Working Paper. Central Bankers as Supervisors: Do Crises Matter?

Transcription

1 Università Commerciale Luigi Bocconi BAFFI CAREFIN Centre for Applied Research on International Markets, Banking, Finance and Regulation Working Paper By Donato Masciandaro and Davide Romelli Central Bankers as Supervisors: Do Crises Matter? BAFFI CAREFIN Centre Research Paper Series No This Paper can be downloaded without charge from The Social Science Research Network Electronic Paper Collection:

2 Central Bankers as Supervisors: Do Crises Matter? Donato Masciandaro 1,2 and Davide Romelli 3,4 1 Department of Economics and Paolo Baffi Center, Bocconi University, Milan. 2 SUERF The European Money and Finance Forum. 3 THEMA Université de Cergy Pontoise, Cergy Pontoise, France. 4 ESSEC Business School, Cergy Pontoise, France. Abstract Following the global financial crisis many countries have changed their financial supervisory architecture by increasing the involvement of central banks in financial supervision. This has led many scholars to argue that financial crises are an important driver in explaining the evolution of the role of central banks as supervisors, supporting the traditional theory of central banking. In this paper, we formally test whether there is any link between supervisory reforms and the occurrence of financial crises. We study the evolution of financial sector supervision in a large sample of countries by constructing a new database that captures the full set of supervisory reforms implemented during the period We find that systemic banking crises drive reforms in supervisory structure, but at the same time we highlight an equally important bandwagon effect, i.e. reforms are more likely to happen if other countries are reforming their supervisory architecture, as well as if a particular institutional setting is more commonly found among peers. We also show that the degree of central bank independence is an important determinant of the decision to concentrate financial services supervision in the hands of the central bank; in countries with more independent central banks, politicians are less likely to unify financial sector supervision in the hands of the central bank, since they fear the creation of a superpower bureaucratic institution. Our results support the view that the traditional theory of central banking has to be integrated with political economy considerations. Keywords: Financial Supervision, Central Banking, Central Bank Independence, Political Economy, Banking Supervision. JEL classification: E58, E63, G18. address: donato.masciandaro@unibocconi.it. Corresponding author. address: davide.romelli@essec.edu. 1

3 1 Introduction What explains the reforms in the architecture of financial supervisory authorities? The creation, throughout the 1990s and early 2000s, of unified supervisory institutions independent from the central bank has generally been associated with the reputational failures of many central banks during banking crises (Masciandaro, 2006; Masciandaro and Quintyn, 2009). Yet, following the global financial crisis, 14 countries actually increased their degree of central bank involvement in financial supervision, creating a sort of great reversal towards prudential supervision in the hands of central banks (Dalla Pellegrina et al., 2013). A classic example of this reversal is the evolution of supervisory architecture in the United Kingdom between 1997 and In 1997, when the UK parliament voted to give its central bank operational independence with a clear objective of price stability, the responsibility for banking supervision was transferred from the Bank of England to the Financial Services Authority (FSA). However, the supervisory failure of the FSA during the recent crisis led to its dismissal in 2013 and the supervisory authority was assigned to the newly established Prudential Regulation Authority, a subsidiary of Bank of England. This evolution in supervisory architecture towards more central bank involvement is not necessarily a European Union trend. In 2013, for example, Russia and Qatar also unified the supervisory structure of financial services in the hands of their central banks. Figure 1 depicts this trend towards a more unified supervision in the hands of central banks in a sample of 100 countries between 1996 and 2013, where higher central bank involvement corresponds to darker colors. In this paper, we look at what triggers such reforms in the architecture of financial sector supervision, with a special interest on the importance (if any) of systemic financial crises. To do so, we create a new dataset containing information on who supervises the financial sector including banking, insurance and financial markets in a large sample of 100 countries during the period We identify the full set of reforms implemented in supervisory architecture in our sample of countries and develop a new index of concentration of supervisory power in the hands of the central bank. Previous literature on the institutional structure of supervisory authorities has mainly been concerned with two issues: (i) whether banking supervision should be assigned to the central bank or to an independent authority (Goodhart and Schoenmaker, 1995; Bernanke, 2007) and (ii) which are the drivers that motivate the choice of countries to adopt an unified or a decen- 2

4 Figure 1: Evolution of unified supervision inside the central bank ( ) Notes: Figure presents the evolution of the E-CBFA index of central bank involvement in supervision constructed in this paper. Darker colors correspond to higher involvement. tralized financial sector supervision (Masciandaro, 2006, 2009; Melecky and Podpiera, 2013). In this paper we build a bridge between these two approaches by investigating why and how countries adopt a unified financial sector supervision inside the central bank. The new index of supervisory concentration we propose measures the degree to which central banks are involved in the supervision of the banking, insurance and securities markets sectors. We find that, over the period , a large number of countries undertook reforms in financial supervision structure. Consequently, our main concern rests in understanding what drives countries to modify their supervisory architecture over time. Our analysis points to two main drivers for these reforms. First, we find that financial crises episodes significantly increase the probability that a country reforms its supervisory structure. Second, we highlight an equally important bandwagon effect among central banks. We show that countries are more likely to undertake reforms when their peers do so or when their institutional setting is less commonly found among other central banks. We also investigate the determinants of a particular architecture of financial supervision. Previous literature has identified a series of institutional and economic factors that might ex- 3

