PERFORMANCE OF FAMILY-CONTROLLED BANKS: DO POLITICAL CONNECTIONS MATTER?

This paper investigates the effects of political connection on the relationship between familycontrol and organizational performance. A prediction is advanced hypothesizing that those political connections might help to mitigate the agency problem of family ownership on corporate outcome. A dataset of Indonesian banks over the 2001-2008 periods are analyzed. The data reveals that ownership and market competition display an insignificant changes during the period of observation that eventually prevent the study to explore time effect and individual effect as well. The main findings are that the family-controlled banks have a lower performance than that of their counterparts.


INTRODUCTION
Family-owned firm is a form of business organization which is prevalent around the world.Many studies have investigated the different impact between family-controlled and non-family controlled firms on the field of organizational performance (e.g.Maury 2006;Prabowo and Simpson 2011;Miller et al. 2007), investment policy (e.g.Boubakri and Ghouma 2010;Anderson et al. 2012); agency cost of debt (e.g.Anderson et al. 2003); acquisition decisions (e.g.Caprio et al. 2011); capital structure (King and Santor 2008); earnings management (Yang 2010).However, most of empirical papers working on family ownership are conducted in the context of non-financial firms.Only few studies have been done on family-controlled banks (e.g.Bunkanwanicha et al. 2006;Barry et al. 2011), even though in fact many banks are family-owned.
The present paper attempts to investigate the performance difference between family-controlled and non-family controlled banks.Bhaumik and Gregoriou (2010) explain that family ownership could reduce the transaction cost which ultimately improves performance.On the other hand, however, it could also lead to expropriation to minority shareholders.Moreover, I deepen the study by looking at a contingent factor that mitigates the impact of family ownership on performance.As banking is a tightly regulated industry, a strong link to environment, especially the regulators, is quite important which is in line with the resource dependence theory (Pfeffer and Salancik 2003) suggesting that external resources of organizations strongly affect the behavior of the organization.Arguably, banks may establish political connections (for example, by having political figures in their board) to have a close relationship with regulators, and ultimately to improve performance.
Some papers have studied political connections of banks (e.g.Carreta et al. 2012;Nys et al. 2013;). Carreta et al. (2012) study politically connected banks in Italy.They find that having politicians on the board of directors seem to exert a negative impact on banking activity.Another impact of being politically connected banks is found by Nys et al. (2013) revealing that in Indonesia, politically connected banks are benefited in term of receiving more deposits because they are perceived as less risky by depositors more so after the explicit deposit insurance with limited guarantee system implemented1 .Polsiri and Jiraporn (2012) examine the difference probability failure between politically connected and non-politically connected financial institutions.They do not find evidence that political connections through controlling families and state connection determine the failure likelihood.Literature has indicated that there is a mixed result on the impact of having political figures on bank performance.
This paper studies the effect of family ownership on performance and the role of political connections in the context of Indonesia because family-owned firms are prevalent in this country.Claessens et al. (2000) show that 57.7 percent of market capitalization in the stock exchange was controlled by 10 families.Fisman (2001) exhibits that 25 politically connected conglomerates contributed around 30 percent of total GNP.However, If the literature on family firms in Indonesia for the data during Soeharto's period is well-documented (e.g.Dieleman 2010;Dieleman and Sachs 2008), little is known about current develop-ment (after the institutional reforms) of family firms in Indonesia as only few papers have done on this issue (e.g.Tsamenyi et al. 2008;Prabowo and Simpson 2011).This paper discusses several unique features of familyowned banks in Indonesia.First, I test whether such banks have better performance than their counterparts.Secondly, I test the role of political connections of such banks on their performance.
The remainder of the paper is organized as follows.The next section discuss theoretical framework that serve as a basis for rationalizing the prediction.The following section presents data and model.Section 4 analyses and explore empirical test.The last section resumes the investigation.

