New Financial Stability Governance Structures and Central Banks*
Abstract:
JEL Codes: H19, G18, G28, E58.
Keywords: Financial stability committees, Macroprudential policy, Countercyclical capital buffer, Central bank independence.
1 Introduction
Since the global financial crisis, countries have been strengthening their microprudential policy regimes and creating or enhancing frameworks for macroprudential policies directed at system-wide or macro-level risks. Such macroprudential frameworks were laid out in some previous discussion documents (see, for example, the IMF, 2011, CGFS, 2010, and Bank of England, 2009). These documents emphasize that the ultimate objective of macroprudential policy is the stability of the financial system as a whole across a wide range of likely macroeconomic and credit market backdrops. The documents describe three components of macroprudential policy frameworks, specifically: (1) measuring and monitoring systemic risk; (2) implementing policies to mitigate identified systemic risks; and, (3) establishing an institutional and governance structure for implementing policy.
In this paper, we present new evidence on the institutional and governance structures for macroprudential policies. Institutional design is especially important for macroprudential policies because, in contrast to microprudential policies, macroprudential policies consider financial risks that can span many types of financial intermediaries, as well as interactions between the financial system and the real economy. In addition, time-varying macroprudential policies are expected to build resilience in anticipation of possible future shocks, not in response to current shocks, and so need to be preemptive. As a result, effective implementation should involve mechanisms to cover the broad financial system, and to avoid the risk of policy inertia. As emphasized by Peter Conti-Brown in describing the Federal Reserve, "Having the right institutional design...isn't a side show to the real questions of monetary policy and financial regulation. Governance may in fact be the whole show." (2016, p. 26).
We build a dataset of financial stability governance structures for 58 countries from official sources that are available to the public, updated through mid-2018. We document a dramatic growth in multi-agency financial stability committees (FSCs) since the global financial crisis (figure 1). Only 11 countries had FSCs for macroprudential purposes in 2008, whereas 47 countries had FSCs in 2018. Over that same period, no countries created a new single regulatory agency with sole authority for macroprudential policies. This dramatic growth in FSCs indicates it is important to understand the role of these committees in implementing macroprudential policies.
There are several reasons for why multi-agency committees would be an effective governance structure for macroprudential policy. Prudential regulators should be involved since most policies will affect and be put into effect through regulated financial institutions. Central banks should be involved for their expertise in macroeconomic forecasting and in setting countercyclical monetary policy given they have some degree of operational independence for setting monetary policy. Elected officials could also have a role since macroprudential policies that are viewed as having elements of credit allocation, which would raise political economy considerations. For example, there may be a tradeoff between expanding homeownership and reducing rapid mortgage debt growth, by tightening loan-to-value ratios or raising the countercyclical capital buffer.
International organizations usually advocate for a prominent role for central banks. Central banks are designed to have longer time horizons than politicians or market participants and are therefore more willing to bear the near-term costs of policies when gains may not be realized within an election cycle. For example, the IMF argues central banks foster policy coordination between macroprudential and monetary policy and can "help shield macroprudential policy from political interference that can slow the deployment of tools" (2014, p. 34). The ESRB recommends that "the national central banks should have a leading role in macroprudential oversight because of their expertise related to setting policies for price and exchange rate stability, and existing responsibilities in the area of financial stability" (2011, p.2).
But there also are potential costs of prominent roles for central banks. Excess concentration of power in unelected officials is a principal concern (Goodhart, 2010, Tucker, 2014, 2016). In evaluating the changing responsibilities of central banks following the crisis, Goodhart (2010) writes that the "combination of operational independence to set interest rates and liquidity management together with prospective macroprudential regulation just vests too much power in a non-elected body."1 Tucker (2014, 2016) also argues that while central banks have the skills and the independence, institutional structures should have a role for elected officials because macroprudential policies may affect credit cycles or have important distributional consequences. In contrast, the IMF has raised concerns about a prominent role for elected officials because it could risk delaying macroprudential actions and compromise the independence of regulatory agencies and the central bank that serve on the same committee (2014, p. 35).
We evaluate the institutional structures for macroprudential policy that exist in the 58 countries in our sample in light of these rationales and ultimately with a goal to determine their effectiveness to implement time-varying macroprudential policies. We document the institutional arrangements. No FSC exists in 11 of the 58 countries in our sample, which are mostly smaller countries, and instead either the central bank or the prudential regulator is designated the macroprudential authority. A FSC exists in 47 countries, and both prudential regulators and central banks are on all but one, consistent with a coordination motive. The ministry of finance, representing the elected government, is a formal member of all but a few FSCs, and is an informal observer in other FSCs where it is not a formal member. Moreover, the ministry of finance is more often the chair of the FSC than the central bank, suggesting countries value strengthening the political legitimacy for setting macroprudential policy and coordination across agencies. In any case, political leadership relative to a central bank leadership suggests a greater tendency to delay actions.
While most countries have established FSCs, they have generally not made significant changes to how the country would take macroprudential policy actions. Only three FSCs can directly implement countercyclical policies, and these three and ten others can issue "comply or explain" directives, in which an agency is expected to respond by taking the directed action or explain why it did not.2 Instead, most FSCs are either advisory - with the ability to issue warnings and non-binding recommendations but without the ability to take or direct actions of the member agencies - or operate purely to facilitate information sharing, communication, and policy coordination across agencies.
We use logit analysis to examine the probability a country creates a multi-agency FSC, and find that FSCs are more likely in larger countries and less likely in countries where the central bank is also the prudential regulator for banks and other parts of the financial sector. These results suggest that better communication and coordination is a main motive for creating FSCs since larger countries with more regulators will have more interested parties to coordinate. To provide additional insight into a country's decision to set up a multi-agency FSC, we look only at those countries where the central bank is also a prudential regulator, and find that the same factors determine the decision. We find that a central bank - rather than a FSC - is more likely a macroprudential authority in smaller countries and when the central bank is a regulator for a broader swath of the financial system, and when there are likely to be fewer regulators to coordinate.
Given the various characteristics of FSCs, we use cluster analysis to group countries according to the ability of their institutional structures to take actions. We find a cluster of 12 FSCs that appear to be designed to effectively take or direct actions, based on whether the FSC is formal, has a chair, a voting process, or tools to take actions. But 23 FSCs, almost half, lack two or more of these basic features, indicating they are less effective. We document that the 12 FSCs that are "more able" are in countries that are wealthier, have more developed financial sectors, stronger Rule of law, and more Checks and balances for government actions. These indicators suggest the more effective FSCs are set up in countries with the resources and infrastructure to do so.
Given that many countries lack effective FSCs, we then look at whether this might reflect that countries have another agency, with the macro expertise and independence to overcome policy inertia and be an effective macroprudential authority, namely the central bank. We find a distinct cluster of 18 central banks that do not have macroprudential authorities - based on the fact that they cannot set the countercyclical buffer (CCyB) or loan-to-value ratios (LTVs) - and are not a prudential bank regulator or a prudential regulator for more than banks. However, there is less clarity about a distinct cluster of strong macroprudential-oriented central banks. Instead, central banks that already are broad prudential regulators and have more political independence for monetary policy are not likely to be able to set LTVs, and those that are able to set LTVs are not a broad microprudential regulator. These characteristics are consistent with political economy considerations to not place multiple powers into a central bank. In addition, in a cross-tabulation of the CBs and FSCs, we do not find an inverse relationship between weaker FSCs and more able macroprudential CBs, suggesting many countries lack effective macroprudential authorities.
Our study is the first to focus specifically on the FSCs that have become very prevalent since the global financial crisis. It sheds light on countries' revealed preferences for creating FSCs that can more effectively implement policies versus political economy considerations, such as coordinating among existing regulators, avoiding concentration of power for financial policies in an independent central bank, and potential conflicts between macroprudential policies and other social objectives. We infer from the choices of structure, governance, and tools of institutional structures how countries value the competing considerations for implementing macroprudential policies.
Our results suggest that an effective policy implementation motive is supported by decisions to establish FSCs with both processes and good tools in one-quarter of the countries with FSCs, as well as decisions to designate a sole central bank authority in nine countries that are smaller and have fewer financial regulators. But most countries with FSCs appear to establish them mainly to facilitate information sharing and improve communication and coordination among agencies rather than to implement policies, and tools are controlled by the agencies raising issues about whether there is clear accountability among agencies and FSCs for financial stability. Moreover, we do not find evidence that central banks are especially able to take macroprudential actions when FSCs are not set up to do so, suggesting that institutional structures are not in place in many countries to effectively implement time-varying macroprudential policies. Rather countries' decisions appear to be consistent with preventing placing additional powers in central banks that already exercise independence in microprudential supervision and monetary policy.
Finally, we document an increased and substantial role of the political sector into macroprudential policymaking. While this greater role may reflect that objectives of macroprudential policies, such as to reduce credit growth, may come into conflict with other social objectives, it also suggests that many institutional arrangements are not set up to take unpopular actions and reduce the risk of policy inertia or inaction that can arise with time-varying macroprudential policies. An important implication of these results is that central banks when setting monetary policy should not assume that macroprudential authorities can also exercise independence.
We first review existing studies on best practices for implementing time-varying macroprudential policies and competing political economy considerations in section 2. We then describe the key characteristics of the institutional arrangements of macroprudential authorities in 58 countries, focusing heavily on the membership, leadership, and tools available to implement policies of FSCs in section 3. We present results from logit analysis of the decisions to establish FSCs, and cluster analysis for the ability of FSCs and central banks to take macroprudential policies in section 4. Section 5 concludes.
2 Ability to Implement Macroprudential Policies
The ability to implement macroprudential policies depends on the governance structure and the available tools of the responsible authorities. For this paper, we are interested in macroprudential policies that involve the dynamic adjustment of the parameters of financial regulatory policies, which combined with structural requirements such as enhanced capital standards, can reduce the economic costs of recessions and financial crises. While the arrangements to implement structural macroprudential policies are tied closely to existing microprudential regulators, the governance framework for time-varying macroprudential policies is new and less well-established. Given the increased prevalence of multi-agency FSCs, we look at whether FSCs appear to be set up to implement policies or to foster communication and coordination across agencies. We then infer from countries' decisions about FSC structures and central bank (CB) authorities whether the institutional framework can overcome the risk of policy inertia or inaction, and whether authorities and tools are aligned in a way to enhance accountability and effectiveness.