5 plain a unified supervisory architecture (see Melecky and Podpiera, 2013). We focus on the determinants of a unified supervision inside the central bank. In line with previous findings, our results show that the degree of central bank independence, economic development and several country specific characteristics are relevant in influencing the decision to concentrate, over time, the supervision of the financial sector in the hands of the central bank. The outline of the paper is as follows. Section 2 overviews the literature on the topic and motivates our approach. Section 3 discusses the methodology followed in building the index of supervisory unification inside the central bank. In Section 4 we discuss the empirical strategies followed and the data. Section 5 presents the main results, while Section 6 concludes. 2 Supervision and central banking: state of the art A large literature has studied the consequences of central banks involvement in banking supervision on monetary policy outcomes and central bank independence. This literature, however, does not provide a clear solution on the optimality of assigning banking supervision to the monetary authority or to an institution different from the central bank. Historically, banking supervision has not always been entrusted to central banks (Ugolini, 2011). However, cyclical patterns do exist. In the decades before the Great Moderation, several central banks were strongly involved in supervisory activities, which were considered thoroughly integrated with the overall responsibility of central banks to manage liquidity (see Toniolo, 2011). Throughout the Great Moderation period, however, a decrease in the involvement of central banks in banking supervision was mostly seen (Masciandaro and Quintyn, 2009). Recently, and in particular after the financial crisis, the reputational failures of many supervisory institutions have reinforced the idea that banking supervisors need the market expertise and professional economists of central banks and could be more efficient as a built-in function of central banking (Goodhart, 2008). Thus, a shift in the general perception of monetary policy institutions also occurred, with central banks being nowadays perceived as public policy institutions with the goal to promote both monetary and financial stability. Given this historical evolution of banking supervision, two conflicting views regarding the merger of monetary and supervisory functions inside the central bank exist: a) the integration and b) the separation view (Masciandaro, 2012). In the first view, central banks involvement in banking supervision is usually supported by arguments related to the informational advantages 4

6 and economies of scale derived from bringing all functions under the authority of the central bank. Furthermore, given that macroeconomic and supervisory goals are interdependent, a single agency responsible for both objectives might be better able to consider these interdependencies (Bernanke, 2007). Finally, the human capital employed in central banks (highly skilled) could implement a more effective supervision. On the opposite side, according to the separation view, the risk of policy failures is higher if the central bank has supervisory responsibilities. At times, central banks might deviate from implementing the optimal monetary policy in order to avoid possible adverse effects on the banking sector (Goodhart and Schoenmaker, 1995). Empirical literature is also inconclusive on the superiority of either the integration or the separation view. For example, Arnone and Gambini (2007) find evidence in support of the integration view by highlighting the positive link between the compliance with the Basel Core Principles of supervision and the integration of supervisory powers inside the central bank. Moreover, Peek et al. (1999) show that having supervisory information available improves the efficiency of the monetary policy function, while, more recently, Barth et al. (2013) find that assigning banking supervision to an institution independent from the government enhances banks efficiency. On the other hand, Di Noia and Di Giorgio (1999) support the separation view by showing that inflation rates are higher and more volatile in countries where only the central bank is in charge of banking supervision. Looking at the United States, Ioannidou (2005) finds that the FED s monetary policies do alter its banking supervisory activity. Dincer and Eichengreen (2012) find evidence that nonperforming loans are lower if banking supervision is assigned to an independent authority different from the central, while capital ratios are higher when the central bank is the supervisor. Apart from these conflicting predictions regarding the optimal involvement of central banks in banking supervision, there is little research investigating the role of central banks as supervisors of the entire financial sector, i.e. banking, insurance and securities markets. Given the growing international financial integration and the creation of financial conglomerates, the concept of supervision cannot be focused exclusively on banking supervision. The interplay between banks, insurance companies and financial markets poses, indeed, new challenges for the institutional settings of financial supervisors (Masciandaro and Quintyn, 2013). As a result, several observers suggest that central banks should be involved in the supervision and regulation of all the institutions able to create credit and liquidity (De Grauwe, 2008). 5