LITERATURE REVIEW AND HY-POTHESES DEVELOPMENT Family Controlled Firms and Performance
The effect of family ownership on firm performance has been quoted as having two opposite arguments.In the one hand, family ownership might weaken shareholder-manager conflict (agency problem type I) that eventually reduce transaction cost (La Porta et al. 1999).Therefore, the performance of such firms would be higher than their counterparts.On the other hand, the presence of family ownership increases the exposure of the firms to suffering from majority-minority conflict (Type 2 agency problem).The substantial shareholding of controlling family will deliver a higher incentive to expropriate minority shareholders (Bhaumik and Gregoriou 2010).Some studies also reveal that family member involvement in the board of directors is more likely to exacerbate the effect of family ownership on performance.A number of papers also document that better governance would reduce the negative impact of family ownership on performance.
Another strand of literature on family firms deepens the analysis of family firms by decomposing the method of exercising corporate control.Specifically, the method of control is decomposed into two groups such as immediate ownership and ultimate ownership.This view is grounded on the common feature that most of family firms are affiliated with a business group which is commonly controlled by a very rich family (conglomerates).Khanna and Palepu (2000) and Bhaumik and Gregoriou (2010) argue that affiliation with business groups might add to a firm's performance and market value, especially in contexts where markets for capital and other factors of production are imperfect.
Accordingly, the family-controlled banks in this paper follow such grouping.The first is Family Direct Ownership defined as individuals who have ownership at least 10%.The second is Family Indirect Ownership, those are affiliated with business groups which are controlled by family.The second type of family ownership may exacerbate the agency conflict between controlling shareholders and minority shareholders (tunneling hypothesis) more so if divergence between control right and cash flow right is quite high.Some papers, however, argue that family indirect ownership (through a business group) can lead to a better performance which is known as the propping hypothesis (the antithesis of tunneling hypothesis) contending that the ultimate owner will prop up the firms when they encounter financial or non-financial difficulties.
The family-controlled bank is relatively less received attention, especially with regards to performance of such banks.Barry et al. (2011), in a cross country study of European banks, find no significant effect of family ownership on bank performance.Bunkanwanicha et al. (2006) study family-controlled banks in Thailand especially those affiated with a family group business (indirect family ownership).They find that bottom tier banks in the pyramids have lower performance due to risky loans as they are assigned to undertake risky investment.
As a civil (French) law country, Indonesia is categorized as a weak investor protection country.Weak investor protection can create incentives for controlling shareholders to expropriate the minority which means that Type 2 agency is dominant than Type 1 agency problem.It is therefore supposed that family ownership of banks in Indonesia has a negative impact on performance.

The Role of Political Connections
Political connections are less studied in the banking literature.Most of empirical papers examining the intersection between politics and banking focus on the government ownership of banks (e.g.La Porta et al. 2002;Sapienza 2004;Dinc 2005;Micco et al. 2007).However, recently, some papers have reached politically private banks.For example, Carreta et al. ( 2011), using sample of Italian cooperative banks, document that having politicians in their board of directors has a negative impact on banking activity such as net interest income, loan portfolio quality and capitalization levels.However, they also find that efficiency is improved for banks having politicians in the influential positions.Nys et al. (2012), using a dataset of Indonesian commercial banks, find evidence that politically connected banks are perceived less risky by depositors because depositors believe that these banks will be rescued by the government when they face financial problem.These banks, therefore, have a higher proportion of deposits than nonconnected banks.
On the particular issue which is the focus of this paper, we expect that political connections can mitigate the negative impact of family ownership of banks.Arguably, as Nys et al. (2012) explain, politically connected banks have a greater access to deposits even with a lower rate which improve their performance.It is therefore supposed that political connections can reduce the negative impact of family ownership.

RESEARCH METHODOLOGY Measurement of Variables
Family Direct Ownership (FBDIRECT) is individuals (or family) who has ownership at least 10%.Family Indirect Ownership (FBINDIRECT), those are affiliated with business groups which are controlled by family.Bank Performance, measured by the ratio of net income to total assets (ROA).Political connections (POLCON) is the banks which are majority owned by the Indonesian government or banks which having political figures on their board of commissioners or board of directors 2 .Political figures are politicians, bureaucrats, or former of them as suggested by Nys et al. (2012).
I included several bank characteristics that serve as control variables.Bank size is proxied by the natural log of total assets (LNTA).Risk aversion is measured by the ratio of equity to total assets (EQTA).Diversification (DIVER) is calculated using the bank diversification index developed by Elsas et al. (2010).Listed bank (LISTED) is a dummy variable that takes 1 if the bank is listed in Indonesian Stock Exchange and zero otherwise.Foreign bank (FOB) is a dummy variable that takes 1 if the bank has foreign ownership and zero otherwise.Banking market structure is measured using the Herfindahl Hirschman Index (HHI).
To test the moderating effect of political connections, I create interaction variables between family ownership and political connections.FBDIR_POLCON and FBIN-DIR_POLCON stand for interaction between political connections and family direct ownership as well as political connections and family indirect ownership, respectively.