To assess best practices for governance, we draw on existing research about governance of financial policymakers. There is a general recognition that financial prudential regulators (PR) should have operational or instrument independence (though not necessarily political or goal independence) and their resources should not be subject to appropriations (Basel Core Principles of Effective Bank Supervision). Those principles should also be relevant for implementing macroprudential policies. The ESRB's guidance to countries on their macroprudential policy framework emphasizes the authorities should be shielded against outside pressures through independence.3
For CBs and monetary policy, the key issue in governance is how it can solve the time-inconsistency and inflation bias problem. In this case, the problem is that politicians have a preference for pushing the unemployment rate below its natural rate, perhaps because it makes them more likely to win elections (Kydland and Prescott, 1977; Barro and Gordon, 1983; Fischer, 1995). A CB that in setting monetary policy is heavily influenced by politicians would try to achieve this by generating more inflation than wage earners had expected (because doing so will mean that real wages will be lower and firms will employ more workers, unemployment rate will be lower than its natural rate). But inflation expectations are not fixed and the public would adjust upward their inflation expectations knowing that politicians have this incentive. As such, a CB heavily influenced by politicians will end up with higher inflation expectations and higher inflation - and thereby an upward inflation bias - without the benefit of fewer unemployed workers. An empirical negative correlation that has been documented for advanced economies between a CB's operational independence and inflation provides empirical support for CBs that are independent of political officials as a best practice for monetary policy (Arnone et al., 2007). For time-varying macroprudential policy greater political influence would lead policymakers to permit more buoyant financial conditions and higher levels of financial system vulnerabilities given that such developments would likely accompany lower unemployment rates. The public would expect this, though it is not clear that this expectation would obviate the effects on the unemployment rate as increased inflation expectations do for monetary policy.
Balls et al. (2018) offer a proposal for implementing macroprudential policy which builds from the governance for setting monetary policy. CBs have operational (instrument) independence but not political (goal) independence for monetary policy (there is political accountability in terms of mandate-setting, appointment of officials, and delegation of tools). A similar framework could be the basis for effective macroprudential policy. A committee chaired by the government, represented by the ministry of finance (MoF) could be responsible for identifying risks to financial stability and setting the priorities (or goals) for macroprudential policy. The CB bank would be operationally independent and responsible for setting policies (that is, using its macroprudential policy tools) to achieve a desired degree of financial stability set by the government. The two bodies would allow financial stability goals to be decided by politicians, providing legitimacy and accountability, and implementation by the CB would be more insulated from short-term political pressures and therefore mitigate policy inertia. Moreover, the CB is positioned to internalize tradeoffs between monetary and macroprudential policies.
However, there are arguments that greater involvement of CBs into macroprudential policy would compromise the ability of the CB to achieve its monetary policy objectives. In particular, the incentives of a CB that makes simultaneous monetary policy and macroprudential decisions can be distorted because of time-inconsistency and political pressures (see Ueda and Valencia, 2012, and Smets, 2014). Policymakers in this setting minimize a quadratic loss function for inflation and output variability, augmented with a loss term for leverage variability.4 When the CB has price stability as its sole objective, policy will be set to achieve the optimal level of inflation, and macroprudential policy also delivers the optimal level of output and leverage, knowing that if it is lax and allows debt to become excessive, monetary policy will not inflate away the debt by delivering higher inflation. But if monetary policy also is expected to target financial stability as well as price stability, it will have an incentive to inflate away debt. Knowing this, macroprudential policy will be lax, which will lead to higher optimal debt and again an upward inflation bias.
Smets (2014) argues that these time-inconsistency risks can be mitigated if there are separate objectives, instruments, communications, and accountability for price stability and financial stability, and there is information sharing between the two bodies. Another argument in favor of separating policies is that since crises cannot be eliminated despite good policymaking, the actual occurrence of a crisis could compromise the credibility and, in turn, independence, of the CB.
Clearly, there are arguments on both sides for the role of the CB as the macroprudential authority or as a member of a multi-agency committee. In this paper, we document the institutional structures that countries have adopted to meet new responsibilities. We update and expand results from an earlier study based on a 2010 survey (Nier et al., 2011), and we focus heavily on new multi-agency FSCs given their dramatic rise since that earlier study, rather than the role of the CB.5 Most previous papers have assumed that a stronger role for the CB, either as the main macroprudential authority or as a chair of a FSC, implied a more effective structure for macroprudential policies (see Lim et al., 2013, Mosciandaro and Volpicello, 2016, and Lombardi and Siklos, 2016). Below we look also at the combination of membership, leadership, and tools to evaluate the strength of the institutional structure.
There is mixed empirical evidence on benefits from a CB also having banking regulatory and supervisory authorities. Nier et al. (2011) find that the group of countries with close integration between CB and banking supervisory agencies have lower costs, in terms of failed banking assets, capital injections, and guarantees, than countries with separate arrangements. Goodhart and Schoenmaker (1995) found that there were significantly lower actual and expected bank failure rates in the 11 countries with an integrated regime than in the 13 countries with a non-integrated regime for a sample of 104 large bank failures in the 1980s to early 1990s. Merrouche and Nier (2010) found that the buildup of banking imbalances (measured by the ratio of loans to deposits) was less severe where the CB had full control of supervision and regulation. However, Koetter, at al. (2014) finds no improvement in the credit risk or non-performing loan ratio at banks when the CB is also the PR in 44 countries. None of these studies focus on the effects of CB interaction on the financial condition on the entire financial system. Lim et al. (2013) looks at the strength of institutional arrangements for macroprudential policy on how quickly actions are taken to moderate credit growth from 2008 to 2011 for a sample of 39 countries. They find a negative correlation between policy response time and the involvement of the CB, suggesting that including the CB is conducive to reducing policy response time. However, they focus on the strength of the CB without considering the role of the FSC as an authority.
3 Characteristics of Governance Structures for Macroprudential Policy
We collect data on governance structures for a sample of 58 countries (listed in table A.1 of appendix A), which is current as of June 2018. A brief outline of how we collected our data is provided in the first subsection and findings are described in the subsequent four subsections.
a. Sample and data sources
We started with the sample of 64 countries in the macroprudential policy tool database of Cerutti et al. (2016) that are identified as countries having used macroprudential policies in a time-varying way. We dropped seven countries from their sample because of limited information about the use of tools or governance structure (as highlighted in Cerutti et al.), but kept Saudi Arabia in order to preserve the full set of G-20 countries. We also added Cyprus, not in the Cerutti et al. database, in order to include the full set of EU countries. Note that in our analysis, we include separately each of the 19 countries in the euro area rather than treat the euro area as a single entity. As we document in subsections 2.b and 2.c, there is considerable heterogeneity in governance structures across these countries, so this treatment does not bias the results. Our process ultimately results in a sample of 58 countries, of which 28 are advanced economies and 30 are emerging market or developing economies, as categorized Arnone and Romelli (2013), which are consistent with the IMF's 2007 WEO report.6 This sample includes a significantly higher share of advanced economies than previous studies of macroprudential authorities (e.g, Lim et al., 2013).
The main sources for our information on countries' financial stability governance structures, safety and soundness authority responsibilities, and tool availability were national authorities' websites (and further documents referenced therein), national authorities' financial stability reports, IMF Article IV reports, and, where available, IMF financial sector assessment program (FSAP) reports. In addition, we undertook various cross checks, including comparing what we inferred about financial stability governance structures from our sources with Lombardi and Siklos (2016) on macroprudential policies, with Nier et al. (2011) on safety and soundness for microprudential policies, and with an appendix table on institutional structure in a recent IMF/FSB/BIS report (2016).7 For information about the availability of tools, we additionally consulted responses to the IMF's Global Macroprudential Policy Instrument (GPMI) survey for 2013 data. A large reason for our preference for national authority websites is the very recent nature of financial stability governance structures, which in turn means continued ongoing changes in these structures. Indeed, some changes have occurred as recently as the past year and since the first draft of our paper.8Building this database ourselves by drawing on national authority websites also allows us to apply a consistent categorization of governance structure characteristics across countries, which is not the case with self-reported survey responses.
b. Financial stability committees
We find that 47 of the 58 countries have formal or de facto financial stability committees (FSCs), and 11 countries do not have FSCs (see figure 2 and table 1). Of the 47 countries with FSCs, 35 countries have a FSC that has been created formally by legislation and 12 countries have a de facto FSC, which means that a committee exists and meets regularly but exists only from non-legal arrangements between the agencies, such as memorandums of understanding (MOUs). Of the 11 countries that do not have formal or de facto committees, all have assigned, at least in practice, macroprudential responsibilities to an existing agency, and of these, nine countries have the central bank (CB) as the macroprudential authority and two countries have the PR.9 These two cases are Peru and Finland, for which the PR is separate from the CB.
As we noted earlier there is considerable heterogeneity in governance structures across euro area countries, so that treating them as separate countries in our analysis does not bias the results. For example, with regard to whether euro area countries have a FSC, of the 19 in our sample, 14 have a formal or de facto FSC. Of the five that do not have formal or de facto committees, four have assigned macroprudential responsibilities to the CB and one has assigned it to the PR.
Even as FSCs have become more prevalent, their importance in terms of global activity is even greater since they tend to be more likely in the larger economies. When we weight FSC by economy size (real GDP), nearly all of the collective GDP of the 58 countries would be in the countries with FSCs (see figure 2). In particular, the 47 countries with FSCs account for 97 percent of the GDP of the 58 countries combined. That also means that the 11 countries without a FSC account for a very small share, 3 percent, of collective GDP.
