’Italia, Research Department.
1. Introduction^1
The role of bank capital in the monetary transmission mechanism has been largely neglected by economic theory. The traditional interpretation of the “bank lending channel” focuses on the effects of reserve requirements on demand deposits, while no attention is paid to bank’s equity; bank capital is traditionally interpreted as an “irrelevant” balance sheet item (Friedman, 1991; Van den Heuvel, 2003). Moreover, in contrast with the wide literature that analyzes the link between risk aversion and wealth, there is scarce evidence on the relationship between a bank’s risk attitude and her level of capitalization. This lack of attention contrasts with the importance given, both at an empirical and theoretical level, to the macroeconomic consequences of the Basle Capital Accord that designed risk-based capital requirements for banks.^2
The main aim of this paper is to study how bank capital may influence the response of lending to monetary policy and GDP shocks. There are two ways in which bank capital may affect the impact of monetary shocks: through the traditional “bank lending channel” and through a more “direct” mechanism defined “bank capital channel”. Both channels rest on the failure of the Modigliani-Miller theorem of the financial structure irrelevance but, as we will discuss, for different reasons.
Bank’s capitalization influences the “bank lending channel” due to imperfections in the market for debt. In particular, bank capital influences the capacity to raise uninsured form of debt and therefore bank’s ability to contain the effect on lending of a deposit drop. The mechanism is the following. After a monetary tightening, reservable deposits drop and banks raise non-reservable debt in order to protect their loan portfolios. As these non-reservable funding are typically uninsured (i.e. bonds or CDs), banks encounter an adverse selection problem (Stein, 1998); low capitalized banks, perceived more risky by the market, have
(^1) The authors are grateful to Giorgio Gobbi, Simonetta Iannotti, Francesco Lippi, Silvia Magri, Alberto Franco Pozzolo, Skander Van den Heuvel and an anonymous referee for useful comments. The model owes a lot to discussions with Michael Ehrmann, Jorge Martinez-Pagès, Patrick Sevestre and Andreas Worms. The usual disclaimer applies. The opinions expressed in this paper are those of the authors only and in no way involve the responsibility of the Bank of Italy. Email gambacorta.leonardo@insedia.interbusiness.it; mistrulli.paoloemilio@insedia.interbusiness.it. (^2) See Basle Commitee on Banking Supervision (1999) for a reference on the subject.
find that lending of undercapitalized banks suffers more from a monetary tightening, but their results are not significant at conventional values for the main European countries.
This paper presents three novelties with respect to the existing literature. The first one is the definition of capitalization; we define banks’ capitalization as the amount of capital that banks hold in excess of the minimum required to meet prudential regulation standards. This definition allows us to overcome some problems of the capital-to-asset ratio generally used in the existing literature. Since minimum capital requirements take into account the quality of banks’ balance sheet activities, the excess capital represents a cushion that controls for the level of banks’ risk and indicates a lower probability of a bank to go into default. Moreover, excess capital is a direct measure of banks capacity to expand credit because it takes into consideration prudential regulation constraints. The second novelty lies in the tentative to analyze the effects of capitalization on banks response to various economic shocks. In the case of monetary shocks we separate the effects of the “bank lending channel” from those of the “bank capital channel”. We provide a tentative explanation of the effect of GDP shocks on lending based on the link between bank capital and risk-aversion. Exogenous capital shocks that refer to specific solvency ratio that supervisors set for very risky banks are also analyzed. The third novelty is the use of a unique dataset of quarterly data for Italian banks over the period 1992-2001; the full coverage of banks and the long sample period should overcome some distributional bias detected for other public available dataset. To tackle problems in the use of dynamic panels, all the models have been estimated using the GMM estimator suggested by Arellano and Bond (1991).
