Trade-offs in Bank Resolution; IMF Staff Discussion Note 18/02

February 2018

SDN/18/02

I M F S T A F F D I S C

U S S I O N N O T

Trade-offs in Bank Resolution

Giovanni Dell’Ariccia, Maria Soledad Martinez Peria,

Deniz Igan, Elsie Addo Awadzi, Marc Dobler, and

Damiano Sandri

DISCLAIMER: Staff Discussion Notes (SDNs) showcase policy-related analysis and research being

developed by IMF staff members and are published to elicit comments and to encourage debate. The

views expressed in Staff Discussion Notes are those of the author(s) and do not necessarily represent the

views of the IMF, its Executive Board, or IMF management.

TRADE-OFFS IN BANK RESOLUTION

2 INTERNATIONAL MONETARY FUND

Trade-offs in Bank Resolution

Prepared by Giovanni Dell’Ariccia, Maria Soledad Martinez Peria, Deniz Igan,

Elsie Addo Awadzi, Marc Dobler, and Damiano Sandri

†

Authorized for distribution by Maurice Obstfeld, Tobias Adrian, and Ross Leckow

February 2018

DISCLAIMER: Staff Discussion Notes (SDNs) showcase policy-related analysis and research

being developed by IMF staff members and are published to elicit comments and to encourage

debate. The views expressed in Staff Discussion Notes are those of the author(s) and do not

necessarily represent the views of the IMF, its Executive Board, or IMF management.

JEL Classification Numbers: G21, G28, H81, K20

Keywords: Bank resolution, Spillovers, Financial regulation

Authors’ E-mail Addresses:

gdellaricc[email protected]; digan@imf.org;

mmartinezperia@imf.org; dsandri@imf.org;

eaddoawadzi@imf.org; mdobler@imf.org

†

The authors are grateful to John Bluedorn, Enrica Detragiache, Rishi Goyal, Anna Ilyina, Marina Moretti, Alvaro Piris

Chavarri, Mahmood Pradhan, Miguel Savastano, and Miguel Segoviano for valuable comments and discussions. Nicola

Babarcich and Antoine Malfroy-Camine provided outstanding research support and Gabriela Maciel invaluable

administrative support.

TRADE-OFFS IN BANK RESOLUTION

INTERNATIONAL MONETARY FUND 3

CONTENTS

EXECUTIVE SUMMARY ________________________________________________________________________ 4

INTRODUCTION _______________________________________________________________________________ 5

I. KEY INSIGHTS FROM A SIMPLE MODEL __________________________________________________ 7

II. EMPIRICAL EVIDENCE ____________________________________________________________________ 11

III. RESOLUTION FRAMEWORKS IN PRACTICE ______________________________________________ 24

IV. CONCLUSION _____________________________________________________________________________ 26

Figures

1. Optimal Resolution Framework as a Function of Spillover Intensity __________________________ 9

2. Bank Size and Reaction to Events Altering Bail-out Expectations ____________________________ 14

3. Holdings of Bank Equity, Contingent Capital, and Bail-in-able Debt_________________________ 23

Tables

1. Market Reaction to Bail-out Events _________________________________________________________ 13

2. Borrowing Costs and Risk Metrics for Large versus Small Banks ____________________________ 15

3. Size and Risk Taking ________________________________________________________________________ 16

4. Market Reaction to EU Bail-in Events ________________________________________________________ 18

5. Market Reaction to US Bail-in Events________________________________________________________ 19

6. Reducing the Cost of Resolutions ___________________________________________________________ 25

Boxes

1. Fiscal Implications of Bail-outs ______________________________________________________________ 28

2. What Is Bail-in? _____________________________________________________________________________ 29

3. Resolution Regimes in Key Large Home Jurisdictions _______________________________________ 30

4. Making Bail-in a Credible Policy Option for Systemic Banks _________________________________ 31

5. Good International Practice in Public Solvency Support _____________________________________ 32

Appendix I. A Simple Model of Bail-ins and Bail-outs __________________________________________ 33

Appendix II. Sample Coverage and Empirical Specifications ___________________________________ 35

Appendix Table 1. Bail-in Events in Europe ____________________________________________________ 38

Appendix Table 2. Bail-in Events in the United States __________________________________________ 39

References _____________________________________________________________________________________ 40

TRADE-OFFS IN BANK RESOLUTION

4 INTERNATIONAL MONETARY FUND

EXECUTIVE SUMMARY

During the global financial crisis, national authorities faced a steep policy trade-off in dealing with

systemic bank failures. On the one hand, they knew bail-outs would reinforce expectations of future

public support for distressed financial institutions. This could then undermine market discipline and

lead to excessive risk taking—potentially seeding the ground for the next crisis. On the other hand,

partly due to the absence of legal powers to resolve systemic failures through bail-ins, the use of

public resources seemed necessary to prevent distress in one bank from spreading to others and

becoming system-wide and to contain the economic and social costs of the crisis. Thus, in most cases,

failing banks were bailed out. Most of the costs and risks were borne by taxpayers, sometimes to such

a degree that the financial standing of the sovereign was threatened.

Since then, reforms have aimed to reduce the likelihood of crises and minimize costs should a crisis

occur, including by shifting the burden to private investors and improving the trade-off between bail-

outs and bail-ins. First, given the fiscal and moral hazard risks associated with government-funded

resolution, the consensus is that bail-outs need to be the exception rather than the rule. With this

objective in mind, reforms have aimed at formally delimiting the role of fiscal resources in the context

of crisis resolution. Several countries have imposed strict conditions on the use of public funds in

support of ailing banks and have introduced measures aimed at minimizing moral hazard. Second,

new frameworks provide comprehensive powers to resolve a financial institution, including by bailing

in private stakeholders (equity holders and unsecured and uninsured creditors). These include

statutory bail-in powers, which enable resolution authorities to terminate or write down unsecured

liabilities of failing banks and to convert claims of unsecured creditors into equity. The reforms have

also sought to contain potential financial stability risks stemming from bail-ins by ensuring that banks

(especially, large and complex ones) are subject to adequate loss-absorbing capacity requirements,

and have aimed to make these banks more resolvable via effective resolution planning.

This note revisits the trade-off entailed in a policymaker’s decision on the relative role of bail-ins of

private stakeholders and public bail-outs. It does this by presenting an illustrative welfare-based and

micro-founded model that juxtaposes the spillover effects from bail-ins and the moral hazard

consequences of bail-outs. It also provides empirical evidence consistent with the existence of moral

hazard effects associated with bail-outs and of spillovers associated with bail-ins. Finally, it discusses

progress in shifting the burden of a crisis to private investors, strengthening resolution frameworks,

and improving the bail-in/out trade-offs by enhancing resolvability and increasing loss absorbency.

The note supports the ongoing reform agenda to provide resolution authorities with effective bail-in

powers and stresses that frameworks should aim at minimizing moral hazard associated with bail-

outs. Nonetheless, it also emphasizes the need to allow for sufficient, albeit constrained, flexibility to

be able to use public resources in the context of systemic banking crises—when spillovers are

significant and deemed likely to severely jeopardize financial stability. Furthermore, the analysis calls

for continued efforts to enhance loss-absorbing capacity, ensure that holders of bail-in-able debt are

those best situated to absorb losses, and improve arrangements for cross-border resolution. This is

essential to further boost the effectiveness of bail-in powers and contain the risk of spillovers.

INTRODUCTION

During the global financial crisis, governments in the United States and Europe resorted extensively

to public bail-outs to prevent bank failures from destabilizing the financial sector and the economy.

This strategy fueled strong public resentment against using scarce fiscal resources to rescue banks,

especially given the fiscal consolidation efforts that followed. Furthermore, the use of bail-outs

reignited the well-known debate about their moral hazard impact on the behavior of financial

institutions. The perception that there would be few consequences for those responsible for the

banks’ losses reinforced this concern and public frustration about the handling of the crisis.

The academic literature and policy debate have long recognized that bail-outs entail a policy trade-

off between ex ante and ex post efficiency. On the one hand, expectations of public financial

support for distressed financial institutions may undermine market discipline and lead to excessive

risk taking—essentially seeding financial vulnerabilities that may precipitate a crisis. Expectations

of a bail-out may also trigger a leverage cycle, where lending standards are relaxed and leverage

becomes too high in boom times and too low in crisis times (Geanakoplos 2010). On the other

hand, in a crisis, the use of public resources to support the financial sector may sometimes be

necessary to contain the effects of system-wide financial distress.

A key question for policymakers

is how to balance these two effects—where to position the system along the trade-off (how much

to bail in and how much to bail out)—and how to improve the trade-off itself.

Against this backdrop, recent regulatory reforms and international standards on resolution have

placed significant emphasis on reducing the need for and mitigating the risk of future bail-outs,

including by improving the viability of bail-ins. Put differently, bail-outs need to become the

exception rather than the rule. To this end, a new international standard—the Financial Stability

Board’s Key Attributes (KA) of Effective Resolution Regimes, issued in 2011—advocates for

stronger resolution powers, available at early stages of distress, combined with better planning,

resolvability assessments, and cross-border cooperation. The KA aim to make systemic financial

institutions resolvable “without severe systemic disruption and without exposing taxpayers to loss.

. .” while making “it possible for shareholders and unsecured and uninsured creditors to absorb

losses.” Bail-in powers, which enable resolution authorities to recapitalize financial firms by

reducing the nominal value (that is, a “haircut”) of bank liabilities or converting them into equity,

are a key feature of the KA.

Resolution regimes that can allocate losses effectively among bank stakeholders through bail-ins

are beneficial for several reasons. First, stakeholders that could be exposed to loss in the event of

failure are likely to impose greater discipline on managers, thus reducing leverage and excessive

risk taking. This should in turn reduce the likelihood that banks fail. Second, by recognizing the

potential for loss and by calling for adequate loss-absorbing capacity, these frameworks may

reduce the risk of systemic spillovers. Third, by clarifying ex ante how losses would accrue to private

creditors in resolution, these frameworks may help address cross-border burden-sharing issues.

Not least, they reduce the direct fiscal cost of the crisis and may weaken the feedback effects

between sovereign and bank vulnerabilities (the “sovereign-bank nexus”).

TRADE-OFFS IN BANK RESOLUTION

6 INTERNATIONAL MONETARY FUND

Reform efforts since the crisis have improved the trade-off between bail-ins and bail-outs by

seeking to make bail-ins a credible option and bail-outs less likely. In some countries (especially in

Europe), the default approach to handling distressed banks before the crisis was bail-outs, and this

had fiscal and moral hazard costs. The reforms aimed to change this faulty approach and establish

frameworks that allow for orderly bail-ins, while maintaining some flexibility to provide public

funding to preserve financial stability and contain the macroeconomic consequences of a systemic

crisis. Yet some fear that resolution reforms may have gone too far and overly restrict a

government’s ability to use fiscal resources when necessary (see, for instance, Geithner 2014). For

example, a bail-in could create expectations for further bail-ins in other banks. This could

undermine investor confidence even in healthy banks. Spillovers could also arise if those

experiencing losses are forced to rebalance their portfolio and sell their claims on the initially

nondistressed banks, threatening the stability of these other banks. The key feature of these

spillovers is that they involve externalities that may not be fully priced by banks and private

investors and therefore provide a rationale for policy intervention.