A few of the 11 countries do not have a FSC and have made an existing agency responsible for macroprudential policy, nonetheless these countries' have informal information sharing and coordination arrangements in place among agencies. To our knowledge there are four such countries. In three - Belgium, New Zealand, and Singapore - of these four countries, informal arrangements exist between the CB, which is the authority with macroprudential responsibilities, and the MoF, indicating they do not act solely on their own. For example, in Belgium, the MoF needs to approve regulation for financial stability that is issued by the CB. Indeed, in Belgium the MoF rejected regulation to raise risk weights on some loans (according to the risk profile) that was proposed to it by the CB and asked the CB to undertake more analysis in the issue.10 In New Zealand there is a written MOU between the CB governor, who is responsible for macroprudential policy, and the minister for finance, which says that the CB governor must consult with the minister when macroprudential policy actions seem likely. In Singapore, where "stamp duties" have been an important policy tool to address rapidly increasing house-price valuations, informal consultative arrangements are in place between the CB and MoF. For Finland, where the PR is the existing agency with macroprudential responsibility, the PR and CB jointly conduct systemic risk monitoring and jointly prepare vulnerabilities analyses and preliminary recommendations on macroprudential tools for the PR director.
c. Financial stability committee membership and leadership
Most of the FSCs (39 of 47) have three to five member agencies, with a range of 2 to 9. A PR is always represented, either independently or as part of the CB, which is not surprising since most macroprudential tools would apply to regulated financial firms (table 1). The CB is on the FSC in all but one country - specifically Chile - although it has observer status. The MoF is on fewer FSCs, although in four of the seven countries where it is not a member - specifically, Malta, Portugal, Slovenia, and the U.K. - it participates in meetings as an observer. Market regulators, insurance company regulators, pension fund regulators, and deposit insurers also are frequent FSC members. Additionally, a handful of countries have other agencies on the FSC - such as, the country's accounting standards authority (France), its consumer protection regulator (South Africa and the United States), or other government departments (India).
The literature on monetary policy committees - such as that by Blinder (2007, 2008) - emphasized the need to capture a range of expertise and the desirability for consensus on the committee as considerations influencing the number of members on a policy committee. For FSCs, however, the exact number of agencies represented seem to largely reflect the number financial regulatory authorities that are already in existence in a country rather than these considerations. That is, if, in a country, there are a large number of agencies each with authority over a relatively limited part of the financial sector - that is, a separate bank regulator, insurance company regulator, pension fund regulator, and market regulator and a separate authority administering the deposit insurance fund - FSCs typically have a large number of members. In contrast, if, in a country, there are only a small number of agencies with authority for multiple parts of the financial sector - that is, the bank regulator also has authority for other financial institutions, perhaps also financial markets, and perhaps also the deposit insurance fund - FSCs typically have a small number of members. In both cases the range of expertise on the FSC would be about the same, and, although the ability to obtain consensus would be different, this seems to be a secondary consideration to financial sector coverage.
In addition, seven countries have independent members on the FSC; that is, committee members who are unaffiliated with any agency on the committee but have expertise on various financial sector topics. Independent members can bring additional expertise and an outside perspective. Their inclusion may also indicate that the FSC was created for the purpose of taking actions, rather than just coordinating among existing agencies.
Next, we look at the chair of the FSC to measure leadership, since the chair sets the agenda and often serves as the government's official voice on macroprudential policies. The MoF is the most frequent chair, either chair or co-chair of 25 FSCs. We interpret this role for the MoF as strengthening the political legitimacy of macroprudential policy relative to if the CB or PR had this role. Thus, the high frequency of MoF chairs would be at odds with a recommendation by the IMF to limit the participation of politicians due to the possibility of them delaying the implementation of time-varying policies. CBs are the next most frequent chairs, specifically the chair or co-chair of 18 FSCs, where in three cases they are co-chair with the MoF and in one they are co-chair with the PR. In no country is the PR the sole chair, but it is a co-chair with the CB in Switzerland. In seven other countries, there is no chair or the chair rotates, suggesting weaker leadership: In Romania and Brazil, the chair rotates between all members, in Ireland, Israel, Japan, and the Philippines, there is no FSC chair, while for Saudi Arabia we could not find any information about the FSC's chair.11
d. Financial stability committee powers
FSCs differ in terms of direct powers. Based on the Tinbergen separation principle (that Carrillo et al., 2017 document to be quantitatively material), we expect that a FSC with its own tools, could achieve better outcomes than if authority over tools were to remain at member agencies. While member agencies could implement macroprudential policies, they may also have other mandates which could be in conflict at times with financial stability. Moreover FSCs with their own powers increases transparency and accountability for financial stability from when tools are dispersed among member agencies.
Direct powers. Few FSCs have what the IMF/FSB/BIS report (2016) and the IMF's (2013) Key Aspects of Macroprudential Policy would consider as "hard" or "semi-hard" powers: Hard powers give policymakers direct control over macroprudential tools or the ability to direct other regulatory authorities. Semi-hard powers enable policymakers to make formal recommendations to other regulatory authorities, coupled with a "comply or explain" requirement. Comply or explain requirements can be used to influence the wide range of regulatory actions that would ultimately be undertaken by other supervisory and regulatory agencies.
Only 13 of the 47 FSCs have semi-hard or hard powers to direct countercyclical actions (table 1). Three FSCs - France's High Council for Financial Stability (HCFS), Malaysia's Financial Stability Executive Committee (FSEC), and UK's Financial Policy Committee (FPC) - have hard powers over time-varying macroprudential tools.12 In the case of France, the HCFS has authority over setting the CCyB, while in the case of the U.K., the FPC has authority over a range of time-varying macroprudential policy tools, including the CCyB and LTV ratios. For Malaysia, the FSEC decides on specific measures to be taken by the CB, which is the PR for banks and insurance companies, to avert or reduce risks to financial stability and can issue orders to any person or financial institution not supervised by the CB. Ten others have only semi-hard powers, which is the authority to make recommendations with formal comply or explain authority.13 The remaining 34 FSCs in our sample have either only "soft" powers, which enable policymakers to express an opinion or issue a warning or non-binding a recommendation but without any comply or explain requirements, or to serve only an information-sharing or policy-coordination function across agencies, which is an even softer power. Thus, it appears that few FSCs function to implement policies.
Comply or explain powers for FSCs are well-suited to situations where further judgment by the member agency implementing the policy is important, and where a policy action is expected to face considerable political pressure, such that broad support and transparency for an agency's actions are needed (see IMF, 2013 for this view). Comply or explain powers may be more practical for addressing the structural component of systemic risk since they may be better suited to macroprudential policy interventions that are less frequent in nature. An example of this is the U.S. FSOC's recommendation to the market regulator in 2014 to eliminate the fixed net asset value in order to reduce the risk of investor runs in prime money market funds that were permitted to invest in instruments with credit risk.
That said, more recent experience suggests that FSC comply or explain instructions can also be directed at cyclical risks. For example, in June 2014 the U.K.'s FSC made recommendations to microprudential authorities in relation to cyclical developments in owner-occupied mortgage lending. Similarly, in January 2016, September 2016, and April 2018, Iceland's FSC recommended to the PR - the agency with ultimate authority for setting the CCyB - that the tool be activated to 1 percent and then increased to 1.25 percent and again 1.75 percent, which the PR then implemented. Likewise, in December 2017, Denmark's FSC recommended to the Minister for Industry, Business, and Financial Affairs that the tool be activated to 0.5 percent and about six months later recommended an increase to 1 percent. In November 2016 the ESRB issued comply or explain warnings on medium-term vulnerabilities in the residential real estate sector to the MoFs of eight EU Member States (specifically, Austria, Belgium, Denmark, Finland, Luxembourg, the Netherlands, Sweden and the United Kingdom).14MoFs in seven of the eight countries replied in writing, most often citing they were already monitoring the situation and some had already taken actions, though one country wrote that the warning was not justified.
Voting. We also collected information on voting, which we interpret as a process that would facilitate taking actions. We find that 24 of the 47 FSCs take votes, where we also consider FSCs that explicitly state that they aim to reach decisions by consensus as voting. All but two of these 24 FSCs that take votes are formal FSCs. All FSCs that have hard powers over time-varying macroprudential policy tools vote, while of all the FSCs that have semi-hard tools and all but one - specifically, Hong Kong's FSC - vote. Thus, voting appears to be related to powers.
Still, there are 12 FSCs that vote but do not have either hard or semi-hard tools. We find that in the majority (about seven) of these cases, the FSC can issue warnings about financial stability risks and make recommendations - albeit non-binding ones - to agencies. One of these FSCs in discussing their voting practices emphasize the desirability of achieving consensus and unanimity among FSC members in arriving at their decisions in preference to just achieving a majority, since without comply or explain powers achieving unanimity strengthens the effect of the FSC's recommendation. This means that five FSCs that do vote function only to undertake activities like monitoring, and sharing information or coordinating policies across agencies. These FSCs likely vote on issues like information sharing agreements between agencies, workplans for interagency groups, or public communications. From the other perspective about seven of the 23 FSCs that do not take votes can issue warnings about financial stability risks and make non-binding recommendations.
Communication. Communication is a soft tool that authorities could use to raise public awareness of risks and understanding of the need for authorities to take mitigating actions (see IMF, 2013 and 2014, and CGFS, 2016). A principal form is a financial stability report (FSR), but in our dataset, nearly all are published by the CBs and only a few FSCs publish FSRs, including those in the U.S. and Mexico. Cihak et al. (2012) document the rapid growth in the number of CBs that published FSRs, from 1 to 80, between 1996 and 2011, but also that there is a general lack of "forward-lookingness" in FSRs, which would make them less capable of assessing systemic risks. Correa, et al. (2017) document that while the sentiment conveyed by CBs in the FSRs correlates with the financial cycle, communications in the FSR have little influence on the financial cycle.
e. Macroprudential powers outside of financial stability committee powers
Since our review found that very few FSCs have hard tools, we looked further at whether individual agencies have authorities to implement time-varying tools. We focus on LTVs, which has been documented to be the most frequently-used tool in Cerutti et al. (2016) and Akinci and Olmstead-Rumsey (2017), and the new CCyB. While surveys suggest that the use of macroprudential tools has been growing and now is substantial, the actual frequency of change for most tools is very limited, suggesting most are not used in a time-varying way to address cyclical vulnerabilities. Cerutti et al. (2016) show that only LTVs and reserve requirements (for purposes other than monetary policy) are correlated with credit growth in a way to suggest they have been used to reduce a boom-bust credit cycle.15 Akinci and Olmstead-Rumsey (2017) document that LTVs are the most frequently-used macroprudential tool, and were either tightened or loosened in 57 countries about 80 times in the fourteen years from 2000 to 2013.16 Debt service-to-income ratios were also used, but less frequently, about 30 times, and most were in emerging market economies.
