Politicians ‘on Board’: Do Political Connections Affect Banking Activities in Italy?

Authors


Vincenzo Farina, Department of Economics and Finance, University of Rome Tor Vergata, Via Columbia 2, 00133 Rome, Italy. E-mail: vincenzo.farina@uniroma2.it Tel: +39 06 72595433; Fax: +39 06 2040219

Abstract

In this article, we provide an empirical analysis to investigate whether politicians serving on the board of directors for banks influence performance, lending, and risk taking behaviours. We refer our analysis to all politicians (almost 160,000) belonging to a political body in Italy. Overall, our dataset contains 1,808 board members referring to 123 cooperative banks. Our results suggest that politicians who hold executive roles on the board of directors seem to exert a negative impact on banking activity. Therefore, in the current debate on the reform of the statutes of the Italian cooperative banks, we argue that the problem is not having politicians on the board of directors for banks, but rather allowing politicians to hold executive positions.

Introduction

In various countries, politicians at both the local and national level are able to continue working in the private sector after elections (Gagliarducci et al., 2010). When this occurs, the link between politics and business is quite widespread (Faccio, 2006). In this article, we provide an empirical analysis to investigate whether politicians serving on the board of directors for banks influence performance, lending, and risk taking behaviours.

Our work is unique in that it determines the political connection of banks by matching politicians and board members to measure the percentage of directors that are national and local representatives. In fact, existing studies of the impact of politics in the banking industry (La Porta et al., 2002; Sapienza, 2004; Dinç, 2005; Khwaja and Mian, 2005; De Nicolò and Loukoianova, 2007; Micco et al., 2007), only identify politically connected banks using the dummy variable of state or private ownership. Since banks are no longer state-owned in many countries, we want to test whether or not political presence on the board could act as a substitute for state-ownership.

We focus on Italian cooperative banks, which are an interesting case study for several reasons. First, because Italian banks do not have tradable participations and are not listed on the stock exchange, takeovers and other control mechanisms relying on the share price do not act as an automatic disciplining channel. Second, local political pressure is a matter of concern owing to regional limits on expansion and the strong connection to the territory (Ferri et al., 2001; Bos and Kool, 2006; Battaglia et al., 2010). Third, the fact that these banks are cooperatives does not reduce the importance of assuring sound and prudent management practices. However, as demonstrated by some recent scandals in Italy, these banks are currently faced with increased political pressure.

Our results suggest that politicians who hold executive roles on the board of directors seem to exert a negative impact on net interest revenues, loan portfolio quality and capitalization level; and a positive impact on banks' efficiency in terms of overhead costs. Consequently, we present a novel perspective to analyse this issue. We argue that the problem is not having politicians on the board of directors for banks, but rather allowing politicians to hold executive positions.

The rest of this paper is organized as follows. In the next section, we present the details of our study. Section 3 describes the data used in the analysis. Section 4 lays out the econometric model and estimation results. Finally, Section 5 concludes our analysis.

Literature review

What are the effects of having politicians on the board? According to the resource dependence theory, the need for environmental linkages is a function of the levels and types of dependence facing the firm (Pfeffer and Salancik, 1978; Boyd, 1990). One of the possibilities to reduce uncertainty is to appoint politicians to the board of directors (Mahon and Murray, 1981; Lang and Lockhart, 1990). In fact, these political linkages could positively affect a firm's performance by influencing: (1) stock value (Faccio 2006; Leuz and Oberholzer-Gee, 2006; Claessens et al., 2008;); (2) actions of legislators in passing relevant regulation (Agrawal and Knoeber, 2001; Mobarak and Purbasari, 2006; Mian et al., 2010); (3) the probability of a financial bailout (Faccio et al., 2006); (4) access to financial resources at more convenient conditions (Gomez and Jomo, 1999; Fraser et al., 2006; Claessens et al., 2008; Boubakri et al., 2009; Infante and Piazza, 2010); and (5) market power (Cingano and Pinotti, 2009).

