Intra-competitiveness and inter-competitiveness among mutual banks: the case of Trento

Transcription

1 Intra-competitiveness and inter-competitiveness among mutual banks: the case of Trento Gian Paolo Barbetta, Luca Colombo, Stefano Colombo, Michele Grillo Università Cattolica del Sacro Cuore, Milano DRAFT April 2011 Gian Paolo Barbetta Professor of Economic policy, Università Cattolica di Milano, largo Gemelli, Milano. Gianpaolo.barbetta@unicatt.it, Luca Colombo Professor of Economics, Università Cattolica di Milano, largo Gemelli, Milano. Lucava.colombo@unicatt.it, Stefano Colombo Researcher of Economics, Università Cattolica di Milano, largo Gemelli, Milano. Stefano.combo@unicatt.it, Michele Grillo Professor of Political economy, Università Cattolica di Milano, largo Gemelli, Milano. Michele.grillo@unicatt.it - Corresponding author. 1

2 Intra-competitiveness and inter-competitiveness among mutual banks: the case of Trento Gian Paolo Barbetta, Luca Colombo, Stefano Colombo, Michele Grillo Università Cattolica del Sacro Cuore, Milano Introduction Mutual banks are often relevant players in the banking industry. This is the case, for example, in the Italian province of Trento, where mutual banks represent more than a half of the whole banking industry in the same province, both in terms of and deposits. In the last twenty years, following the liberalization of the banking industry, Italian banks have undergone a reorganization process that has greatly impacted both on the workings of individual firms and on the overall structure of the banking industry. In contrast with the previous period, when both internal and external growth of individual banks were hindered by a number of administrative constraints, liberalization allowed banks to expand in a number of directions, thus letting them benefit from scale and scope economies. With competition in the banking industry becoming more intense, the traditional distinction of different types of banks waned in many instances, inducing a substantial change of the industry structure that resulted in a reduction in the total number of banking firms and in an increase in the number of branches. Local banking markets have been particularly affected by the liberalization. Today, many credit institutions traditionally serving local markets such as popular banks and the former saving banks belong to credit conglomerates operating at the national level. This has been seen as a source of concern, as the benefits of larger scale and scope economies have likely been obtained at the cost of weaker links with the territory and of reduced capability to serve local financial needs. However, the picture is somewhat different for a third type of local banks, namely mutual cooperative banks (the so-called Banche di Credito Cooperativo, or BCCs). On the one hand, these banks too have been affected by the aggregation process and the connected increase in the number of branches. On the other hand, aggregations have taken place only within the cooperative credit system, without determining relevant changes in bank size and ownership structure. This has allowed mutual cooperative banks to maintain their local roots basically unchanged, in such a way that today mutual banks still remain largely focused on the collection of local savings and on the provision of financial services for locally established small- and sometimes even medium-sized firms. Nonetheless, especially at the beginning of the last decade, the specificity of mutual cooperative banks was seen as an additional source of concern. In what appeared to be an irreversibly changing market environment, the persistence of the traditional model of local banking was seen as a weakness. It was feared that, being unable to fully exploit the benefits of scale and scope economies with respect to their competitors, local mutual banks would eventually be put at serious cost disadvantage. It was also feared that local mutual banks would find growing difficulties in the provision of those innovative and dynamic financial services that have been covering the most competitive arena in modern financial markets. Even in the typical credit activity, it was feared that mutual banks would experience a worsening in the quality of, as a combined result of the disappearing of other types of local banks and of 2