The results indicate that well-capitalized banks can better shield their lending from monetary policy shocks as they have, consistently with the “bank lending channel” hypothesis, an easier access to non-deposit fund raising. In this respect, banks’ capitalization effect is larger for non-cooperative banks, which are more dependent on non-deposit forms of external funds. Capitalization also influences the way banks react to GDP shocks. Again, the credit supply of well-capitalized banks is less pro-cyclical. This result indicates that well- capitalized banks are more risk-averse and, as their borrowers are less risky, suffer less from economic downturns via loan losses. Moreover, well-capitalized banks can better absorb temporarily financial difficulties on the part of their borrowers and preserve long term lending relationships. Exogenous capital shocks, due to the introduction of a specific (higher
than 8 per cent) solvency ratio for highly risky banks, determine an overall reduction of 20 per cent in lending after two years. This result is consistent with the hypothesis that it costs less to adjust lending than capital.
The remainder of the paper is organized as follows. The next section reviews the literature and explains the main link between capital requirements and banks’ loan supply. Section 3 indicates some stylized facts concerning bank capital in Italy. In Section 4 we describe the econometric model and the data. Section 5 presents our empirical results and the robustness checks. The last section summarizes the main conclusions.
2. Bank capital and the business cycle
There are several theories that explain how bank capital could influence the propagation of economic shocks. All these theories suggest the existence of market imperfections that modify the standard results of the Modigliani and Miller theorem. Broadly speaking, if capital markets were perfect a bank would always be able to raise funds (debt or equity) in order to finance lending opportunities and her level of capital would have no role.
The aim of this Section is to discuss how bank capital may influence the reaction of bank lending to two kinds of economic disturbances: monetary policy and GDP shocks.
The first kind of shock occurs when a monetary tightening (easening) determines a reduction (increase) of reservable deposits and an increase (reduction) of market interest rates. In this case, there are two ways in which bank capital may influence the impact of monetary policy changes on lending: through the traditional “bank lending channel” and through a more “direct” mechanism defined as “bank capital channel”.
Both mechanism are based on adverse selection problems that affect banks fund- raising: the “bank lending channel” relies on imperfections in the market for bank debt (Bernanke and Blinder, 1988; Stein, 1998; Kishan and Opiela, 2000), while the “bank capital channel” concentrates on an imperfect market for banks’ equity (Thakor, 1996; Bolton and Freixas, 2001; Van den Heuvel, 2001a).
According to the “bank lending channel” thesis, a monetary tightening has effect on bank lending because the drop in reservable deposits cannot be completely offset by issuing
availability of the bank to expand credit because it directly controls for prudential regulation constraints.
The “bank capital channel” is based on three hypotheses. First, there is an imperfect market for bank equity: banks cannot easily issue new equity for the presence of agency costs and tax disadvantages (Myers and Majluf, 1984; Stein, 1998; Calomiris and Hubbard, 1995; Cornett and Tehranian, 1994). Second, banks are subject to interest rate risk because their assets have typically a higher maturity with respect to liabilities (maturity transformation). Third, regulatory capital requirements limit the supply of credit (Thakor, 1996; Bolton and Freixas, 2001; Van den Heuvel, 2001a).
The mechanism is the following. After an increase of market interest rates, a lower fraction of loans can be renegotiated with respect to deposits (loans are mainly long term, while deposits are typically short term): banks suffer therefore a cost due to the maturity mismatching that reduces profits and then capital. If equity is sufficiently low and it is too costly to issue new shares, banks reduce lending, otherwise they fail to meet regulatory capital requirements.
The “bank capital channel” can also be at work even if capital requirement is not currently binding. Van den Heuvel (2001) shows that low-capitalized banks may optimally forgo lending opportunities now in order to lower the risk of capital inadequacy in the future. This is interesting because in reality, as shown in Section 3, most banks are not constrained at any given time. It is also worth noting that, according to the “bank capital channel”, a negative effect of a monetary tightening on bank lending could be generated also if banks face a perfect market for non-reservable liabilities.
