Abstract
Macroprudential policies – caps on loan to value ratios, limits on credit growth and other balance sheets restrictions, (countercyclical) capital and reserve requirements and surcharges, and Pigouvian levies – have become part of the policy paradigm in emerging markets and advanced countries alike. But knowledge is still limited on these tools. Macroprudential policies ought to be motivated by market failures and externalities, but these can be hard to identify. They can also interact with various other policies, such as monetary and microprudential, raising coordination issues. Some countries, especially emerging markets, have used these tools and analyses suggest that some can reduce procyclicality and crisis risks. Yet, much remains to be studied, including tools’ costs ? by adversely affecting resource allocations; how to best adapt tools to country circumstances; and preferred institutional designs, including how to address political economy risks. As such, policy makers should move carefully in adopting tools.
I. Introduction
This paper reviews existing research on the motivations for macroprudential policies, possible specific tools, actual usage, and lessons from experiences. The recent wave of financial crises has led to a greater recognition of the large and at times adverse real economic effects of finance. It has also made clear that existing tools – whether microprudential, monetary, fiscal, or other policies – even when conducted properly and effectively in their own ways, do always not suffice to assure financial stability. Combined with a broader rethinking of macroeconomic and financial policies (e.g., Akerlof et al., 2014), this has led to a call for macroprudential policies, i.e., those policies aiming to reduce systemic risks arising from “excessive” financial procyclicality and from interconnections and other “cross-sectional” factors.
While the need for macroprudential policies is now largely accepted, many questions remain, starting from their motivations. 2 In principle, macroprudential policies should be motivated by externalities and market failures arising from various financial frictions and market imperfections that exist even when microprudential supervision and monetary policy are conducted effectively (which regrettably is often not the case in practice). Few theoretical analyses exist, however, to guide macroprudential policies this way and hardly any have been formally tested. Consequently, most often the design of policies has not started from first principles, but more arising from generic concerns. Related, the set of policies currently being considered is mostly based on existing microprudential and regulatory tools (i.e., caps on loan to value ratios, limits on credit growth, additional capital adequacy requirements, reserve requirements and other balance sheets restrictions), which have been given additional, macroprudential objectives, with forms of “Pigouvian” taxes and levies added.
Even though both the motivations and expected effectiveness of various policies are not well known, usage has often proceeded on an ad-hoc or experimental bases, especially in emerging markets. Evaluations of usage to date, mostly aimed at affecting developments in credit and housing markets, suggest that some tools can help reduce financial procyclicality and lower crisis risks. Notably, caps on loan to value and debt service to income ratios seem to help in reducing booms, and thereby busts, in real estate markets, major sources of instability. Reserve requirements and targeted levies on foreign exchange exposures also help in reducing system-wide vulnerabilities. And progress is being made to reduce the systemic risks created by large financial institutions using, among others, macroprudential policies.
Still, it is not well-known how policies are most effectively calibrated to circumstances (e.g., when and how much to raise or lower a countercyclical capital requirements) and adapted to country characteristics (e.g., which tools to use given specific financial market structures). Knowledge on which policies and how to use them for risks in capital markets is very limited. Neither is much known about the costs of policies. By definition, macroprudential policies distort some behaviors. Unless perfectly targeted at the source, i.e., where the externalities or market failures arise, which is unlikely, policies can worsen some resource allocations. And by constraining actions of agents, they can increase overall systemic risks.
Questions also arise on the best institutional design for usage, e.g., who is made in charge of macroprudential policies. Should these policies be conducted by the central bank, an existing microprudential or market conduct supervisory agency, a new “macroprudential” agency, or a committee composed of various agencies and others (such as representatives from the ministry of finance)? Institutional designs matter as the conduct of macroprudential policies can interfere with the primary objective of some agencies. A central bank may have more difficulty communicating its monetary policy stance when also in charge of macroprudential policy. And a microprudential authority may be less able to execute its goals when its staff needs multiple skills and is confronted with (at times conflicting) goals. A big concern would be if adopting these policies reduces the importance given to assuring properly conducted monetary policy – e.g., as when these policies become a substitute for monetary policy – or improving microprudential supervision, as these policies are essential in their own right and likely more important overall for reducing systemic risks.
A major issue, closely related to institutional design, is how the political economy of macroprudential policies will play out. By involving the government more directly into the allocation of resources, it (or the specific agencies in charge) will become (more) exposed to outside pressures. This risk needs to be acknowledged explicitly and addressed in institutional design(s), including accountability and transparency. Some policies may for example need to be presented to parliament for broader, public approval, to avoid exposing regulatory agencies to political risks (of ex-post not having “prevented” a crisis when it decided ex-ante not to set too low a LTV so as to allow first-time buyers to acquire a home).
Overall, while the greater system-wide focus is clearly welcome in light of recent crises, many unknowns still exist and a large research agenda remains. In the meantime, policy makers may want to move cautiously with adopting macroprudential tools and prioritize the ones they do and clearly taste their objectives. If in the end, only few policies are actually adopted, a key objective and possible main achievement could still be attained: which is greater appreciation of a more system-wide view of finance in all its aspects and of the various policies that can reduce the risk of crises and lower any excessive procyclicality.
The paper proceeds as follows. Section 2 presents the literature analyzing the motivations for macroprudential policies, considering both time-series, “procyclicality,” and cross-sectional, systemic risk, dimensions. Section III reviews the knowledge on the interactions of macroprudential policies with other policies, notably monetary, and international dimensions. Section 4 describes possible tools, choices and calibration strategies, and use to date. Section 5 presents findings of existing research and case-studies. Section 6 concludes with lessons and outstanding policy issues, including thinking on institutional design and for research.
II. Motivation for macroprudential policies
Financial crises have led to (renewed) attention on macroprudential policies. The recent global financial crisis and its aftermath have been painful reminders of the multifaceted interactions between macroeconomic and financial market developments, “macro-financial linkages.” They have led to (a call for) the adoption of macroprudential policies (as well as the review and reform of other financial policies and institutions).3 The fundamental rationales behind such policies, however, are not always clearly articulated. Proponents do not always start from the key externalities and market failures associated with activities of financial intermediaries and markets that can lead to excessive procyclicality and the buildup of systemic risk. While proper underlying focuses, procyclicality and systemic risks can arise from many factors, including aggregate shocks (e.g., commodity price shocks) and policy deficiencies (arguably, procyclicality and systemic risks mostly largely relate to weaknesses in the conduct of microprudential and monetary policies). These causes require their own approaches, including fixing deficiencies. These are not all causes that need to be addressed by macroprudential policies. Even though macroprudential policies can mitigate say the general financial or business cycle, or the presence of insufficiently disciplined large financial institutions, only externalities justify a macroprudential approach.