3 the difficulties of diversifying their credit risk, due to the local concentration of borrowers. The overturning, in those same years, of the ratio of bad to total for mutual banks with respect to the entire banking system (traditionally more favourable to the former) was taken as a serious negative signal. Along the decade, however, a number of facts have forced observers to attenuate those assessments. The provision of credit to local borrowers proved to be a distinctive feature of mutual cooperative banks for which the rest of the banking system offered poor substitutability. As Guiso, Sapienza and Zingales (2004) argue, distance is an important barrier to lending and a main cause of segmentation of local credit markets. The spatial proximity of mutual banks with their borrowers proves to be the source of important comparative advantages in screening, monitoring and enforcement (as the local bank is better informed when compared to nonmutual banks - on the quality of local borrowers, can better monitor the use of loaned funds and is better equipped in recovering them). Moreover, the mutual cooperative bank, operating in the same community to which its customers also belong and establishing with them long-term relations, is better equipped to contrast the moral hazard that typically arises in credit markets. The control of moral hazard is still tighter when borrowers are members of the mutual bank, as team incentives apply in such cases (since the losses for unrecovered are borne by all members and the threat of social sanctions, as well as the pressure of voice and loyalty, are stronger). At the end of the decade, the prospects of cooperative credit involve a typical trade-off: on the one hand, mutual banks appear to be a largely not substitutable provider of to local borrowers; on the other hand, mutual banks are subjected to specific disadvantages, mainly attributable to small size, specialization and the higher correlation of the default risks of debt owners. This paper analyses the above trade-off from a particular, albeit relevant, perspective. The basic idea is, that the solution of the trade-off may be significantly affected by the competitive conditions in the markets in which mutual banks operate. To start with, consider that, historically, mutual banks were created as cooperative organizations among self-producers. Marginal firms in local markets were compelled to establish mutual credit organizations in order to escape the specific credit rationing that might cause them disadvantages as a consequence of typical market failures in normal credit markets. Therefore, at the beginning, mutual banks were mostly natural monopolist in their own territory. It is true that, as time passed, mutual banks also started operating in larger territories, where they competed with other types of banks. Under those circumstances, however, mutual banks kept being tied to their original raison d être, as they adapted to a sort of natural market segmentation: even in larger markets, in fact, mutual credit cooperatives mainly generated to marginal borrowers that would find excessively costly to obtain, or would be totally denied, credit from other types of banks operating in the same markets. Moreover, market segmentation and the specificity of mutual banks led, as a natural consequence, to the circumstance that in normal conditions only one mutual bank was operating in a given territory. As a matter of fact, before liberalization, this result was also enforced by the current law. However, after 1993, under the new Banking Law (Testo Unico Bancario), the territorial overlapping of mutual banks was no longer hampered. On the contrary, in recent years, both the Bank of Italy and the Italian Competition Authority have repeatedly argued that competition among mutual banks ( intra-competitiveness ) is to be seen as a valuable component of the competitive process in banking markets, along with competition between the mutual banks and other (non-mutual) types of banks ( inter-competitiveness ). In contrast with the views held by the Bank of Italy and the Italian Competition Authority, we ask whether the territorial overlapping of mutual banks might be seen 3