Bank capitalization may also influence the way lending supply reacts to output shocks. Bank capitalization, that is bank wealth, is linked to risk taking behavior and then to banks’ portfolio choices; this means that lending of banks with different degrees of capitalization (or risk aversion) may react differently to economic downturns. While a wide stream of literature on financial intermediation has analyzed the relation between bank capitalization
and risk taking behavior,^5 the nature of this link is still quite controversial. A first class of models (Kim and Santomero, 1988; Rochet, 1992; Hellman, Murdock and Stiglitz, 2000) argue that well-capitalized banks are less risk averse. In the presence of a solvency regulation, well-capitalized banks detain a higher level of capital just because their lending portfolio is riskier. In this case we should observe that well-capitalized banks react more to business cycle fluctuations because they have selected ex-ante a lending portfolio with higher return and risk.
In Kim and Santomero (1988), the introduction of a solvency regulation entails an inefficient asset allocation by banks. The total volume of their risky portfolio will decrease (as a direct effect of the solvency regulation), but its composition will be distorted in the direction of more risky assets (recomposition effect). In this model, the probability of failure increases after capital requirements are introduced because the direct effect is dominated by the recomposition of the risky portfolio. On the same line, Hellman, Murdock and Stiglitz (2000) argue that higher capital requirements are the cause of excessive risk-taking by banks. Since capital regulation increase banks’ cost of funding (equity is more costly than debt) and lower the value of the bank, the management of the bank reacts by increasing the level of credit portfolio risk.^6
The main implications of this class of models are three. First, well-capitalized banks are less risk averse because regulation creates an incentive in doing so. Second, risk-based capital standards would become efficient only if the weights that reflect the relative riskiness
(^5) The relation between wealth and attitude towards risk is central to many fields of economics. As far as credit markets are concerned, this relation has been largely employed in analyzing the role of collateral in mitigating asymmetric information problems between banks and borrowers (see Coco, 2000 for a recent survey on this subject). (^6) A different explanation is given by Besanko and Kanatas (1996). They depart from Kim and Santomero (1988) and Hellman, Murdock and Stiglitz (2000) by allowing for outside equity (owned by shareholders who are not in control of the firm) and by stressing the role of managerial incentive schemes in a moral hazard framework. Modeling at the same time asset-substitution (among assets with different risk profiles) and effort aversion moral hazard, they show that while a higher capital requirement reduces asset-substitution problems, it lowers the incentive to exert the optimal amount of effort. This result rests on what they call a “dilution effect”: if bank insiders are wealth-constrained or risk-averse, more stringent capital standards dilute insiders’ ownership share, and thus their marginal benefit of effort. The main conclusion of the model is that, if the effort aversion effect is larger than the asset-substitution effect, higher capital standards induce banks to take on average more risk. Gorton and Rosen (1996) argue that excessive risk taking among well-capitalized banks could also reflect exogenous conditions such as managerial incompetence or a lack of lending opportunities.
supervisors (Dewatripont and Tirole, 1994; Repullo, 2000; van den Heuvel, 2001a). Therefore, banks choose a certain excess capital at time t taking into account the possibility that in the future they could not be able to meet regulatory standards. The amount of capital banks hold in excess to capital requirement depends on their (global) risk aversion that is independent of the initial level of wealth.^9 Differences in (global) risk aversion among banks may emerge not only for heterogeneity in corporate governance but also, and more substantially, for institutional reasons. In Italy, as we will discuss in the following section, the institutional characteristics of credit cooperative banks (CCBs) are very different with respect to that of limited companies. If we allow for heterogeneity in (global) risk-aversion among banks the excess capital becomes a crucial measure to capture differences in the risk profile of banks’ portfolios. The simple capital-to-asset ratio is no longer informative because it does not capture the constraint due to regulation.
3. Some stylized facts on bank capital
The 1988 Basle Capital Accord and its subsequent amendments require capital to be above a threshold that is defined as a function of several types of risk. In other words, it is possible to distinguish between the default risk (credit risk) and the risk related to adverse fluctuations in asset market prices (market risk). 10 In Italy, the capital requirements for credit risks have been introduced in 1992, those for market risks in 1995.