7 Arguments in favor of a unified supervision of the financial sector rest on the fact that a unified regime could be associated with a higher degree of compliance with internationally accepted standards in banking, insurance and securities regulation (Arnone and Gambini, 2007; Cihak and Podpiera, 2007). Melecky and Podpiera (2013) find that more developed, small open economies characterized by better public governance are more prone to integrate their supervision. At the same time, they find that countries where financial sector supervision appears less integrated are associated with a higher degree of central bank independence. Evidence against central bank involvement in supervision is provided in Gaganis and Pasiouras (2013) who survey 3386 commercial banks in 78 countries and find that banks efficiency decreases as the number of sectors supervised by the central bank increases. However, their analysis does not cover recent financial crisis; therefore it is difficult to infer if higher efficiency was connected to better or worse supervision. Finally, Masciandaro et al. (2013) analyze empirically the impact of supervisory architecture and governance on economic resilience for a broad sample of 100 countries. Their results suggest that the unification of supervision inside an independent authority is negatively correlated with resilience, while no significant impact is found when looking at the degree of central bank involvement in supervision. Overall, previous approaches do not provide clear evidence as to whether a unified financial sector supervision ought to be assigned to the central bank or another authority. In this paper, we take the analysis one step forward, by investigating the main drivers of reforms that modify the involvement of central banks in supervision. In doing so, we bring together the literature on the role of central banks in supervision with the one on the unification of financial sector supervision, by analyzing how and why financial sector supervision might be focused in the hands of the central bank. In a political economy model, Masciandaro (2009) argues that the choice of supervision inside the central bank depends on the type of policymakers involved. If the central bank is already highly independent, granting the unified supervisory power to this institution would increase the risk of bureaucratic misconduct. Our empirical investigation provides a direct test for this hypothesis by looking at how central bank independence might influence the degree of financial sector supervisory unification inside the central bank. This complements recent research that generally finds a negative link between the degree of central bank independence and supervisory unification inside or outside the central bank (Masciandaro, 2006, 2009). 6

8 3 Supervision and central banking: metrics A recent literature on financial supervision has proposed several indices of financial sector supervision unification, as well as measures of central banks involvement in financial supervision. In this paper, we follow the approach proposed by Masciandaro (2006) who builds two indexes: one measuring the degree of financial sector supervision unification (dubbed FSU index) and one characterizing the degree of central bank involvement in financial supervision (CBFA index). We extend the CBFA index by (i) enlarging the different options for supervisory architecture and (ii) tracking the evolution of central banks supervisory involvement over time. We construct this new index, called E-CBFA (Extended Central Bank as Financial Authority Index), for a sample of 100 countries from 1996 to Our methodological steps are as follows. First, we identify which is the authority in charge of the supervision of the following three sectors: a) banking, b) insurance, and c) securities markets. Whenever we find that the central bank is the supervisor of the banking sector, we ask whether its responsibilities are shared or not with other authorities. Next, we look at whether the central bank is also involved in the supervision of one or both of the other sectors. Finally, we transform this qualitative information into quantitative indicators, by assigning a value to the degree of supervisory unification in the hands of the central bank. The new E-CBFA index distinguishes among the following levels of unification: A) A unified supervision inside the central bank (7 points). B) A unified supervision of the banking and securities markets sectors inside the central bank (6 points). C) A unified supervision of the banking and insurance sectors inside the central bank (5 points). D) Only banking supervision is in the hands of the central bank (4 points). E) The central bank shares the supervision of the whole financial system with another authority (Twin Peaks system) (3 points). F) Banking supervision is shared between the central bank and another authority (2 points). G) The central bank is not involved in supervision (1 point). 7

9 These levels of unification represent an extension to Masciandaro s (2006) FSU and CBFA indices, by combining the level of detail of the first and adopting a hierarchy similar to the second one, i.e. a higher score for higher central bank involvement in supervision. The E-CBFA index takes maximum scores in countries where the central bank is the unique supervisor and lower scores if supervision is assigned to an authority different from the central bank. Similarly to the FSU index, the E-CBFA assigns a higher value to a unified supervision of the banking and securities markets sectors inside the central bank given the predominant importance of these two sectors over insurance in every country. Finally, our index takes into consideration not only the presence of a twin peaks system in which there exist a separation between business and prudential supervision, but also the case in which banking supervision is shared between the central bank and another authority. Melecky and Podpiera (2013) also build an index that distinguishes between unified prudential supervision in the hands of a financial supervision authority or of the central bank. Their measure of supervisory unification assigns lower values for a sectorial supervision outside (1 point) or inside (2 points) the central bank and reaches the maximum value for unified supervision in the hands of the central bank (4 points). However, previous research shows that countries are more likely to follow a path dependence in assigning financial supervision inside or outside the central bank (Masciandaro, 2006). More specifically, countries characterized by a sectorial supervision outside the central bank are more likely to reform their supervisory architecture towards a unified supervision outside the central bank. On the other hand, whenever the monetary policy authority is already responsible for banking supervision, the move towards a unified financial sector supervision tends to place full supervisory powers in the hands of the central bank. For this reason, we consider a unified financial sector supervision inside or outside the central bank to be at the extreme opposite points of our index, an element that is not clearly distinguishable in Melecky and Podpiera s (2013) categorization. Looking at the evolution of the E-CBFA index during , we find that 75% of the countries included in our sample chose to reform their financial supervisory structure by establishing a new supervisory authority and/or changing the power of at least one of the already existing supervisors. In 35 of these reforms, the involvement of central banks in banking and financial supervision has been modified. 1 Figure 2 plots the changes in supervisory architecture, measured using the E-CBFA index, between 1996 and A clear trend towards an increas- 1 Appendix Table A1 shows the list of countries that modified their supervisory architecture by re-shaping the central bank involvement in financial sector supervision. 8