4 as a matter of serious concern from the point of view of the economic theory of cooperative credit. Under general conditions, mutual cooperation is intrinsically exposed to the risk of free-riding. As is well known, to avoid free-riding, the logic of mutual cooperation demands that long-term relationships can be established among the participating agents. In other words, a tension can arise between market cooperation and mutual cooperation. Whereas the market is the place for impersonal exchanges, mutual cooperation involves that personal exchanges, relying upon inter-temporal relationships, can be established. The point is that competition between cooperative organizations could jeopardize cooperation within cooperative organizations. This may happen whenever the possibility of joining an alternative team opens an exit option for an agent in a team, inducing him to deviate from cooperating with its current partners in order to exploit (higher) short-term benefits. Those considerations obviously hold also for cooperation among self-producers in the credit market. It is therefore important for credit co-operators to be all connected in a unique and common network of long term credit relationships, as a necessary condition for the sharing of losses and gains to act as an appropriate incentive for peer monitoring and the control of moral hazard. To provide an answer to our research question, in this paper we analyse the working of mutual banks and their performance in the province of Trento, taking explicitly into account the different competitive conditions in the local markets in which they have operated in the decade In particular, we focus on the effects of intra-competitiveness (i.e., competition among mutual banks) and intercompetitiveness (i.e., competition between the mutual banks and other non-mutual banks) of mutual banks, and we analyse how those conditions have affected the mutual banks strategic choices and economic performances. To this purpose, we have divided the set of mutual banks operating in the province of Trento into three subgroups. The first group is composed of those mutual banks that act as monopolists in their own territory. The second group includes those mutual banks that compete both with other mutual banks and with other (namely, non-mutual) banks. Finally, the third group consists of those mutual banks that, in their territory, compete only with nonmutual banks. We find that, in the decade , the three groups have shown significant differences both in the strategies implemented, and in their economic performance. We take these results arising from both a statistical descriptive analysis and an econometric exercise as suggesting that, for mutual banks, intra-competitiveness and inter-competitiveness matter. Specifically, we find that, at least for medium and large sized mutual banks, the loan to deposit ratio tends to be significantly higher for mutual banks belonging to the second group (i.e., for mutual banks that compete only with non-mutual banks, but do not compete with other mutual banks). Moreover mutual banks belonging to the second group also show a higher loan to total asset ratio. We interpret these results as suggesting that both the ability of mutual banks to transform local savings into local (that we label as local effectiveness ), and the ability of mutual banks to manage the credit risk (that we label as mission efficiency ) is better achieved when the degree of inter-competition is high, but the degree of intra-competition is low. A further, consistent, result shows that the ratio of bad to total is significantly lower for mutual banks belonging to the second group. 1. The mutual bank industry in the province of Trento The number of mutual banks in the province of Trento decreased during the decade , from 65 in 2000 to 46 in This is the result of mergers and 4

5 acquisitions within the mutual bank sector. During the same decade, the weight of the mutual banks over the entire banking system of the province increased substantially, particularly when considering the share of, which grew from 50% in 2000 to 66% in As already stated in the Introduction, in order to analyse the impact of different competitive environments on the economic performance of mutual banks, it is useful to classify the 46 banks operating in 2009 into three groups: 1) The first group is composed of the mutual banks that in 2009 were monopolists in their own territory; this group includes nine small banks (with about 6.47% of the total assets of the mutual banks sector) operating in small municipalities. 