The CCyB is of special interest because it is a new tool and is strictly a macroprudential rather than an individual bank safety and soundness tool. It is calibrated generally to system-wide rather than bank-specific risks and allows for cross-border reciprocity arrangements, although it would apply only to regulated banks. Because of these features, a country could decide that the tool could be given to the PR that regulates banks, the CB with skills in system-wide risk analysis, or the FSC because it is strictly a macroprudential tool.
Overall, our tabulation suggests that while most countries have the authorities for CCyB and LTV adjustments, FSCs almost never directly controls these tools (table 2). For the CCyB,
53 countries have established the authority, but only three FSCs can set or direct the CCyB. The CB has the power in 32 countries, the PR in 17, and the MoF (including other government) in four, albeit most with a strong role for the CB in providing advice. This tabulation indicates that countries generally have assigned this new tool to existing regulators for capital.
In contrast, use of LTVs likely involves more political economy considerations than the CCyB, because they apply to borrowers rather than lenders, and tightening LTVs may be in more direct conflict with other social objectives, such as expanding homeownership. We tabulate that 39 countries have established the authority for LTVs, less than for CCyB, although we recognize that countries may be able to establish a new authority if they were to want to use LTVs as a macroprudential tool. Again, we find that FSCs do not direct the setting of this tool. The FSC has authority in two countries, while the CB has it in 22, the MoF (and other government) in nine, and the PR in seven.
Of the time-varying macroprudential policy tools, LTVs have the most cases of authority being assigned to the MoF. This outcome is consistent with Tucker (2014, 2016) that policies like LTVs that have distributional consequences should not be directed by unelected officials in independent agencies. A statement by Belgium's CB illustrates the issue. Belgium's CB, the appointed macroprudential authority in that country, writes in its CCyB framework documents that while the CB as the PR would be the appropriate authority to set the CCyB, it would be more appropriate for the government to set LTV ratios because of their distributional impacts.
The Bank was endowed with a wide range of macroprudential instruments which may be activated to mitigate emerging systemic risks. The Bank can impose additional capital or liquidity requirements, but also has tools beyond capital- and liquidity-based ones at its disposal. In view of their distributional impact, the Bank nevertheless has no responsibility for activating lending limits. In particular, imposing ceilings on the amount of mortgage debt in relation to the value of property and the level of debt repayments relative to income is a competence of the federal government.17
As noted above, very few FSCs publish FSRs while nearly all CBs publish them. Since nearly all CBs publish an FSR, we look at when they started to publish relative to when the FSC was created. We assume that CBs that published FSRs before the FSC was created have some degree of responsibility, either explicit or implicit, in at least communicating about financial stability risks. In our dataset, 38 of 47 CBs began publishing FSRs at least a year before the FSC in the country was established, suggesting a high share of CBs had some pre-existing stake in financial stability.
In summary, FSCs rarely have direct authorities for these time-varying tools and members retain the authorities for setting the tools Such an arrangement of FSCs in which the traditional agencies participate as members could still - per the findings of Bodenstein et al. (2014) - produce benefits from improved communication or information sharing.18 However, conflicts for policy use could arise if the agencies have only microprudential mandates and do not also have a financial stability mandate, and there is less clear accountability for financial stability.
4 Empirical Analysis of FSC Structures and Characteristics
In this section, we investigate the motivations for countries to set up their financial stability governance structures; that is, whether FSCs are primarily formed to take or direct actions, or to improve information sharing, communication, and coordination of policies across agencies. We take two approaches to consider this question. First, we use logit regressions to evaluate which countries have set up a FSC as opposed to not having set up a FSC and consider whether characteristics of these countries seem more consistent with FSCs, in general, being set up to coordinate versus to take action. We also look at FSC processes and authorities that would put a FSC in a stronger position to take action rather than just coordinate using cluster analysis. Consistent with the implications of our logit regressions, we find that a relatively small portion - about one-quarter - of FSCs appear positioned to be able to take actions in the face of building financial stability risks given their existing authorities.
Lastly, we consider whether FSCs being set up largely to coordinate reflects the fact that other agencies are in a strong position to take action. We focus on the CB, given that it has specialized skills stemming from its monetary policy function to undertake time-varying macroprudential policy in a preemptive way. We find generally that FSCs that are less able to take actions are not this way because the CB is especially strong in this regard, suggesting that many countries do not have either a strong FSC or a strong CB for macroprudential actions.19
a. Motivation for setting-up a financial stability committee - Logit regression analysis
In the logit regressions for a country's decision to set up a FSC or not, the dependent variable in our main set of regressions is called "FSC exists" (FSC exists) and is assigned a value for each country according to the following definition:
- FSC exists = 0, if no FSC exists; and,
- FSC exists = 1, if a formal or de facto FSC exists
This variable is defined for all 58 countries in our dataset. In our dataset, FSC exists = 0 for 11 countries and FSC exists = 1 for 47 countries.
If better communication and coordination is a motive, we would expect that countries with more financial regulators and that are larger would benefit most from the formation of a FSC. To approximate the number of regulators that countries would want to be involved in setting macroprudential policies, we use variables designating whether the central bank is a PR for banks (CB is a PR) and whether the central bank is also a regulator for other part of the financial sector (CB is a wide PR). If coordination is a motive, a CB that is also a PR would require fewer agencies to coordinate, so we would expect a negative coefficient on these variables. Country size may reflect more agencies or interests to consider, and so we would expect a positive coefficient. To approximate country size we use the variable log (GDP). Our logit regression specification is given by:
where X includes CB is a PR, CB is a wide PR, and Log GDP in the baseline specification. The sample characteristics of the regressors in our baseline characterization are given in table 3, panels A and D. The CB is a PR in 60 percent of the countries, and is a wide regulator for more than banks in 31 percent. Note also that the CB's role in the microprudential regulatory structure were largely in place when countries set up their financial stability governance structures.20
Table 4, column 1 shows coefficient estimates for these variables, which are highly supportive of a coordination role for FSCs. A country for which the CB is the PR as well as the PR for more than banks has a 60 percent probability of having a FSC (assuming Log GDP at the sample average), which is low given the unconditional probability is 81 percent (that is, 47 divided by 58 countries). A country for which the CB is only the bank PR (which would imply more separate financial regulators) has a 90 percent probability of establishing a FSC, while a country for which the CB has no microprudential authorities has a 93 percent probability. Larger countries also have higher probabilities of having a FSC.
We include additional variables to show robustness of these results in Columns 2 through 11. All variables are described in table 3. We add these variables one-by-one since many variables are fairly highly correlated and we have only 58 countries in our dataset. A first set of variables reflect the depth of a country's financial system, where greater financial depth might proxy for a more complex financial system that then has more regulators and thus a greater need for coordination. We use two variables to consider financial depth; an index of financial development (Svirydzenka, 2016) and the nonfinancial credit-to-GDP ratio. Both variables have positive coefficients, though only the credit-to-GDP ratio is significant at the 10 percent level, and is not economically significant. That is, a one standard deviation higher credit-to-GDP ratio relative to its average only boosts the probability of a country having a FSC by 1 or 2 percentage points. Additionally, adding the IMF's index of financial development and the credit-to-GDP ratio does not change our coefficient estimates for CB is a PR, CB is a wide PR, and Log GDP.
We also include some variables to consider whether the riskiness of an economy might affect the formation of a FSC because these countries might want stronger governance for more effective policies. Specifically, we include the standard deviation of GDP relative to average GDP - that is, the coefficient of variation of GDP - over previous decades, and the number of financial crises that the country has suffered since 1970. Per Bruno and Shin (2014), we also add a variable reflecting the variability of capital inflows - specifically, the coefficient of variation of log capital inflow to GDP - and include a measure of a country's wealth, as measured by per-capita GDP. None of these four variables are significant and none change the coefficients that we obtain on CB is a PR, CB is a wide PR, or Log GDP.
To reflect the soundness of a country's overall governance, we include Rule of law and Checks and balances to represent countries with structures in place to protect rights and processes for policy-making. While these characteristics could be consistent with creating FSCs for coordination - because coordination may be more effective in this environment - they could also be consistent with instead allowing an existing agency - primarily the CB - to have authority for macroprudential policy, because there are fewer concerns about abuses of concentrated power when these practices are in place. Finally, we also include characteristics about the independence of the CB. Greater CB independence might make countries want a FSC, to coordinate rather than let a CB act on its own, since doing so would reduce the risk of an excess concentration of power. None of these governance variables are significant either and none change the coefficients that we obtain on CB is the PR, CB is a wide PR, or Log GDP. 21
We provide additional insight into a country's decision to set up a FSC and look only at those countries where the CB is the PR and consider why some of these countries form a FSC while others do not set up an FSC and designate the CB as the macroprudential authority. (Note that the CB is the designated agency in nine of the 11 countries without FSCs). The dependent variable in this regression, called "CB in PR and FS" (CB in PR and FS), is assigned a value for each country in which CB is the PR according to the definition:
- CB in PR and FS = 0, the CB is a PR and is not a macroprudential authority because a FSC has been set up; and,
- CB in PR and FS = 1, the CB is a PR and is the economy's macroprudential authority (and a FSC has not been set up).
This variable is defined for only the 35 countries in our dataset for which the CB is the PR. In our dataset CB in PR and FS equals 0 for 25 countries and equals 1 for 9 countries. Our logit regression is:
where X includes CB is a wide PR and Log GDP in the baseline specification. Relative to the equation for FSC exists, the variable CB as a PR has been dropped because all CBs are PRs in this regression. This CB logit regression is essentially the opposite of the FSC exists logit regression but conditioned on only countries where CBs are PRs, and so we expect the coefficient on CB is a wide PR to be positive and the coefficient on Log GDP to be negative if coordination is a motive for FSCs.