Leuz and Oberholzer-Gee (2006) argue that in relationship-based systems, global financing and strong political connections are alternative means to create firm value. Examining this point in greater detail, Faccio (2006) finds that the announcement of a new political connection results in a significant increase of stock value, especially when a board director enters politics (but not for appointments of politicians to corporate boards). Additionally, many firms show a widespread overlap of controlling shareholders and top officers who are connected with national parliaments or governments. This is particularly prevalent in countries with higher levels of corruption, barriers to foreign investment, and more transparent systems. Claessens et al. (2008) show that Brazilian firms providing contributions to elected federal deputies experienced higher stock returns at the time of the 1998 and 2002 elections but increased their bank leverage after each election. With regard to legislator's conditioning problems that may arise from political linkages, Agrawal and Knoeber (2001) find that outside directors with backgrounds in politics, government, or law are more common on the boards of firms for which politics matters more. Mobarak and Purbasari (2006), analysing the case of developing countries, find that politicians in power are more interested in protecting the business interests of particular individuals connected to them as opposed to industries as a whole. Finally, Mian et al. (2010) examine the determinants of congressional voting behaviour on the American Housing Rescue and Foreclosure Prevention Act (AHRFPA) of 2008 and the Emergency Economic Stabilization Act (EESA) of 2008. They find evidence that constituent interests and special interests influenced voting patterns during the crisis. Specifically, representatives from districts experiencing an increase in mortgage default rates were significantly more likely to vote in favour of the AHRFPA. Moreover, increased campaign contributions from the financial services industry was associated with a higher likelihood of voting in favour of the EESA. Regarding the increased probability of a financial bailout, Faccio et al. (2006) analyse the likelihood of government bailouts in a sample of 450 politically connected firms from 35 countries between 1997 and 2002. They find that politically connected firms are significantly more likely to be bailed out than similar non-connected firms are. Additionally, politically connected firms are disproportionately more likely to be bailed out when the IMF or World Bank provides financial assistance to the firm's home government. Furthermore, of those firms that are bailed out, those that are politically connected exhibit significantly worse financial performance than their non-connected peers do at the time of the bailout and over the following two years.

Regarding the access to financial resources at more convenient conditions, Gomez and Jomo (1999) and Fraser et al. (2006) find a positive and significant link between political connections and leverage of firms in Malaysia. This is noteworthy because the government in Malaysia exerts a significant influence over the corporate sector through listing restrictions, direct equity ownership of listed firms, control of the banking sector, and through government-sponsored institutional investors. Boubakri et al. (2009) provide evidence that investors require a lower cost of capital for politically connected firms, suggesting that these firms are generally considered less risky than non-connected firms.

Finally, Infante and Piazza (2010) find evidence that politically connected firms benefit from lower interest rates only when the political connection is at a local level. They show that this effect is generally stronger when politically connected firms borrow from banks with politicians on their boards and when the degree of autonomy granted to local loan officers is higher. Regarding market power, Cingano and Pinotti (2009) find greater levels of influence for politically connected firms. However, this power is not driven by higher productivity, but by greater sales to the public administration.

It could be argued that the presence of politics could make firms less profitable. This may occur because these firms have lower managerial incentives or because they inefficiently cater to politicians' wishes such as the pursuit of individual goals or the transfer of financial resources to their supporters (Shleifer, 1998; Fan, Wong, Zhang 2007; Boubakri et al., 2008; Claessens et al., 2008; Yeh et al., 2010). Shleifer (1998) argues that private ownership is preferable to public ownership when the incentives to innovate and to contain costs are strong. Moreover, the case for private provision only becomes stronger when factors such as competition between suppliers, reputational mechanisms, the possibility of provision by private not-for-profit firms, and political patronage and corruption are at play. Therefore, one could expect that firms have a greater inclination to search for political connections when the external governance structure is ineffective. In this regard, Boubakri et al. (2008) show that political connections in newly privatized firms are prevalent in countries with lower judicial independence.

Fan et al. (2007) show that the appointment of politically connected CEOs does not enhance a firm's efficiency; rather, it fulfils the political goals of politicians. Claessens et al. (2008) indicate that access to bank finance is an important channel through which political connections operate.

Finally, Yeh et al. (2010) find that the quality of corporate governance is negatively associated with the likelihood of firms engaging in political connections, while firm size is positively associated with these connections.