2) The second group is composed of the mutual banks that in 2009 competed both with other mutual banks and with non-mutual banks; this group includes eighteen mutual banks representing 57,58% of the total assets of the whole system of mutual banks in the province. Most banks belonging to this group are large in size. 3) The third group includes the mutual banks that in 2009 competed with nonmutual banks only. It embraces nineteen banks representing 36.25% of the total assets of the whole system. Most banks belonging to this third group are intermediate in size with respect to the banks belonging to the first and the second group. Table 1: key indicators for the three groups of mutual banks (BCCs) all MBs first group second group third group 2,000 2,009 2,000 2,009 2,000 2,009 2,000 2,009 Total assets 6,904,693,216 15,049,943, ,160, ,970,296 3,802,996,298 8,665,151,027 2,654,536,909 5,455,821,863 growth rate ( ) ratio of the group over all BCCs Loans to customers 4,313,408,644 11,727,802, ,787, ,377,855 2,433,007,815 6,786,968,578 1,644,613,603 4,260,456,038 growth rate ( ) ratio of the group over all BCCs Deposits and bank bonds 5,426,332,123 12,697,213, ,438, ,375,456 2,983,695,037 7,372,170,963 2,094,199,026 4,548,667,364 growth rate ( ) ratio of the group over all BCCs of which: bank bonds 1,553,265,221 5,979,454, ,712, ,450, ,197,498 3,331,403, ,354,964 2,320,601,404 growth rate ( ) ratio of the group over all BCCs Bank bonds/ deposits and bank bonds Off-balance sheet assets under management 4,337,941,506 3,349,313, ,352, ,580,376 2,578,192,607 2,010,176,244 1,432,396,308 1,062,556,868 growth rate ( ) ratio of the group over all BCCs Table 1 summarizes some key indicators and their growth rates over the decade. First of all, when considering total assets of the whole mutual banks system, the share of groups 1 and 3 decreased, while that of group 2 grew. Second, the share of made by banks belonging to groups 1 and 2 increased, while that of group 3 shrank. Third, the share of deposits of group 2 increased, but that of groups 1 and 3 declined. Fourth, the (off-balance sheet) assets under management diminished for all groups, and in particular for group Competition and performance of the mutual banks in the province of Trento As argued in the Introduction, the specific mission of a mutual bank is to provide to local borrowers, which mainly consist of small and medium-size 5

6 enterprises. Indeed, it is widely recognized that mutual banks have a relative advantage in screening, monitoring and enforcement over large banks operating at the national scale. Thus, we interpret as the key variable to measure the compliance of mutual banks to their mission. More precisely, we adopt two different criteria: 1) Local effectiveness, measured by the ratio of to deposits (in which, with a slight language abuse, we include also own bank bonds): the higher is this ratio, the greater is the ability of mutual banks to transform local savings into to local borrowers. 2) Mission efficiency, measured by the ratio of to total assets. It is widely recognized that the portfolios of mutual banks are - on average - riskier than those of other banks, due to a larger risk correlation (as mutual bank customers are geographically concentrated). Therefore, assuming that mutual banks manage risks efficiently, a larger -to-assets ratio would indicate a higher efficiency in managing credit risk Local effectiveness We observe a clear ranking of the three groups as far as the -to-deposits ratio is concerned. In 2009 the banks of group 3 show an average -to-deposits ratio of 93.7%, followed by the banks of group 2 (92.1%) and finally by those of group 1 (87.6%). Moreover, we observe that during the decade the ratio increased at a faster rate for the banks belonging respectively to groups 1 and 3 (from 67.7% to 87.6%, and from 78.5% to 93.7%) than for banks of group 2 (from 81.5% to 92.1%). The descriptive evidence on the local effectiveness of mutual banks is summarized in Table 2. However, we also observe significant differences within the banks belonging to group 2. In particular, the smallest banks of this group follow a pattern that closely resembles that of the banks of group 1. Moreover, the banks of the second group operating in the territory characterized by the highest degree of competition 1 among mutual banks showed a faster increase of the -to-deposits ratio than the average bank of group 2. Summarizing, the descriptive evidence suggests that: - Banks of group 1 are characterized by a lower degree of