Identifying precisely the source of externalities operating through the financial system thus help determine the corresponding, specific macroprudential policies. While many, policy-oriented papers, notably at the BIS, had drawn attention to the need for a macroprudential approach (e.g., Borio, 2003, Borio and White 2003, White, 2006), these mostly did not adopt the formal perspectives of externalities. Several recent papers, however, have identified the externalities that can give rise to procyclicality and systemic risk, with Brunnermeier et al. (2009) one of the first to do so. De Nicolò, Favara and Ratnovski (2012), on which this section heavily draws, classify the ones known as follows (see also Allen and Carletti, 2011, Bank of England, 2011, Schoenmaker and Wierts, 2011):
1. Externalities related to strategic complementarities, that arise from the strategic interactions of banks and other financial institutions and agents, and which cause the build-up of vulnerabilities during the expansionary phase of a financial cycle;
2. Externalities related to fire sales and credit crunches, that arise from a generalized sell-off of assets causing a decline in asset prices, a deterioration of balance sheets of intermediaries and investors, and a drying up of financing, especially during the contractionary phase of a financial (and business) cycle; and
3. Externalities related to interconnectedness, caused by the propagation of shocks from systemic institutions or through financial markets or networks (“contagion”).
While one can classify externalities in other ways, the literature generally makes a similar distinction, that is, between externalities that are more of a time-series nature, i.e., give rise to procyclicality in good and bad times (under 1 and 2), and those more of a cross-sectional nature, i.e., due to interconnectedness (under 3). Table 1 organizes these along the vertical axis, with specific groups of tools (to be reviewed in section 4) along the horizontal axis. I review specific externalities under each heading next.
The Macroprudential Toolkit
The Macroprudential Toolkit
| Policy Tool | |||||
|---|---|---|---|---|---|
| Restrictions related to borrower, instrument, or activity | Restrictions on financial sector balance sheet (assets, liabilities) | Capital requirements, provisioning, surcharges | Taxation, levies | Other (including institutional infrastructure) | |
| Expansionary phase | Time varying caps/limits/rules on:
| Time varying caps/limits on:
| Countercyclical capital requirements, leverage restrictions, general (dynamic) provisioning | Levy/tax on specific assets and/or liabilities |
|
| Contractionary phase: fire-sales, credit crunch | Adjustment to specific loan-loss provisioning, margins or hair-cuts (e.g., through the cycle, dynamic) | Liquidity limits (e.g., Net Stable Funding Ratio, Liquidity Coverage Ratio) | Countercyclical capital requirements, general (dynamic) provisioning | Levy/tax (e.g., on non-core liabilities) |
|
| Contagion, or shock propagation from SIFIs or networks | Varying restrictions on asset composition, activities (e.g., Volcker, Vickers) | Institution-specific limits on (bilateral) financial exposures, other balance sheet measures | Capital surcharges linked to systemic risk | Tax/levy varying by externality (size, network) |
|
| Enhancing resilience | |||||
| Dampening the cycle | |||||
| Dispelling gestation of cycle | |||||
The Macroprudential Toolkit
| Policy Tool | |||||
|---|---|---|---|---|---|
| Restrictions related to borrower, instrument, or activity | Restrictions on financial sector balance sheet (assets, liabilities) | Capital requirements, provisioning, surcharges | Taxation, levies | Other (including institutional infrastructure) | |
| Expansionary phase | Time varying caps/limits/rules on:
| Time varying caps/limits on:
| Countercyclical capital requirements, leverage restrictions, general (dynamic) provisioning | Levy/tax on specific assets and/or liabilities |
|
| Contractionary phase: fire-sales, credit crunch | Adjustment to specific loan-loss provisioning, margins or hair-cuts (e.g., through the cycle, dynamic) | Liquidity limits (e.g., Net Stable Funding Ratio, Liquidity Coverage Ratio) | Countercyclical capital requirements, general (dynamic) provisioning | Levy/tax (e.g., on non-core liabilities) |
|
| Contagion, or shock propagation from SIFIs or networks | Varying restrictions on asset composition, activities (e.g., Volcker, Vickers) | Institution-specific limits on (bilateral) financial exposures, other balance sheet measures | Capital surcharges linked to systemic risk | Tax/levy varying by externality (size, network) |
|
| Enhancing resilience | |||||
| Dampening the cycle | |||||
| Dispelling gestation of cycle | |||||
Externalities related to strategic complementarities. These externalities arise during the buildup of risks in the boom period and are due to various reasons, albeit many not well understood. Historical experiences suggest that financial intermediaries tend to assume exposures to common credit and liquidity risk in the upswing of a business cycle, amplifying financial cycles and contributing to asset price volatility.
One reason is because of strategic complementarities arising in market interactions between rational agents, meaning that the payoff from a certain strategy increases with the number of other agents undertaking the same strategy. One source relates to increased competition in boom times, which can affect economy-wide credit standards. In the presence of imperfect information, banks need incentives to assess borrowers’ risk. In boom times, incentives are less to screen potential borrowers due to lower rents prompted by fiercer competition. As a result, they reduce screening intensity and increase lending, worsening the pool of borrowers (Ruckes, 2004; Dell’Ariccia and Marquez, 2006; Gorton and He, 2008). This reverses with lower credit origination and less competitive pressures in the contractionary phase.
Other sources are reputational concerns and incentive structure for financial managers. When managers care about perceptions of ability, their credit, investment or other policies may be influenced by those of others (Rajan, 1994). Excessive long-term risk-taking can arise under pay for short-term performance (Acharya, Pagano, and Volpin, 2013). Benchmarking in various forms can lead to externalities, e.g., an institution reporting poor performance will be evaluated more leniently if many others do so similarly. Institutions then have incentives to maintain risky lending, hide losses or otherwise copy each others’ behavior until the buildup of risks forces them to “coordinate” to a strategy of loss recognition and external financing contracts (Allen and Saunders, 2003, review). Complementarities can come from institutional rules, such as mark-to-market (fair value) accounting or the required use of (similar) “value-at-risks” models (Adrian and Shin, 2010, 2014). Other sources appear more behavioral, as when investors chase similar investment opportunities (Shleifer 2000, and Barberis, 2013, review) or ‘neglect’ the possibility of rare but large shocks (Gennaioli et al., 2013).
Externalities can also arise from the optimal ex-ante response of agents to ex-post government interventions. The prospect of a bailout can mean strategic complementarities, as it can lead institutions, especially banks, to engage ex-ante in correlated asset choices. Anticipating that simultaneous failures trigger a bailout (to prevent a financial meltdown), banks may find it optimal to correlate risks to maximize the probability that any failure is a joint failure (Farhi and Tirole, 2012; Acharya and Yorulmazer, 2007). As firms mimic each others’ strategy, overall vulnerabilities increase whether through correlated asset choices, maturity and exchange rate mismatches or otherwise (Ratnovski, 2009; Allen and Carletti, 2011). These vulnerabilities in turn can lead to or deepen a financial bust.
Externalities Related to Fire Sales and Credit Crunches. A fire sale occurs when an investor is forced to liquidate an asset at a time when potential buyers are also troubled. Given limited buyers, the asset is sold at a price below its fundamental value, causing losses to the seller (Shleifer and Vishny 1992; Allen and Gale 1994). Not only does this asset fetch a lower price, but similar assets held by other financial institutions may also decline in value. This reduces the capitalization and ability to post assets as collateral of all financial institutions, forcing them to liquidate other assets. The new round of selling triggers further losses, new selling, etc., thus creating a pecuniary externality.