In this note, we examine the economic forces that determine the relative costs and benefits of bail-

ins and bail-outs. We use the term “bail-in” in a generic sense, as the ability of resolution authorities

to impose losses on private stakeholders in order to recapitalize a failing bank. Bail-in-able claims

comprise equity and unsecured and uninsured liabilities with certain features. We use “bail-out” to

refer to the ability of governments to provide public funds to restore the solvency of banks. We

consider a model that provides a framework to assess under what circumstances losses may need

to be borne by the public sector rather than private investors. The answer crucially hinges on the

trade-off between the moral hazard costs associated with bail-outs and the potential spillovers

arising from bail-ins. We then examine the pre-reform trade-off by presenting empirical evidence

on these costs and spillovers. By reviewing the literature and performing some new analysis, we

confirm that the expectation of bail-outs lowers funding costs and, in some instances, generates

greater risk taking. We assess the extent of spillovers by looking at pre-reform events during which

bank stakeholders suffered losses and find evidence consistent with financial spillovers, though we

recognize that the evidence lends itself to alternative interpretations. Finally, we discuss the

progress made under the reformed frameworks to improve the bail-in/out trade-offs.

The key takeaways from the note are as follows:

• Recent regulatory reforms have improved the trade-off between bail-ins and bail-outs.

Effective resolution frameworks (as contemplated by the KA) are likely to reduce the potential

spillovers from bail-ins. Better-defined constraints on the use of public funds have likely

reduced the expectation of future bail-outs and the associated potential for moral hazard.

Greater loss-absorbing capacity has reduced the probability of financial distress.

• Bail-outs should be the exception not the rule—their use justified as a last resort, exclusively

when financial stability is gravely threatened, and structured to mitigate the associated costs.

They should occur only alongside loss sharing with private stakeholders of the troubled bank

and time-bound restructuring plans that address the underlying weaknesses and help restore

the bank’s long-term viability.

INTERNATIONAL MONETARY FUND 7

• Yet the framework should allow for the use of some public funds, with appropriate safeguards,

during systemic crises if and when this is necessary to protect financial stability. Public funds

may be needed if imposing extensive losses on private stakeholders would unleash large

spillovers. In these exceptional cases, the moral hazard and fiscal costs of bail-outs would be

preferable to the disruptive effects that spillovers associated with bail-ins could have on

financial stability and the economy at large.

• To further reduce the recourse to bail-outs, policymakers should continue ongoing efforts to

enhance resolvability and minimize the risk that bail-ins may result in large spillovers. Further

clarifying which financial instruments can be subject to bail-in and increasing the loss-

absorbing capacity of major financial institutions is key. Regulation should ensure that holders

of loss-absorbing capacity in banks are those most capable of understanding and absorbing

losses with a low risk of transmitting further shocks to the financial system and the economy.

Furthermore, more progress is needed to improve cross-border resolution aspects.

The rest of this note is structured as follows. Section I presents the theoretical model. Section II

discusses the empirical evidence. Section III provides a brief discussion of the recent reforms to

enhance resolution frameworks. Section IV concludes.

I. KEY INSIGHTS FROM A SIMPLE MODEL

To gain clear insights into the optimal use of bail-ins and bail-outs, we develop a simple banking

model that builds on Sandri 2015 and Cordella, Dell’Ariccia, and Marquez 2016. This section

provides an overview of the model and presents its key implications; we refer the reader to

Appendix I for the formal details.

The model features a continuum of banks that are heterogeneous in size. Banks raise deposits and

issue debt to finance loans. Loan repayments are subject to both idiosyncratic and aggregate

shocks. Banks can increase the likelihood that loans are repaid by exercising a costly monitoring

effort that can be interpreted as an inverse measure of risk taking. They choose the level of

monitoring to maximize expected profits. Thanks to limited liability, banks are not responsible for

losses beyond their capital buffer. Losses exceeding capital are covered with a bail-in of private

creditors, a public bail-out, or a combination of the two.

The key feature of the model is that bail-ins may entail systemic spillovers; that is, they may impose

negative externalities on society at large in addition to the losses borne by the bank’s stakeholders.

These externalities are modeled in reduced form, possibly proportional to the total size of bail-ins

in the banking sector. They can involve bankruptcy costs, heightened macro-financial uncertainty,

and a deterioration of the economic outlook. In practice, spillovers can spread through several

channels. They can operate mechanically through balance sheet transmission, whereby a bail-in

can jeopardize the solvency of exposed stakeholders and trigger bankruptcy chains. Or a bail-in

The combination or mix of bail-ins and bail-outs can be interpreted as corresponding in practice to burden

sharing, where part of the losses is borne by private stakeholders and the remaining part is assumed by the

government and, ultimately, taxpayers.

TRADE-OFFS IN BANK RESOLUTION

8 INTERNATIONAL MONETARY FUND

can act as a wake-up call and lead to a sudden reassessment of bank risk. This can in turn

undermine banks’ market access and have negative repercussions for credit supply and GDP. The

strength of spillovers varies across time and depends on broader economic and financial

conditions, including the amount and structure of loss-absorbing capacity. For example, bail-in of

a midsize bank may entail no spillovers if it happens during an economic upswing, but may become

destabilizing if it occurs at a time of severe distress in the banking sector. The strength of spillovers

may also vary across countries, reflecting different macro-financial and institutional characteristics.

For instance, countries with more developed or larger financial sectors may be prone to spillovers

due to greater interconnectedness across institutions and stronger macro-financial linkages.

Whereas bail-ins can be socially costly due to spillovers, bail-outs entail costs as well. The model

recognizes that bail-outs may have administrative fixed costs (for example, operating a program

of government investment and divestment). Bail-outs also involve fiscal costs and may threaten

sovereign debt sustainability. The model leaves aside this aspect since bail-ins can also have fiscal

consequences when spillovers endanger output and fiscal revenues. Box 1 shows that assessing

net fiscal costs of bail-outs based on up-front outlays may be misleading because mitigating the

output effects of a financial crisis can prevent even bigger tax revenue losses.

Crucially, the model emphasizes that bail-outs generate moral hazard by leading to insufficient

monitoring by banks; that is, to excessive risk taking. This can occur through two main channels.

First, bail-outs can undermine market discipline. When bank stakeholders expect to be bailed out,

they do not penalize banks for taking on excessive risk by charging higher interest rates on their

liabilities.

By weakening market discipline, the expectation of bail-outs can also lead to higher

leverage, possibly magnifying dangerous leverage cycles (Geanakoplos 2010; Adrian and Shin

2014). This can further increase risk taking because of a stronger risk-shifting motive: due to limited

liability, the bank neglects the losses associated with its default, so that it finds it optimal to exercise

less monitoring the more leveraged it is. Second, bail-outs can generate moral hazard by providing

transfers to bank shareholders and managers. For example, this can happen if bail-outs prevent

the ousting of bank management or preserve equity values. Then, by providing some

compensation even in the case of large losses, bail-outs promote risk taking.

In designing a framework to address bank losses through bail-ins or bail-outs, policymakers need

to carefully consider the relative costs of these two instruments. Specifically, the model highlights

the key trade-off between the possible spillover effects associated with bail-ins and the moral

hazard consequences and fixed costs of bail-outs. Figure 1 illustrates the model implications about

how an optimal resolution approach should depend on the intensity of spillovers. The figure shows

how losses exceeding bank capital should be covered through bail-ins and bail-outs (panel 1), the

implications for bank risk taking (panel 2), and the consequences for social welfare (panel 3).

This argument hinges on the assumption that bank stakeholders can observe how much risk the bank is taking.

If they cannot, say, because of the complexity of a bank’s operations, they cannot penalize additional risk taking

by proportionally raising funding costs. So, market discipline is absent. Bail-outs then merely reduce funding costs

by lowering or eliminating the risk of bail-ins. In turn, this increases the bank’s franchise/charter value, thus

strengthening incentives for prudent risk management. See Section II for more on this counterargument.

Careful bail-out design could mitigate this effect. See Box 5.

INTERNATIONAL MONETARY FUND 9

Figure 1. Optimal Resolution Framework as a Function of Spillover Intensity

Consider first the optimal resolution approach in the absence of spillovers. In this case, bail-ins

entail no social welfare costs since they simply reallocate losses from the bank to its creditors

without imposing externalities. By contrast, bail-outs are costly not only due to fixed costs but also

because they generate moral hazard and increase risk taking. So when spillovers are not a concern,

it is best to avoid bail-outs and absorb losses entirely through bail-ins.

When instead spillovers are present, bail-ins are socially costly since they not only involve a transfer

of losses from the bank to its creditors but also impose negative externalities on society at large.

In choosing whether to absorb bank losses with bail-ins or bail-outs, it is crucial to trade off these

externalities with the moral hazard consequences of bail-outs. When spillovers are relatively small,

it is preferable to suffer their consequences than generate moral hazard by providing bail-outs.

If spillovers are instead severe, bail-outs become more justified, possibly entirely replacing bail-ins

in exceptional cases. Panel 2 shows that such a large use of bail-outs has the potential to generate

serious moral hazard by increasing risk taking. Nonetheless, when spillovers are particularly severe;

for example, when an aggregate shock triggers a systemic banking crisis, it is preferable to tolerate

the consequences of moral hazard than suffer the destabilizing effects associated with bail-ins.

These insights call for a resolution framework that allows use of public funds if and when the risks

to macro-financial stability from bail-ins are exceptionally severe. Embedding the option to use

public funds in such exceptional cases completes the framework and also makes it more resilient

to the issues that may arise in practice (see below and Section III for more on this point).

The model provides three additional insights. First, even if bail-ins and bail-outs are used optimally

in line with the model prescriptions, social welfare is declining as spillovers become more severe

(panel 3). It is therefore crucial for policymakers not only to tailor bank resolution to the extent of

spillovers, but also to design resolution regimes that can reduce the potential for spillovers in the

first place.

This agrees with recent reform efforts to enhance the resolvability of large and complex

financial institutions via effective resolution powers and planning and adequate loss-absorbing

capacity. Furthermore, spillovers can be contained by restraining cross holding of bank debt within

the financial sector and applying enhanced prudential standards to financial institutions that are

too systemic to fail. Ample liquidity provision is also essential to contain the risk of spillovers, even

though at times of severe financial distress it is often difficult to ascertain whether a financial

institution is confronting a liquidity or solvency crisis. Finally, authorities should be mindful of

Measures to reduce spillovers should be designed in a way that does not impair risk sharing. For example, blunt

restrictions on interbank linkages may backfire by undermining risk and funding diversification.

TRADE-OFFS IN BANK RESOLUTION

10 INTERNATIONAL MONETARY FUND

financial institutions’ perverse incentives to become more systemic to benefit from future bail-

outs. This can occur both at the level of individual banks that may try to become more systemic—

for example, by increasing their balance sheets despite weak lending opportunities—or at the level

of the whole banking sector, for example, as banks correlate their exposures to increase systemic

risk (Schneider and Tornell 2004; Farhi and Tirole 2012).

Second, the model shows that when spillovers are relatively weak, the use of bail-outs should be

restrained. This raises a time-consistency problem. Policymakers may pledge ex ante to limit bail-

outs and use them only when spillovers are high. But, once bank failures occur, they may be

tempted to use them if spillovers exceed the fixed costs of bail-outs, neglecting the moral hazard

effects. To contain this time-consistency problem, it is imperative to put in place a resolution

framework that induces policymakers to rely on bail-ins when systemic concerns are absent or

moderate, and have a transparent approach for determining “systemic-ness.”

Importantly,

frameworks that induce the use of bail-ins should be credible. Otherwise investors and banks may

still expect to be bailed out, while being in fact bailed in during a crisis. This is a very dangerous

scenario since it entails both moral hazard costs and spillover effects.

Third, the model can also be used to analyze the role of leverage limits. By boosting capital ratios,

these limits reduce the likelihood that banks will face capital shortfalls and thus require socially

costly bail-ins or bail-outs. This benefit should be traded off, however, against the possible

negative effect of leverage limits on credit supply. Also noteworthy is that leverage limits reduce

the moral hazard effects of bail-outs since they lower the likelihood that a bank will benefit from

them. This improves the trade-off policymakers face and may provide more room to use bail-outs

if needed.