local effectiveness than those belonging to groups 2 and 3. However, as the smallest banks belonging to group 2 do not behave differently from those of group 1, one need to check carefully whether different group performances are the result of size differences rather than of different competitive environments. - Banks of group 2 are characterized by a lower degree of local effectiveness than those belonging to group Mission efficiency Recall that we measure mission efficiency through the ratio of to total assets: the higher is this ratio, the more efficient is credit risk management. Table 3 summarizes the descriptive evidence on mutual banks mission efficiency. Looking at the whole of mutual banks, we note that the - to-total- assets ratio increased from 63.4% to 77.9% during the decade In this context, the banks belonging to group 1 show the lowest -to-total-assets ratio, which is easily explained recalling that these mutual banks typically operate in small municipalities characterized by more correlated risks. As for the second and third group of mutual banks, although they show very similar ratios (78.3% and 78.1%, respectively), their dynamics has been different over the decade. In fact, for group 2, the -to-totalassets ratio (65.7% in 2000) increased at a slower pace than for group 3 (62% in 2000). 1 This sub-group is composed of six banks operating in the area of Garda Lake and Idro Lake. 6

7 Table 2: local effectiveness All BCCs Loans / Deposits and bank bonds Loans 4,313,408,644 4,781,253,575 5,664,075,808 6,604,150,139 7,492,703,815 8,269,782,110 9,348,816,517 10,483,007,351 11,326,048,551 11,727,802,471 Deposits and bank bonds 5,426,332,123 6,272,538,024 7,047,869,440 7,892,971,233 8,642,682,652 9,455,501,015 10,010,555,011 10,790,383,265 12,042,774,797 12,697,213,782 Group 1 Loans / Deposits and bank bonds Loans 235,787, ,345, ,920, ,842, ,486, ,350, ,182, ,368, ,525, ,377,855 Deposits and bank bonds 348,438, ,663, ,350, ,705, ,690, ,833, ,122, ,086, ,101, ,375,456 Group 2 Loans / Deposits and bank bonds Loans 2,433,007,815 2,733,376,079 3,277,847,449 3,874,313,199 4,365,252,118 4,805,385,023 5,440,609,027 6,132,864,146 6,597,752,341 6,786,968,578 Deposits and bank bonds 2,983,695,037 3,487,743,143 3,985,013,400 4,457,881,846 4,910,226,897 5,391,765,018 5,714,995,963 6,208,735,021 6,991,210,638 7,372,170,963 of which: * second-group small banks Loans / Deposits and bank bonds Loans 160,863, ,398, ,725, ,338, ,622, ,066, ,708, ,254, ,743, ,731,027 Deposits and bank bonds 236,838, ,944, ,730, ,187, ,536, ,247, ,090, ,015, ,168, ,414,087 * second-group highly competitive banks Loans / Deposits and bank bonds Loans 669,830, ,837, ,306,490 1,082,128,212 1,187,803,515 1,303,102,564 1,462,771,887 1,663,757,765 1,831,124,414 1,898,771,627 Deposits and bank bonds 861,977,429 1,015,282,622 1,149,277,911 1,245,782,620 1,369,135,187 1,475,052,799 1,532,463,789 1,701,632,869 1,908,391,918 2,031,547,259 Group 3 Loans / Deposits and bank bonds Loans 1,644,613,603 1,789,532,485 2,071,307,832 2,360,993,968 2,696,965,391 2,973,046,868 3,354,025,086 3,744,775,000 4,075,770,560 4,260,456,038 Deposits and bank bonds 2,094,199,026 2,380,131,411 2,594,505,818 2,910,383,429 3,159,765,584 3,437,902,024 3,625,436,407 3,884,561,423 4,300,462,629 4,548,667,364 Table 3: mission efficiency All BCCs Loans / Deposits and bank bonds Loans 4,313,408,644 4,781,253,575 5,664,075,808 6,604,150,139 7,492,703,815 8,269,782,110 9,348,816,517 10,483,007,351 11,326,048,551 11,727,802,471 Deposits and bank bonds 6,807,211,673 7,715,191,845 8,561,348,263 9,444,286,738 10,269,365,925 11,094,655,600 12,001,242,364 13,134,257,469 14,390,624,068 15,049,943,185 Group 1 Loans / Deposits and bank bonds Loans 235,787, ,345, ,920, ,842, ,486, ,350, ,182, ,368, ,525, ,377,855 Deposits and bank bonds 447,160, ,217, ,605, ,733, ,075, ,261, ,778, ,920, ,582, ,970,296 Group 2 Loans / Deposits and bank bonds Loans 2,433,007,815 2,733,376,079 3,277,847,449 3,874,313,199 4,365,252,118 4,805,385,023 5,440,609,027 6,132,864,146 6,597,752,341 6,786,968,578 Deposits and bank bonds 3,705,514,755 4,258,073,366 4,785,031,255 5,271,667,038 5,754,572,997 6,247,545,610 6,855,323,097 7,513,788,355 8,288,194,946 8,665,151,027 of which: * second-group small banks Loans / Deposits and bank bonds Loans 160,863, ,398, ,725, ,338, ,622, ,066, ,708, ,254, ,743, ,731,027 Deposits and bank bonds 