10 ing supervision in the hands of the central bank can be seen, given that positive changes in the value of the E-CBFA Index correspond to an increased concentration of supervisory powers inside the central bank. At the same time, this trend appears even stronger after the global financial crisis. Prior to this, most supervisory reforms undertaken reduced the degree of central bank involvement. However, this trend is reverted after the crisis, with most countries moving towards a higher concentration in the hands of the central bank (upper right-hand side quadrant). In fact, if we look at the 17 reforms that took place since the beginning of the recent crisis, we find that 14 of them increased the involvement of central banks in financial supervision. This is, of course, in line with the belief that financial crises might largely influence the decision to implement reforms in the supervisory architecture. Figure 2: Magnitude of reforms in E-CBFA Index ( ) Similarly, Figure 3 shows the institutional setting adopted by the countries that reformed their financial sector supervision between 1996 and Interestingly, in more than 40% of the reforms implemented before the crisis (8 out of 18), the central bank has been replaced by an independent unified supervisor or sectorial supervisors. After 2007, however, only two countries moved to supervision outside the central bank. On the other hand, out of the 17 reforms implemented since 2007, in 8 countries the central bank has become the unique supervisor of the financial sector. 9

11 Figure 3: Magnitude of reforms in E-CBFA Index ( ) Notes: Values reported next to each point indicate the number of countries that reformed their supervisory architecture in the year and adopted the specific level of central bank involvement in financial sector supervision. 4 Supervision and central banking: empirics Our empirical investigation has two aims. First, we are concerned with identifying the main drivers of reforms in supervisory architecture. To that end, we consider three sets of factors that could potentially impact the probability of reforming: (i) financial crises, (ii) bandwagon effects and (iii) domestic factors. We estimate the role of these factors on the conditional probability of having a reform in the architecture of supervisory authorities using the following specification: Prob(e it = 1) = F(φ Crises t β C + φt Bandwagon β B + φt Domestic β D ), (1) where e it is a reform dummy variable that takes the value 1 if country i is experiencing a supervisory reform that modifies the E-CBFA index in year t; φ Crises t episodes; φ Bandwagon t is a dummy for crises captures different proxies for bandwagon effects; and φ Domestic t is a vector of country-specific characteristics. The appropriate methodology to estimate Equation (1) is determined by the distribution of the cumulative distribution function, F( ). Because episodes occur irregularly (97.5% of the sample is zeros), F( ) is asymmetric. Therefore, we estimate 10

12 Equation (1) using the complementary logarithmic (or cloglog) framework 2, which assumes that F( ) is the cumulative distribution function (cdf) of the extreme value distribution. In other words, this estimation strategy assumes that: F(z) = 1 exp[ exp(z)]. (2) The choice of explanatory variables reflects the theoretical and empirical literature reviewed in Section 2. The first set of explanatory variables is represented by a crisis dummy that signals the presence of a systemic banking crisis in the last two or five years (as in Masciandaro, 2009). The date of the crisis comes from Laeven and Valencia (2013). Financial crises might indeed signal the possible supervisory failure of a certain architecture and previous literature considers these events as a reason for reforming the supervisory institutional setup. However, the impact of financial crises on the degree of supervisory consolidation is still a puzzling issue. Second, we argue that the probability of reforming financial supervision architecture is connected to a so-called, bandwagon effect (Masciandaro et al., 2008). Indeed, a high level of cooperation between central banks might stimulate a process by which these institutions learn from and follow the policy changes implemented by their peers (Borio et al., 2011). We construct four proxies for this effect. The first measure of bandwagon effects, called Same E-CBFA Index (World) indicates the number of countries all around the world that are characterized by the same financial supervision architecture as country i in year t. This variable can be interpreted in the following way: whether there exist any sort of bandwagon effects in the architecture of financial sector supervision, the probability that country i undertakes a reform in year t is negatively related with the number of countries that are currently adopting the same supervisory architecture as country i (same E-CBFA). To put it differently, when the supervisory system of country i is the same in many other countries, country i is less likely to implement a reform. The more fashionable a supervisory architecture is, the less interested a country will be in implementing a reform to modify it. Similarly, the variable Same E-CBFA Index (Continent) looks at the number of countries, in the same continent, that are characterized by the same financial supervision as country i in year t. 2 This methodology represents an alternative to logit and probit models and is typically used when the positive (or negative) outcome is rare (i.e. the number of zeros is large). This is also the case here since reforms do not happen that often in our sample of 1400 country-year observations. 11