Four additional considerations may be relevant when it comes to deciding between bail-ins and

bail-outs, but they are not explicitly captured in the model. First, the case for bail-outs may seem

weaker if the sovereign faces debt sustainability problems because of the fiscal costs involved. But,

this concern must be weighed against the risk that the spillovers from bail-ins could themselves

endanger the fiscal position by lowering GDP and tax revenues. Therefore, whether fiscal concerns

should tilt the balance against bail-outs or bail-ins depends again on the strength of the spillovers.

Second, since banks most likely to receive bail-outs benefit from lower funding costs, they enjoy a

competitive advantage relative to others. This can generate significant distortions within the

banking sector, possibly leading to inefficient credit allocation and positions of market power. To

level the playing field, policymakers may thus consider imposing capital surcharges on those

institutions that are more likely to benefit from bail-outs. Further, levies could be imposed on a

resolution fund (where one exists) to mitigate bail-out benefits for large banks.

We use the term “systemic-ness” not necessarily in the context of individual financial institutions but also to

encompass situations that may raise systemic concerns (for example, clustered failures of many small banks). A

sound financial stability assessment framework is critical in making this determination. Such determination could

be made based on the growing toolkit to assess systemic risk (see, for instance, IMF-BIS-FSB 2009 and Blancher

and others 2013).

INTERNATIONAL MONETARY FUND 11

Third, the model neglects potential implementation costs and challenges associated with bail-ins

(Avgouleas and Goodhart 2015). For example, the bail-in process may be time demanding and

subject to legal challenges, especially in the context of cross-border resolution. Moreover, bail-ins

may face political opposition if they entail the transfer of bank ownership to foreign investors. Last

but not least, the extent of state ownership of the banking sector, depth of financial markets, and

availability of sophisticated private investors that can absorb the losses could affect the trade-off

a country may find itself facing.

Finally, the model considers the trade-offs from the point of view of a single policymaker. In some

cases, multiple policymakers from different jurisdictions or a college of policymakers would be

involved. For instance, in a currency union, one member state’s bail-out could affect its public debt

and may raise the perception that banks in other member states would also be bailed out in the

event of trouble. Then, what may seem optimal based on purely domestic considerations may end

up shifting risks and vulnerabilities to other countries. This points to the importance of cross-

border aspects of resolution frameworks, which are discussed in Section III.

II. EMPIRICAL EVIDENCE

The discussion so far has highlighted the theoretical pros and cons of bail-ins and bail-outs. These

are based on assumptions―moral hazard effects of bail-outs and spillover risks of bail-ins―that

we seek to verify in this section. We rely largely on the existing literature but present new evidence

where needed. Appendix II provides information on the details.

A critical caveat to consider is the fact that resolution frameworks explicitly featuring statutory bail-

in powers are new and have not yet been fully tested. Furthermore, adequate loss-absorbing

capacity in sufficient quality and quantity is not yet in place. This hinders the feasibility of direct

comparisons of costs and benefits of bail-ins and bail-outs in today’s regulatory environment.

Rather, this section focuses on the trade-offs policymakers faced at the time of the crisis. Consistent

with this, we use a broad definition of bail-in encompassing potentially any resolution that

imposed losses on private stakeholders. Despite these limitations, we believe the empirical analysis

presented can be useful in considering the trade-offs entailed. As reforms since the crisis have

likely reduced the spillovers stemming from bank resolution, the estimates could be considered an

upper bound for the spillover effect associated with bail-ins in the current environment.

Do bail-outs entail moral hazard?

From a theory perspective, it is ambiguous whether bail-outs motivate banks to behave less or

more responsibly. On the one hand, bail-outs encourage risk taking in a classic moral hazard

interpretation. If bank managers and shareholders know that they will not have to bear the full

consequences of the risk they are taking when negative outcomes are realized, they have stronger

incentives to take on more risk. Moral hazard also weakens the incentives of shareholders,

creditors, and depositors to discipline banks. Less monitoring and underpricing of risk may, in turn,

result in riskier portfolios and higher leverage. On the other hand, there is also an argument in the

literature that bail-outs may result in higher franchise/charter values because they reduce funding

TRADE-OFFS IN BANK RESOLUTION

12 INTERNATIONAL MONETARY FUND

costs and, hence, discourage risk taking (Sarin and Summers 2016).

The lower rates promised to

creditors/depositors increase payoffs, conditional on success, to managers/shareholders (as

residual claimants). This enhances their incentives to choose safer portfolios.

It is an empirical question whether the former effect—banks take on more risk because they don’t

have to bear all the losses—or the latter—banks take less risk because bail-outs help them boost

their charter values—dominates.

The evidence is far from clear-cut: the expectation of bail-outs

does seem to reduce funding costs, but it is unclear if this results in higher or lower risk taking.

Indirect evidence on the effects of bail-outs is provided by cross-country studies that analyze the

system-wide association between government guarantees and risk taking—and, ultimately,

financial system fragility. Using deposit insurance, government ownership, and bank concentration

as proxies for the scope of the public financial safety net,

some report destabilizing effects (Caprio

and Martinez Peria 2002; Demirguc-Kunt and Detragiache 2002), while others find no significant

impact or even a stabilizing effect (Barth, Caprio, and Levine, 2004; Beck, Demirguc-Kunt, and

Levine 2006). The mixed findings can in part be attributed to differences in controlling for factors

such as charter values and institutional quality. If banks earn rents through limits on competition,

relationship lending, or reputation building, the effects of the public financial safety net on risk

taking may be mitigated because these rents increase charter value and lower moral hazard (Gropp

and Vesala 2004). If there is strong contract protection/regulation/supervision and low corruption,

incentives and opportunities for excessive risk taking may be reduced (Hovakimian, Kane, and

Laeven 2003). The macro-financial backdrop also plays a role: guarantees induce risk taking in

tranquil times but dissuade it during crises (Anginer, Demirguc-Kunt, and Zhu 2014). One

interpretation is that, by shielding banks from macroeconomic and contagion risks, guarantees

increase charter values and reduce risk taking.

With the recognition that cross-country studies may suffer from endogeneity problems, we turn

to studies that exploit banks’ varying likelihood of benefiting from bail-outs. Moral hazard effects,

if present, should be particularly visible for larger financial institutions—which expect or are

perceived as more likely to be bailed out. The literature shows that large banks indeed benefit from

lower borrowing costs and credit default swap (CDS) spreads.

Since the crisis, the funding

advantage of large banks—the too-big-to-fail subsidy—has been reduced but not eliminated (IMF

Franchise value is defined as the present value of the stream of profits that a firm is expected to earn as a going

concern. In banking, given the important role of regulation, charter value—the present value of being able to

continue to do business in the future and earn rents in a highly regulated environment with high barriers to entry—

is often used synonymously with franchise value (Demsetz, Saidenberg, and Strahan 1996; Furlong and Kwan 2006).

The model in Section I assumes that the former effect dominates to illustrate the trade-off between moral hazard

costs of bail-outs and spillover effects of bail-ins. Also note that the literature is mostly based on guarantees (in

many cases for depositors), which are distinct from bailing out equity holders or managers.

The reasoning for the last proxy is that an individual bank is more likely to be systemically important in a

concentrated than in a dispersed sector and, thus, more likely to benefit from public support measures.

See Acharya, Anginer, and Warburton 2014; Ueda and Weder di Mauro 2013; GAO 2014; and Santos 2014 for

evidence of the funding advantage enjoyed by large banks. Kroszner 2016 provides a thorough and critical review

of the literature.

(continued)

INTERNATIONAL MONETARY FUND 13

2014a; GAO 2014). But many studies focus on the gross subsidy. It is plausible that the subsidy net

of the higher regulatory costs now faced by larger banks—including enhanced capital and liquidity

requirements and closer supervision—is lower today.

To provide stronger evidence that the lower funding costs enjoyed by large banks are linked to

the expectations of bail-outs, we look at how funding costs change around episodes that affect

the likelihood of bail-outs. Concretely, we look at the change in CDS spreads and equity prices

following the failure of Lehman Brothers and the passage of the Emergency Economic Stabilization

Act of 2008―which established the Troubled Asset Relief Program (TARP)―as well as the

beginning of the European bail-outs, distinguishing large banks from small banks.

We assume

that the first of these events—the Lehman

bankruptcy—sent a signal that policymakers’

appetite for bail-outs was low. The latter two

would signal the opposite. Figure 2 presents

the raw data of daily changes in spreads and

prices. As anticipated, large banks suffered

more than small banks with the Lehman

failure—consistent with a decline in their

funding cost advantage—and enjoyed better

outcomes when TARP was approved as well

as when Dexia was recapitalized. Importantly,

these relationships survive in an event-study

setup (Table 1). In fact, when changes are

calculated controlling for market

benchmarks, the relationship between size

and equity returns is statistically significant

and positive in the European event as well,

while the reaction in CDS spreads is still

insignificant.

Whether lower funding costs due to bail-outs

translate into more risk taking is not clear-

cut. Using mainly precrisis data, some studies

A net subsidy may still exist if market perceptions of bail-out probability have not changed yet. See Elliott 2014

for further discussion.

The event we use for Europe is the capital injection to Dexia in September 2008. There were other events between

October and December of the same year involving government interventions (capital injections and asset/debt

guarantees) in other European banks (see Fratianni and Marchionne 2013 for a list). The key insights from examining

these are similar. Note that comparing the failure of Lehman Brothers to the recapitalization of Dexia (and others)

is difficult, given the systemic nature and shock value of the former.

This insignificance may be due to pooling the banks located in the country where the event took place with

those in other countries. To explore this, we split the sample and execute the event studies separately for US and

non-US banks in the Lehman and TARP events and for EU and non-EU banks in the Dexia event. The findings are

comparable to those obtained in the pooled sample: even though they are smaller in magnitude, the coefficient

on bank size has the same sign and is statistically significant in the subsamples of non-US and non-EU banks.

(1) (2) (3) (4) (5) (6)

Lehman Lehman TARP TARP Dexia Dexia

Size 4.147* 2.816 -8.641*** -5.723* -1.394 0.086

(2.405) (3.650) (2.588) (3.061) (2.007) (2.233)

Constant -36.62 -28.82 94.21*** 133.8*** 20.00 5.016

(29.12) (48.92) (30.93) (41.03) (26.03) (29.93)

Observations 79 79 79 79 79 79

R squared 0.331 0.295 0.478 0.621 0.454 0.500

Country FE NO YES NO YES NO YES

(1) (2) (3) (4) (5) (6)

Lehman Lehman TARP TARP Dexia Dexia

Size -1.565** -0.536** 2.033*** 2.412** 1.651*** 2.014***

(0.678) (0.233) (0.549) (0.916) (0.426) (0.499)

Constant 18.68* -28.52*** -14.63* -43.96*** -10.47* -28.30***

(9.934) (2.728) (7.359) (10.72) (5.695) (5.844)

Observations 461 461 461 461 461 461

R squared 0.586 0.581 0.464 0.494 0.454 0.511

Country FE NO YES

YES NO YES

Panel 1. Cumulative Abnormal Changes in the Cross Section of CDS spreads

Table 1. Market Reaction to Bail-out Events

Note: Cumulative abnormal CDS spread and equity price changes are computed

over a seven-day (t−3 to t+3) event window for TARP and Dexia and a four-day (t

to t+3) event window for Lehman, defined relative to the bail-out event date (t),

over 2008:Q1–2016:Q3. Bail-out event dates are recorded on September 15, 2008

(Lehman bankruptcy), October 30, 2008 (TARP passage), and September 30, 2008

(Dexia capital injection). Size is the log of total assets. Standard errors clustered at

the country level are displayed in parentheses; statistical significance at the 10

percent, 5 percent, and 1 percent levels is denoted by *, **, and ***, respectively.