290,685, ,523, ,199, ,009, ,237, ,710, ,738, ,207, ,950, ,628,583 * second-group highly competitive banks Loans / Deposits and bank bonds Loans 669,830, ,837, ,306,490 1,082,128,212 1,187,803,515 1,303,102,564 1,462,771,887 1,663,757,765 1,831,124,414 1,898,771,627 Deposits and bank bonds 1,050,169,784 1,210,139,553 1,362,975,991 1,454,689,881 1,581,171,383 1,686,096,526 1,846,164,424 2,053,189,769 2,264,435,328 2,393,628,614 Group 3 Loans / Deposits and bank bonds Loans 1,644,613,603 1,789,532,485 2,071,307,832 2,360,993,968 2,696,965,391 2,973,046,868 3,354,025,086 3,744,775,000 4,075,770,560 4,260,456,038 Deposits and bank bonds 2,654,536,909 2,951,901,368 3,206,711,353 3,537,885,967 3,830,717,127 4,105,848,100 4,353,140,597 4,772,548,789 5,201,846,262 5,455,821,863 The analysis of mission efficiency can be complemented by the analysis of the ratio of bad to total, that we take as a proxy for the credit risk faced by a bank. Table 4 summarizes the descriptive evidence. 7

8 Table 4: bad all BCC bad / total bad 70,687,922 74,567,668 76,703,866 81,454,198 87,491,353 92,720,958 92,500, ,962, ,732, ,658,462 4,313,408,644 4,781,253,575 5,664,075,808 6,604,150,139 7,492,703,815 8,269,782,110 9,348,816,517 10,483,007,351 11,326,048,551 11,727,802,471 all banks (BCCs+Non- N.A. N.A BCCs) Group 1 bad / total bad 3,205,206 4,027,547 4,992,982 5,074,032 6,549,308 6,844,565 6,294,640 9,013,767 9,651,648 16,414, ,787, ,345, ,920, ,842, ,486, ,350, ,182, ,368, ,525, ,377,855 Group 2 bad / total bad 45,114,064 47,387,585 42,201,020 44,856,753 51,746,004 53,427,058 52,415,664 61,712,514 91,731, ,183,052 2,433,007,815 2,733,376,079 3,277,847,449 3,874,313,199 4,365,252,118 4,805,385,023 5,440,609,027 6,132,864,146 6,597,752,341 6,786,968,578 of which: * second-group small banks bad / total bad 2,686,746 2,809,989 3,586,979 3,692,036 3,684,727 3,660,477 2,906,645 3,175,952 7,039,045 11,008, ,863, ,398, ,725, ,338, ,622, ,066, ,708, ,254, ,743, ,731,027 * second-group highly competitive banks bad / total bad 18,240,400 15,905,805 13,590,889 14,986,023 19,920,827 16,813,600 16,489,898 21,330,969 33,464,367 45,380, ,830, ,837, ,306,490 1,082,128,212 1,187,803,515 1,303,102,564 1,462,771,887 1,663,757,765 1,831,124,414 1,898,771,627 Group 3 bad / total bad 21,078,657 21,884,397 27,585,282 29,552,741 26,669,528 29,281,086 31,327,325 39,972,838 46,168,524 69,404,884 1,644,613,603 1,789,532,485 2,071,307,832 2,360,993,968 2,696,965,391 2,973,046,868 3,354,025,086 3,744,775,000 4,075,770,560 4,260,456,038 We observe that this ratio is lower for mutual banks than for the average of all banks, with the difference being particularly large in the years of the financial crisis ( ). All mutual banks, regardless of their competitive environment, have seen a decrease in the ratio until 2006, and an increase thereafter. Within groups, the only systematic difference concerns the lower value of the ratio of bad to total for group 3 (as illustrated in Figure 1). Figure 1: bad Bad / Total Group 1 Group 2 Group

9 3. Empirical analysis The descriptive evidence illustrated in the previous sections highlights the existence of differences (on average) among mutual banks operating in the different competitive environments characterizing the three different groups identified above. The econometric exercise in this section is aimed at assessing quantitatively the differences between the three groups. In particular, we address three issues: 1. the relationship between and deposits, as an indicator of local effectiveness ; 2. the relationship between and total assets, as an indicator of mission efficiency ; 3. the relationship between bad and total. 3.1 Local effectiveness In order to estimate the degree of mutual banks local effectiveness, we regress on deposits, a series of dummies indicating the group to which each bank belongs (using group 2 as the benchmark) and interaction terms between deposits and group dummies. Moreover, we control for fixed time effects by means of year dummies, using year 2009 as the benchmark. Our results are summarized in Table 5, which reports three models. Our base specification is given by the pooled crosssection analysis in Model 1. We then check the robustness of our findings with Model 2, which accounts for the presence of repeated observations for each mutual bank, therefore controlling for cluster effects; and with Model 3, which explicitly controls for the panel structure of our data