Fire sales and credit crunches are an obvious possibility for banks since they issue liquid liabilities to fund illiquid assets, exposing them to the risk of having to liquidate investments prematurely, as happened in the Great Depression (Rajan and Ramcharan, 2014). Although guarantees and central bank support, such as deposit insurance and liquidity facilities, reduce the likelihood of fire sales, their effectiveness can be limited when banks also rely on wholesale funding, as many did before the crisis. Or when other important players in the intermediation process, such as broker-dealers and ‘shadow banks,’ that do not (formally) benefit from such support, have to sell assets they can also depress the values of (similar) assets banks hold.
Fire sales can also trigger an external financing, credit crunch with adverse real consequences. As banks balance sheets are impaired they will cut back on their financing. And as asset prices decline and collateral becomes less valuable, final borrowers (corporations, households and sovereigns) have less access to finance, which worsens the real economy (Goldstein, Ozdenoren and Yuan, 2013). Even more generally, small financial shocks can trigger demand and other real sector externalities, including from capital flows, and aggravated by the zero lower bound on interest rates (see Korinek, 2011; Schmitt-Grohe and Uribe, 2012; Farhi and Werning, 2013; Korinek and Simsek, 2014).
Even though fire sales and credit crunch externalities manifest themselves in a downturn, the imbalances that sow the risks are often built up in booms. The reason is that atomistic agents take prices as given, but on aggregate the equilibrium price depends on their joint behavior. As they do not internalize the possible effects of a generalized fire sale on ex-post borrowing capacity, agents may overborrow, leading to excessive leverage and inflated asset prices (Bianchi, 2011, Caballero and Krishnamurthy, 2003, 2004, Lorenzoni, 2008, Jeanne and Korinek, 2010, and Stein, 2012; Manconi et al., 2012; Merrill et al., 2012; see Brunnermeier, Eisenbach and Sannikov, 2013, for a review).
Externalities Related to Interconnectedness. Banks and other financial institutions are very interconnected, with distress or failure of one affecting others. Spillovers can arise because of bilateral balance sheets (interbank) and other exposures (Allen and Gale, 2000; Diamond and Rajan, 2011; Perotti and Suarez, 2011), asset price movements (as discussed above), or aggregate feedback from the real economy (Bebchuk and Goldstein, 2011). Financial institutions can reduce but not entirely eliminate these risks as interconnectedness is often beyond their individual control and actors do not internalize their implications for systemic risk (Acemoglu et al., 2013). Also, interconnectedness may arise for genuine mutual hedging and diversification motives (Wagner, 2011). Related, as the financial networks literature (Allen and Gale, 2007; Gaia et al., 2011) has shown, while high interconnectedness mitigates the impact of small shocks by spreading them, it can amplify large shocks since they can reach more counterparties.
Interconnectedness externalities are particularly strong for systemically important financial institutions (SIFIs).4 Unlike smaller institutions, distressed SIFIs cannot easily be wound down, since they are often complex, operate internationally, provide unique services, or are backbones of the financial infrastructure, making them “too big to fail” (Strahan 2013 reviews). Historically, most interventions in SIFIs are then also de facto bailouts, which protect creditors (and sometimes even shareholders and often management) from the full scale of losses. The anticipation of bailouts perversely affects the (risk-taking) incentives for SIFIs and other market participants. It introduces a race among institutions to become systemically important, as this lowers the cost of funding, and reduces market discipline for creditors of SIFIs, especially the riskiest ones (Ueda and Weder di Mauro, 2012). In turn, these behaviors lead to aggregate risk-shifting and distorted competition (see IMF, 2014b).
The externalities of a time-series nature can interact with those of a cross-sectional nature to create systemic risks. Rapid growth of large financial institutions during a boom means procyclicality gets reinforced by contagion risks. Conversely there can be complementarities in the tools to be used to mitigate either source of externalities.
III. Interactions with other policies and international dimensions
Macroprudential policies are not the only policies aimed at economic (including price) and financial stability. Others include monetary, microprudential, fiscal, as well as competition policies. Macroprudential policies interact with these. Furthermore, some macroprudential policies can be motivated by the need to correct for the “distortions” introduced by other policies. Macroprudential policies can also have international spillovers, both inward and outward, and consequently there can be overlap and interrelationships with capital flow management (CFM) policies. How to coordinate macroprudential policies with these other policies? I review the (limited) literature on this briefly next, focusing mostly on interactions between monetary and macroprudential policies as most relevant and most studied.
A. Macroprudential and monetary policies
Both macroprudential and monetary policies are useful for countercyclical management: monetary policy primarily aimed at price stability; and macroprudential policies primarily aimed at financial stability. Since these policies interact with each other, each may enhance or diminish the effectiveness of the other. IMF (2013a and 2013b) reviews the (limited) literature on the conduct of both policies in the presence of these interactions. It first presents an ideal, but unrealistic benchmark, in which both policies perfectly achieve their objectives. It then addresses three questions: If macroprudential policies work imperfectly, what are the implications for monetary policy? If monetary policy is constrained, what is the role for macroprudential policies? And with institutional and political economy constraints, how can both be adjusted?
Benchmark world, when policies work perfectly. Monetary policy alone cannot be expected to achieve financial stability effectively or efficiently because its causes are not always related to the interest rate level or the degree of liquidity in the system (which monetary policy can affect). For mitigating the effects of financial distortions or when financial distortions are more acute in some sectors of the economy than in others, monetary policy is too blunt a tool. Pricking an asset price bubble for example can require large changes in the policy rate (Bean and others, 2010). Similarly, using macroprudential policies primarily for managing aggregate demand may create additional distortions by imposing constraints beyond where financial instability originates. For example, to limit general credit growth may be too harmful from an aggregate economic perspective. It is thus desirable, when both policies are available, to keep monetary policy primarily focused on price stability and macroprudential policies on financial stability.
Monetary policy, however, does affect financial stability: (i) by shaping ex-ante risk-taking incentives of individual agents, affecting leverage and short-term or foreign-currency borrowing (Dell’Ariccia and Marquez, 2013, review); or (ii) by affecting ex-post the tightness of borrowing constraints, possibly exacerbating asset price and related externalities and leverage cycles. Similarly, macroprudential policies can affect overall output by constraining borrowing and hence expenditures in one or more sectors. These side effects imply that one needs to consider how the conduct of both policies is affected. Most analytical papers to date find the sole presence of side effects to have no major implications for the conduct of both policies when policies operate perfectly.5 In particular, most Dynamic Stochastic General Equilibrium (DSGE) models suggest that monetary policy not to change markedly when macroprudential policies are also used, even when different types of shocks are considered. A big caveat is that most models employ limited representations of financial systems and related financial frictions, and often use assumptions that imply linear relationships, making both policies operate mostly similar (see Benes, Kumhof, and Laxton, 2014a and 2014b, for DSGE-models with non-linearities).