Within the theoretical framework of the resource dependency theory, government is an important source of external dependency for regulated firms such as banks. Consequently, various studies focus on the impact of politics in the banking industry (La Porta et al., 2002; Sapienza, 2004; Dinç, 2005; Khwaja and Mian, 2005; De Nicolò and Loukoianova, 2007; Micco et al., 2007). In principle, because having politicians on the board improves the coordination between banks and government, one could expect positive effects from political connections. Conversely, political presence could have potentially negative consequences on performance, loan quality, efficiency, and overall risk for banks.

Micco et al. (2007) analyse performance in this regard and find that state-owned banks tend to have lower profitability and higher costs than their private counterparts. Their research shows that this differential is driven by political considerations. La Porta et al. (2002) claim that politically connected banks are inefficient because they are under the influence of politicians who are only interested in pursuing their personal objectives.

However, examining a sample of European banks during 1986–1989, Molyneux and Thornton (1992) find that government ownership has a positive impact on bank profitability. Moreover, Altunbas et al. (2001) indicate that state-owned banks have slight cost and profit advantages over their private commercial bank counterparts in the German banking market. This can be explained by their lower cost of funds.

Dinç (2005), analysing lending behaviours, finds that private banks increase their lending in the year before an election, while government-owned banks decrease their lending. On the contrary, private banks decrease their lending in election years, while state-owned banks increase their lending.

Khwaja and Mian (2005) show that state-owned banks tend to lend more to firms with politically connected directors. Sapienza (2004) compares the interest rate charged to two sets of companies with identical characteristics and finds that state-owned banks offer lower interest rates than private banks. Moreover, the lending behaviour of state-owned banks is affected by the electoral results of the party affiliated with the bank. The stronger the political party in the area where the firm is borrowing, the lower the interest rates charged. Finally, considering risk taking profiles, De Nicolò and Loukoianova (2007) find that the relationship between bank concentration and bank's risk of failure is significantly higher in state-owned banks with a sizeable market share than in private banks.

The data

Our analysis is focused on 2006, which is a politically significant year because of the various elections at different administrative levels. Moreover, by studying the data from 2006, we can reasonably consider that the recent financial crisis is not a factor.

According to Federcasse (Italian Federation of Cooperative and Rural Banks), in 2006 there were 438 cooperative banks with 3,758 branches, more than four million clients, around one million members, 111 billion Euros of deposits (7% market share), and 94 billion Euros of loans (8% market share). These banks were based in 2,493 municipalities (in 556 municipalities, they were the only banking institution). Their presence was particularly significant in the northeastern regions, driven mainly by Trentino Alto Adige, but was also relevant in some central and southern regions.

Compared to commercial banks, cooperative banks share the mutualistic principle of cooperatives. According to Italian Banking Law, they must primarily grant credit to their members. In fact, credit operations with members need to account for at least half of the risk weighted assets of a cooperative bank. The cooperative bank is required to allocate at least 70% of annual net profit to reserves. In addition, 3% of the net profit must be paid into a special mutual aid fund for the promotion and development of the cooperation.

Other distinctive traits of cooperatives include regional limits on expansion, non-tradable shares, decision making based on the ‘one person one vote’ principle, limited ownership rights, and board of directors' approval of new membership applications.

Bank-level financial information comes from the Bankscope database. We extracted data about loans (specifically amount and quality), profitability, efficiency, and capitalization. In detail, the calculation of loans over total assets (L/TA) allows us to test for a relationship between politicians on the boards and lending volumes of cooperative banks. The variable net interest revenue over total assets (NIR/TA) allows us to test for a link between politicians on the boards and the interest margin of cooperative banks. Since higher levels of this variable could come from either a lower rate on bank deposits or a higher rate on loans, we need to examine the risk-taking behaviour of politically connected banks. Therefore, we define the variable loan loss provisions over total assets (LLP/TA) in order to test the existence of a relationship between politicians on the boards and the level of risk of cooperative banks. The ratio overhead costs over total assets (OH/TA) allows us to test the existence of a relationship between politicians on the boards and the efficiency of cooperative banks. Equity over total assets (E/TA) allows us to test the existence of a relationship between politicians on the boards and the capitalization of cooperative banks.