For sample coverage and further details on the empirical analysis, see Appendix II.

Panel 2. Cumulative Abnormal Changes in the Cross Section of Equity Prices

Sources: Bloomberg Finance L.P.; SNL Financial; Thomson Reuters Datastream; and

IMF staff calculations.

TRADE-OFFS IN BANK RESOLUTION

14 INTERNATIONAL MONETARY FUND

Figure 2. Bank Size and Reaction to Events

Altering Bail-out Expectations

Sources: Bloomberg Finance L.P.; Thomson Reuters Datastream; and IMF staff calculations.

Notes: Daily changes in spreads and prices around the events affecting bail-out expectations are

shown. The event dates are recorded as September 15, 2008 (Lehman bankruptcy), October 30, 2008

(TARP passage), and September 30, 2008 (Dexia capital injection). For sample coverage and further

details on the empirical analysis, see Appendix II.

y = 8.6352x - 76.159

R² = 0.3051

-40

-20

100

8 10 12 14 16

CDS spreads, percent change

Total assets (log)

Lehman Bankrutpcy

CDS spreads

y = -2.8168x + 23.622

R² = 0.1166

-80

-60

-40

-20

6 8 10 12 14 16

Equity prices, percent change

Total assets (log)

Lehman Bankruptcy

Equity prices

y = -6.6207x + 82.936

R² = 0.1528

-80

-60

-40

-20

8 10 12 14

CDS spreads, percent change

Total assets (log)

TARP Passage

CDS spreads

y = -0.8678x - 4.1854

R² = 0.0202

-80

-60

-40

-20

6 8 10 12 14 16

Equity prices, percent change

Total assets (log)

TARP Passage

Equity prices

y = -1.0101x + 22.626

R² = 0.0051

-80

-60

-40

-20

8 10 12 14 16

CDS spread, percent change

Total assets (log)

Dexia Capital Injection

CDS spreads

y = 0.4106x - 8.986

R² = 0.0089

-100

-80

-60

-40

-20

6 8 10 12 14 16

Equity prices, percent change

Total assets (log)

Dexia Capital Injection

Equity prices

INTERNATIONAL MONETARY FUND 15

show that larger banks take on larger positions on riskier loan segments, greater reliance on riskier

funding, lower buffers against potential loan losses, and greater loan losses after a shock).

Examination of our sample covering a set of global banks since 2008 gives some indication that

large and small banks respond differently to a bail-out event: large banks increase risk taking while

small ones decrease it (see Black and Hazelwood 2012 for a similar finding). This is, however, not

robust across specifications. The evidence on whether large banks take on more risk in general is

similarly mixed: large banks tend to have less

core capital to total assets relative to risk-

weighted assets—pointing to riskier

portfolios—but also higher distance-to-

default and lower nonperforming loan ratios

(Table 2).

Of course, plotting various

indicators against a proxy for bail-out

probability does not account for other factors

that may be at play. Yet regressions

controlling for bank characteristics and fixed

effects paint a similar picture: not all risk-

taking measures point in the same direction

(Table 3).

A possible explanation is that

tighter postcrisis supervision has prevented

large banks from taking excessive risk.

The impact of bail-outs on funding costs and

risk taking is analyzed also in a series of

recent studies using German data that allow

for better identification. Some find a positive

link between bail-out expectations—

predicted using regional political factors—

and probability of distress (Dam and Koetter

2012). Furthermore, in a natural experiment

See, among others, Boyd and Gertler 1994; Hovakimian and Kane 2000; and Schnabel 2009 for evidence of

excessive risk taking by large banks. In addition to having higher stand-alone risks, large banks contribute more to

systemic risk. See Laeven, Ratnovski, and Tong 2016 and the references therein.

Large banks may extend riskier loans on an individual basis, but this may not be reflected in a riskier loan portfolio

or balance sheet because of greater diversification. Empirical support for such an effect of diversification on bank

risk, however, is limited: better diversification resulting from larger size can generate both scale economies and

risk-taking incentives (Hughes and Mester 2013), and large banks may use their diversification advantage to

operate with lower capital ratios and pursue riskier activities (Demsetz and Strahan 1997).

Based on a comparison of return on assets and return on equity between large and small banks, there is no

obvious sign of potential charter-value effects (Table 2). Controlling for bank and time fixed effects (and a measure

of bank risk) reveals a positive, albeit not robust, relationship between bank profitability and size (Table 3). This is

in line with findings reported elsewhere (see, for instance, Goddard, Molyneux, and Wilson 2004).

Recent research argues that the relationship between bank size and risk taking changed in the aftermath of the

crisis—as did the relationship between bank size and funding cost advantage. See, for instance, Bhagat, Bolton,

and Lu 2015, who report that the positive correlation between total assets and z-scores breaks down in 2010–12.

Obs. Mean Std. Dev. Min. Max.

Large Banks

CDS spread 2,361 157.7 144.0 5.9 1623.0

Senior unsecured spread 101 169.0 150.5 13.1 901.8

Subordinated spread 108 350.6 164.8 45.4 860.9

CoCo spread 411 581.2 248.9 159.4 1837.8

Z-score 1,044 3.9 0.9 1.6 7.9

Tier 1 ratio 1,426 11.7 2.8 6.7 26.4

Leverage ratio 936 9.1 13.1 2.8 71.7

NPL ratio 959 3.2 3.1 0.0 21.4

ROA 1,209 0.5 0.6 -2.8 2.4

ROE 1,223 9.1 8.4 -36.7 30.0

Small Banks

CDS spread 1,240 183.0 166.3 1.2 1698.7

Senior unsecured spread 41 181.9 165.0 20.1 641.7

Subordinated spread 18 322.5 122.7 215.6 667.6

CoCo spread 137 667.3 290.8 146.1 2050.6

Z-score 10,741 3.6 1.0 -2.4 6.0

Tier 1 ratio 11,699 12.4 3.2 6.7 27.9

Leverage ratio 10,856 12.8 16.6 2.7 82.0

NPL ratio 11,010 3.5 4.9 0.0 85.4

ROA 12,609 0.9 0.8 -2.8 3.6

ROE 12,540 9.2 7.5 -35.9 34.9

Sources: Bloomberg Finance L.P.; SNL Financial; Thomson Reuters Datastream;

and IMF staff calculations.

Table 2. Funding Costs, Risk, and Profitability in Large versus Small Banks

Note: Large banks are defined as those with assets in the top percentile. CoCo

refers to contingent convertible bonds. Z -score is computed as equity capital

ratio plus return on assets divided by the standard deviation of return on

assets (calculated over a rolling window of 10 quarters). Tier 1 ratio is defined

as the ratio of Tier 1 capital to risk-weighted assets, and leverage ratio is

defined as the ratio of Tier 1 capital to total tangible assets. NPL ratio is the

ratio of nonperforming loans to total loans. ROA is return on average assets

and ROE is return on average equity. For sample coverage and further details

on the empirical analysis, see Appendix II.

TRADE-OFFS IN BANK RESOLUTION

16 INTERNATIONAL MONETARY FUND

where government guarantees for savings banks were removed due to litigation, banks that lost

their guarantees are shown to cut off the riskiest borrowers, adjust their liabilities away from risk-

sensitive debt instruments, and see their bond yield spreads increase (Gropp, Gruendl, and Guettler

2014).

However, others report that capital support—instrumented by exploiting geographically

driven differences in insurance and acquisition frameworks across Germany—reduces bank risk

taking as well as lending (Berger and others 2016).

A growing body of literature on the US experience with TARP arrives at mixed conclusions. Some

argue that banks receiving government funds shifted to riskier segments within the same asset

class, resulting in increased volatility and default risk, despite improved capital ratios (Duchin and

Sosyura 2014). Others deliver a more nuanced message that weak banks—those with low charter

values and high leverage—increased risk taking, but strong banks—those with high charter values

and low leverage—decreased it (Schenk and Thornton 2016) and that TARP reduced systemic risk,

especially in larger, safer banks located in economically stronger areas (Berger, Roman, and

Sedunov 2016).

Analysis using changes in bank ratings to gauge the perceived likelihood of government support

suggests that nuances exist through time in addition to across banks: higher bail-out probabilities

result in higher risk taking in tranquil periods but in less risk taking when there is a crisis (Damar,

Gropp, and Mordel 2012).

This is consistent with the argument that the charter value effect may

dominate when the economy is hit by a systemic shock and contagion risk is high.

They interpret this as support for the theory showing that capital reduces moral hazard (Morrison and White

2005) and strengthens banks’ monitoring incentives (Allen, Carletti, and Marquez 2011; Mehran and Thakor 2011).

In October 2006, Dominion Bond Rating Service introduced a change to its rating system to explicitly account

for the potential of government support. This led to rating changes that were not a result of changes in the

respective banks’ credit fundamentals. Damar, Gropp, and Mordel (2012) exploit this natural experiment.

(1)

(2)

(3)

(4) (5)

(6) (7)

(8) (9)

(10) (11) (12)

-score Z

-score

Tier 1

ratio

Tier 1

ratio

Leverage

ratio

Leverage

ratio

NPL

ratio

NPL

ratio

ROA

ROE ROE

Size

-0.1481***

-0.1348***

-3.1073***

-1.7066***

-0.9142***

-0.6241***

-1.6868***

-0.5621***

0.6799*** 0.0275

7.9816*** 0.2720

(0.0238)

(0.0220) (0.3982) (0.2557) (0.3459) (0.1424) (0.5101)

(0.1991) (0.1099) (0.0370) (1.4334) (0.3393)

Loan ratio

-0.0059***

-0.0061***

-0.1233***

-0.1344***

0.0312** 0.0194** -0.0139 -0.0089 0.0164***

0.0004 0.2060*** -0.0098

(0.0012) (0.0011)

(0.0250)

(0.0223) (0.0134) (0.0097) (0.0201) (0.0143) (0.0044) (0.0031) (0.0559) (0.0321)

Deposit ratio -0.0001

0.0000

-0.0245 -0.0275*

-0.0161 -0.0180**

-0.0024

0.0059 0.0201***

0.0119*** 0.2111*** 0.1401***

(0.0011) (0.0011) (0.0189) (0.0165) (0.0104) (0.0082) (0.0229) (0.0175) (0.0044) (0.0037) (0.0567) (0.0384)

-score

3.6889*** 1.1025***

43.9870***

9.8422***

(0.2904) (0.1382) (3.8339)

(1.2880)

Observations 11,259

11,259 12,317 12,317 1,762

1,762 12,145 12,145 11,242 11,242

11,183 11,183

squared

0.160 0.172

0.263 0.070 0.365 0.310

Number of banks

583

583 634 634 312

312

634 634 584 584

581

Bank FE YES YES YES YES YES

YES

Quarter FE YES YES YES YES YES YES YES YES YES

YES

YES YES

Country FE

YES

YES YES

YES

Table 3. Size and Risk Taking

Note: Size is the log of total assets. Loan ratio refers to gross loans in percent of total assets. Deposit ratio refers to total deposits in

percent of total assets.

Z-score is computed as equity capital ratio plus return on assets divided by the standard deviation of return on

assets (calculated over a rolling window of ten quarters). Tier 1 ratio is defined as the ratio of Tier 1 capital to risk-weighted assets, and

leverage ratio is defined as the ratio of Tier 1 capital to total tangible assets. NPL ratio is the ratio of nonperforming loans to total

loans. ROA is return on average assets and ROE is return on average equity. Standard errors are displayed in parentheses; statistical

significance at the 10 percent, 5 percent, and 1 percent levels is denoted by *, **, and ***, respectively. For sample coverage and further

details on the empirical analysis, see Appendix II.