When either macroprudential or monetary policies work imperfectly. In the real world policies do not operate perfectly. Furthermore, neither policy is immune to political pressures and time inconsistency issues. As such, the conduct of both may need to be adjusted to consider the weaknesses in the other, but how is conceptually and empirically unclear. Weaknesses in macroprudential policies mean monetary policy more likely needs to respond to financial conditions. Indeed, in models where macroprudential policy is absent or time invariant, but with financial “distortions” still present, optimal monetary policy responds to some degree to financial conditions, in addition to output and inflation (Curdia and Woodford, 2009, Carlstrom and Fuerst, 2010, Adam and Woodford, 2013). By extension, with imperfectly targeted or effective macroprudential policies, monetary policy may need to respond to financial conditions and “lend a hand” in achieving financial stability, also because of its more general reach (e.g., as “it gets in all of the cracks;” see Stein, 2013). This “leaning against the wind” argument is, however, not generally accepted (e.g., compare Bernanke and Gertler, 2001, with BIS, 2014; see also Yellen, 2014).
Similarly, macroprudential policy may need to respond to aggregate developments related to financial activities when monetary (and other) policies are constrained, as with economies pegging their exchange rate or in currency unions. The case of the euro area shows the economic (and financial) risks that arise when booms are not (or cannot) mitigated at the national level. When the effective monetary stance gives rise to macroeconomic imbalances or excessively strong overall risk-taking incentives, national macroprudential policies may need to be used, especially when other policies are imperfectly coordinated internationally (e.g., as when foreign lenders are not constrained from lending to the country).6 Of course, the macroeconomic risks need to be related to financial activities (e.g., a housing boom that is of macroeconomic concern, even when completely financed internationally). And when monetary arrangements are not adequate, strengthening monetary policy’s effectiveness will likely be better than using macroprudential policies as imperfect substitutes.
B. Interactions with other policies
There are, besides monetary policy, many policies that can interact with or condition the use of macroprudential policies. These include fiscal, microprudential, and other structural policies. I review the research in these areas briefly.
Fiscal policy. Tax policies can contribute to systemic risk when they encourage leverage, as when interest payments are tax deductable, or affect asset prices (see De Mooij, 2011, Keen and De Mooij, 2012). Macroprudential authorities have therefore an interest in the correction of such biases. Even when not contributing directly to risks, taxes can affect the conduct of macroprudential policies. Real estate taxes (property taxes, stamp duties) can be capitalized into house prices (e.g., Van den Noord, 2005), making (future) tax policies possibly relevant for financial stability. Since various Pigouvian taxes and levies can address systemic externalities (IMF 2010), coordination between macroprudential and fiscal agencies may be needed. Little is known though on the quantitative importance of these aspects. And fiscal policy in the aggregate matters as it can counter (or be a source of) procyclicality.
Microprudential. Macroprudential policies presume effective microprudential regulation and supervision. Most often, when conducted properly microprudential objectives will be aligned with macroprudential policies, but there can be conflicts (Osiński, Seal, and Hoogduin, 2013; Angelini, Nicoletti-Altimari, and Visco, 2012). This is most clear in bad times when a macroprudential perspective may suggest relaxing regulatory requirements – as they impede the provision of credit to the economy or contribute to fire-sale effects, while the microprudential perspective may seek to retain or tighten requirements – so as to protect the interest of depositors of individual banks or investors. In good times, conflict of interests are less likely, e.g., both authorities will ask banks to build up buffers, but the macroprudential perspective will likely still call for greater prudence. Some of this conflict is institutionally related. For example, accounting indicators, more often used by microprudential authorities, likely give a more positive picture of an institution’s balance sheet in boom time than a system’s view would. While recognized, how to address these issues largely remains an open question. And, as also argued by Jeanne and Korinek (2013), an ex-post strategy of cleaning up after a crisis can be part of an efficient approach to “managing” risks, thus calling for crisis management to coordinate with ex-ante policies.
Other, structural policies. Conflicts can also arise in the design of structural policies, as when risks arise from how microprudential policies are conducted. For example, a very high loan to value ratio is likely to increase the incidence of real estate booms. Even when set optimally from a microprudential perspective, capital requirements can increase overall procyclicality (Angelini et al. 2010; Repullo and Suarez, 2013). Or a public safety net, including deposit insurance, while reducing the risk of runs on individual institutions, can give rise to greater system risks (Demirguc-Kunt and Detragiache, 2002; Demirguc-Kunt, Kane and Laeven, 2008). The use of ratings may introduce (more) procyclicality (Amato and Furfine, 2004). And accounting rules aimed at greater transparency and fostering more market discipline can mean more procyclicality as chances of fires-sales increase when institutions mark asset to market (Leuz and Laux, 2010; Ellul et al., 2012,). Also, by affecting incentives for risk-taking, there can be an inverse U-shaped relationship between bank competition and financial stability (Allen and Gale 2004; see further Beck, 2008, Ratnovski, 2013). And house price developments will be importantly affected by land use and construction policies. These examples show that macroprudential policies need to be coordinated with many policy areas, in part as the need for them arises exactly from these other policies.
C. International coordination
International financial and policy spillovers. The de-facto international financial integration of most countries affects the desired use of and effectiveness of macroprudential policies. Given financial integration, cross border spillovers may arise as when the financial cycle is in an upswing in one country but in a downswing in another, or because countries are (or are not) using macroprudential policies.7 As argued by Shin (2012) and shown by Rey (2013) and others, there is much commonality to financial cycles globally, suggesting policies are naturally coordinated. The cycle appears largely driven by conditions in major advanced countries, however, and it is thus not obvious that the commonality itself or addressing it from the major countries’ perspectives alone is optimal for all countries. Regardless, being financial integrated means countries have less control over their own financial stability.
Policy spillovers can also arise (more likely) when countries vary in policies or calibrations to deal with similar risks, or in policy effectiveness. Aiyar, Calomiris, and Wieladek (2014) show that foreign bank branches increased their lending in the UK in response to tighter measures applied to local banks, a sign of cross-border competition and regulatory arbitrage. Or when policies are not effective at the source country to stem risks related to outflows, recipient countries can be negatively affected if they cannot stop inflows. Spillovers can arise when institutions adjust to local restrictions by decreasing or increasing cross-border activities. Aiyar, Calomiris, Hooley, Korniyenko and Wieladek (2014) show that, as supervisors required UK-based banks and subsidiaries to meet higher capital requirements during the 2000s, local banks lend less abroad which may or not have been optimal. Spillovers can also arise when institutions from country A reduce cross-border flows to country B in response to its rules and increase flows to country C (see Forbes, Fratzscher, Kostka, and Straub, 2012 for the case of capital controls).