13 If these first two variables provide information on the number of countries that adopt the same financial supervisory architecture in a certain year, the other two provide an indication of the popularity of undertaking reforms. More specifically, the variable Reforms in E- CBFA (World), measures the number of countries that are undertaking a supervisory reform (that modifies E-CBFA) all around the world in year t. If reforms in financial supervision architecture are fashionable in a certain year, this should make it more likely for a country to reform its supervisory architecture as well. Similarly, we define the variable Reforms in E- CBFA (Continent), that indicates the number of countries that are undertaking a supervisory reform in year t and are located in the same continent as country i. Thus, given the four different measures that capture a bandwagon effect, we expect a negative relationship between the probability of a reform and the first two variables and a positive correlation between the likelihood of a reform and the last two variables. Among the set of country characteristics, we consider the degree of central bank independence (CBI) and its evolution over time as a potential determinant of the probability of reforms. To account for the effect of CBI, previous literature mainly employs the CBI index constructed by Arnone et al. (2009) which assesses the degree of CBI at the end of 2003 (see Masciandaro, 2006; Melecky and Podpiera, 2013; Dalla Pellegrina et al., 2013, among the others). Given our panel setting, we employ the dynamic CBI index computed by Romelli (2014) based on the Grilli et al. (1991) (GMT) index. 3 Furthermore, given that reforms in central bank legislation might also influence the degree of central banks involvement in supervision, we also compute two dummy variables (CBI reform) that capture changes in the degree of central bank independence. Masciandaro (2009) builds a political economy model to study the determinants of supervisory architectures and finds that, in general, the quality of public sector governance plays an important role in shaping supervisory institutional architecture. The model acknowledges that the behavior of policymakers (politicians) might influence supervisory unification. Based on these arguments one can expect that changes in the political orientation of the government might stimulate the implementation of reforms. We capture this effect through a dummy variable that proxies changes in the political orientation of the government which took place up to two years prior to a reform in supervisory structure. We further consider a governance and a democracy index as two other political economy variables that might influence the likelihood 3 Details on how the different CBI indices are computed are provided in Appendix Table A2, while the advantage of using the Grilli et al. (1991) CBI index is stressed in Dalla Pellegrina et al. (2013). 12

14 of supervisory reforms. Other country-specific controls include a proxy for the degree of economic development and two variables capturing the possible effect of the legal origins hypothesis. We include a dummy variable that indicates the set of countries that belongs to the OECD to disentangle if more developed economies experience a higher probability of reforming the degree of central bank involvement in supervision. Finally, in line with previous research such as Masciandaro et al. (2008), we also consider the legal origin hypothesis and introduce several dummies for countries legal systems (La Porta et al., 1999). Our second methodological approach looks at the determinants of central banks involvement in financial supervision. As compared to the previous approach we are now interested in the determinants of the level of the E-CBFA index and not of the probability that it is changed. Previous literature has proposed several factors that could explain why countries might prefer a certain degree of supervisory concentration. We follow the empirical strategy of Dalla Pellegrina et al. (2013) and consider a set of economic, geo-political and cultural elements as determinants of the level of the E-CBFA index. Most importantly, we consider to what extent the degree of central bank independence and its evolution over time impacts the choice of supervisory architecture. All explanatory variables considered are detailed in Table A2 in the appendix. 5 Empirical results In this section, we first discuss the results obtained by analyzing the drivers of the likelihood of financial sector supervision reforms. In the second part, we present the estimations for the determinants of the level of central bank involvement in financial supervision. Finally, we present several robustness checks for the second empirical strategy. 5.1 Reforms in financial supervision Tables 1 and 2 present the cloglog estimation of the probability of reforms of the E-CBFA index in Equation (1). Columns (1) and (2) in both tables provide the baseline regressions which include the impact of financial crises and the four alternative bandwagon variables on the probability of implementing a reform. Table 1 considers the first two bandwagon variables that capture the number of countries characterized by the same financial supervision architecture as 13