Sources: Bloomberg Finance L.P.; SNL Financial; Thomson Reuters Datastream; and IMF staff calculations.

INTERNATIONAL MONETARY FUND 17

When interpreting the mixed evidence, it is also important to consider that the moral hazard effects

of bail-outs might be attenuated by the fact that banks that are more likely to receive government

support might be subject to tighter scrutiny by supervisors and regulators. In this respect, the

postcrisis regulatory changes (for example, leverage ratios) may have helped limit risk taking by

systemically important financial institutions in the presence of implicit and explicit government

guarantees. While it is too early to judge, existing evidence on the role played by regulation and

governance in determining risk taking (Laeven and Levine 2009) supports cautious optimism.

Do bail-ins entail systemic spillovers?

We now turn to the spillovers associated with bail-ins before the regulatory reforms. Since these

reforms have likely reduced the magnitude and scope of these spillovers, the evidence should be

read as a rationale for the reforms, rather than as an estimate of potential spillovers under the new

resolution frameworks (see Section III and Box 3). We consider whether, during the crisis and in its

aftermath, bail-in of a financial institution led to a repricing of (bail-in-able) unsecured debt and

equity of other banks. Providing an accurate measure of the extent of spillovers is, however, an

arduous task for several reasons.

On the one hand, the analysis may underestimate the extent of spillovers because the decision to

impose losses on unsecured creditors is endogenous: aware of the possible spillovers,

policymakers may use bail-outs rather than bail-ins in the context of a systemic crisis or failure of

a large, complex, interconnected institution. Further, there have been very few events during which

bail-in of private stakeholders has not been combined with (or has not been a precondition for)

injection of public funds, thus containing the potential for spillovers.

On the other hand, comovement in bank equity and bond prices may not accurately reflect the

kinds of spillovers highlighted in the model that are relevant for the bail-in/out trade-off (for

example, those due to fire sales or a sudden shift to a worse equilibrium driven by heightened risk

aversion). Comovement may instead arise because a bail-in reveals genuine information about a

deterioration of the macro-financial outlook or because it leads investors to update their beliefs

about the credibility of a bail-in regime. Distinguishing between spillovers and aggregate shocks

is particularly challenging because the potential for spillovers is much greater in the context of a

deteriorating outlook. So what in normal times would be an idiosyncratic bail-in may turn into an

event that leads to panic among holders of claims on unaffected banks.

With these significant caveats, we start our analysis by looking at some European resolution cases

where losses were imposed on private stakeholders (see list in Appendix Table 1). Given the

heterogeneity of the surrounding circumstances (applicable resolution framework, extent of losses

imposed and of public support, type of instrument or investor bailed in, macro-financial backdrop,

etc.), we analyze each event separately. It is important to note that these cannot be regarded as

examples of good resolution practice. Most occurred before the adoption of the new regulatory

standards and resolution frameworks compliant with the KA. As such, the new frameworks are

likely to deliver better outcomes, especially once they become fully operational. That said, these

examples provide insights into the challenges the new frameworks are meant to tackle.

TRADE-OFFS IN BANK RESOLUTION

18 INTERNATIONAL MONETARY FUND

(1) (2) (3)

(4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16)

Austria Cyprus

Denmark

(1)

Denmark

(2)

Greece (1) Greece (2) Greece (3) Italy (1) Italy (2) Netherlands

Portugal

(1)

Portugal

(2)

Slovenia Spain (1) Spain (2)

United

Kingdom

Country -1.059*** - -7.590*** -3.618** -16.76***

-5.239*** 11.09 0.390 -0.827 -6.236*** -13.18*** -5.369*** - -5.739** -5.032*** 1.963

(0.209) (0.969) (1.455) (2.995) (1.620) (9.602) (0.913) (1.334) (0.961) (1.709) (0.133) (2.254) (1.446) (1.653)

EU less country -0.915*** -2.971*** -1.097*** 2.083** -2.498** -3.515*** 0.626 -0.339 -2.393 1.744 -1.098** -0.519** -1.808** -1.338* -2.613*** -2.359***

(0.337) (0.914) (0.367) (0.851) (1.099) (0.845) (0.542) (0.493) (1.522) (1.184) (0.557) (0.256) (0.705) (0.686) (0.938) (0.856)

Russia -2.802**

(1.277)

Constant 1.008*** -0.906*** 0.441** -1.590*** 0.217 0.938*** 0.146 -0.858*** -1.292*** -0.407 -0.252 0.361*** 0.475** -1.020*** -0.978*** 2.345***

(0.0853) (0.257) (0.214) (0.260) (0.325) (0.267) (0.193) (0.213) (0.186) (0.961) (0.241) (0.0896) (0.229) (0.220) (0.219) (0.309)

Observations 582 553 538 578 545 551 580 577 579 552 562 580 557 550 553 554

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16)

Austria Cyprus

Denmark

(1)

Denmark

(2)

Greece (1) Greece (2) Greece (3) Italy (1) Italy (2) Netherlands

Portugal

(1)

Portugal

(2)

Slovenia Spain (1) Spain (2)

United

Kingdom

Country 3.459** -

-9.034*** -7.411*** - - - 6.986*** 0.153 8.874** 7.840* -1.161 - -3.640 4.328** 1.848***

(1.476) (1.713) (1.300) (1.460) (1.506) (3.628) (4.333) (1.478) (3.937) (1.797) (0.360)

EU less country 0.845 1.356* -5.785*** -1.956 4.457** 4.898*** 0.976 4.146*** -0.762 -0.245 2.667* 3.593*** 4.156** 1.280 2.488*** 1.325***

(1.816) (0.740) (1.925) (1.602) (2.232) (1.427) (0.973) (1.177) (1.053) (1.558) (1.497) (0.779) (1.796) (1.127) (0.900) (0.420)

Russia 0.244

(1.103)

Constant -4.608*** 3.136*** 0.876 -0.579 -9.867*** 1.094 -2.301*** -1.600** -0.193 -1.433 0.800 0.270 -0.282 -1.150** 3.685*** -0.411

(1.476) (0.394) (1.713) (1.300) (1.372) (0.963) (0.695) (0.667) (0.790) (1.254) (0.723) (0.380) (1.649) (0.539) (0.467) (0.360)

Observations 97 98 92 97 98 99 97 97 97 99 97 97 70 97 98 99

Panel 2.

Cumulative Abnormal Changes in the Cross Section of CDS Spreads

Panel 1. Cumulative Abnormal Changes in the Cross Section of Equity Prices

Table 4. Market Reaction to EU Bail-in Events

Note: Cumulative abnormal changes in equity prices and in CDS spreads are computed over a ten-day (t+1 to t+10) event window, defined relative to the event date (t), over the period between 2008:Q1 and

2016:Q3. Event window for Denmark (1) and Portugal (1) in Panel 1 and for Spain (2) and Cyprus in Panel 2 is (t+1 to t+5) while that for the United Kingdom in Panel 2 and Portugal (2) in both panels is (t to

t+1). Standard errors are displayed in parentheses; statistical significance at the 10 percent, 5 percent, and 1 percent levels is denoted by *, **, and ***, respectively. For sample coverage and further details on

the empirical analysis, see Appendix II. For a list and description of the events, see Appendix Table 1.

Sources: Bloomberg Finance L.P.; SNL Financial; Thomson Reuters Datastream; and IMF staff calculations.

INTERNATIONAL MONETARY FUND 19

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16)

Wamu Silverton Independent Waterfield Midwest Tierone Enterprise Community Bankeast Easternshore Dupage Union Syringa Multiple1 Allendale Aztec

In-state

36.03

***

10.30

2.574

4.382

***

-4.671

***

-3.415

***

-1.362

***

-0.572 -2.545

-0.470

-0.483 -4.623** -1.733*

2.372

***

-2.327

***

-2.078***

(0.000) (0.012) (0.015) (0.000) (6.89-e05) (0.002) (3.06-e6) (0.599) (0.139) (0.057) (0.448) (0.022) (0.055) (0.000) (0.003) (1.88-e06)

Out-of-state

21.67

***

3.111

3.422

***

2.805

***

-7.971

***

-1.329

-0.934

***

-1.332

***

-5.160

***

-0.654

***

-0.821 -5.841*** -2.499***

2.459

***

-0.923

-1.511***

(0.000) (0.047) (6.81-e08) (0.002) (1.44-e06) (0.093) (1.13-e05) (0.007) (0.000) (1.82-e05) (0.156) (0.000) (3.51-e09) (9.2-e07) (0.051) (0.000)

Constant

3.463

***

1.503

***

-0.525

***

0.534

-1.238

***

-0.386 0.003

0.699

2.934

***

-0.132

-1.105*** -0.230 -1.038***

0.762

-0.667

***

1.408***

(1.83-e05) (0.002) (0.007) (0.031) (4.32-e05) (0.104) (0.977) (0.021) (0.000) (0.090) (1.59-e05) (0.323) (1.67-e05) (0.022) (0.009) (7.10-e06)

Observations

517 518 522 530 533 533 538 538 547 548 557 557 557 557 559 561

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16)

Wamu Silverton Independent Waterfield Midwest Tierone Enterprise Community Bankeast Easternshore Dupage Union Syringa Multiple1 Allendale Aztec

In-state

2.978

19.00

***

-9.390

***

1.725

4.333

***

-5.046

2.604

1.380 1.018

3.736

0.273 -12.32*** 1.167

2.978

3.078

1.097

(0.017) (3.67-e07) (1.49-e05) (0.429) (4.94-e05) (0.448) (0.017) (0.584) (0.550) (0.036) (0.956) (0.000) (0.776) (0.017) (0.087) (0.525)

Out-of-state

2.775

21.41

***

-5.117

1.823

3.121

***

1.698 0.613

4.656

7.735

***

4.256

***

-5.686** 0.293 5.094 2.775 0.385 -1.717

(0.411) (7.08-e06) (0.041) (0.028) (0.007) (0.417) (0.741) (0.014) (0.006) (2.4-e05) (0.012) (0.948) (0.178) (0.411) (0.749) (0.222)

Constant

1.767

-16.60

***

2.236

***

-2.738

***

-1.891

***

-3.171

***

-0.858

-3.398

***

-2.308

-2.272

***

9.287*** 2.932*** -1.387

1.767

-3.676

***

-1.076

(0.033) (0.000) (0.001) (0.000) (0.000) (0.002) (0.164) (0.006) (0.124) (8.99-e06) (0.000) (0.000) (0.112) (0.033) (6.64-e07) (0.106)

Observations

97 72 100 73 101 101 91 91 99 99 70 70 70 97 97 97

Table 5. Market Reaction to US Bail-in Events

Panel 1. Cumulative Abnormal Changes in the Cross Section of Equity Prices

Panel 2. Cumulative Abnormal Changes in the Cross Section of CDS Spreads

Event window for Easternshore, Enterprise, and Silverton in Panel 1 is (t+1 to t+5) while that for Midwest and Waterfield in Panel 1 and Easternshore and Enterprise in Panel 2 is (t to t+1). Event labeled “Multiple1”

represents two bail-in events which occurred on the same day (Millennium Bank NA and Vantage Point Bank). Standard errors are displayed in parentheses; statistical significance at the 10 percent, 5 percent, and 1

percent levels is denoted by *, **, and ***, respectively. For sample coverage and further details on the empirical analysis, see Appendix II. For a list of the events, see Appendix Table 2.

Sources: Bloomberg Finance L.P.; SNL Financial; Thomson Reuters Datastream; and IMF staff calculations.