As shown in table 5, column 1, the coefficients for CB is a wide PR and Log GDP in this regression are essentially just the opposite of those in the FSC exists regression and, as such, provide additional support for the view that countries form FSCs primarily for coordination purposes. We also add a number of additional variables to reflect financial depth, the riskiness of the economy, and overall governance to our baseline specification. In all cases, these variables are insignificant in our CB logit regressions, as they were mostly for the FSC logit regressions. The coefficients on these variables in the CB regressions, while statistically insignificant, are in most cases similar in size but of the opposite sign of the coefficient on the same variable in the FSC regressions.
b. Motivation for setting-up a financial stability committee - cluster analysis
We also look at whether FSCs are set up to coordinate rather than to take action by defining FSC processes and authorities that would put a FSC in a strong position to take actions when financial stability risks are rising, and then consider the countries' characteristics. We focus on four processes and authorities that would put a FSC in a strong position to take action in the face of building financial stability risks: whether the FSC has been created formally created by legislation, whether the FSC has a chair, whether the FSC votes, and whether the FSC has tools. We use these variables in combination to summarize the FSC's ability to act since each on its own may be a narrow measure of strength. These variables are defined in table 3.
Whether the FSC has been created formally by legislation, as opposed to through more informal means, such as MOUs between agencies (FSC is formal), influences a FSC's ability to act because any action of a formal FSC will have greater legitimacy than that of n FSC that operates via inter-agency MOUs. Whether a FSC has a chair (FSC has a chair) also influences a FSC's ability to act, since a FSC without a chair is less likely to be able to move forward initiatives to address building risks. Additionally, by having a FSC chair, there may be more accountability for both the FSC's actions and inactions, which should create a greater incentive for it to respond to building risks. Whether the FSC takes votes, where this includes FSCs that reach their decisions by consensus (FSC votes) should also influence a FSC's ability to act, because voting means that the FSC does take a decision, which, all else equal, makes taking action more likely. Lastly, whether the FSC has hard or semi-hard tools (FSC has good tools) is also important, since employing tools is ultimately the way that a FSC would respond to building risks.
We use cluster analysis to consider FSCs' ability to take action in response to building financial stability risks because we have multiple variables that can signify ability to take action, and we do not want to impose priors that any variable would be sufficient on its own. While we believe that having "more" of the characteristics is stronger than having "none," we chose not to make an index because it would require weights on each variable and would obscure any correlations between variables.
Cluster analysis has been used elsewhere to study characteristics of governance structures of corporations (see Kaplan and Stromberg, 2002, and Gillan, Hartzell, and Starks, 2007).22 In our case, we use cluster analysis to divide the FSCs into groups - or clusters - where the FSCs in each cluster are similar in terms of the four FSC characteristics defined above. We use a hierarchical iterative process that starts with each FSC as its own cluster and then progressively merges up individual clusters until some desired criteria is met. We adopt standard criteria for defining clusters, which is to minimize the dissimilarities of FSCs within each cluster - measured by squared Euclidian distances or weighted sum of squares - and to maximize the dissimilarities across clusters. A second criteria for choosing clusters is broad similarity in the weighted sum of squares in each of the clusters.
Cluster analysis yields tree-like diagrams, called dendrograms, and the dendrogram for our analysis on FSCs is shown in figure 3. The hierarchical iterative process for forming clusters can be seen from the splits in the branches in the tree. The number of clusters in each figure is our choosing and determines when our algorithm stops merging clusters. Our criteria suggest four clusters each with about 12 countries. Our within (cluster) sum of squares' (WSS) values, which we report in the accompanying table, suggest that our clusters are quite compact. Indeed, for one cluster - the blue cluster - the WSS is zero, indicating that the countries in the cluster are identical across the variables used to construct the dendrogram. This is also evident from the dendrogram diagram itself, which shows no further branches for that cluster in the dendrogram.
We use a pseudo F statistic to select number of clusters. The pseudo F statistic for the cluster analysis describes the ratio of between-cluster variance to within-cluster variance (Calinski and Harabasz, 1974):
Pseudo F = (BSS / k-1) / (WSS / N - k)
Where k is the number of clusters and N is the number of observations. Larger values of pseudo F indicate clusters that are more close-knit and more separated from others.
The pseudo F-statistics for the FSC dendrogram overall with four clusters is 48.6, indicating significance of clusters based on these four characteristics.23Panel A of the accompanying table reports for each cluster the averages of the characteristics that we used to form the clusters, and reports that they are jointly statistically significantly different across the clusters based on the Kruskal-Wallis test.
The clusters in the dendrogram in figure 3 are ordered starting from most able on the left to less able on the right. We view the FSCs in the blue cluster in figure 3 as being the most able of the FSCs to take actions. The 12 FSCs in this cluster all are formal, have a single chair, have a voting process, and have tools. The FSCs in the next group of 12 - the teal cluster - are mostly all formal, mostly have a single chair, and all have a voting process, but none of them have good tools, and so are not as able to take actions as FSCs in the blue cluster. We consider the next two groups - the red and olive clusters - as even less able to act. In particular, the FSCs in the olive cluster are mostly defacto FSCs, none vote, and none have tools.
Overall, our cluster analysis finds that only 12 out of 47 of the FSCs - that is, only about one-quarter - appear to be in a position to directly take actions in the face of building risks. The vast majority of the FSCs - that is, about three-quarters - have little ability to take action.
The remaining panels of the accompanying table consider other FSC and country variables that are not used in determining FSC ability to act but could vary across our clusters and provide insight into why countries choose to create strong or weak FSCs. Panel B reports other FSC characteristics, panel C reports other prudential governance variables, panel D reports broader country governance variables, and panel E report economic characteristics of the country. In terms of other FSC variables, the strongest FSCs appear to be more likely to have independent members than the other clusters, consistent with stronger governance.
A few country variables appear to be significant across clusters. Countries with higher per capita GDP and higher financial market development are associated with stronger FSCs, as are countries with stronger Rule of law and Checks and balances. These indicators suggest countries that are wealthier, with strong governing mechanisms, and that have more developed financial systems are more likely to set up FSCs with ability to take actions reflecting both have the resources and infrastructure to do so as well as a need for good policies given the significance of the financial sector. Notably, however, while these variables seem to weaken initially as we move from stronger to weaker FSC clusters, some of them increase for the weakest group, such that the most able and least able FSC clusters share common country characteristics. Thus, we explore the relationship of FSC strength and the ability of existing agencies to take macroprudential actions, as countries with similar characteristics may both want a strong macroprudential policy framework but choose different arrangements to achieve that objective.
c. Central bank ability to act - cluster analysis
In terms of existing agencies with the ability to take actions, we focus on the CB given it has political independence and specialized skills relating to its monetary policy responsibilities to consider the economy and financial system as a whole, which is relevant for macroprudential policy. Using cluster analysis in a similar way, we evaluate whether CBs can be divided into groups based on their ability to take actions. We are especially interested in whether the strength of a CB is inversely related to the strength of a FSC - that is, FSCs are weaker in countries where the CB already can be a strong macropru authority, and vice versa that FSCs are strong when the CB is not a strong macropru authority. This latter case could reflect that countries that want strong, independent central banks for setting monetary policy do not also want to give the CB authority for macroprudential powers.
We focus on five characteristics that would put a CB in a strong position to take action in the face of building financial stability risks. First, a CB that is a PR (CB is a PR) will likely have better information and tools related to their regulated entities that can help to reduce financial stability risks. For example, CBs that also are PRs will have authorities for setting structural capital requirements. Second, a CB that is a wide PR (CB is a wide PR) may be even stronger since it has responsibility, at least for microprudential powers, for a broader swath of the financial system. The next two variables measure whether the CB has the authority for setting the CCyB (CB has CCyB) or for setting the LTV (CB has LTV). As discussed above, the CB has the authority in 32 of 47 countries with FSCs to set the CCyB, and in 22 to set LTVs. These are direct measures of ability to take actions. A fifth variable is based on whether the CB publishes a Financial Stability Report and whether it started publishing the report before the FSC was created (CB early FSR). We interpret a CB having published a FSR before the FSC was created as reflecting some responsibility, either explicit or implicit, for at least communicating about financial stability. In our dataset, 80 percent of CBs began publishing before the FSC was created. We use the above-mentioned five variables in combination to summarize the CB's ability to act since each on its own may be a narrow measure of strength.
The definitions of the five characteristics and the dendrograms for the CBs are shown in figure 4. As with the FSCs, the hierarchical iterative process for forming clusters can be seen from the splits in the branches in the tree. Our criteria suggest that three clusters is a good representation. Our within (cluster) sum of squares' (WSS) values, which we report in the accompanying table, suggest that one of the groups (the least able, the black cluster in the dendrogram) is very compact with a WSS of 0.9, indicating that the 18 countries in the cluster are nearly identical across the five variables used to construct the dendrogram.
The pseudo F-statistic for the dendrogram overall is 30.5, indicating significance. Panel A of the accompanying table reports for each cluster the averages of the characteristics that we used to form the clusters, and reports that the averages for each of the five variables are jointly statistically significantly different across the clusters based on the Kruskal-Wallis test.
The clusters in the dendrogram are shown with the least able - the black cluster - on the left. A striking feature of the least able to act CB cluster is the commonality of the 18 CBs. None of the 18 FSCs in this cluster are PRs or wide PRs, and none have the authority to set CCyB or LTVs. At the same time, nearly all of the CBs in this cluster published an FSR before the FSC was created, suggesting CBs that already had a role in promoting financial stability and are not a PR do not get new authorities for CCyB or LTVs.
Looking at additional macroprudential governance characteristics for this least able cluster, the chair of the FSC is more likely to be the MoF, and setting CCyB and LTVs are done by the PR and the MoF (or government more generally). Moreover, these countries have stronger Rule of law. Combined, this cluster represents countries that have set up governance structures for monetary policy and macroprudential policy that are highly separate.
The next two clusters - the blue and teal clusters - include all the CBs that are also a PR or a wide PR, and many have CCyB and LTV authorities, and thus are more able macroprudential authorities than the first cluster. The teal cluster represents CBs with more limited microprudential authorities as not all are PRs and none are a wide PR, but they are more likely to be able to set LTVs. In contrast, the blue cluster includes CBs that are stronger microprudential authorities but are unlikely to set LTVs. (There is little difference between the two groups in whether the CB has CCyB, likely because it is tied to whether the CB is a bank regulator.) These results suggest a tradeoff for the CB of either a strong microprudential regulator or strong macroprudential powers. Consistent with a tradeoff interpretation, CBs that are early FSR publishers - those as having an existing stake in financial stability - are less likely to set LTVs.