Even though the scope for (policy) spillovers is large, the case for international coordination and cooperation depends on the presence of negative externalities. While the welfare gains from coordinating macroprudential policies have not yet been much analyzed (see Jeanne and Korinek, 2014 for some thoughts), analysis on multilateral aspects of CFM tools (e.g., Ostry, Ghosh, and Korinek, 2012) relates. Building on this, Korinek (2014) argues that spillovers can lead to inefficiencies under three circumstances: if policies are “beggar-thy-neighbor;” if policy instruments to deal with externalities operate imperfectly; and if global markets are incomplete or restricted (see also Brunnermeier and Sannikov, 2014). And Jeanne (2014) shows the need to coordinate when some countries are in a liquidity trap as the global real interest rate cannot adjust sufficiently.
While there can be some (limited) scope in principle, policy coordination is hard in practice (see Ostry and Ghosh, 2013). And indeed so far, coordination has been limited, with instruments and mechanisms only defined for the countercyclical and systemic capital surcharges in Basel III. While more progress can be envisioned, (policy) spillovers are likely to remain. For individual countries, CFM tools may then sometimes be part of a useful policy response (IMF 2012c). This raises how to coordinate between CFM tools and macroprudential policies. Here Korinek and Sandri (2014) provide a useful dichotomy: macroprudential policies should address externalities related to domestic credit and CFM tools those related to exchange rate movements. How to make this operational, however, remains to be determined (see further Ostry et al., 2012).
IV. Possible macroprudential tools and actual uses
This section first reviews the toolkit available in principle, and then the actual use of policies.
A. The Toolkit Available
Many macroprudential tools have been proposed and some have been used, even before the recent crisis. The toolkit available in principle is quite large and includes existing microprudential and other regulatory tools, taxes and levies, and new instruments. Table 1 categorizes these in a 3-by-5 matrix (for other classifications, see CGFS, 2010, IMF 2011b, and ESRB, 2014). Most tools considered to date apply to the banking system, mainly given the existence of microprudential tools adaptable to macroprudential objectives and related more extensive theory and knowledge. A lack of understanding of possible externalities in other financial market segments is, however, also at play (e.g., as in shadow banking, see Claessens, Pozsar, Ratnovski, and Singh, 2012 for a review; and in insurance, see IAIS, 2013). Note further that many instruments (can) also serve other policy objectives, including, besides microprudential, assuring consumer protection or fostering greater competition, and that other tools can be considered.
The matrix covers along the vertical axis the three goals (as per section 3) and along the horizontal axis five set of tools: a) quantitative restrictions on borrowers, instruments or activities; b) capital and provisioning requirements; c) other quantitative restrictions on financial institutions’ balance sheets; d) taxation/levies on activities or balance sheet composition; and e) other, more institutional-oriented measures, such as accounting changes, changes to compensation, etc. Except for a), which aims to affect demand for financing, all can be seen as affecting the supply side of financing. The first four measures are meant to be time-, institution-, or state-varying, while the fifth one is more structural.8
Tools under the 15 (3*5) combinations include those correcting (for factors that can give rise to) externalities and market failures or compensating for policies that can contribute to adverse financial dynamics (such as the pro-cyclicality introduced by microprudential capital requirements). Besides mapping each tool to specific externalities, with some tools possibly mitigating more than one, tools are ideally also mapped to intermediate targets, such as changes in credit, leverage, asset prices, interconnections and the like. Knowledge on what intermediate indicators to use and how to calibrate tools is still limited, however (see further IMF, 2013c, 2013d).
Use and Calibration of Macroprudential Policies. The preferred use of policies, in their extensive (whether or not to use a specific tool) and intensive margins (how much to use it), will vary by the degree of amplification in the financial (and real) sector cycles, exposures to systemic shocks and risks, and the effectiveness of (specific) policies. As such, many dimensions come into play, including a country’s structural, institutional and financial market characteristics. Models provide some limited guidance for use and calibrations.
DSGE-models with financial frictions can suggest an optimal mix of macroprudential and monetary policies (e.g., Kannan, Rabanal, and Scott, 2009; Quint and Rabanal, 2013). Or some historically derived indicators of (excessive) procyclicality and systemic risks, e.g., a notion of a “credit gap,” can suggest specific dynamic provisioning surcharges (Drehmann et al. 2011). And a Pigouvian tax on SIFIs can be made to depend on measures reflecting the size of interconnectedness externalities (Kocherlakota, 2013).
Many questions exist, however, on what measures reliably indicate systemic risk build-up, with both Type I and II errors, and on the time horizon at which risks can be detected. Notably, how to account for country circumstances and characteristics is still unclear. Obviously, some factors are likely relevant: the overall depth of a country’s financial system, which differs vastly; financial structure, e.g., the importance of banks versus capital markets, with institution-based measures likely of greater importance than borrower-based measures when most financing comes from a regulated system;9 the industrial organization and ownership structure, since a more concentrated system makes the application of tools easier, or because domestic, state-owned and foreign banks react differently to policies.
International financial integration and exchange rate regime matter as well. Openness affects exposures, both directly, as regards to say capital flows risks, and indirectly, given the strong links between behavior of capital flows and banking vulnerabilities (e.g., Hahm, Shin, and Shin, 2013; Cerutti, Claessens, and Ratnovski, 2014). Financial integration also affects how effective policies may be. A very open capital account and large foreign bank presence make circumvention more likely. And with a fixed exchange rate, monetary policy cannot be a possibly complementary tool. These and other considerations will affect which policy is best and whether CFM tools can complement (e.g., Hahm, Mishkin, Shin, and Shin, 2011).
Preferred use could also vary by the availability and effectiveness of fiscal and microprudential policies. High public debt makes countercyclical fiscal policy harder, making macroprudential policies more important. Microprudential supervision may face greater challenges in some markets. Institutional (e.g., lack of data, know-how and skills in supervisory agencies), political economy, and other constraints may lead countries to adopt macroprudential policies in specific ways. Use could also vary with other tools available to mitigate systemic risks. Stress tests could complement macroprudential policies.10
Furthermore, financial reforms are proceeding in various ways, some coordinated (e.g., new liquidity requirements) and some country-specific (e.g., Vickers, Volcker and Liikanen rules), making overall institutional environments in flux and requiring further adaptations.
B. Actual Use of Macroprudential Policies
Information on the actual use of macroprudential policies is limited, in part because (the use of) tools are not always clearly identified (some countries have adopted more explicit frameworks, but most have not yet). Some data have nevertheless been collected for some 65 countries by the IMF (see Lim et al., 2011, and Cerutti et al., forthcoming, for exact coverage and data definitions). The seven specific instruments reviewed here are: caps on loan-to-value (LTV) and debt-to-income (DTI) ratios, limits on credit growth (CG), limits on foreign lending (FC), reserve requirements (RR), dynamic provisioning (DP), and counter-cyclical requirements (CTC). One can organize these measures along the categories of Table 1: those aimed at borrowers (caps on LTV and DTI ratios); those aimed at financial institutions’ assets (CG and FC) and liabilities (RR); and those aimed at building buffers (DP, CTC).