TRADE-OFFS IN BANK RESOLUTION

20 INTERNATIONAL MONETARY FUND

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16)

Columbia Slavie Freedom Eastside Greenchoice NBRS Chicago Frontier Northernstar Crestview Highland Capitol Doral Edgebrook Premier Multiple2

In-state

-0.388 2.146** -0.978 1.109* 1.493*** -1.107** 2.811*** -2.579*** -0.574

2.042

***

4.496***

-1.924

***

3.141

***

1.193*** -1.114* -1.120**

(0.158) (0.039) (0.256) (0.577) (0.015) (0.035) (3.13-e05) (1.53-e07) (0.364) (0.001) (0.000) (3.95-e05) (0.000) (0.002) (0.078) (0.034)

Out-of-state

-1.223*** 2.044*** -0.608 -0.068 0.511 -1.987*** 2.637*** -2.679***

-0.708

0.884 5.686***

-2.359

***

5.113

***

1.247*** -1.613*** -0.395

(2.09-e05) (0.000) (0.211) (0.954) (0.780) (1.05-e08) (8.83-e08) (0.000) (0.028) (0.132) (0.000) (6.5-e07) (0.000) (6.57-e05) (5.61-e08) (0.455)

Constant

0.141 0.381** -0.875*** 0.872*** -0.526* 0.878*** 0.777** -0.594**

0.599

***

-1.063

***

-1.238***

0.764

***

1.117

***

0.112 0.868*** -1.179***

(0.450) (0.044) (0.000) (0.001) (0.058) (1.15-e08) (0.019) (0.015) (0.005) (0.001) (0.000) (0.007) (0.001) (0.620) (5.70-e07) (0.002)

Observations

561 561 562 562 562 568 568 568 570 570 570 572 573 576 576 578

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16)

Columbia Slavie Freedom Eastside Greenchoice NBRS Chicago Frontier Northernstar Crestview Highland Capitol Doral Edgebrook Premier Multiple2

In-state

5.528* 10.11*** 1.232*** 0.754 -7.502*** 4.185*** -0.232 -6.486*** 0.808 0.478 -2.367

5.431

-8.613

***

1.303** 1.629 2.197*

(0.069) (0.002) (0.820-e09) (0.584) (1.19-e05) (2.52-e10) (0.725) (4.28-e05) (0.331) (0.588) (0.429) (0.021) (0.000) (0.044) (0.599) (0.067)

Out-of-state

4.821*** -1.020 -0.007 4.040 -3.358 2.613*** 2.088* -5.442*** 0.327 1.329 -2.571***

8.308

-2.057 1.962*** 1.314 2.615

(0.000) (0.862) (0.995) (0.179) (0.146) (0.008) (0.057) (4.50-e07) (0.769) (0.298) (0.005) (0.043) (0.330) (0.000) (0.588) (0.127)

Constant

-6.580*** -6.353*** 0.841*** 1.340* 4.417*** -1.014** -1.214*** 5.544***

2.508

***

-2.820

***

2.259**

-5.988

***

-0.268 -1.386*** -3.630*** -1.232***

(0.000) (0.000) (2.96-e07) (0.062) (1.68-e07) (0.013) (0.008) (0.000) (8.72-e09) (7.41-e05) (0.010) (4.81-e06) (0.790) (0.000) (6.49-e07) (3.36-e08)

Observations

97 97 97 97 97 97 97 97 97 96 96 97 87 97 97 97

Event window for Chicago, Edgebrook, Freedom, Multiple2, and NBRS in Panel 1 is (t to t+1) while that for Columbia, NBRS, and Premier in Panel 2 is (t+1 to t+5). Event labeled “Multiple2” represents two bail-in events

which occurred on the same day (the Bank of Georgia and Hometown National Bank). Standard errors are displayed in parentheses; statistical significance at the 10 percent, 5 percent, and 1 percent levels is denoted by

*, **, and ***, respectively. For sample coverage and further details on the empirical analysis, see Appendix II. For a list of the events, see Appendix Table 2.

Table 5. Market Reaction to US Bail-in Events (continued)

Panel 1. Cumulative Abnormal Changes in the Cross Section of Equity Prices

Panel 2. Cumulative Abnormal Changes in the Cross Section of CDS Spreads

Sources: Bloomberg Finance L.P.; SNL Financial; Thomson Reuters Datastream; and IMF staff calculations.

INTERNATIONAL MONETARY FUND 21

The empirical analysis suggests that imposing losses on unsecured creditors may have implications

not only for the bailed-in bank but also for other banks, especially when there is a considerable degree

of uncertainty and when macro-financial risks are high (Table 4). In 11 out of 16 cases, equity prices

for banks in EU countries other than the country in which the event took place show negative and

significant changes. This is notable because equity holders would suffer losses or be wiped out one

way or another independently of which debt securities are bailed in. In eight of these cases, CDS

spreads show significant increases.

Almost all of these cases (the asset transfer and bail-in in Cyprus;

two resolutions in Greece where shareholders and subordinated debtholders were significantly

diluted; Portuguese and Spanish bridge bank operations; and resolutions and recapitalizations in

Slovenia affecting five banks) took place in environments of heightened systemic risk.

Noteworthy in the European cases was a lack of significant prior experience with bank resolutions that

would impose losses on private stakeholders, insufficient clarity ex ante on how losses would be

allocated, and lack of explicit loss-absorbing capacity—as enshrined in the recent reforms. In

particular, which funding instruments get bailed in and by how much likely has a bearing on the extent

of spillovers.

An interesting case in point is Cyprus, where bail-in of unsecured bondholders and

depositors had negative effects in other EU and non-EU countries (primarily Russian banks). These

effects were mostly transitory, with CDS spreads and equity stabilizing about a week or so after the

shock. That said, there may be other costs associated with bailing in unsecured creditors, particularly

depositors, under an ad hoc regime (for example, bail-in of depositors in Cyprus led to claims for

compensation and may have also encouraged strategic defaults on bank loans, exacerbating the

nonperforming loan problem).

We also look at the resolutions executed by the Federal Deposit Insurance Corporation (FDIC) in the

United States. It is worth noting that many of these were handled through a purchase and assumption,

and the FDIC has a track record of imposing losses on bank creditors. We examine the cases that

involved relatively large banks—those ranked in the top 30 by total assets among the more than 500

bank failures the FDIC handled since 2000 (Appendix Table 2)—and find mixed results on the equity

prices and CDS spreads of other banks (Table 5). In half of the cases we find negative coefficients for

equity prices of banks located in a state other than that of the failed bank, while in roughly 30 percent

of the cases the coefficients are positive (the rest are insignificant). For CDS spreads, a third have

positive coefficients, and for about 10 percent they are negative. These are smaller percentages than

in the European case (of which two-thirds have negative significant coefficients for equity prices and

more than half have positive significant coefficients for CDS spreads). The lack of firm evidence is not

surprising given that the US cases involve relatively small, local banks.

With the caveat that the analysis does not control for any other factors that might have affected these

groups of banks differently around the event dates, the dynamics examined are consistent with the

We also examine senior bond spreads and observe that they rose as well, but these increases are not statistically

significant. One reason could be that the sample size is much smaller for bonds. Other data limitations could also be

an explanation—for example, the 10-year bonds we are looking at may or may not be subject to bail-in under the rules

of a particular jurisdiction. Yet another reason could be that bondholders believed that they would ultimately be wholly

compensated through litigation (in many cases, investors that were subject to losses took their case to courts).

Spillovers may also be affected by a bank’s liability structure and creditor hierarchy. See IMF 2013.

TRADE-OFFS IN BANK RESOLUTION

22 INTERNATIONAL MONETARY FUND

possibility that bailing in some banks may have had spillovers for creditors and equity holders in other

banks that were not yet subject to bail-ins. That said, the evidence is also consistent with markets

revising their beliefs about the macro-financial outlook or on the credibility of a bail-in regime. For

instance, negative effects in the European cases could reflect increased market discipline resulting

from the imposition of bail-ins or from signaling to investors that policymakers have become less

inclined to bail them out (Schafer, Schnabel, and Weder di Mauro 2016). There is no easy way to tell

the two interpretations apart. It is worth noting that the latter interpretation is less likely to apply in

the US case, given the commonality in macro-financial factors across states and the existence of a

federal resolution authority applying the same standards across state lines for FDIC-insured banks.

The new resolution frameworks in key jurisdictions explicitly aim at preventing spillovers by ensuring

that holders of bail-in-able securities can withstand and understand the risks involved (see Section III).

In practice, this may require more time.

Taking a closer look at who holds bank securities could

inform the debate: if securities that are hit first are held by unsophisticated or leveraged investors,

distress in one bank can quickly translate into instability across a system or across borders. A stock-

taking exercise on who holds bank equity, contingent capital (in the form of contingent convertible

bonds), and bail-in-able debt (which, for illustrative purposes only, is assumed to be composed of

senior unsecured and subordinated debt securities) shows cross-country variation. But a considerable

share of bail-in-able debt is held by credit institutions, while the public sector and nonfinancial

institutions are major holders of equities (Figure 3). The extent of cross holdings among banks and

within the financial sector suggests that shocks to one or a group of banks may spread to others

through losses on bail-in-able debt holdings.

In some countries, the extent of holdings by retail

investors may render resolution through bail-in politically more difficult.

Additional analyses produce results that seem to be more consistent with the spillover explanation than the belief-

updating explanation. In regressions that are not reported for the sake of brevity but available on request, we examine

whether there is a differential response by large banks. If the market reaction is due to an updating of beliefs on the

likelihood of bail-outs, it would be expected to vary by bank size reflecting too-big-to-fail considerations. We do not

find any robust evidence of such a differential for large banks. We also look at the time windows over which the effects

are present. Markets would be quick to update their beliefs and, once they reassess their expectations of bail-outs,

they would have no reason to go back—suggesting a very short-lived market reaction. We, however, observe the

reaction to last for up to two weeks. Finally, we include the size of bilateral cross-border banking flows between the

country where a bail-in event takes place and the country where a bank is headquartered as an additional variable in

the empirical specification. If spillovers are present, we would expect them to be stronger for banks located in countries

with more connections to the event country. The coefficient on this variable indicates, in half of the studied cases, a

significant negative equity price reaction where bilateral flows are larger. Interestingly, the number of cases with

significant negative coefficients increases from a third in the one-day window to a half in the ten-day window.

Note that indirect evidence on spillovers comes from the examination of fiscal costs of systemic banking crises (Box

1). If larger losses are imposed on the private stakeholders when the size of the bail-out is small, larger output losses

in these cases (when up-front fiscal cost is less than 1 percent of GDP) could be interpreted as an indication that these

losses generate spillovers to other parts of the economy.

The new EU resolution regime was recently applied for the first time in the context of a Spanish bank.

Simulations using confidential supervisory data show that contagion from bailing in a large euro-area bank is limited

under plausible scenarios, but in rare cases where the financial system is already weak, bail-in of a bank may lead to

failure at a counterparty (Huser and others 2017). This underlines the importance of continuing to implement measures

that limit interbank holdings of bail-in-able securities.

INTERNATIONAL MONETARY FUND 23

Figure 3. Holdings of Bank Equity, Contingent Capital, and Bail-in-able Debt

Sources: CapitalIQ; Dealogic Loan Analytics; ECB Securities Holdings database; and IMF staff

calculations.

Note: Information on holders is available only for a subset of the contingent convertible bonds

(CoCos) issued. Bail-in-able debt refers to senior unsecured and subordinated debt securities.