15 country i in year t, while Table 2 captures the other two proxies for the bandwagon effect, i.e. the number of reforms in E-CBFA in year t. We find a positive correlation between the financial crisis dummy and the likelihood of reforms in central banks involvement in financial supervision in most of the estimations. Thus, in line with popular belief, experiencing a systemic banking crisis in the two years prior to a reform increases the probability that countries change their supervisory architecture. 4 Our results also show an interesting fashion effect among central banks. The bandwagon effects variables in Table 1 are negatively correlated to the E-CBFA reform dummy and strongly significant across all specifications. Thus, countries seem more inclined to change their supervisory architecture if their institutional setting is less used among their peers. At the same time, the bandwagon variables in Table 2 are positively correlated with the probability of a supervisory reform. This implies that more reforms will be undertaken in periods in which other countries are also reorganizing their supervisory architecture. Columns (3) to (6) in Tables 1 and 2 augment the basic estimation by introducing the dummy that captures the central bank legislative reforms that modified the degree of central bank independence (CBI reform), as well as an electoral cycle dummy. We find that central bank legislative reforms that modify the GMT independence index increase the likelihood of reforms. These results might be, however, influenced by the fact that this index also provides information on the involvement of central banks in banking supervision. In order to overcome these possible endogeneity problems, we provide a restricted version of this index (GMT bis), which is not influenced by the dynamics of the banking supervision architecture. The positive and statistically significant relationship between this index and the likelihood of reforms, present in most of the regressions, confirms the important role played by central bank legislative reforms in explaining the occurrence of supervisory changes. Finally, the last columns of these tables include other country characteristics commonly used in the literature. We find no evidence that government changes, good political governance, the level of democracy or the degree of economic development play an important role in explaining supervisory reforms. The legal origin hypothesis appears to matter in some cases with both civil and common law systems having a lower probability of reforming their central banks involvement in financial supervision. It should also be noted that the significance of the variables of interest remains unchanged after introducing all these additional controls. 4 We have also checked the robustness of the estimations in Tables 1 and 2 when looking at the occurrence of financial crisis in the last five years. Results are qualitatively unchanged and are available upon request. 14

16 Table 1: Determinants of Supervisory reforms (Bandwagon effects = Same system) Expl. vbs: Dependent variable: E-CBFA Reform (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) 15 Financial crisis impact (2y) * ** * * ** ** ** ** *** (0.482) (0.485) (0.517) (0.530) (0.484) (0.488) (0.514) (0.525) (0.491) (0.498) Same E-CBFA Index (World) *** *** *** *** *** (0.018) (0.019) (0.018) (0.018) (0.017) Same E-CBFA Index (Continent) *** *** *** *** *** (0.060) (0.059) (0.060) (0.052) (0.052) CBI reform (GMT) *** *** *** *** (0.445) (0.451) (0.438) (0.429) CBI reform (GMT bis) ** ** * * (0.639) (0.643) (0.632) (0.634) Government change (2y) (0.025) (0.023) (0.033) (0.033) (0.025) (0.022) (0.029) (0.032) Governance (0.491) (0.512) (0.491) (0.521) Polity (0.062) (0.061) (0.061) (0.061) OECD (0.736) (0.724) (0.713) (0.711) Common Law ** *** *** *** (0.685) (0.733) (0.669) (0.719) Civil Law *** *** *** *** (0.569) (0.600) (0.546) (0.573) Nr of observations 1,377 1,377 1,254 1,254 1,254 1,254 1,212 1,212 1,212 1,212 Nr of countries LR chi 2 /Wald chi Log likelihood Pseudo R Notes: Standard errors in parentheses; *** p<0.01, ** p<0.05, * p<0.10. Constant terms are included in all regressions, but not reported.

17 Table 2: Determinants of Supervisory reforms (Bandwagon effects = Nr of reforms) Expl. vbs: Dependent variable: E-CBFA Reform (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) 16 Financial crisis impact (2y) * ** ** ** * * (0.485) (0.512) (0.485) (0.519) (0.484) (0.522) (0.506) (0.529) (0.490) (0.514) Reforms in ECBFA (World) *** *** *** *** *** (0.166) (0.178) (0.168) (0.185) (0.169) Reforms in ECBFA (Continent) *** *** *** *** *** (0.212) (0.221) (0.215) (0.232) (0.222) CBI reform (GMT) *** *** *** *** (0.421) (0.436) (0.446) (0.455) CBI reform (GMT bis) * * (0.630) (0.642) (0.639) (0.641) Government change (2y) (0.025) (0.024) (0.031) (0.025) (0.021) (0.024) (0.026) (0.024) Governance (0.439) (0.431) (0.446) (0.441) Polity (0.054) (0.055) (0.055) (0.056) OECD (0.697) (0.709) (0.705) (0.711) Common Law (0.579) (0.603) (0.606) (0.628) Civil Law *** * *** ** (0.506) (0.540) (0.529) (0.563) Nr of observations 1,377 1,377 1,254 1,254 1,254 1,254 1,212 1,212 1,212 1,212 Nr of countries LR chi 2 /Wald chi Log likelihood Pseudo R Notes: Standard errors in parentheses; *** p<0.01, ** p<0.05, * p<0.10. Constant terms are included in all regressions, but not reported.