The composition of the boards is obtained from the Annual Relations of the cooperative banks. Following Infante and Piazza (2010), we identify politically connected banks by matching the names of the members on the boards with the names of the members of the Italian political body during the year 2006. We consider the national Parliament and all the local councils (in the 20 Italian regions, the 103 provinces, or in the 8,094 municipalities). In 2006, the Italian political body consisted of almost 160,000 members.

The data on local politicians come from the Italian Ministry of Interior's website, while data on national politicians are drawn from the websites of the two Houses (Camera and Senato) of the Italian Parliament.

The data is quite rich and includes a multitude of personal and political information. Among other things, we found information regarding birth date, birthplace, gender, qualification, profession, political party, and election date. In order to obtain a meaningful match, we restricted the sample of banks to those for which we also had the birth dates of the members. This generated a subset of 123 banks with 1,808 board members. Overall, we found 225 distinct individuals classified as politicians on the boards of our sample of cooperative banks. We present the comparison of descriptive statistics for banks in and out of the sample in Table 1. Table 2 shows the correlations for the variables of our sample.

Table 1. Descriptive statistics
 Banks in the sampleBanks out of the sample
 MeanSDMinMaxMeanSDMinMax
L/TA0.653250.132290.321550.893420.662360.138160.019610.89391
NIR/TA0.031020.006740.018550.063290.030230.005250.017240.04854
LLP/TA0.002490.00298−0.004220.016900.002350.00267−0.003840.01617
OH/TA0.026920.006830.017000.067070.026330.006340.009810.07843
E/TA0.123700.033380.072660.225320.127730.068030.064100.87931
log(TA)12.128440.935459.7050414.4457212.288680.986988.5370015.46261
NSub5.219513.805810225.394644.41081036
NEmp57.8780553.31189526971.6572686.530893980
P0.026870.0432500.15385
P*0.145280.1575001
Pexp0.124660.0962100.46154
Table 2. Correlations
 L/TANIR/TALLP/TAOH/TAE/TAlog(TA)NbraNempPP*
  1. *Significant at 10%, **Significant at 5%, ***Significant at 1%.

L/TA          
NIR/TA−0.41739***         
LLP/TA−0.15340*0.28219***        
OH/TA−0.25818***0.66846***0.25107***       
E/TA−0.30226***0.29454***−0.25754***0.19281**      
log(TA)0.39339***−0.41100***0.05125−0.43491***−0.35375***     
Nbra0.11760−0.16939*−0.10345−0.18631**−0.093370.37332***    
Nemp0.31900***−0.22741**0.06415−0.20809**−0.24003***0.86417***0.33851***   
P−0.25201***−0.026660.16218*−0.01365−0.06768−0.08960−0.03876−0.03066  
P*−0.06600−0.03199−0.048460.01040−0.03394−0.11593−0.19938**−0.123260.04593 
Pexp−0.29344***0.14991*−0.047880.16930*0.13247−0.17570*0.03515−0.16089*0.33352***−0.16766*

Evaluated in the full scope of Italian cooperative banks, our sample does not seem to be evenly distributed geographically throughout Italy. The northeast region shows high values. The provinces of Trento and Bolzano are relevant in the case for Trentino Alto Adige and Vicenza for Veneto. Nevertheless, the presence of cooperative banks in the other provinces of central and southern Italy is also significant.

Two facts confirm the link between cooperative banks and the territory. First, no board member for our sample of banks is also a representative in the parliament. Second, parties that are represented at a national level have little representation on the boards. Politicians on the boards are essentially an expression of local parties, with no national counterpart: 182 out of 225 members were elected in civic parties. Other members are an expression of centre-right parties (18 members), centre-left parties (17 members), and other parties (8 members). Despite the fact that the macro areas of the country have different numbers of cooperative banks, there is little variance in the presence of politically connected banks compared with the presence of non-connected ones (Table 3).