Other characteristics (shown in panels B to E) support not placing macroprudential powers at CBs that already have broad powers. The blue cluster with broader microprudential authorities also has more political independence for monetary policy - but not LTVs - consistent with not granting more authorities to an already powerful CB. The teal cluster that arguably represents the more able macroprudential CBs with LTVs, are in countries where the MoF is more likely to be the chair than the CB, consistent with countries putting in checks against a CB with both LTVs and CCyB. These results overall point to a pattern where countries do not grant additional macroprudential tools to CBs that already have other related authorities, suggesting political economy considerations are important factors in granting CBs macroprudential authorities.
d. Cross-tabulation of financial stability committees and central banks
We examine a cross-tabulation of FSC clusters and CB clusters. When countries have a weak FSC, there would be less concern for effective policymaking if the CB were strong than if it were also weak. However, we do not find an inverse relationship. Instead, weak FSCs appear randomly distributed across the CBs by ability to act. For the 23 weak FSCs, 10 are paired with CBs that lack any macroprudential authorities, six with CBs that are broad microprudential regulators and could be moderately strong, and seven with more able macroprudential CBs. Thus, we do not find compelling evidence that suggests that FSCs are weak because the CBs are strong.24 Conversely, of the 12 countries with strong FSCs, 5 countries have weak CBs for macroprudential authorities, again not compelling evidence of an inverse relationship.
To summarize effectiveness, we tally up effective authorities based on the criteria specified above - whether they have tools, accountability, and can reduce policy inertia. There are 12 effective FSCs and 11 CBs. One country has both an effective FSC and CB, so there are a total of 21 distinct countries. There are an additional five countries that are in the moderately able FSC cluster and moderately able CB cluster, which combined could be effective, which could raise the total to 26 countries. This tally suggests that 21of 47 countries with FSCs have neither a strong FSC nor a strong CB, suggesting many countries lack effective institutional structures for dynamic macroprudential policies. Of course, we have looked only at institutional structures at this point to evaluate effectiveness, but as more experience with macroprudential actions develops over time, we will be able to relate country experiences to their institutional structures and draw more definitive conclusions.
5 Conclusion
Most countries now have FSCs, but most are not designed to take or direct macroprudential actions to promote financial stability. Only one-quarter have attributes - formal, voting process, single chair, and good tools - that would make them a "more able" macroprudential authorities, and one-half lack two or more of these attributes. Other FSCs that lack tools and clear accountability still may produce benefits by facilitating better communication and coordination across agencies. FSCs generally also broaden the political legitimacy of macroprudential policy by involving the ministry of finance, often as the chair of the FSC, though that may introduce a greater tendency to delay taking actions in response to building risks.
Some countries with FSCs have granted macroprudential authorities to the CB, but we do not find that countries with less effective FSCs are more likely to have CBs with more macroprudential authorities. We find that CBs that are broad PRs and have higher political independence for monetary policy are not likely to also set LTVs. Rather LTVs are granted to CBs that are not broad microprudential regulators nor have high political independence for monetary policy, and for which the MoF is more likely the chair of the FSC rather than the CB as chair. We infer that countries generally do not want CBs to simultaneously have strong powers for microprudential policy, monetary policy, and macroprudential policy. We tabulate overall that up to 26 countries with FSCs can take or direct some macroprudential actions without a high risk of policy inertia based on both FSC and CB institutional arrangements. This suggests that in many countries, FSCs or FSCs and CBs combined may not set macroprudential policy in a way that it provides an alternative to monetary policy to address cyclical systemic risks. Of course, we have looked only at institutional structures at this point to evaluate effectiveness, but as more experience with macroprudential actions develops over time, we will be able to relate country experiences to their institutional structures and draw more definitive conclusions.
This study provides a snapshot of institutional structures and practices of macroprudential authorities, which continue to evolve. Even as countries continue to develop their frameworks, this study is useful for countries as a benchmark of current practices and what those practices reveal about competing objectives for macroprudential policy, including effectiveness but also concerns about interactions with existing microprudential regulators, excess concentration of power in a CB, and possible conflicts with other social objectives. We believe that the highly varied legal, regulatory, and financial structures across countries makes it unlikely that a single institutional arrangement would be optimal for all countries. But we believe it is important to highlight that the evidence suggests that countries are placing relatively low weight on the ability of policy institutions to take action and relatively high weight on political economy considerations.
References
Akinci, O., and J. Olmstead-Rumsey, 2018. "How Effective are Macroprudential Policies? An Empirical Investigation." Journal of Financial Intermediation 33, pp. 33-57.
Arnone, M., B. Laurens, J. Segalotto, and M. Sommer, (2007). "Central Bank Autonomy: Lessons from Global Trends," IMF Working Paper WP/07/88, April.
Arnone, M., and D. Romelli, 2013. "Dynamic central bank independence indices and inflation rate: A new empirical exploration." Journal of Financial Stability 9, pp. 385-98
Balls, E., J. Howat, and A. Stansbury, 2018. "Central Bank Independence Revisited." M-RCBG Associate Working Paper Series, No. 67
Bank of England, 2009. "The Role of Macroprudential Policy" Discussion Paper, Nov.
Basel Committee on Banking Supervision, 2017. "Implementation: Range of Practices in Implementing the Countercyclical Capital Buffer Policy." Bank for International Settlements.
Blinder, A., 2007. "On the Design of Monetary Policy Committees." CEPS Working Paper No. 153, Center for Economic Policy Studies.
Blinder, A., 2008. "Making Monetary Policy by Committee." CEPS Working Paper No. 167, Center for Economic Policy Studies.
Bodenstein, M., L. Guerrieri, and J. LaBriola, 2014. "Macroeconomic Policy Games." Finance and Economics Discussion Paper No. 2014-87.
Bruno, V., and H. Shin, 2014. "Cross-border Banking and Global Liquidity." BIS Working Paper No. 458.
Calinski, T., and J. Harabasz, 1974. "A Dendrite Method for Cluster Analysis: Communications in Statistics." Theory and Methods, 3, pp. 1-27.
Carrillo, J., E. Mendoza, V. Nuguer, and J. Roldan-Pena, 2017. "Tight Money-Tight Credit: Coordination Failure in the Conduct of Monetary and Financial Policies." NBER 23151.
Cerutti, E., S. Claessens, and L. Laeven, 2015. "The Use and Effectiveness of Macroprudential Policies: New Evidence." Journal of Financial Stability forthcoming.
Cerutti, E., R. Correa, E. Fiorentino, and E. Segalla, 2016. "Changes in Prudential Policy Instruments - A New Cross-Country Database." International Finance Discussion Papers 1169, Board of Governors of the Federal Reserve System.
Cihak, M., S. Munoz, S. Sharifuddin, and K. Tintchev, 2012. "Financial Stability Reports: What are they Good For?" IMF Working Paper 12/1, International Monetary Fund.
Committee on the Global Financial System, 2010. "Macroprudential Instruments and Frameworks." CGFS Paper No. 38, Bank for International Settlements.
Committee on the Global Financial System, 2016. "Objective-setting and Communication of Macroprudential Policies." CGFS Paper No. 57, Bank for International Settlements.
Conti-Brown, P. (2016), "The Power and Independence of the Federal Reserve," Princeton University Press Princeton, New Jersey.
Correa, R., K. Garud, J. Londono, and N. Mislang, 2017. "Sentiment in Central Banks' Financial Stability Reports." International Finance Discussion Paper No. D2017‐9.
Elliot, D., G. Feldberg, and A. Lehnert, 2013. "The History of Cyclical Macroprudential Policy in the United States." Finance and Economics Discussion Series Paper No. 2013-29.
European Systemic Risk Board, 2011. "Recommendation on the Macro Prudential Mandate of National Authorities." (ESRB/2011/3)
Fischer, S., 1995. "Central Bank Independence Revisited," American Economic Review Papers and Proceedings, Vol. 85, pp. 201-206
Goodhart, C. and D. Schoenmaker, 1995. "Should the Functions of Monetary Policy and Banking Supervision be Separated?" Oxford Economic Papers 47, Oxford University Press (4), pp. 539-560.
Goodhart, C., 2010. "The Changing Role of Central Banks." Financial Markets Group, London School of Economics.
International Monetary Fund, 2011. "Macroprudential Policy: An Organizing Framework."
International Monetary Fund, 2013. "Key Aspects of Macroprudential Policy."
International Monetary Fund, 2014. "Staff Guidance Note of Macroprudential Policy."
International Monetary Fund, 2016. "Fostering Stability in a Low-Growth, Low-Rate Era." October Global Financial Stability Report.
International Monetary Fund, 2017. "Low Growth, Low Rates, and Financial Intermediation." April Global Financial Stability Report.
International Monetary Fund, 2018. "The IMF's Annual Macroprudential Policy Survey: Objectives, Design, and Country Responses."
International Monetary Fund, Financial Stability Board, and Bank for International Settlements, 2016. "Elements of Effective Macroprudential Policies: Lessons from International Experience.
Koetter, M., K. Roszbach, and G. Spagnolo, 2014. "Financial Stability and Central Bank Governance." International Journal of Central Banking, December, pp 31-67.
Kuttner, K., and I. Shim, 2013. "Can Non-Interest Rate Policies Stabilize Housing Markets? Evidence from a Panel of 57 Economies." NBER Working Paper 19723.
Laeven, L. and F. Valencia, 2012. "Systemic Banking Crises Database: An Update." IMF Working Paper 12/163. International Monetary Fund.
Lim, C., I. Krznar, F. Lipinsky, A. Otani, and X. Wu, 2013. "The Macroprudential Framework: Policy Responsiveness and Institutional Arrangements." IMF Working Paper 13/166. International Monetary Fund.
Lim, C., A. Klemm, S. Ogawa, M. Pani, and C. Visconti, 2017. "Financial Stability Reports in Latin America and the Caribbean." IMF Working Paper 17/73. International Monetary Fund.
Lombardi, D. and P. Siklos, 2016. "Benchmarking Macroprudential Policies: An Initial Assessment." Journal of Financial Stability 27, pp. 35-49.
Masciandaro, D. and A. Volpicella, 2016. "Macroprudential Governance and Central Banks: Facts and Drivers." Journal of International Money and Finance 61, pp. 101-119.
Merrouche, O. and E. Nier, 2010. "What Caused the Global Financial Crisis? - Evidence on the Drivers of Financial Imbalances 1999-2007." IMF Working Paper 10/265. International Monetary Fund.