18 This first analysis provides strong evidence of both an effect of crises as well as a bandwagon effect among central banks as important determinants of reforms in the central bank involvement in financial sector supervision during the period As previously argued, financial crises might indeed show the supervisory failure connected to the adoption of a particular financial supervision architecture. Our results confirm this belief. However, our analysis also highlights an equally important element in the choice of supervisory structure in a country, which is the peer effect among central banks. 5.2 Determinants of supervision inside the central bank Results regarding the determinants of the degree of central bank involvement in financial sector supervision are presented in Table 3. We follow previous approaches such as Dalla Pellegrina et al. (2013) who investigate the factors that influence the likelihood of policymakers assigning banking supervision to central banks. In contrast to Dalla Pellegrina et al. (2013) who look at the central bank s involvement in banking supervision in 2010, we employ a panel analysis that looks at the determinants of the concentration of financial sector supervision, not only the banking sector, in the hands of central banks over the period Since this approach might be subject to possible endogeneity problems, we lag most time series variables by one period. We present our main results in Table 3. In Columns (1 4) we employ Dalla Pellegrina et al. s (2013) empirical strategy using our new measure of financial system supervision concentration as dependent variable. We also re-estimate their analysis using the CBFA Index proposed by Masciandaro (2006) in a panel setting in Columns (5 8). 5 Table 3 provides evidence of the role played by central bank independence in explaining the level of central bank involvement in financial supervision. We use several definition of central bank independence including the classical political (GMTp) and operational (GMTo) indices in Columns (1) and (5), as well as more restricted versions of the two in Columns (2 4) and (6 8), respectively. Our results show the negative effect of operational independence on the degree of central banks involvement in financial supervision. On the contrary, higher political independence has a positive and significant effect only when considering the broad GMTp indicator. These results support the idea that more powerful the supervisor (as 5 We consider however a different specification for the religion dummies, since we take into consideration the share of adherents to a particular religion in the total population of the country. See Appendix Table A2 and Dalla Pellegrina et al. (2013) for more details on all these control variables. 17

19 Table 3: Determinants of Unified Supervision inside the Central Bank Expl. vbs: Dependent variable: E-CBFA Index Dependent variable: CBFA Index (1) (2) (3) (4) (5) (6) (7) (8) L.GMTp ** ** (0.523) (0.517) L.GMTo *** *** (0.529) (0.523) L.GMTp bis (0.507) (0.504) (0.508) (0.506) L.GMTo bis *** ** (0.474) (0.480) L.GMTp ter (0.410) (0.413) L.GMTo ter *** *** *** *** (0.408) (0.395) (0.411) (0.400) L.Unemployment * * * (0.020) (0.020) (0.020) (0.020) (0.020) (0.020) (0.020) (0.020) OECD (0.509) (0.524) (0.523) (0.522) (0.497) (0.507) (0.507) (0.508) L.CB Assets/GDP (0.019) (0.020) (0.020) (0.019) (0.019) (0.019) (0.019) (0.019) L.Corruption * *** *** *** (0.218) (0.224) (0.224) (0.224) (0.217) (0.224) (0.224) (0.224) Buddhist % ** ** (2.156) (2.327) (2.329) (2.321) (2.020) (2.161) (2.162) (2.165) Catholic % *** *** *** *** *** *** *** *** (1.216) (1.290) (1.291) (1.282) (1.229) (1.303) (1.302) (1.302) Muslim % ** *** *** *** ** ** ** ** (1.105) (1.195) (1.194) (1.189) (1.053) (1.130) (1.128) (1.129) Orthodox % ** ** ** ** *** *** *** (1.121) (1.139) (1.141) (1.137) (1.084) (1.105) (1.107) (1.107) Protestant % *** *** *** *** *** *** *** *** (1.581) (1.601) (1.607) (1.597) (1.663) (1.675) (1.676) (1.672) Latitude (2.799) (3.094) (3.097) (3.083) (2.613) (2.829) (2.827) (2.832) Observations 1,117 1,117 1,117 1,117 1,117 1,117 1,117 1,117 Number of ctyid LR chi 2 /Wald chi Log likelihood Pseudo R Notes: Standard errors in parentheses; *** p<0.01, ** p<0.05, * p<0.10. Constant terms and continent dummies are included in all regressions, but not reported. an independent central bank is supposed to be), the greater the fear of powerful institutions or bureaucratic misconduct (Masciandaro and Quintyn, 2013). This implies that in countries with more independent central banks, politicians are less likely to unify financial sector supervision in the hands of the central bank, since they fear the creation of a super-powerful bureaucratic institution. Regarding the other possible determinants of the level of supervision concentration, we find that past unemployment rates have a positive impact on the level of E-CBFA. These results, read together with the negative sign of the OECD dummy variable, show that developing countries are more likely to assign supervisory powers to central banks. Finally, the negative sign of the corruption index signals how, in more corrupted countries, central banks also tend to assume financial services supervision responsibilities. In such environments, the central bank, given its international exposure, might be considered as one of the few institutions able to escape 18