estimating a random-effect panel model 2. Focusing on Model 1, it is immediate to observe that for the mutual banks belonging to group 2 - deposits have a positive and statistically significant (at the 1% level) effect on : an increase of 1 million Euros in deposits determines a growth of slightly more than 900,000 Euros in. As the intercept of the model is also positive (and statistically significant at the 10% level), one can conclude that for banks belonging to the benchmark group the ratio of to deposits takes decreasing values as the size of the mutual banks increases, and it is therefore lower for the largest banks in the group. Looking at the results for the third group of banks, Table 5 shows that the overall effect of deposits on is even larger: an increase of 1 million Euros in deposits determines now a growth of about 977,000 Euros in. At the same time, however, the model intercept for the banks belonging to this group becomes negative. Therefore, as it can be appreciated in Figure 2, above a threshold of 225,125,881 million Euros in deposits, the ratio of to deposits which increases in bank size takes larger values than those characterizing the banks belonging to the second group. The estimated amount of at this threshold is 211,006,964 Euros, entailing a -to-deposits ratio of 93.7%. These are plausible values for the banks in our sample. In fact, looking at the 2009 data for the third group, one could see that 8 out of the 19 banks in the group have deposits and exceeding the estimated threshold. When considering the banks belonging to group 2, 10 out of 18 banks in the group exceed the threshold. These results suggest that absence of competition among mutual banks improves local effectiveness only if these banks (i.e., banks belonging to group 3) reach a size close to the average group size (which is equal to 239 million in deposits). 2 Note that we can t rely on a fixed-effect estimator, as by definition there is no variability over time in the three groups of mutual banks. We rely instead on a random-effect model exploiting cross-sectional variability to estimate the effects of covariates on the dependent variable. 9

10 Loans Third group Second group 211 MLN 225 MLN Deposits and bank bonds Figure 2: Third group versus second group BCCs Dependant variable: I(a) II(b) III(c) cons group 1 group 3 deposits and bank bonds deposits and bank bonds*group 1 deposits and bank bonds*group * *** *** *** *** ** *** * *** *** * ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) Year dummies Yes Yes Yes Obs. nr Prob>F no R-square no Wald Chi-Sq. no no Prob>Chi-Sq no no Notes. Significance levels: *>90%; **>95%; ***>99%. Robust standard errors in parenthesis (a) pooled cross section (b) pooled cross section and cluster (c) panel model with random-effects Table 5: localistic effectiveness estimates Looking finally at group 1, it emerges that the effect of an increase in deposits on is smaller than the one observed for groups 2 and 3, with an increase of 1 million Euros in deposits determining only an increase of about 753,000 Euros in. Although it is not possible to draw a definite conclusion in terms of the to-deposit ratio as the bank size changes (since the intercept for group 1 is not 10

11 statistically significant), our econometric estimates unambiguously establish that the local effectiveness of group 1 banks is much smaller than that of the other two groups. Therefore, the lower value of the -to-deposits ratio is due not only to the smaller size of these banks, but also to their specific characteristics, that we relate to the competitive environment in which they operate. It is finally worth noting that our results are largely confirmed by the different econometric specifications of the models in Columns 2 and 3 of Table 5, both in terms of coefficients magnitude and statistical significance. 