As far as credit risk is concerned, capital must be at least equal to 8 per cent of the total amount of risk-weighted assets (solvency ratio).^11 A bank-specific solvency ratio (higher
(^9) A simple way to say that bank i is globally more averse than bank j is to assume that the objective function of bank i is a concave transformation of bank j objective function. (^10) In general the need of capital requirements arises to overcome moral hazard problems inducing banks to detain a “socially optimal” amount of capital. In the event of a crisis, the lower the leverage ratio is, the higher the probability that a bank will fail to pay back its debts. The moral hazard problem is amplified in the presence of a deposit insurance system. For a more detailed explanation of the rationale for capital requirements, see among others, Giammarino, Lewis and Sappington (1993), Dewatripont and Tirole (1994), Vlaar (2000) and Rime (2001). (^11) In Italy, regulation establishes a minimum capital requirement as a function of the amount of risk- weighted assets (and certain off-balance sheet activities). Assets are classified into five buckets with different risk weights. Risk weights are zero for cash and government bonds, 20 per cent for bank claims on other banks, 50 per cent for mortgage lending, 100 per cent for other loans on the private sector, 200 per cent for participating in highly risky non financial firms (firms that have recorded losses in the last two years). Till
than 8 per cent) can be disposed in case of a poor performance in terms of asset quality, liquidity and organization. On the contrary, the ratio decreases to 7 per cent for banks that belong to a banking group that meets an 8 per cent solvency ratio on a consolidated basis. Capital requirements on market risks are related to open trading positions in securities, foreign exchange and commodities.^12
Banks have to hold an amount of capital that must be at least equal to the sum of credit and market risk capital requirements.^13
One of the objectives of the 1988 Basle Accord was to increase banks’ capitalization (Basle Committee on Banking Supervision, 1999). We observe that banks’ capitalization increased during the period that preceded the implementation of the Basle Accord, Italian (Fig. 1) and it was slightly declining afterwards. It seems therefore that banks have constituted sufficient capital and reserves endowments before risk-based capital requirements were implemented. This seems to support the thesis that bank capital is sticky.
Large banks’ capitalization has been constantly lower than the average.^14 At the opposite, credit cooperative banks (CCBs), typically very small, are better capitalized than
September 1996 bad loans weight was also equal to 200 per cent. For any bank j its capital requirement is defined as:
kj ⋅ WAj = kj ⋅ ∑ i =^51 α iAij
where k j is the solvency ratio , WAj the total amount of risk-weighted assets, α i is the risk weight for asset
type i and Aij is the unweighted amount of the i -type asset bank j holds. (^12) Market risk capital requirements are computed on the basis of a quite complex algorithm. Regulation distinguishes between a “specific risk” and a “general risk”. The former refers to losses that can be determined by market price fluctuations, which are specifically related to the issuer economic condition. The latter is related to asset price fluctuations correlated to market developments ( systematic risk ). The capital requirement depends on issuer characteristics and on the asset maturity. Ceteris paribus , the capital requirement on market risks is lower for banks belonging to a group. 13 Prudential regulation allows banks to meet capital requirements by holding an amount of capital that is defined as the sum of the so-called Tier 1 and Tier 2 capital (regulatory capital). Tier 1 o r core capital includes stock issues, reserves and provisions for general banking risks ; Tier 2 or supplementary capital consists of general loan loss provisions, ibryd instruments and subordinated debt. Tier 1 capital is required to be equal at least to the 50 per cent of the total. Subordinated debts must not exceed 50 per cent of Tier 1 capital. Recently, banks have been allowed to issue subordinated debts specifically to face market risk requirements (the so- called Tier 3 capital). (^14) We have considered large banks those with total assets greater than 10 billions euro at September 2001. To control for mergers we have assumed that consolidation happened at the beginning of the period (see Appendix 2 for further details on merger treatment).