Empirical Model
The empirical models incorporate two main variables: the main variables of interest and control variables.I included several bank characteristics that serve as control variables.The hypothesis testing relies on the following specifications: 2 Indonesia has a dual board system; the board of commissioners performs supervisory roles and board of directors act as executives (Nam & Nam, 2004).where: ROA = Bank Performance, measured by the ratio of net income to total assets ().FBDIRECT = individuals (or family) with at least 10% ownership FBINDIRECT = Individuals (or family) that are affiliated with business groups which are controlled by family.POLCON = Banks which are majority owned by the Indonesian government or banks which having political figures on their board of commissioners or board of directors.LNTA = the natural log of total assets.EQTA = the ratio of equity to total assets.DIVER = bank diversification index developed by Elsas et al. (2010).LISTED = dummy variable that takes 1 if the bank is listed in Indonesian Stock Exchange and zero otherwise.FOB = dummy variable that takes 1 if the bank has foreign ownership and zero otherwise.HHI = Banking market structure is measured using the Herfindahl Hirschman Index.
The empirical models are tested using OLS.Yet, I was unable to include individual as well as time effect as there is a singular matrix within the proxies of family ownership and the HHI.

Data
Data are gathered from various sources.Financial statements, financial performance, and ownership structure comes from the database of Indonesian central banks (Bank Indonesia).I rely on annual report and OneSource database to identify political connections.Yet, internet searching is also employed to facilitate data collection and confirmation, particularly, in the final stage.

RESULTS AND DISCUSSION
Descriptive statistics of variables is presented in Table 1.The mean (median) of ROA as dependent variable is 2.708 (2.520).20% of sample are direct family ownership, while the proportion of indirect family ownership is 20.1%.58.7% of sample are classified as politically connected banks, either state or private.Foreign banks include 21.4% of sample.16.2% of observations are publicly traded banks.
Table 2 exhibits the correlation matrix of variables.As expected, direct and indirect family ownership are negatively correlated with the proxy of performance which is return on assets.
Regressions results of model 1 and model 2 are presented in table 3. Results of control variables are relatively consistent with the previous studies.Large banks have a higher performance than small banks because they are benefited of economies of scope.The ratio of equity to total assets as well as HHI are positively associated with bank performance, while listed banks are found to have a lower performance compare to privately-held banks.I do not find evidence on the impact of foreign banks as well as listed banks on performance.
Turning to the variables of interest as the advanced prediction, family ownership, either direct or indirect, has a negative impact on bank performance.This result in line with the finding of Prabowo and Simpson (2011) for Indonesian non-financial firms.They also find that family control (family ownership and family involvement on the board) is negatively related to firm performance.Referring to Bunkanwanicha et al. (2006), the negative effect of indirect family ownership might come from the fact that the family-controlled banks are at the lowest level in the pyramids.The controlling shareholder pushes these banks to undertake risky investment.As shown in column 2 of table 3, political connections have a positive impact on bank performance.However, our argument on the moderating effect political connections in the link between family ownership and performance is not supported.None of the interaction variables are significant.

CONCLUSION
I investigate performance of family-controlled banks in Indonesia compare to their counterparts.Our results suggest that performance of family-owned banks in Indonesia, either direct or indirect family ownership is lower than non-family controlled banks.The results also reveal that politically connected banks have a higher performance than non-connected banks.However, I do not confirm the mitigating role of political connections in the relationship between family ownership and performance.
Nevertheless, several caveats are in order.First, I do not disentangle political connections into different types of connections such as connections with executives, parliament or political party.Also, I do not separate the connections into ownership connection and board connection.Second, I do not test using the GMM technique as one could expect that family ownership might be an endogenous factor. α 2 FBINDIRECT i,t + α 3 POLCON i,t-1 + α 4 FBDIR_POLCON i,t + α 5 FBINDIR_POLCON i,t + α 6 LNTA i,t + α 7 EQTA i,t + α 8 DIVER i,t + α 9 FOB i + α 10 LISTED i,t + α 11 HHI i,t + ε i,t ..... (2)