Usage of policies in general. In the sample, 42 countries – of which 28 are emerging and developing and 14 advanced – implemented at least one instrument once during 2000-2013, while 23 never used any. Most usage is by emerging markets, consistent with their greater needs, being more exposed to external shocks, including from volatile capital flows, and having more “imperfect” and generally less liberalized financial systems with more “market failures.”
Countries use LTV ratios the most (Table 2, column 1): 24 used it at least once. Next are DTI (23), FC (15), RR (10), DP (7), CG (6), and CTC (5). Weighting by the length of time and relative to overall use (column 2), most often used is again LTV, 28% of country-year combinations when a policy was used. Following closely behind is DTI (24%), then RR (15%), FC (14%), CG (9%), DP (8%), and finally CTC (2%).
Overall Use of Macroprudential Instruments
Overall Use of Macroprudential Instruments
| Type of Instrument | Total Countries | Frequency of Use | Emerging Markets | Advanced Countries | Frequency of EMs-year | Frequency of ACs-year |
|---|---|---|---|---|---|---|
| (1) | (2) | (3) | (4) | (5) | (6) | |
| LTV | 24 | 28% | 13 | 11 | 20% | 55% |
| DTI | 23 | 24% | 16 | 7 | 25% | 20% |
| CG | 6 | 9% | 6 | 0 | 12% | 0% |
| FC | 15 | 14% | 12 | 3 | 14% | 15% |
| RR | 10 | 15% | 10 | 0 | 20% | 0% |
| DP | 7 | 8% | 6 | 1 | 7% | 9% |
| CTC | 5 | 2% | 2 | 3 | 2% | 2% |
| Total by classification | 42 | 100% | 28 | 14 | 100% | 100% |
Overall Use of Macroprudential Instruments
| Type of Instrument | Total Countries | Frequency of Use | Emerging Markets | Advanced Countries | Frequency of EMs-year | Frequency of ACs-year |
|---|---|---|---|---|---|---|
| (1) | (2) | (3) | (4) | (5) | (6) | |
| LTV | 24 | 28% | 13 | 11 | 20% | 55% |
| DTI | 23 | 24% | 16 | 7 | 25% | 20% |
| CG | 6 | 9% | 6 | 0 | 12% | 0% |
| FC | 15 | 14% | 12 | 3 | 14% | 15% |
| RR | 10 | 15% | 10 | 0 | 20% | 0% |
| DP | 7 | 8% | 6 | 1 | 7% | 9% |
| CTC | 5 | 2% | 2 | 3 | 2% | 2% |
| Total by classification | 42 | 100% | 28 | 14 | 100% | 100% |
Usage of policies by country groupings. Use varies among countries (columns 3 and 4). In advanced countries, LTV and DTI ratios are used the most, while other policies are rarely used. Differences are starker considering how long policies are used (columns 7 and 8). Emerging markets use more policies and longer than advanced countries do and tend to favor more foreign exchange and liquidity related policies (FC, RR), maybe due to their concerns with large and volatile capital flows and related systemic risks. But they also use CG more often, possibly as their systems are less liberalized. Advanced countries prefer the demand for credit related LTVs (55%) and DTIs (20%), perhaps out of concern with excessive leverage. The increased usage since the late 1990s by more countries reflects the growing recognition of the policies. Overall though policies were used four times more intensively by emerging markets than by advanced countries right before the crisis, with this ratio declining to 3.3 as advanced countries started to use them.
V. Research and other evidence on experiences
This section reviews the literature on the effectiveness of macroprudential policies. Most are cross-country, aggregate analyses, investigating cyclical aspects and notably in credit and housing markets. Some more micro, case studies exist, also largely focused on cyclical aspects, and some work focuses on cross-sectional, systemic risk aspects.
A. Aggregate, Cross-Sectional Studies, Focusing on Procyclicality
Several papers have analyzed effects of policies on various measures of financial vulnerability and stability (see also ECB, 2012, IMF, 2013a-d; notably IMF 2013d, Table 4 and 5, and ESCB, 2014, for reviews). Lim et al. (2011) document, using cross-country regressions, some policies being effective in reducing the procyclicality of credit and leverage. Specifically, tools such as LTV and DTI, ceilings on credit growth, RR, and dynamic provisioning rules can mitigate procyclicality. IMF (2013b) investigates, also in a cross-country context, how (changes in) policies affect financial vulnerabilities (credit growth, house prices, and portfolio capital inflows) and the real economy (output growth, and sectoral allocation, i.e., the share of residential investment), considering also whether effects are symmetric between tightening and loosening. Overall, both (time-varying) capital requirements and RRs significant affects credit growth, LTV limits and capital requirements (but not RRs) strongly affects house price appreciation rates, and RRs reduce portfolio inflows in emerging markets with floating exchange rates. They find no significant indication of asymmetric responses. LTVs appear to impact overall output growth, maybe through reducing construction investment, but no other policies do so.
Crowe et al. (2011) find that policies such as maximum LTV have the best chance to curb a real estate boom. They also argue that their narrower focus reduces their overall costs. And, measures aimed at strengthening the banking system (such as dynamic provisioning), even when failing to stop a boom, may still help to cope with the possible bust. IMF (2011a) finds LTV tools to be effective in reducing price shocks and containing feedback between asset prices and credit. Vandenbussche, Vogel, and Detragiache (2012) find that capital ratio requirements and non-standard liquidity measures (marginal reserve requirements on foreign funding or linked to credit growth) helped slow down house price inflation in Central, Eastern and Southeastern Europe.
Dell’Ariccia et al. (2012) find that macroprudential policies can reduce the incidence of general credit booms and decrease the probability that booms end up badly. Using specific policies, they find credit and interest controls and open foreign exchange position limits to be significant in most regressions. Consistent with a focus on vulnerabilities, policies reduce the probability of a boom that ends up in a financial crisis or subsequent economic underperformance, i.e., policies reduce the risk of a bust, while simultaneously reducing how the rest of the economy is affected by troubles in the financial system.
Claessens et al. (2013) investigate, using panel GMM regressions, how changes in balance sheets of some 2800 banks in 48 countries over 2000-2010 respond to specific policies. Controlling for endogeneity and country characteristics and macroeconomic policies (by including among others countries’ lagged GDP growth and interest rates), they find that measures aimed at borrowers – LTV and DTI caps, and CG and FC limits – are effective in reducing the growth in bank’s leverage, asset and noncore to core liabilities growth. While countercyclical buffers (such as RR, PRD, and DP) also help mitigate increases in bank leverage and assets, few policies help stop declines in adverse times, consistent with the ex-ante nature of macroprudential tools and the challenges in adjusting policies in times of stress (e.g., how quick and far to allow banks to reduce their capital buffers).
Kuttner and Shim (2013), using data from 57 countries spanning more than three decades, investigate whether nine non-interest rate policy tools, including macroprudential, help in stabilizing house prices and housing credit. Using conventional panel regressions, housing credit growth is significantly affected by changes in the maximum debt-service-to-income (DSTI) ratio, maximum LTVs, limits on exposure to the housing sector, and housing-related taxes. But only the DSTI ratio limit significantly affects housing credit growth when they use mean group and panel event study methods. And, of the policies considered, only a change in housing-related taxes discernible impacts house price appreciation.