Figure 4 shows the maturity transformation performed by banks. As we have discussed in the previous section the existence of a maturity mismatching between assets and liabilities is a necessary condition for the “bank capital channel” to be at work. Since loans have always typically a longer maturity than bank fund-raising, the average maturity of total assets is higher than that of liabilities. In this case, as predicted by the “bank capital channel”, the bank suffers a cost when interest rates are raised and obtains a gain vice versa. The difference between the average maturity of assets and that of liabilities is higher for CCBs than for other banks. In fact, CCBs balance sheets are characterized by a higher percentage of long-term loans, while their bonds issues are more limited. For example, at the end of September 2002, the ratio between medium and long-term loans over total loans was 57 per cent for CCBs and 46 per cent for other banks. On the contrary, the ratio between bond and total fund raising was, respectively, 27 and 29 per cent. These differences were even higher at the beginning of our sample period. Therefore, the analysis of the maturity mismatching between assets and liabilities indicates room for the existence of a “bank capital channel” in Italy with a potential higher effect for CCBs.
There is no conclusive evidence about the effects of bank capital on lending behavior of Italian banks. In principle the financial structure of the Italian economy during the nineties makes more likely that a “bank lending channel” was at work (see Gambacorta, 2001). Most empirical papers based on VAR analysis confirm the existence of such a channel in Italy (Buttiglione and Ferri, 1994; Angeloni et al., 1995; Bagliano and Favero, 1995; Fanelli and Paruolo, 1999; Chiades and Gambacorta, 2003). However there is much less evidence on cross sectional differences in the effectiveness of the “bank lending channel” in Italy, due to capitalization (see de Bondt, 1999; Favero et al., 2001; King 2002; that analyze mainly the effect of banks dimension and liquidity; some evidence of the effect of capitalization on lending of Italian banks is detected by Altunbas, 2002). So far no evidence has been provided on the existence of the so-called “bank capital channel”.
Apart from the differences in specification, all these paper use the BankScope dataset that, as pointed out by Ehrmann et al. (2003), suffers of two weaknesses. First, the data are collected annually, which might be too infrequent to capture the adjustment of bank aggregates to monetary policy. Second, the sample of Italian banks available in BankScope is biased towards large banks. For example, in 1998 only 576 up to 921 Italian banks were
included in the BankScope dataset. Moreover the average size of a bank was 3.7 billion euro against 1.7 for the total population. To tackle these problems our analysis will be based on the Bank of Italy Supervisory Reports database, using quarterly data for the full population of Italian banks.
4. The econometric model and the data
The empirical specification, based on Kashyap and Stein (1995), is designed to test whether banks with a different degree of capitalization react differently to a monetary policy or a GDP shock. A simple theoretical framework that justifies the choice of the specification is reported in Appendix 1. 15
The empirical model is given by the following equation, which includes interaction terms that are the product of the excess capital with the monetary policy indicator and the real GDP; all bank specific characteristics (excess capital, cost due to maturity mismatching, etc.) refer to t -1 to avoid an endogeneity bias (see Kashyap and Stein, 1995; 2000; Ehrmann et al., 2003):
1 1 1
4 4 4 4 1 0 0 0 4 4 (^11 )
ln ln ln
it (^ )^ it it ln
it i (^) j j it j (^) j j t j (^) j j t j (^) j j t j
i t (^) j j t j (^) j j t j it it
L L MP y
X MP X MP X y
= −^ = −^ = −^ = − − (^) = − (^) = −
with i =1,…, N ( N = number of banks) and t =1, …, T ( t = quarters) and where:
L it = loans of bank i in quarter t
MPt = monetary policy indicator
y t = real GDP
π (^) t = inflation rate
(^15) The model presented in Appendix 1 is a slightly modified version of the analytical framework in Ehrmann et al. (2003). The main differences are two. First, it introduces bank capital regulation in a static way as in Kishan and Opiela (2000). Second, following the literature on bank capital and risk attitude (see Section
- we model loan losses as a function of bank capitalization.
1
4 j 1 j^ it t^ j
γ X − MP −
=
∑ ∆ and^1
4 1 j (^) it ln t j j
τ X − y −
=
∑ ∆ in equation (1)^ are zero for the average bank
( X it − 1 =0). Second, the coefficients β j and δ j are directly interpretable, respectively, as the
average monetary policy effect and the average GDP effect.