Nier, E.,2009. "Financial Stability Frameworks and the Role of Central Banks: Lessons from the Crisis." IMF 09/70, International Monetary Fund.
Nier, E., J. Osinski, L. Jacome, and P. Madrid, 2011. "Towards Effective Macroprudential Policy Frameworks: An Assessment of Stylized Institutional Models." IMF Working Paper 11/250, International Monetary Fund.
Osterloo, S. and J. deHaan, 2004. "Central Banks and Financial Stability: A Survey." Journal of Financial Stability 1, pp. 257-273.
Smaga, P., 2013. "Assessing Involvement of Central Banks in Financial Stability." Center for Financial Stability Policy Paper.
Smets, F., 2014. "Financial Stability and Monetary Policy: How Closely Interlinked." International Journal of Central Banking, June, pp 263-300.
Svirydzenka, K., 2016. "Introducing a New Broad Based Index of Financial Development." IMF Working Paper 16/05. International Monetary Fund.
Tucker, P., 2014. "The Political Economy of Macroprudential Regimes," in Macroprudentialism, Dirk Schoenmaker (ed.), A VoxEU Book, CEPR Press and Duisenberg School of Finance.
Tucker, P., 2016. "The Design and Governance of Financial Stability Regimes: A Common-resource Problem That Challenges Technical Know-How, Democratic Accountability and International Coordination," in CIGI Essays on International Finance, Volume 3.
Ueda, K., and F. Valencia, 2012. "Central Bank Independence and Macro-Prudential Regulation." IMF Working Paper 12/101. International Monetary Fund.
Table 1: Financial Stability Committees: Membership, leadership, and authorities
| Macroprudential Authorities | No. of countries |
|---|---|
| FSC | 47 |
| Formal | 35 |
| De facto | 12 |
| No FSC | 11 |
| CB is the macroprudential authority | 9 |
| PR is the macroprudential authority | 2 |
| FSCs | No. of countries |
|---|---|
| Membership MoF | 40 |
| Membership CB | 46 |
| Membership PR | 47 |
| Membership Independent | 7 |
| Chair or co-chair MoF | 25 |
| Chair or co-chair CB | 18 |
| Chair or co-chair PR | 1 |
| Chair or co-chair Other | 7 |
| Voting process Yes | 24 |
| Voting process No | 23 |
| Tools Soft at most | 34 |
| Tools Semi-hard at most | 10 |
| Tools Hard | 3 |
Table 2: Authorities for Selected Tools
* Includes "Government" (Argentina)
** Includes where the MoF sets the CCyB with input from the CB or PR
| CCyB | LTVs | |
|---|---|---|
| Country designates an authority: | ||
| Yes | 53 | 39 |
| No | 5 | 19 |
| If Yes, which agency: | ||
| FSC | 3 | 2 |
| Central Bank | 32 | 22 |
| CB as PR | 30 | 19 |
| Ministry of Finance** | 3 | 7 |
| Prudential Regulator | 16 | 7 |
| Other* | 1 | 2 |
Table 3: Regression and cluster analysis variables
| Description | Mean | Std. dev. | 10th pctl. | |
|---|---|---|---|---|
| A. Central bank micro- and macroprudential gov. variables | ||||
| CB is a PR | As collected in our dataset and equal to 1 if the CB is thePR and 0 if not. | .60 | .49 | 0 |
| CB is a wide PR | As collected in our dataset and equal to 1 if the CBis the PR of more than just banks and 0 if not. | .31 | .47 | 0 |
| B. Other central bank governance variables | ||||
| CB political independence | As measured by Grilli, Masciandaro, and Tabellini (1991) and based on the involvement of the government in appointing the CB governor or as a participant for formulating monetary policy. | 0.64 | 0.30 | 0.25 |
| CB operational independence | As measured by Grilli, Masciandaro, and Tabellini (1991) and based on linkages between the CB and government in terms of credit provision by the CB to the government, and also if the CB is a PR. | 0.78 | 0.18 | 0.56 |
| C. Broader country governance variables | ||||
| Rule of law | As measured by the World Bank to capture the traditions and institutions by which authority in a country is exercised and is a measure of the extent to which agents have confidence in and abide by the rules of society, and in particular the quality of contract enforcement, property rights, the police, and the courts. | 0.78 | 0.91 | -0.51 |
| Checks and balances | As measured originally by the World Bank and now the Inter-American Development Bank to capture the institutions by which limits are placed on the actions of one branch of the government by other branches of the government with purview over these actions. | 0.66 | 0.24 | 0.20 |
| D. Economy-characteristic variables. | ||||
| Log GDP | An indicator of the economy's size. As reported by the World Bank and measured by US dollar denominated GDP. | 26.4 | 1.5 | 24.4 |
| GDP per capita | An indicator of a country's wealth. As reported by the World Bank and measured by GDP in US dollars per person. | 29.5 | 16.9 | 6.7 |
| Coef. of variation of GDP | An indicator of the variability of a country's GDP and measured by the ratio of the standard deviation to the average of GDP from 1980 to 2007. | .17 | .05 | .12 |
| Private credit-to-GDP | As reported in the BIS statistics. An indicator of an economy's credit intensity and is measured by the ratio of private nonfinancial credit to GDP. | 3.01 | 1.14 | 1.94 |
| Coef. of variation of log capital inflows-to-GDP | As reported by the IMF WP 13/183 Capital Flows database. An indicator of the variability of logged gross capital inflows to GDP and measured by the standard deviation of logged gross capital inflows to GDP divided by its average. (We drop Luxembourg, an extreme outlier.) | 91.85 | 53.76 | 32.70 |
| Number of crises | As measured by Laeven and Valencia (2012) and is the number of financial crisis experienced by a country since 1970 - defined as a severe impairment of banking intermediation that required some fiscal assistance. | 1.14 | .76 | 0 |
| Financial development index | As measured by Svirydzenka (2016) in IMF WP 16/5 and based on the depth, access, and efficiency of countries financial institutions and markets. | 0.60 | 0.21 | 0.33 |
| Financial market development index | As measured by Svirydzenka (2016) in IMF WP 16/5. See above. | 0.65 | 0.20 | 0.35 |
| Fin. institution development index | As measured by Svirydzenka (2016) in IMF WP 16/5. See above. | 0.55 | 0.26 | 0.19 |
Table 4: Regression for the probability of a country having established a FSC
Note: *p**p***p<0.01
Dependent variable: FSC exists = 1 / FSC exists = 0
| (1) | (2) | (3) | (4) | (5) | (6) | (7) | (8) | (9) | (10) | (11) | |
|---|---|---|---|---|---|---|---|---|---|---|---|
| CB is PR indicator |
-0.43 (1.08) |
-0.31 (1.08) |
-0.29 (1.09) |
-0.36 (1.10) |
-0.18 (1.13) |
-0.54 (1.10) |
-0.39 (1.09) |
-0.54 (1.12) |
-0.50 (1.09) |
-0.27 (1.10) |
-1.03 (1.25) |
| CB is wide PR indicator |
-1.78* (0.94) |
-1.79* (0.95) |
-1.73* (0.96) |
-1.96** (0.99) |
-1.94* (1.01) |
-1.75* (0.94) |
-2.10** (0.99) |
-1.74* (0.94) |
-1.72* (0.95) |
-1.94** (0.97) |
-1.77* (0.96) |
| Log GDP (2007) |
0.46* (0.27) |
0.31 (0.31) |
0.42 (0.31) |
0.54* (0.29) |
0.49* (0.28) |
0.50* (0.28) |
0.67** (0.33) |
0.46* (0.27) |
0.46* (0.27) |
0.46* (0.27) |
0.49* (0.27) |
| Financial development index |
2.44 (2.18) |
||||||||||
| Private credit-to-GDP |
0.02* (0.01) |
||||||||||
| Number of crises |
-0.42 (0.49) |
||||||||||
| Cov(GDP) (2007) |
16.14 (10.66) |
||||||||||
| Per-capita GDP (2007) |
-0.01 (0.02) |
||||||||||
| Log capital inflows-to-GDP (2007) |
0.32 (0.40) |
||||||||||
| Rule of law |
-0.19 (0.45) |
||||||||||
| Checks and balances |
-0.29 (1.57) |
||||||||||
| CB political independence |
-1.35 (1.38) |
||||||||||
| CB operational independence |
-2.65 (2.63) |
||||||||||
| Constant |
-9.52 (6.86) |
-7.10 (7.51) |
-9.98 (8.00) |
-11.11 (7.32) |
-13.07* (7.56) |
-10.03 (6.94) |
-15.92* (8.98) |
-9.44 (6.79) |
-9.29 (6.83) |
-8.76 (7.01) |
-7.92 (6.90) |
| Observations | 58 | 58 | 58 | 58 | 58 | 58 | 57 | 58 | 57 | 58 | 58 |
| Log Likelihood | -23.30 | -22.65 | -21.53 | -22.94 | -21.87 | -23.12 | -21.86 | -23.21 | -23.15 | -22.80 | -22.74 |
| Akaike Inf. Crit. | 54.60 | 55.29 | 53.06 | 55.87 | 53.74 | 56.24 | 53.71 | 56.42 | 56.31 | 55.61 | 55.48 |
Table 5: Regression for the probability of a CB that is a PR being the macroprudential authority
Note: *p**p***p<0.01
Dependent variable: CB in PR and FS = 1 / CB in PR and FS = 0