Zhang and Zoli (2014) review the use of key macroprudential instruments and capital flow measures in 13 Asian economies and 33 other economies since 2000 and study their effects. Their analysis suggests that measures helped curb housing price growth, equity flows, credit growth, and bank leverage, with loan-to-value ratio caps, housing tax measures, and foreign currency-related measures especially effective.
While suggestive, these studies come with many caveats. Many struggle with identification and endogeneity – e.g., as policies are adopted when the cycle is already up – and other biases, which can only partially be addressed by econometric techniques (such as GMM). Almost all face challenges in controlling for other country characteristics, including the quality of microprudential supervision. Few consider both the use of a policy and its intensity (e.g., the presence of a LTV and its level, whether set high or low) or differentiate by phase of the cycle – e.g., to investigate whether policies help most in mitigating booms or building buffers for busts. Almost all focus on credit and housing developments, and none study risks in capital markets and non-bank financial institutions. And, obviously, not one identifies the specific externalities or market failures policies are supposed to address, but rather mostly analyze manifestations of financial cycles, especially asset prices, that are supposedly of “concern” (e.g., studies are less clear in how LTVs reduce systemic risks, rather than controlling house prices per se).
Nevertheless, both analytical reasoning and existing evidence suggest some basic directions. Higher sectoral capital requirements, such as the CCB and other capital surcharges, by definition can help in increasing resilience by creating additional buffers. Direct measures, such as caps on LTV and DTI ratios, can likely limit mechanisms creating positive feedback between credit growth and asset price inflation. Conversely, such caps can enhance resilience and reduce the risks of fire-sale dynamics.
B. Case and Other Studies on Procyclicality and Cross-Sectional Risks
Country-specific “case” studies investigating the role of macroprudential policies in reducing financial procyclicality often focus on specific risks or market segments, and use micro data. Jiménez et al. (2012) find for Spain that dynamic provisioning can be useful in taming credit supply cycles, even though it did not suffice to stop the boom (see also Saurina, 2009). More importantly, during bad times, dynamic provisioning helps smooth the downturn, upholding firm credit availability and performance during recessions. Using sectoral data, Igan and Kang (2012) find LTV and DTI limits to moderate mortgage credit growth in Korea. And policies appear to reduce real estate cycles in Hong Kong (Wong, Fong, Li and Choi, 2011).
Some use of macroeconomic tools can be interpreted with a macroprudential perspective. Dassatti Camors and Peydro (2014) investigate the effects of a large and unexpected increase in RR in Uruguay in 2008 using detailed, bank-firm matched data. Their evidence suggests some ambiguous results. On one hand credit growth declines on aggregate, but at the same more risky firms get more credit. They also document that larger and possibly more systemic banks are less affected. There may thus be tradeoffs using RR, since less credit does not necessarily mean less system risks (RR may still be beneficial as macroeconomic tool).
The UK is a case where the use of microprudential tools over the period 1998-2007 has been interpreted with a macroprudential perspective. Aiyar, Calomiris and Wieladek (2013) show that bank-specific higher capital requirements dampened lending by banks, with quite strong aggregate effects: an increase in requirements of 1% reduced bank lending by between 5.7% and 7.6%, a high multiplier.11 Tighter monetary policy also reduced the supply of lending, but not that of large banks.
A case study analyzing house prices for Israel (IMF, 2014a) shows that macroprudential measures have effects, but only over the six-month period following adoption, with LTVs more effective than DP and CTC. And while policies reduce somewhat transactions, evidence is limited that they contribute to curb house price inflation. Israel also shows that macroprudential policies can create challenges for communication and accountability, even more so when loose monetary policy conditions, proper in their own right, provide opposing forces. And they can have social and political sensitivities, notably when first-time buyers are excluded from the housing markets. Other countries, like Canada and Sweden, have been facing similar challenges, with, in environments of low interest rates, strong increases in house prices and household debt, even though they use some macroprudential policies.
To limit systemic liquidity risks in Korea, a macroprudential stability levy on short-term foreign exchange lending and a core funding ratio were imposed (Shin, 2010). Analysis (IMF, 2012a) suggests that these measures contributed to a shift away from short-term foreign exchange funding, mostly driven by shifts of foreign branches towards longer term funding. This in turn may have made, as Bruno and Shin (2014) show, interbank capital flows less sensitive to global financial conditions compared to other Asian countries. Aregger, Brown and Rossi (2013) find for Switzerland, where taxes vary across cantons, that higher capital gains taxes exacerbate house prices dynamics while transaction taxes have no impact.
Basel III includes a countercyclical buffer, CCB, and BCBS (2010a) has suggested a methodology for setting it, with bodies such as the ESRB (2014) providing further guidance for their regional (EU) jurisdictions. The CCB is loosely calibrated on the probability and cost of systemic crises (see Drehmann and others, 2010). The guidance suggests increasing capital if credit to GDP rises substantially above its trend value, e.g., up to 2.5 percent of risk-weighted assets if the so-called credit-to-GDP gap rise above 10 percentage points, with room for discretion whether and when to invoke (and an ability to impose a higher CCB). Some countries (UK, Switzerland, India, and New Zealand) are implementing the CCB.12 As its incentive effects are likely limited, its value derives mainly from providing higher buffers in bad times. Questions remain though, notably on what basis to release the CCB when the cycle turns (some favoring adverse developments in asset prices, which are timelier; others in credit markets, which are less subject to interpretation), but also on how to adapt the CCB when credit is a small part of overall financial intermediation (as in the US).
In terms of reducing systemic risk of a cross-sectional nature, the BCBS (2013) has agreed on a methodology for the systemic capital surcharges for G-SIBs and D-SIBs and determined (and published) the individual surcharges (from 1% up to 3.5%) for the 29 G-SIFIs identified (FSB, 2013). Some individual countries, such as Austria, Denmark, Singapore, and Sweden, have gone beyond the BIII standards and put in place higher capital requirements for their large domestic banks, and a number of other countries plan to do so as well. And the US has adopted a more stringent leverage requirement for large banks, while Switzerland has additional contingent capital and leverage requirements for its large institutions. No studies exist, however, on the impact and effectiveness of these measures or of the CCB.
Otherwise, work has mostly focused on the identification and measurement of systemic risks due to contagion and other spillovers in interbank and other financial markets (Bisias et al. 2012, and Adrian, Covitz, and Lang, 2013, reviews tools for financial system monitoring, as applied largely to the US; see Arsov et al. 2013 for a more general review). Many central banks, supervisory agencies and international agencies now also supervise their large financial institutions, including insurance corporations, more closely (as the “too big to fail” problem is still prevalent; see further IMF, 2014b, and Laeven, Tong and Ratnovski, 2014). Many also conduct (regular) stress tests, which, inter alia, help to identify those institutions more likely to cause systemic distress.13 Other relevant efforts underway are new regulations for shadow banking (Adrian and Ashcraft, 2012) and financial infrastructure reforms.