Table 3. Banks' distribution across Italian macro-areas
Macro-areasSampleTotal
  1. The definition of macro-areas is based on ISTAT (The National Institute of Statistics) classification of Italian regions: North (Piemonte,Valle D'Aosta, Lombardia, Trentino Alto Adige, Veneto, Friuli Venezia Giulia, Liguria, Emilia-Romagna). Center (Toscana, Umbria, Marche, Lazio). South + Islands (Abruzzi, Molise, Campania, Puglia, Basilicata, Calabria, Sicilia, Sardegna).

North68243
Centre2085
South + Islands35110
Italy123438

In general, political membership on the boards is very significant in some areas in the central region (such as the province of Ascoli Piceno) and southern region (such as the provinces of Campobasso, Cosenza, and Taranto). However, political membership is also prevalent in northern Italian provinces (such as Bolzano, Como, Cuneo, Mantova and Reggio Emilia).

Finally, we obtain the following distribution (Table 4) by classifying members of the boards by position (executive or non-executive) and the status of the politician. It is worth noting that, out of our sample of 1,808 board members, a non-negligible fraction is represented by members who are also politicians (12.4%). Moreover, 50 of them hold an executive position. Since it is reasonable to expect different levels of influence of executive and non-executive politicians on banking activities, we use two measures of political connection. The first measure is the percentage P of tout court board members appointed in a local and/or national government, without distinguishing among executive or non-executive positions. The second measure of political connection is the percentage P* of board executive members appointed in a local and/or national government. We are aware that these definitions of political connectedness cannot detect the presence of members who are only indirectly influenced by political events. In other words, it is possible that influence is also exercised by individuals not necessarily on the board (such as loan officers), who could be directly or indirectly politically influenced.

Table 4. Members of the boards with respect to the position (executive or non-executive) and the status of politician
PositionBoard executiveBoard membership
Non-politician3461,237
Politician50175

Model and results

To infer the effect of political presence on the board over the variables of interest, we estimate a series of regression models of the form:

image

For each regression, y represents one of the variables of interest defined in the previous section: y1 = Loans/Total Assets; y2 = Net Interest Revenue/Total Assets; y3 = Loan Loss Provisions/Total Assets; y4 = Overheads/Total Assets and; y5 = Equity/Total Assets.

We maintain the same covariates in all models. Our variable of interest P represents the percentage of politicians on the boards of cooperative banks. Two important control variables are as follows: the percentage Pexp of board members with more than five years of political experience and the logarithm of the total assets log(TA) that accounts for the dimension of cooperative banks.

Other control variables include the number of branches of the cooperative banks Nbra, the number of employees Nemp, and a geographical dummy MA indicating the macro-areas (see Table 3) of the banks. Results for these models are presented in Table 5.

Table 5. Testing the effect of tout court political connections
 L/TANIR/TALLP/TAOH/TAE/TA
  1. Standard errors in parentheses.*Significant at 10%, **Significant at 5%, ***Significant at 1%.

Const0.50878**0.07334***−0.002630.09429***0.34915***
(0.22400)(0.01098)(0.00663)(0.01270)(0.07188)
P−0.00019−0.00005−0.00002−0.00002−0.00013
(0.00058)(0.00003)(0.00002)(0.00003)(0.00019)
PolExp−0.00191**−0.00002−0.000030.000040.00019
(0.00096)(0.00005)(0.00003)(0.00005)(0.00031)
log(TA)0.01937−0.00391***0.00043−0.00606***−0.01930***
(0.01965)(0.00096)(0.00058)(0.00111)(0.00631)
Nbra−0.00133−0.00010−0.00015*−0.000110.00054
(0.00257)(0.00013)(0.00008)(0.00015)(0.00082)
Nemp0.000010.00006***0.000000.00008***0.00013
(0.00033)(0.00002)(0.00001)(0.00002)(0.00011)
MA (c)−0.036840.00442***0.00222***0.00255*−0.01655**
(0.02531)(0.00124)(0.00075)(0.00144)(0.00812)
MA (s + i)−0.18002***0.00923***0.00194***0.00489***0.00211
(0.02251)(0.00110)(0.00067)(0.00128)(0.00722)
AdjR20.466270.505860.075670.356080.13706
p–value(F)0.000000.000000.023500.000000.00103
N° obs123123123123123

Subsequently, we run the same models using the second measure of political connection P*. We test the hypothesis that, to modify the politics of the bank, one must not only be a member of the board and a politician but also in an influential position. Results for this new definition of political connection are presented in Table 6.