| (1) | (2) | (3) | (4) | (5) | (6) | (7) | (8) | (9) | (10) | (11) | |
|---|---|---|---|---|---|---|---|---|---|---|---|
| CB is wide PR indicator |
1.88* (0.98) |
1.84* (0.97) |
1.77* (0.97) |
2.23** (1.09) |
1.94* (1.02) |
1.89* (0.98) |
2.30** (1.07) |
1.83* (0.99) |
1.87* (1.00) |
2.06** (1.02) |
1.94* (1.02) |
| Log GDP (2007) |
-0.59 (0.36) |
-0.45 (0.43) |
-0.51 (0.40) |
-0.76* (0.42) |
-0.58 (0.36) |
-0.58 (0.37) |
-0.81* (0.43) |
-0.68* (0.38) |
-0.61* (0.36) |
-0.62* (0.37) |
-0.68* (0.37) |
| Financial development index |
-1.77 (2.71) |
||||||||||
| Private credit-to-GDP |
-0.01 (0.01) |
||||||||||
| Number of crises |
0.63 (0.55) |
||||||||||
| Cov(GDP) (2007) |
-11.79 (10.81) |
||||||||||
| Per-capita GDP (2007) |
-0.004 (0.03) |
||||||||||
| Log capital inflows-to-GDP (2007) |
0.05 (0.48) |
||||||||||
| Rule of law |
0.64 (0.60) |
||||||||||
| Checks and balances |
0.76 (1.80) |
||||||||||
| CB political independence |
1.40 (1.60) |
||||||||||
| CB operational independence |
3.10 (2.98) |
||||||||||
| Constant |
13.38 (9.25) |
10.54 (10.30) |
12.17 (10.14) |
16.87 (10.27) |
14.86 (9.35) |
13.14 (9.38) |
18.58* (11.14) |
15.24 (9.57) |
13.22 (9.08) |
12.91 (9.50) |
13.35 (9.14) |
| Observations | 34 | 34 | 34 | 34 | 34 | 34 | 33 | 34 | 33 | 34 | 34 |
| Log Likelihood | -16.38 | -16.16 | -15.74 | -15.74 | -15.68 | -16.37 | -15.02 | -15.75 | -16.17 | -15.98 | -15.78 |
| Akaike Inf. Crit. | 38.76 | 40.33 | 39.47 | 39.48 | 39.37 | 40.73 | 38.04 | 39.49 | 40.34 | 39.96 | 39.55 |
Figure 3: FSC Ability to Act Dendrogram and Characteristics of Clusters
- FSC is formal, which
- equals 1 if the FSC has been created formally by legislation
- equals 0 if the FSC exists only through non-legal arrangements between agencies
- FSC has a chair votes, which
- equals 1 if the FSC has a chair or co-chair
- equals 0 if the FSC does not have a chair or co-chair
- FSC votes, which
- equals 1 if the FSC takes votes
- equals 0 if the FSC does not take votes
- FSC has good tools, which
- equals 1 if the FSC has either hard or semi-hard tools
- equals 0 if the FSC has neither of these two types of tool
F-Stat for clusters = 47.14
Least able FSC Most able FSC
Table 6: Characteristics of Clusters in the FSC Ability to Act Dendrogram
| Least able Olive (4) |
Red (1) | Teal (3) | Most able Blue (2) |
|
|---|---|---|---|---|
| A. Dendrogram info. & avg. values of variables | ||||
| No. of countries | 10 | 13 | 12 | 12 |
| Within (cluster) sum of squares (WSS) | 3.4 | 0.9 | 3.9 | 0.0 |
| Between (cluster) sum of squares (BSS) | 8.8 | 5.0 | 3.8 | 10.2 |
| Formal FSC | 0.1*** | 1*** | 0.75*** | 1*** |
| Single chair | 0.5*** | 1*** | 0.83*** | 1*** |
| FSC votes | 0*** | 0*** | 1*** | 1*** |
| Good tools | 0*** | 0.08*** | 0*** | 1*** |
| B. Average values of other FSC variables | ||||
| CB is chair | 0.20 | 0.31 | 0.50 | 0.50 |
| MoF is chair | 0.30 | 0.62 | 0.50 | 0.58 |
| CB and MoF are members | 0.80 | 1.00 | 0.67 | 0.83 |
| Indep. members on FSC | 0** | 0.08** | 0.08** | 0.42** |
| No. agencies on FSC | 3.40 | 4.31 | 4.00 | 4.08 |
| C. Avg. values of other pru.. gov. variables | ||||
| CB is PR | 0.60 | 0.54 | 0.42 | 0.58 |
| CB is wide PR | 0.30 | 0.23 | 0.17 | 0.25 |
| CB has CCyB | 0.40 | 0.46 | 0.67 | 0.42 |
| CB has LTVs | 0.50 | 0.38 | 0.50 | 0.08 |
| PR has CCyB | 0.20 | 0.38 | 0.25 | 0.42 |
| PR has LTVs | 0.00 | 0.23 | 0.08 | 0.17 |
| MoF or other has CCyB | 0.20 | 0.00 | 0.00 | 0.17 |
| MoF or other has LTV's | 0.20 | 0.23 | 0.08 | 0.17 |
| Early FSR | 0.70 | 0.69 | 0.92 | 0.92 |
| D. Average values of other governance variables | ||||
| CB political independence | 0.53 | 0.58 | 0.61 | 0.75 |
| CB operational independence | 0.76 | 0.72 | 0.76 | 0.87 |
| Rule of law | 1.06** | 0.35** | 0.44** | 1.32** |
| Checks and balances | 0.64* | 0.58* | 0.64* | 0.79* |
| E. Avg. values of economy variables | ||||
| Log GDP (2007) | 26.78 | 26.60 | 26.19 | 26.69 |
| Cov(GDP) (2007) | 0.17 | 0.18 | 0.19 | 0.16 |
| Per-capita GDP (2007) | 40.2*** | 22.3*** | 20.92*** | 41.0*** |
| Private credit-to-GDP (2007) | 111.97 | 82.63 | 84.14 | 120.88 |
| Log capital inflows-to-GDP (2007) | 2.84 | 2.85 | 2.65 | 3.64 |
| Nunmber of crises | 0.80 | 1.31 | 1.17 | 1.17 |
| Financial development index | 0.72* | 0.56* | 0.53* | 0.72* |
| Financial institutions development index | 0.72* | 0.57* | 0.62* | 0.78* |
| Financial markets development index | 0.72* | 0.53* | 0.44* | 0.66* |
Figure 4: CB Ability to Act Dendrogram and Characteristics of Clusters (for countries with FSCs)
- CB is a PR, which
- equals 1 if the CB is a bank prudential regulator
- equals 0 if the CB is not a bank prudential regulator
- CB is a wide PR, which
- equals 1 if the CB is the prudential regulator of more than just banks
- equals 0 if the CB is not a prudential regulator or just regulates banks
- CB has CCyB, which
- equals 1 if the CB has authority for setting the CCyB
- equals 0 if the CB does not have authority for setting for the CCyB
- CB has LTV, which
- equals 1 if the CB has authority for setting LTV ratios
- equals 0 if the CB does not have authority for setting LTV ratios
- CB early FSR, which
- equals 1 if the CB was publishing an FSR before its FSC was formed
- equals 0 if the CB does not publish an FSR or started publishing it after its FSC was formed
F-Stat for clusters = 30.49
Least able CB Most able CB
Table 7: Characteristics of Clusters in CB Ability to Act Dendrogram (for countries with FSCs)
| Least able Black (1) |
Blue (2) | Most able Teal (3) |
|
|---|---|---|---|
| A. Dendrogram info. & avg. values of variables | |||
| No. of countries | 18 | 18 | 11 |
| Within (cluster) sum of squares (WSS) | 0.9 | 12.6 | 7.5 |
| Between (cluster) sum of squares (BSS) | 13.1 | 8.8 | 7.2 |
| Early FSR | 0.94*** | 0.89*** | 0.45*** |
| CB has CCyB | 0*** | 0.83*** | 0.73*** |
| CB has LTVs | 0*** | 0.33*** | 1*** |
| CB is PR | 0*** | 1*** | 0.64*** |
| CB is wide PR | 0*** | 0.61*** | 0*** |
| B. Average values of FSC variables | |||
| Formal FSC | 0.72 | 0.83 | 0.64 |
| Single chair | 0.94 | 0.72 | 0.91 |
| FSC votes | 0.44 | 0.67 | 0.36 |
| Good tools | 0.28 | 0.39 | 0.09 |
| CB is chair | 0.39 | 0.44 | 0.27 |
| MoF is chair | 0.67** | 0.28** | 0.64** |
| CB and MoF are members | 0.83 | 0.78 | 0.91 |
| Independent members on FSC | 0.11 | 0.17 | 0.18 |
| C. Avg. values of other pru.. gov. variables | |||
| PR has CCyB | 0.67*** | 0.11*** | 0*** |
| PR has LTVs | 0.28* | 0.06* | 0* |
| MoF or other has CCyB | 0.22** | 0** | 0** |
| MoF or other has LTV's | 0.33* | 0.11* | 0* |
| D. Average values of other governance variables | |||
| CB political independence | 0.58* | 0.73* | 0.5* |
| CB operational independence | 0.91*** | 0.72*** | 0.66*** |
| Rule of law | 1.1* | 0.61* | 0.5* |
| Checks and balances | 0.69 | 0.67 | 0.59 |
| E. Avg. values of economy variables | |||
| Log GDP (2007) | 26.66 | 26.80 | 25.99 |
| Cov(GDP) (2007) | 0.18 | 0.17 | 0.19 |
| Per-capita GDP (2007) | 36.7 | 29.39 | 22.36 |
| Private credit-to-GDP (2007) | 109.33 | 91.96 | 93.71 |
| Log capital inflows-to-GDP (2007) | 3.14 | 2.99 | 2.81 |
| Number of crises | 1.11 | 1.28 | 0.91 |
| Financial development index | 0.69 | 0.62 | 0.54 |
| Financial institutions development index | 0.72 | 0.66 | 0.60 |
| Financial markets development index | 0.65 | 0.57 | 0.48 |
Appendix A. Country coverage
Table A.1. Countries in our Dataset
Argentina, Finland, Luxembourg, Singapore, Australia, France, Malaysia, Slovak Republic, Austria, Germany, Malta, Slovenia, Belgium, Greece, Mexico, South Africa, Brazil, Hong Kong, Netherlands, South Korea, Bulgaria, Hungary, New Zealand, Spain, Canada, Iceland, Norway, Sweden, Chile, India, Peru, Switzerland, China, Indonesia, Philippines, Thailand, Colombia, Ireland, Poland, Turkey, Croatia, Israel, Portugal, Ukraine, Cyprus, Italy, Romania, United Kingdom, Czech Republic, Japan, Russian Federation, United States,Denmark, Latvia, Saudi Arabia,Estonia, Lithuania, Serbia
Footnotes
that has more microprudential authorities. Moreover, we do not find that weak FSCs are offset substantially by strong CBs. Return to Text