So far, however, these exercises are mainly aimed at supervisory actions – e.g., asking for more capital or winding down of weak institutions – or institutional and structural changes – e.g., greater use of central clearing counterparties. They could though also be linked to the use and intensity of macroprudential tools. Network and interconnection models, and other such cross-sectional tools, for example could help with designing and calibrating tools or infrastructures. So far, however, the mapping between risks and tools (e.g., how to map risk of contagion into preventive measures) seems not clear enough for policy applications. These country case and other studies advance over the cross-country analyses in that they are better at identifying specific channels. At the same time, they come with the caveat that they do not control for, or allow one to explore, the role of different country circumstances and conditions. Being focused on one segment of the financial system, many often do not analyze circumventions and risk transfers to other, possibly less regulated parts. The ideal, analyzing comprehensive micro data for many countries, however, has so far been largely elusive.
C. Costs and Tradeoffs
By constraining borrowing and hence expenditures in one or more sectors of the economy, macroprudential policies can affect overall output. Conceptually, the transmission of macroprudential policies to financial and real variables could vary across tools (Kashyap, Berner and Goodhart, 2011, provide a model to assess the costs and benefits of various policies). Obvious examples are CTC and RR that may affect overall lending and output, whereas LTV limits have more sectoral impacts. Policies can in principle also affect the overall price setting process (i.e., by making the allocation of resources across sectors less flexible). And these effects may differ not just by tool used but also by the stage of the country’s financial or economic cycles.
Actual quantitative effects of policies on the real economy, however, are not well known, in large part because due to a lack of data and experiences. Some papers nevertheless try to assess these impacts. The, necessarily preliminary, empirical analysis finds the short-run effects on overall output to be small, even for broad-based tools, such as capital and liquidity requirements (IMF, 2013b).14 Moreover, some of the real sector costs can be countered by appropriate variations in monetary policy (unless it itself is constrained). But these findings remain very tentative with work on the relative strength of the effects across tools even more limited (see further CGFS, 2012).
Importantly, as with other “risk”-based policies (e.g., monetary policy which takes actions under uncertainty), macroprudential policies will have to weigh Type I and Type II errors. Analytical frameworks for assessing the associated costs and benefits (as laid out in IMF 2012b, De Nicolò, Gamba and Lucchetta 2012, Blancher et al. 2013; and Arregui and others 2013a, b), while sometimes still basic (e.g., Elliott, 2011), can help to assess tradeoffs of policies in terms of specific parameters (to be estimated or judged) – like the probability of crisis, the loss given a crisis and the cost of a policy decision – and thereby offer some guidance.
VI. Broader lessons and remaining issues for research and policy making
The recent crisis has led to a reexamination of policies for macroeconomic and financial stability. Part of the evolving thinking involves the adoption of a macroprudential approach, to mitigate boom-bust patterns and systemic risks in financial markets. Many countries, advanced and emerging, have signed on to this new paradigm. Its objectives, conceptual foundations and exact features, however, are still to be determined. I highlight some major knowledge gaps and where practices at times are confused.
On the conceptual side, what the debate, and some of the literature, not always recognizes is that correcting externalities needs to be seen as an intermediate target. Only by adopting policies that control or reduce externalities can one mitigate the market failures that lead to systemic risk. As such, each externality ideally is corrected by a specific tool (of course, tools need not differ by externality and can complement each other; capital (surcharges) for example may be important in reducing several externalities). Regardless, the start is a clear recognition of the causes for systemic risk. Here much more analytical work on specific externalities arising in financial intermediation is needed. Without this, the danger arises that macroprudential policies are used for general management of business and financial cycles, which introduces distortions, adversely affecting resource allocation, undermines transparency and accountability, and (further) exposes regulators to political pressures.
With actual experiences still limited, evidence on the effectiveness of specific tools is only slowly accumulating and comes with many (economic and econometric) caveats, making it difficult to determine which policies to use and when to tighten or loosen them. Furthermore, while addressing one distortion may reduce some manifestations of risks, it can also worsen overall financial stability. Also tools may not be able to reach some activities that can lead to systemic risks, and tighter regulations create stronger incentives for circumvention, risking vulnerabilities building up outside of the regulatory perimeter and policymakers’ sight. Moreover, institutional constraints may impede the optimal deployment of instruments. Cooperation and coordination with microprudential supervisory agencies and international may be legally or institutionally difficult. Furthermore, while more data are being collected (e.g., Financial Soundness Indicators (FSI) and SIFI data by the IMF and BIS respectively),15 deficiencies in quantity and quality of data can hinder analyses and calibrations (Cerutti, Claessens and McGuire, 2014; Heath, 2013; and Brunnermeier and Krishnamurthy, 2014).
Many of these factors will vary across countries, with developing countries for example likely facing more institutional and data hurdles as well as greater risks of discretionary policy implementations. Overall the best approaches given specific country conditions and characteristics remain thus largely open questions (see Acharya, 2013, and Shin, 2013).
Besides challenges in measuring risks and calibrating tools are political economy pressures and risks. These of course relate in part to limited knowledge – on the effectiveness, costs and distortions of tools, challenge in calibrations, adaptations, perimeter (e.g., shadow banking), interactions among policies and conflict of interests with other goals, and (international) coordination – with rules-based policies thus far off. As such institutional design should allow for sufficient analytical capacity and lessons to be learned. As the recurrence of crises show, however, even with more knowledge (as exists on say microprudential policies), robust policies aimed at financial stability are not easy to implement. Similar to other attempts, macroprudential policies may fall thus short.
This makes an institutional design robust to both ex-ante pressures and ex-post risks all the more important. Design involves the location of the macroprudential policy function, with different models – centralized, inside the central bank or outside it, or using a committee structure – being considered. Each model has various benefits, but also risks (see Nier et al. 2011). Regardless of model, there is a need for transparency and accountability in the conduct of macroprudential policy as well as operational independence. While research has addressed the benefits of independence in the monetary policy function, and identified some modalities for achieving it, sounds governance arrangements for the macroprudential policy function (as well as for micro-prudential regulation and supervision) often remain to be adapted (see further IMF, 2013e).
All in all, given these and other limits on current knowledge, one should proceed with some modesty. A “Bayesian” updating approach, where those tools for which impact is well known are used while others are only used as one learns more, may then be attractive, also as it reduces some of the political economy risks (see also Calomiris, 2013). Institutional designs also have to proceed with caution. This more gradual approach does not mean progress. Policy prioritization would help avoid too much discretion, and too little transparency and accountability. And this path could still achieve a key and possible main objective of the new paradigm: a system-wide financial stability view accepted in all aspects of policy making, including microprudential, monetary, fiscal and competition areas. This change in mind-set is needed, and should proceed anyhow. As more data and research come available, one can then further improve the motivations, calibrations, adaptations, and (institutional) designs of macroprudential policies and adopt specific ones.