100

Institutions

Corporations Individuals/insiders

State-owned sh ares

Public and other

Investor Share for Bank Equities

(percent of total)

100

Banks and private banks Central banks Corporations

Fund and asset managers Hedg e funds Insurance and pension funds

Public and other Sovereign

Investor Share for Select CoCo Issuances

(percent of total)

100

Banks (excl. central bank) Other financial institutions Nonfinancial sector

Investor Share for Bail-in-able Debt

(percent of total)

TRADE-OFFS IN BANK RESOLUTION

24 INTERNATIONAL MONETARY FUND

III. RESOLUTION FRAMEWORKS IN PRACTICE

So far we have used the term “bail-in” in a generic sense, to refer to the ability of resolution authorities

to impose losses on private stakeholders to recapitalize a failing bank. In practice, bailing in a bank

involves restructuring its liabilities in resolution on a going-concern basis to allow its recapitalization

from the capital freed up from the exercise. Bail-ins (and other resolution powers) are supposed to be

implemented in a manner that ensures continuity of systemically important financial services and

payment, clearing, and settlement functions, and in a manner that avoids unnecessary destruction of

value. Therefore, they seek to minimize the overall costs of resolution in home and host jurisdictions.

A bail-in can be achieved through various resolution tools, including statutory bail-in powers as a

specific legal tool (Box 2).

Several jurisdictions have introduced bail-in powers as an integral part of

bank resolution regimes (Box 3). Given the intrusive nature of bail-in and possible legal risks from

potential interference with property rights, these are typically applied only when liquidation is not

viable, as authorities seek to preserve the systemically important functions of a financial institution.

Significant attention has been given to ensuring that bail-in is an effective and credible policy option

(Box 4). First, reforms seek to enhance the total loss-absorbing capacity (TLAC) of systemic banks, with

clarity on the types of claims that qualify for loss absorption and the amounts and location within a

financial group’s structure of such claims. Second, it is crucial that unsecured and uninsured bank

creditors understand the risks and can absorb the losses. Global systemically important banks (G-SIBs)

have been issuing significant TLAC, which, to mitigate potential liquidity risk and spillovers, excludes

short-term debt claims. Third, national authorities are expected to ensure that banks are subject to

periodic “resolvability” assessments and that the supervisor or resolution authority has powers to

require banks to remove legal and operational impediments to effective resolution. For cross-border

banks, resolution planning should also encompass ex ante agreement between home and key host

jurisdictions on the resolution strategy and the amount and location of loss-absorbing capacity.

The reforms have improved the trade-off between bail-outs and bail-ins, in addition to aiming to

make the need for recapitalization less likely. These reforms, along with the use of emergency liquidity

assistance,

potentially help reduce spillovers and lend credibility to measures aimed at constraining

public solvency support in a crisis, thus reducing moral hazard.

Yet there may still be extreme circumstances under which the social costs of a bail-in exceed those of

a bail-out. In other words, there remains a residual risk that public solvency support will be needed to

preserve financial stability. Emergency liquidity assistance may end up being offered to potentially

Statutory bail-in powers (Zhou and others 2012) are distinct from debt issued with contractual terms, which allow

for their conversion at a predetermined trigger (for example, contingent capital) and, unlike a negotiated restructuring,

do not require shareholder or creditor consent. Private contractual contingent capital instruments with write-off or

conversion features could complement the statutory power of the resolution authority.

Emergency liquidity can help mitigate spillovers—for example, across banks and from banks to markets (Dobler and

others 2016).

(continued)

INTERNATIONAL MONETARY FUND 25

insolvent banks on terms or against collateral that leave the central bank with some risk of loss.

some cases, state guarantees of bank liabilities (or loss protection on asset portfolios) may be needed

to ensure continued market access. Or it may become apparent that loss-absorbing capacity is

insufficient to cover potential losses or that imposing losses on certain claim holders could cause

contagion. Public funding may also be needed to capitalize bridge banks that acquire systemically

important liabilities and assets of a bank in resolution, to acquire temporary ownership of failed banks,

and to resort to other mechanisms to help restore the solvency of banks.

Given these challenges, building flexibility into resolution regimes would seem prudent. Even if

policymakers believe the risk of exceptional circumstances that would justify a bail-out is very low,

there is value in having the flexibility embedded in the resolution framework: such a framework would

allow authorities to cogently respond if and when necessary (that is, when spillovers are particularly

severe) and transparently set the appropriate safeguards to mitigate the moral hazard costs.

But

there is no international consensus on the optimal degree of flexibility. In addition, determining in

advance the conditions under which a bail-out is justified is a challenge, given their extraordinary and

difficult-to-anticipate nature. Spillovers will likely depend on observable variables, such as a bank’s

size and leverage. But they will also change with less-well-defined and less-readily-observable factors,

such as a bank’s degree of interconnectedness, the phase of the economic cycle, and the overall

conditions of a financial sector. Regimes that are too constrained may not afford sufficient flexibility

to preserve financial stability—for example, when they do not envisage circumstances under which

exemptions from bail-in are warranted and prove time-inconsistent (if subsequently the legislation

ends up having to be rewritten to deal with a crisis). At the other end of the spectrum, regimes that

are too flexible may perpetuate too-big-to-fail premiums, increase contingent fiscal risks, and

undermine the credibility of the regime. At a minimum, jurisdictions should ensure that the actual use

of such flexibility provided in legislation is subject to adequate transparency and accountability

mechanisms in order to help avoid potential

abuse.

The recent reforms to bank resolution

regimes have also included measures that

aim to minimize the burden of resolution on

taxpayers, even in cases where public funding

is deemed desirable. Legislation has imposed

conditions so that public funding may be

provided only (1) on a temporary basis and

only in systemic cases; (2) after bail-in of

shareholders and some creditors; (3) after the

use of ex ante resolution funds; and/or (4)

In a crisis it can be difficult to distinguish between illiquid and insolvent banks, and it is possible that liquidity

assistance ends up becoming solvency support and, hence, a bail-out.

As such, flexibility should not be confused with discretion to provide bail-outs at will. Also, note that KA 6.2

discourages but does not altogether prohibit public funding in resolution. Where necessary, the KA recommends that

losses from such funding be recovered from the private sector. For further information, see Box 3 (with examples from

recent resolution regimes) and Box 5.

Bail-outs to be provided:

needed some creditors

within a financial group's structure funds

levies

Improve resolvability

impediments to resolution

Good governance

Strict oversight

Exit planning

Powers to remove legal/operational

Cross-border planning

Minimizing Spillovers in Bail-ins

Only in systemic cases

On a temporary basis

After bail-in of shareholders and

After the use of ex ante resolution

Review and transparency

Periodic assessments and living wills

understand the risks and can absorb

the losses

Limit who can hold these claims

Enhance loss-absorbing capacity

Subject to ex post recovery through

Clarity on:

which claims qualify

how much of these claims are

where these claims are located

Unsecured/uninsured creditors that

Table 6. Reducing the Cost of Resolutions

Minimizing Moral Hazard in Bail-outs

… and accompanied with:

TRADE-OFFS IN BANK RESOLUTION

26 INTERNATIONAL MONETARY FUND

subject to ex post recovery of any potential use of public funds through levies imposed on the banking

sector. International experience and good practices offer further guidance on how best to structure

public solvency support in such cases (Box 5). Effective resolution regimes would combine these

principles with those aiming to make bail-ins a credible option (Table 6).

All these measures go in the direction of providing resolution authorities with effective bail-in powers

and improving the trade-off between bail-ins and bail-outs. It will take some time to complete

implementation and reach these goals, including ensuring that adequate loss-absorbing capacity is in

place for all banks that could prove systemic. Restrictions on who can hold securities that count as

loss-absorbing capacity would over time change the picture depicted in Figure 3, and limits on retail

investors’ exposure to complex securities would seem appropriate. In principle there are no nationality

restrictions as to who can hold TLAC instruments, but political economy considerations might come

into play, if those holding these securities are foreign investors. Potentially, home authorities might

resist the idea of a systemic bank becoming foreign owned through the conversion of their debt claims

into equity as part of a bail-in.

In addition, more remains to be done to ensure effective resolution

planning and orderly resolution across borders (see IMF 2014b). The Financial Stability Board noted

(FSB 2016c) that only a subset of its members has bank resolution regimes with comprehensive powers

in line with the KA. Furthermore, much of bank debt is issued in foreign jurisdictions where the effect

of national resolution powers remains uncertain, and cross-border resolution planning for individual

banks is untested in practice. Finally, only time will tell whether the reforms have established credibility

and struck the right balance in terms of constraints versus flexibility.

IV. CONCLUSION

After the global financial crisis, public bail-outs came to be viewed as too expensive, too inequitable,

and too harmful to market discipline. Consequently, a defining feature of financial reform efforts has

been limiting recourse to bail-outs. The new frameworks allow authorities to recapitalize a failed bank

through the bail-in of private stakeholders. Hence, reforms after the crisis have likely reduced the

potential for spillovers from bail-ins and made bail-outs less likely.

This note laid out some key considerations about the relative costs of bail-ins and bail-outs from a

theoretical and empirical viewpoint. The theoretical section focused on the trade-off between

spillovers associated with bail-ins and moral hazard effects of bail-outs. The model delivered several

insights. First, when spillovers are small, bail-outs should not be used. To avoid time-consistency

problems, this requires that policymakers operate under a constrained resolution framework that

relies on bail-ins whenever systemic concerns are not severe. Second, when the spillover risks from

bail-ins are very high, there may be some role for bail-outs. In such circumstances, the disruptive

effects of bail-ins may outweigh the moral hazard consequences of bail-outs. Third, social welfare

diminishes with the severity of spillovers. Therefore, it is crucial that policymakers continue to adopt

measures to reduce the potential for spillovers—for example, by providing ample liquidity during

Note, however, that KA 7.4 recommends national authorities not to discriminate against creditors based on their

nationality or the location of their claim, among other things. The treatment of creditors should be transparent and

properly disclosed to depositors, insurance policyholders, and other creditors.

INTERNATIONAL MONETARY FUND 27

periods of banking stress, restraining cross holding of bail-in-able debt within the financial sector,

applying enhanced prudential standards on systemically important financial institutions, and

enhancing the resolvability of large and complex financial institutions via effective resolution planning

and adequate loss-absorbing capacity. Further research aimed at measuring spillovers and moral

hazard is warranted, especially in the context of new resolution frameworks.

Pre-reform empirical evidence provided support for the key features of the model—namely, that bail-

outs could give rise to moral hazard and that in some cases bail-ins might entail financial spillovers.

Specifically, the empirical analysis first showed that public guarantees generated a funding advantage

for large banks. Whether this translates into more risk taking and moral hazard seems to depend on

the institutional environment and the macro-financial backdrop. The empirical analysis then provided

suggestive evidence that imposing losses on the private stakeholders of a bank has negatively affected

the market price of equity or debt securities of other banks in some cases. This may be interpreted as

evidence of adverse financial spillovers. The risk of spillovers is likely to be more contained under the

new resolution frameworks, particularly because authorities have stronger resolution powers,

institutions are increasing their loss-absorbing capacity, and there is greater clarity about which

instruments can be subject to bail-in. Nonetheless, it is important to remain alert to the possibility

that severe spillovers may materialize in the context of systemic crises or cross-border resolution.

There are some essential features that recent reforms have introduced to improve the trade-off

between bail-ins and bail-outs. Continued enhancement of these features can help strike an

appropriate balance between bail-ins and bail-outs and ensure that bail-ins operate as intended:

• Credibility and flexibility in resolution frameworks: Regimes that are too constraining may have

adverse consequences on financial stability—for example, if they preclude exempting creditors

from bail-in under any circumstance. But regimes that are too flexible may perpetuate too-big-

to-fail premiums while potentially increasing fiscal risk. Resolution frameworks should provide the

latitude—subject to strict conditions—for the authorities to provide financial support in the event

of bank failures only in exceptional cases where imposing excessive losses on private stakeholders

may give rise to large spillovers. More attention needs to be paid to good practices in the

provision of public funding as a last resort, to safeguard taxpayers’ interests.