Table 6. Testing the effect of influential political connections
 L/TANIR/TALLP/TAOH/TAE/TA
  • Standard errors in parentheses.

  • *

    Significant at 10%.

  • **

    Significant at 5%.

  • ***

    Significant at 1%.

Const0.53099**0.07487***−0.003800.09604***0.35647***
(0.22234)(0.01077)(0.00658)(0.01250)(0.07123)
P*−0.00327−0.00029***0.00012*−0.00026**−0.00121*
(0.00215)(0.00010)(0.00006)(0.00012)(0.00069)
PolExp−0.001380.00004−0.000040.000080.00040
(0.00099)(0.00005)(0.00003)(0.00006)(0.00032)
log(TA)0.01717−0.00413***0.00050−0.00624***−0.02015***
(0.01951)(0.00095)(0.00058)(0.00110)(0.00625)
Nbra−0.00138−0.00008−0.00013*−0.000110.00056
(0.00252)(0.00012)(0.00007)(0.00014)(0.00081)
Nemp0.000060.00006***0.000000.00009***0.00015
(0.00033)(0.00002)(0.00001)(0.00002)(0.00011)
MA (c)−0.035900.00451***0.00219***0.00262*−0.01618**
(0.02508)(0.00121)(0.00074)(0.00141)(0.00803)
MA (s + i)−0.17754***0.00935***0.00177***0.00509***0.00282
(0.02231)(0.00108)(0.00066)(0.00125)(0.00715)
AdjR20.476250.526410.095240.379430.15594
p-value(F)0.000000.000000.009200.000000.00035
N° obs123123123123123

According to the findings by Dinç (2005) and Micco et al. (2007), there is no relation between political influence and lending volumes of banks located in industrial countries. Our results (Tables 5 and 6) similarly show that in Italy, politically connected banks do not differ significantly from non-connected ones in terms of loan volumes as a fraction of total assets. Therefore, having politicians on the board of directors does not affect the lending volumes of the bank.

Conversely, if loan volumes are the same for the two types of banks, one could consider the possibility that those banks with politicians on the board may be more willing to grant preferential treatment to politically connected firms or to public administration. Unfortunately, our data does not allow us to discriminate between loan destinations, and therefore, the presence of preferential treatment for some categories of borrowers. Nevertheless, we are able to further investigate the impact of political presence on net interest revenue as a fraction of total assets in order to identify structural differences among connected and non-connected banks. This is important because, in case of the same amount of loans, their price could be an indicator of political lending behaviour.

Table 5 shows that more tout court politically connected banks do not exhibit significant differences from less connected ones. Yet, when considering politicians in an influential position (Table 6), we observe their negative impact on net interest revenues. Theoretically, the estimated coefficient implies that, ceteris paribus, the interest margin on total assets ratio decreases by 2.8 basis points on average for each 1% increase of politicians in influential positions.

From a bank's perspective, this result is consistent with Sapienza (2004), who finds that state-owned banks charge systematically lower interest rates to similar or identical firms than do privately-owned banks. In detail, she shows that firms that borrow from state-owned banks pay an average of 44 basis points less than do firms that borrow from private banks. From the firm's perspective, the results seem to be consistent with Infante and Piazza (2010), which finds that political connections ensure lower interest rates to politically connected firms (the effect is significant but economically negligible since it could be estimated around 3 basis points), and Claessens et al. (2008), which considers this an important reason for political connections.

It is possible to provide two explanations for our result. On one hand, lower levels of net interest revenue could simply come from higher rates on bank deposits. On the other hand, they could be explained by low rates on loans as a consequence of the political lending behaviour of banks. Therefore, in order to further investigate this subject, we compare loan loss provisions as a fraction of total assets.

Table 5 shows that when this variable is considered, more politically connected banks are not significantly different from less connected ones. In the other case (Table 6), the estimated coefficient implies that, ceteris paribus, loan loss provisions as a fraction of total assets increase by 1.1 basis points on average for each 1% increase of politicians in influential positions. This result would suggest that more politically connected banks tend to lend to firms for which raising capital is otherwise too difficult. In other words, political influence seems to have a negative impact on loan portfolio quality.