20 possible political captures or corruption pressures in the country. 5.3 Robustness checks We verify the robustness of our results in several ways. First, we consider some of the institutional, economic and financial sector characteristics proposed by Melecky and Podpiera (2013) to explain the level of E-CBFA. We present these results in Table 4, while those pertaining the CBFA index are presented in Appendix Table A3. Table 4: Determinants of Banking Supervision inside the Central Bank Robustness checks Expl. vbs: Dependent variable: E-CBFA Index (1) (2) (3) (4) (5) (6) (7) L.Governance *** *** * (0.246) (0.281) (0.417) (0.443) (0.313) (0.329) L.GMTp *** * *** (0.432) (0.755) (0.526) L.GMTo *** *** *** (0.457) (0.733) (0.558) L.GMTp bis (0.423) (0.747) (0.509) L.GMTo bis ** *** *** (0.430) (0.757) (0.499) L.Crisis cumulative *** *** *** *** *** (0.135) (0.247) (0.247) (0.186) (0.185) L.Real GDP per capita *** *** *** *** (0.000) (0.000) (0.000) (0.000) L.Population (0.000) (0.000) (0.000) (0.000) L.Openness *** *** *** *** (0.004) (0.005) (0.003) (0.003) L.Private credit to GDP (0.005) (0.005) L.Non life premium * (0.213) (0.217) L.Stock market capitalization (0.003) (0.003) (0.002) (0.002) L.Nr of listed companies *** *** (0.000) (0.000) L.Concentration (0.006) (0.006) L.Cost to income ratio (0.007) (0.006) L.Private credit to deposit ratio *** *** ** (0.383) (0.399) (0.247) (0.247) Observations 1,496 1,496 1, ,041 1,041 Number of ctyid LR chi 2 /Wald chi Log likelihood Pseudo R Notes: Standard errors in parentheses; *** p<0.01, ** p<0.05, * p<0.10. Constant terms are included in all regressions, but not reported. We find evidence that the level of financial sector supervision concentration in the hands of the central bank is influenced by: the central banks operational independence, previous 19

21 crises, real GDP per capita, openness and financial sector development. The significant and negative coefficient of the degree of operational independence of the central bank suggests, as already shown in Table 3, that higher central bank involvement in supervision is a less preferred outcome from the point of view of a more economic independent central bank. Moreover, contrary to Melecky and Podpiera (2013), we find that a country s level of development (GDP per capita) has a negative effect on unified supervision, meaning that the concentration of supervision in the hands of the central bank is more common in less developed countries. This result can be read together with the positive coefficient found for the degree of openness to trade of a country. Indeed, smaller economies which tend to have also a higher degree of openness to trade are more likely to have fewer institutional authorities and thus integrate the supervision inside the central bank. Supporting this idea, we also find that less financially developed countries are characterized by a higher concentration of supervisory powers in the hands of the central bank. As a final empirical exercise, we also re-estimate Dalla Pellegrina et al. s (2013) empirical strategy in a panel setting. This will allow us to compare whether the determinants of banking supervision inside the central bank are similar to ones used to explain the concentration of financial services supervision. Results presented in Appendix Table A4 confirm that the indexes of supervisory concentration are related to similar socio-economic characteristics. 6 Concluding remarks The question of whether central banks might be involved in financial sector supervision has become a key issue following the global financial crisis. It is generally argued that crises episodes signal possible supervisory failures of the institutional architecture of financial sector supervision. However previous studies do not find a clear link between the occurrence of financial crises and the design of financial supervisory authorities. In this paper, we overcome some of the limitations of previous research using a novel dataset that tracks the evolution of reforms in the supervisory architecture of the financial sector including banking, insurance and securities markets. We build a new index that measures the degree to which the unified supervision of the entire financial sector is concentrated in the hands of the central bank. Using a panel of 100 countries we find that, over the period , systemic bank- 20

22 ing crises are important drivers of reforms in supervisory architecture. However, a bandwagon effect among central banks appears to be equally important. This shows that the probability of reforming ones institutional setting is positively related to the number of reforms other countries are undertaking in the same period. Similarly, a country is less likely to change its supervisory architecture if this architecture is popular among its peers. Our empirical strategy also extends the literature that looks at the determinants of a unified supervisory authority (Masciandaro, 2009; Melecky and Podpiera, 2013). We depart, however, from previous works by considering the determinants of a unified supervision in the hands of the central bank. We confirm previous findings and highlight the important role of the degree of central bank independence in influencing the decision to concentrate the supervision of financial institutions inside the central bank. While our research highlights important determinants of the probability of reforms in financial supervision, we do not take a stand on whether a higher degree of central bank involvement in supervision might influence the stability or efficiency of the financial sector. Employing our new index of financial supervision architecture, future research could be directed towards understanding the macroeconomic effects of moving towards a unified supervision inside the central bank. 21