To test for the existence of a “bank capital channel” we have introduced a variable
( ρ i ∆ MP ) that represents the bank-specific interest rate cost due to maturity transformation.
In particular ρ i measures the loss (gain) per unit of asset the bank suffers (obtains) when the
monetary policy interest rate is raised (decreased) of one percentage point. We have computed this variable according to supervisory regulation relative to interest rate risk exposure that depends on the maturity mismatching among assets and liabilities.^18 In other
words, if bank’s assets have a higher maturity with respect to liabilities ρ i is positive and
indicates the cost for unit of asset a bank suffers if interest rate are raised by one per cent. To work out the real cost we have therefore multiplied this measure for the realized change in
interest rates. The term ρ i ∆ MP represents the real cost (gain) that a bank suffers (obtains) in
each quarter. As formalized in Appendix 1, this measure influences the level of loans. Since here the dependent variable is a growth rate we have included this measure in first differences.
The set of control variables Φ (^) it include a liquidity indicator, given by the sum of cash
and securities to total assets ratio, and a size indicator, given by the log of total assets. The liquidity indicator has been normalized with respect to the mean over the whole sample period, while the size indicator has been normalized with respect to the mean on each single period. This procedure removes trends in size (for more details see Gambacorta, 2001). As for the other bank specific characteristics also liquidity and size indicators refer to t -1 to avoid an endogeneity bias.
The fact that supervisors can set solvency ratios greater than 8 per cent for highly risky banks (see Section 3), allows us to test for the effects of exogenous capital shocks on bank lending. We analyze the impact of these supervisory actions on lending in the first two years,
(^18) See Appendix 2 for further details.
computing different dummy variables (one for each quarter following the solvency ratio raise) that equal 1 for banks whose solvency ratio is higher than 8 per cent. This allows us to capture bank lending adjustment process. A specific dummy variable controls for the effects of the introduction of market risk capital requirements in the first quarter of 1995.
The sample represents 82 per cent of total bank credit in Italy. Table 1 gives some basic information on what bank balance sheets look like. Credit Cooperative Banks (CCBs) are treated separately because they are significantly smaller, more liquid and better capitalized than other banks. This evidence is consistent with the view that smaller banks need bigger buffer stocks of securities because of their limited ability to raise external finance on the financial market. This interpretation is confirmed on the liability side, where the percentage of bonds is lower among CCBs. The high capitalization of CCBs is, at least in part, due to the Banking Law prescription that limits significantly the distribution of net profits.^19
Within each category, banks have been split, according to their capitalization.^20 Low- capitalized banks are, independently of their form (CCBs or other banks), larger, less liquid and they issue more bonds than well-capitalized banks. While these differences are small among CCBs, they are quite significant among non-CCBs. Among non-cooperative banks, low-capitalized banks are much larger than well-capitalized ones; a higher share is listed and belongs to a banking group. Moreover, they issue more subordinated debt to meet the capital requirement. This evidence is consistent with the view that, ceteris paribus , capitalization is lower for those banks that bear less adjustment costs from issuing new (regulatory) capital; large and listed banks can more easily raise funds on the capital market and they can also rely on a wider set of “quasi-equity” securities that can be issued to meet capital requirements (e.g. subordinated debts); at the same time, banks belonging to a group can
(^19) According to art. 37 of the 1993 Banking Law “Banche di credito cooperativo must allocate at least seventy per cent of net profits for the year to the legal reserve.” (^20) A “lowly capitalized” bank has a capital ratio equal to the average capital ratio below the 10 th (^) percentile, a “highly capitalized” bank, that of the banks above the 90 th^ percentile. Since the characteristics of each bank could change over time, percentiles have been worked out on mean values.
From the first row of the table it is possible to note that the effect of excess capital on lending is always significant and positive: well-capitalized banks are less constrained by capital requirements and have more opportunity to expand their loan portfolio. The effect is higher for CCBs than for other banks because they encounter higher capital adjustment costs: CCBs are more dependent on self-financing and cannot easily raise new regulatory capital.