While profit maximization may not be a primary goal for cooperative banks, cost efficiency is certainly an important objective. Efficient cost management is crucial to guarantee the cooperative banks survival, and consequently, the continuity of ‘services’ provided to their members and customers (Battaglia et al., 2010). Theoretically, if political connections help banks increase ability, knowledge, and experience and overcome bureaucratic obstacles, then politically connected banks would be more efficient and better managed. Conversely, some authors consider politically influenced banks to be less efficient than their private counterparts (La Porta et al., 2002; Sapienza, 2004). In detail, these banks are inefficient because they are guided by politicians who are only interested in maximizing their personal objectives.

Looking at the difference in overhead costs allows us to establish if political connections have an impact on the efficiency level of banks. In this regard, the results in Table 6 show that, ceteris paribus, overheads costs on total assets decrease by 2.5 basis points on average for each 1% increase of politicians in influential positions. This result seems to be consistent with Altunbas et al. (2001) in their study of the German banking market. However, we do not find significant differences when considering tout court political connections (Table 5).

Finally, Table 5 shows that the equity over total assets ratio does not seem to be influenced by tout court political presence on the boards of the banks. However, these results show the negative impact of politicians in influential positions (Table 6). In this case, the estimated coefficient implies that, ceteris paribus, this ratio decreases by 12 basis points on average for each 1% increase of their number. Moreover, in line with regulatory capital requirements defined by the Basel Committee, there appear to also be important differences if we consider the geographical area (and therefore the different risk levels of borrowers) in which the banks have their headquarters.

Conclusions

Our study evaluates the effects of political presence on the board of directors for cooperative banks by considering the special relationships that exist between these institutions and the territory. Results show that politicians on the board have a generally negative impact on banking activity and validate previous findings by La Porta et al. (2002), Sapienza (2004), Dinç (2005), Khwaja and Mian (2005), Micco et al. (2007).

In detail, even if these connections do not exert an impact on lending volume, we find that the percentage of politicians on the board has a significantly negative effect on net interest revenues, loan portfolio quality, and capitalization levels. Our results also show that politicians in influential positions seem to exert a positive effect on the bank's efficiency.

However, these effects are not the consequence of simply having politicians tout court on the board; rather, they are a result of politicians in influential positions. Therefore, in the current debate on the reform of the statutes of Italian cooperative banks, we argue that the problem is not for politicians to be on the boards, but for them to hold executive positions.

This conclusion regarding the implications of politicians serving as executives on the board of directors for banks is both new and relevant. To our knowledge, this is the first paper that considers the problem of political presence on the board of directors by making a distinction based on position (executive or non-executive).

Our result is also relevant for other studies analysing the implication of political connections from managerial and regulatory points of view. In fact, since the behaviour of politicians holding executive positions is difficult to observe and verify, this could amplify the problem of negative incentives (evidenced by the existing literature on this topic). While other studies analyse political impact by considering the variable (state or private) ownership of banks as a dummy, the main advantage of our research is the identification of politically connected banks on the basis of the match of board members and politicians belonging to one of the political bodies in Italy.

Another advantage of this research is the focus on local banks, characterized by a strong connection with the territory. This means that they contribute to the territory's development and are influenced by some local characteristics like political pressure.

It is worth noting that this evidence does not necessarily imply that politicians pursue their private interests in the management of the bank. It seems that the appointment of politically connected executive board members is not beneficial in terms of loan quality and overall profitability of cooperative banks.

Furthermore, at this stage our analysis does not consider the presence of members not necessarily on the board (such as loan officers), who might be directly or indirectly politically influenced. This could be a point to develop in further analyses. A challenge for future researches is also to comprehend if, and under what conditions, interlocking directorates generated by political ties among banks and firms affect lending behaviour of banks. Panel data could be useful in order to monitor board changes and political participation over time (e.g., before and after new elections). In this regard, it is important to analyse if banks with politicians on the board differ in their behaviour on the basis of the strength of the political party in the reference area (such as being the ruling party).

Ancillary