3.2 Mission efficiency As stated at the beginning of this section, we estimate mission efficiency investigating the relationship between to total assets. In doing so, we use the same econometric specifications described in sub-section 3.1, but for the fundamental covariate, which is now total assets. Our results are reported in Table 6. Dependent variable: I(a) II(b) III(c) cons group 1 group 3 total assets total assets*group 1 total assets*group *** *** *** * * *** ** ** *** *** *** ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) Year dummies Yes Yes Yes Obs. nr Prob>F no R-square no Wald Chi-Sq. no no Prob>Chi-Sq no no Notes. Significance levels: *>90%; **>95%; ***>99%. Robust standard errors in parenthesis (a) pooled cross section (b) pooled cross section and cluster (c) panel model with random-effects Table 6: mission efficiency estimates We find that banks belonging to group 3 have, at the same time, a lower model intercept and a larger effect of total assets on when compared with banks belonging to group 2. This indicates that group 3 banks or at least the largest among them have a better ability to control local credit risk than banks belonging to group 2. It must, however, be noted that this result is weaker than the one on local effectiveness, because of the lower statistical significance of the benchmark intercept. On the other hand, focusing on group 1, the results that we obtain in all our econometric specifications concerning mission efficiency are stronger than those obtained for local effectiveness. More precisely, the -to-total assets ratio for these mutual banks is systematically lower than for the mutual banks belonging to the other groups. 11

12 3.3 Bad The results of the econometric analysis concerning the relationship between bad and total, which shares the same features of the models in the previous two sub-sections, is summarized in Table 7. This relationship appears to be positive and increasing for the benchmark group 2. At the same time, the large and positive value of the model intercept and the low value of the slope parameter (both statistically significant at the 1% level) suggest that the fraction of bad out of total is much larger for the smaller banks within the group. Dependent variable: bad I(a) II(b) III(c) IV(d) cons group 1 group 3 *group 1 *group *** *** *** *** *** * *** *** ** *** * ** (465146) ( ) ( ) ( ) ( ) ( ) ( ) (413408) (564826) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) Year dummies Yes Yes Yes Yes Obs. nr Prob>F no R-square no Wald Chi-Sq. no no Prob>Chi-Sq no no Notes. Significance levels: *>90%; **>95%; ***>99%. Robust standard errors in parenthesis (a) pooled cross section (b) pooled cross section and cluster (c) panel model with random-effects (d) pooled cross section with year dummies, excluding 2009 Table 7: bad estimates As for group 3 banks, the descriptive statistics in section 2 (see Table 4) suggest a lower ratio of bad to total when compared with all other groups (as shown in Figure 1). Our econometric analysis is therefore consistent with the descriptive evidence, although only in the specification of Model 1 (Table 7). This notwithstanding, using year 2009 as a reference, we observe that all other year dummies are negative (and statistically significant at the 1% level). This suggests that the economic crisis burst in 2008 following the Lehmann Brothers bankruptcy had a significant impact on Trento s mutual banks starting from Accounting for this observation in our Model 1 econometric specification, by excluding the year 2009 from the analysis, we can confirm all our results for group 2, at the same time finding interesting and significant results also for groups 1 and 3. In particular, the incidence of bad appears to be larger for group 1 than for the benchmark, while it turns out being significantly smaller for the mutual banks belonging to the third group. Therefore, the results for the sub-period are fully consistent with the evidence illustrated in Figure 1. 12

13 References Bofondi, Marcello e Giorgio Gobbi (2003), Bad Loans and Entry in Local Credit Markets, Banca d Italia. Cesarini, Francesco, Ferri, Giovanni e Michele Giardino (1997, a cura di), Credito e sviluppo Banche locali cooperative e imprese minori, Il Mulino, Bologna. Clemente, Claudio (2002, a cura di), Il confronto competitivo nelle banche locali il caso delle Banche di Credito Cooperativo, edizioni Metakom. Ferri, Giovanni (1997), Introduzione a Cesarini, Francesco, Ferri, Giovanni e Michele Giardino (1997). Guiso, Luigi, Paola Sapienza e Luigi Zingales (2004), Does Local Financial Development Matter?, The Quarterly Journal of Economics, August, pp Petersen, Mitchell e Raghuram Rajan (2002), Does Distance Still Matter: The Information Revolution in Small Business Lending, Journal of Finance, pp Smith, Bruce D. e Michael J. Stutzer (1990), Adverse selection and mutuality: The case of the farm credit system, Journal of Financial Intermediation, 1, pp