The response of bank lending to a monetary policy shock has the expected negative sign. These estimates roughly imply that a 1 per cent increase in the monetary policy indicator leads to a decline in lending of around 1.2 per cent for the average bank. The effect is higher for CCBs (-1.8 per cent) than for other banks (-0.2 per cent), that have more access to markets for non-reservable liabilities. Testing the null hypothesis that monetary policy effects are equal among banks with a different degree of capitalization is identical to testing the significance of the long run coefficient of the interaction between excess capital and the monetary policy indicator (see “Excess capital*MP” in table 2). As predicted by the “bank lending channel” hypothesis the effects of a monetary tightening are lower for banks with a higher capitalization, which have easier access to non-deposit financing. Bank capitalization interaction with monetary policy is very high (in absolute value) for non-CCBs, which are more dependent on non-deposit forms of external funds. It is worth noting that well- capitalized non-CCBs are completely insulated from the effect of a monetary tightening (the effect is statistically not different from zero).
The effects of the so-called “bank capital channel” are reported on the eighth row of Table 2. The coefficients have the expected negative sign for all banks groups. These estimates roughly imply that an increase (decrease) of one basis point of the ratio between the maturity transformation cost and total assets determines a reduction (increase) of 1 per cent in the growth rate of lending. The reduction (increase) is bigger for CCBs that, as seen in Section 3, have typically a higher maturity mismatching between assets and liabilities. In fact, CCBs balance sheets are characterized by a higher percentage of long-term loans, while their bonds issues are more limited. Another possible explanation for the higher effect of the “bank capital channel” for CCBs could be their lower use of derivatives for shielding the maturity transformation gap. With these characteristic CCBs suffer a higher cost when interest rates are raised and obtains a higher gain vice versa. To sum up the results indicate
the existence of a “bank capital channel” that amplifies the effects of monetary policy changes on bank lending and asymmetric effects of such a channel among banks groups.
The models show a positive correlation between credit and output. A 1 per cent increase in GDP (which produces a loan demand shift) determines a loan increase of around 0.7 per cent. The effect is lower for CCBs than for other banks. This has two main explanations. First, for CCBs local economic conditions are more important than national ones; second, they have closer customer relationships because they shall grant credit primarily to their members (see “The 1993 Banking Law”, Art. 35).
The interaction term between GDP and excess capital is negative. This means that credit supply of well-capitalized banks is less dependent on the business cycle. This result is consistent with Kwan and Eisenbeis (1997) where capital is found to have a significantly negative effect on credit risk. On theoretical ground our findings are consistent with Flannery (1989) and Gennotte and Pyle (1991) that argue that highly capitalized banks are more-risk averse and select ex-ante borrowers with a lower probability to go into default. Their risk-attitude therefore limits credit supply adjustments in bad states of nature, preserving credit relationships. The latter explanation needs to be discussed with respect to the institutional categories of Italian banks. From the sample split it emerges indeed that the
coefficient of ∆ ( ρ i ∆ MP ) t − 1 is highly significant only for CCBs, while there are no
significant asymmetric effects for the other banks. This is consistent with the stylized fact discussed in Section 3 that CCBs are more risk-averse than other banks. They detain high levels of excess capital and are more able to insulate the effect of an economic downturn. As in Vander Vennet and Van Landshoot (2002) capital provides banks with a structural protection against credit risk changes. Looking at Table 2 well-capitalized CCBs are able to completely insulate the effect of GDP on their lending. On the other hand, non-CCBs seem to be risk-neutral: the effect of a 1 per cent increase in GDP on lending does not differ too much between well-capitalized (1.3 per cent) and poorly capitalized banks (1.5).
As explained in Section 4, the effects of exogenous capital shocks on bank lending are captured by dummy variables related to the introduction of a specific (higher than 8 per cent) solvency ratio. In this case there are not many differences among the three samples. The introduction of specific solvency ratio determines an overall reduction of around 20 per cent