• Enhanced prudential standards and resolution planning: To reduce the scope for bail-outs,

increasing the loss-absorbing capacity of major financial institutions and making them more

resolvable—for instance, through resolution planning—is critical. Progress should continue on this

front, including by linking prudential requirements on loss-absorbing capacity to resolution plans

and by improving cross-border resolution aspects.

• Consumer/investor protection regulations: Clarity regarding which financial instruments may be

subject to bail-in and restrictions on which investors can hold them directly are critical in limiting

potential spillovers and in containing backlash to bail-ins. Retail investors’ exposure to complex

bail-in-able instruments should be limited—for example, only marketed to professional,

institutional, or high-net-worth investors and diversified funds—and cross holding among banks

should be restricted. Enforcing such regulations requires reporting and database upgrades to

make effective real-time monitoring possible.

TRADE-OFFS IN BANK RESOLUTION

28 INTERNATIONAL MONETARY FUND

Box 1. Fiscal Implications of Bail-outs

Bail-outs may involve large fiscal costs. Gross fiscal outlays related to the restructuring of the financial sector can

indeed amount to several percentage points of GDP (Laeven and Valencia 2010) and may, in some cases, jeopardize

the sustainability of public finances. In assessing the fiscal costs of bail-outs, it is, however, important to consider

also the fiscal implications without bail-outs. In the context of a systemic financial crisis, abstaining from bail-outs

may result in a cascade of bank failures and widespread financial spillovers. This can result in a sharp contraction of

GDP and tax revenues, possibly leading to a larger increase in public debt than in the case of up-front and sufficiently

large public support to the banking system.

Testing these implications is unfortunately quite difficult due to endogeneity concerns. This is because the

authorities provide larger bail-outs exactly when crises are more severe. Furthermore, when bail-outs are expected,

the associated moral hazard effects lead to more severe crises. Therefore, we would expect to see a positive

correlation between the size of bail-outs, drop in GDP, and increase in public debt.

Yet the relationship may not be as monotonic as expected: a too-timid response that aims to keep up-front fiscal

costs down may backfire and end up associated with larger fiscal costs. In a sample of 87 systemic banking crises,

there is such an irregularity (Figure 1.1). When up-front fiscal costs are limited to less than 1 percent of GDP, output

losses reach 25 percent of trend GDP. When fiscal costs are between 1 and 5 percent of GDP, the typical output loss

is only 13 percent of trend GDP. Interestingly, this happens as peak nonperforming loan (NPL) ratios reach levels

comparable to cases where up-front fiscal costs are less than 1 percent of GDP. As up-front fiscal costs exceed 5

percent of GDP, output losses become larger and peak NPL ratios rise. Yet the increase in public debt is not

commensurate with the increase in up-front fiscal costs and, hence, the indirect fiscal costs—calculated as the

increase in public debt minus up-front fiscal cost—are lower. In simple ordinary least squares regressions, these

insights are corroborated: controlling for the size of the output loss, the correlation between up-front fiscal costs

and indirect fiscal costs is negative.

These observations show the danger of assessing the fiscal implications of bail-outs based only on their up-front

cost. They suggest that in the case of systemic banking crises, up-front bail-outs may do much to contain the severity

of the crisis and may ultimately also reduce the indirect fiscal costs.

-5

x<1 1<x<5 x>5

Output loss Peak NPLs Increase in public debt Indirect fiscal costs

Figure 1.1. Crisis Severity and Fiscal Costs

(percent)

Sources: Laeven and Valencia 2010; and IMF staff calculations.

Note: x refers to the up-front fiscal cost of a systemic banking crisis, expressed in percent of GDP and defined as

the component of gross fiscal outlays related to the restructuring of the financial sector. It includes costs

associated with bank recapitalizations but excludes asset purchases and direct liquidity assistance from the

treasury. Output loss is computed as the cumulative sum of the differences between actual and trend real GDP

over the period (t, t+3) with t the starting years of the crisis and expressed in percent of trend real GDP. Peak

NPLs are in percent of total loans. Increase in public debt is in percent of GDP and measured over (t-1, t+3).

Indirect fiscal costs are calculated as increase in public debt minus up-front fiscal cost.

INTERNATIONAL MONETARY FUND 29

Box 2. What Is Bail-in?

Statutory bail-in is a resolution tool that enables the recapitalization of a failed financial firm through cancellation,

conversion, transfer, or write-down of claims of equity holders and unsecured and uninsured creditors, to the extent

necessary to absorb losses on its balance sheet.

Following a bail-in, the firm (or a successor firm) can meet

regulatory capital requirements, and its critical functions are preserved. In the analytical sections of this note, we use

the term more broadly to refer to the ability of the authorities to impose losses on private stakeholders of a failing

bank.

Contractual bail-in is another way to impose losses on private stakeholders to recapitalize a bank in resolution. Not

all jurisdictions have adopted mandatory bail-in requirements. In such jurisdictions, implementing a bail-in may not

be legally feasible absent provisions in banks’ debt contracts that allow a resolution authority to bail in claims under

such contracts.

Contingent convertible bonds (CoCos) are hybrid capital securities that absorb losses when the capital of the issuer

falls below a certain threshold. They carry two defining features: (1) a mechanism that specifies how losses will be

absorbed (conversion into common equity or principal write-down) and (2) a trigger that activates this mechanism

(often a given level of the common equity Tier 1 (CET1) ratio and sometimes a “point of nonviability trigger” (PONV)

left at the discretion of the supervisor). An important benefit of CoCos—especially relative to equity—is their lower

after-tax cost. Regulatory capital eligibility considerations determine the other CoCo features, including maturity.

CoCos are supposed to be a first line of defense and to be triggered/depleted by the time a resolution authority

decides to use its bail-in powers.

The scope of bail-in: Bail-in-able claims must be clearly specified in legislation or contract. While some jurisdictions

specify claims that cannot be bailed in, others provide general discretion to exclude on a case-by-case basis claims

that will not be bailed in.

Bail-in must respect the order of priorities of claims established in a given jurisdiction for

bank liquidation, meaning that typically, equity must bear losses first, followed by subordinated debt, with deposits

being the last to absorb losses in jurisdictions that provide for depositor preference.

Ensuring depositor preference

can mitigate the likelihood of contagion by providing depositors with a higher probability of recovering their claims

in a bank failure.

Cross-border context: For cross-border banks, recognition and enforcement of bail-in is achieved through statutory

mechanisms (such as automatic mutual recognition within the European Union or specific powers to recognize and

to give effect to bail-in by foreign resolution authorities) or through contractual approaches to recognition of

resolution actions. With respect to the latter, a number of G20 jurisdictions require their banks to provide for

contractual bail-in recognition clauses in certain debt contracts governed by foreign law.

_____________________________________________

The Financial Stability Board’s Key Attributes (KA) of Effective Resolution Regimes include a “no creditor worse off” rule, designed

to ensure that creditors of a failed bank are not left worse off in resolution than they would have been in liquidation proceedings

against the bank. Mandatory compensation is required for claimants who are made worse off, as determined by an independent

valuation.

For example signatories to the International Swaps and Derivatives Association Resolution Protocol agree to incorporate bail-in

provisions in their financial contracts.

The KA require that this be done on a transparent basis, for limited reasons such as to contain systemic risk or to maximize the

value of the remaining assets for all creditors.

4. There may be instances where the legal framework may permit the resolution authority to depart from the established order

of priorities by exempting certain claims from the imposition of losses in resolution in the interest of financial stability.

Many jurisdictions (including Argentina, Australia, China, India, Hong Kong SAR, Indonesia, Mexico, Russia, Singapore, Switzerland,

Turkey, the United States, and EU member states) give preference to depositors.

For example, EU jurisdictions impose this requirement pursuant to Article 55 of the EU Bank Recovery and Resolution Directive.

TRADE-OFFS IN BANK RESOLUTION

30 INTERNATIONAL MONETARY FUND

Box 3. Resolution Regimes in Key Large Home Jurisdictions

China: The China Banking Regulatory Commission, as banking supervisor, can place a failing bank into receivership,

under which the deposit insurer (currently the People’s Bank of China) acts as receiver and may restructure a failing

bank. The receiver may use the deposit insurance fund to provide financial support for a transfer of assets and

liabilities; for example, through financial assistance, guarantees, and loss sharing. The regime lacks bridge bank, bail-

in, and other powers laid out in the Financial Stability Board’s Key Attributes (KA), which may be needed to resolve

a systemic entity.

European Union: Most EU members had adopted the Bank Restructuring and Resolution Directive (BRRD) by

January 2015 (the deadline for adopting the bail-in powers was a year later). The BRRD established administrative

resolution regimes, with broadly harmonized resolution powers closely aligned with the KA across the European

Union. The mandatory bail-in provisions of the BRRD require that any public funding for a bank in resolution can be

provided only after the bank’s shareholders and creditors have absorbed losses of at least 8 percent of the total

liabilities of the failed bank. Under certain conditions (for example, if a bank is solvent but cannot generate enough

profits to sustain itself during a serious economic disturbance—for instance, the bank has enough capital in the

baseline scenario but falls short in a stress scenario), public support can be provided outside of resolution, subject

to the applicable rules on state aid. These require that capital and subordinated debt holders (at least) absorb losses,

along with other “compensatory” measures to restructure the bank’s operations.

Japan: In 2013 a new regime was introduced, under which the Deposit Insurance Corporation of Japan (DICJ) can

take over control of a bank, under the direction of the bank supervisor, and transfer assets and liabilities to a third-

party purchaser or a bridge bank owned and administered by the DICJ. Resolution actions taken by the DICJ require

court approval in lieu of shareholder approval

. Wider powers to provide financial assistance or temporary

nationalization, can be deployed after a systemic risk determination by the prime minister. The Japanese regime

lacks statutory bail-in power.

United States: There are two separate resolution regimes—one for systemically important nonbanks—for example,

bank holding companies (Title II of the Dodd-Frank Act, introduced in 2010) and another for banks. The Federal

Deposit Insurance Corporation (FDIC) is appointed receiver

under both with resolution powers closely aligned with

those in the KA, which can be exercised administratively and do not require shareholder or creditor consent. Those

under Title II first require that the Treasury secretary determine that the firm represents a systemic risk; and

potentially an expedited court approval process may be triggered. Bankruptcy is the first resolution option in the

event of a failure of a systemic financial company and, to make this prospect achievable, Title I of the Dodd-Frank

Act requires that all large, systemic financial companies submit living wills to demonstrate how they would be

resolved under the bankruptcy code. The Title II regime prohibits taxpayers from bearing the losses of a nonbank

placed into receivership. However, public funds can be lent to the Orderly Liquidation Fund, which in turn can provide

funding to a firm in (but not outside of) resolution, subject to certain conditions. The US regimes lack a statutory

power for bail-in, but may be able to achieve the same economic effect using bridge bank and other powers.

Flexibility in the bank resolution regime was tightened after the global financial crisis—for example, the systemic

risk exception authority of the FDIC can now be used only for banks placed into receivership and wound down.

_____________________________________________

Sources: FSB 2016a, 2016b: and IMF Financial Sector Assessment Program reports.

The terms “receivership” and “receiver” used in the US regime are analogous to “resolution” and “resolution authority.”

INTERNATIONAL MONETARY FUND 31

Box 4. Making Bail-in a Credible Policy Option for Systemic Banks

Several measures have been agreed to at the international level to mitigate the potential spillover risks from the use

of the bail-in tool in resolution. The Financial Stability Board has provided guidance on measures that seek to

improve the outcomes of bail-in for individual countries and in the cross-border context. These include the following:

Recovery and resolution planning (RRP): National authorities are expected to ensure that banks are subject to

periodic resolvability assessments as part of resolution planning exercises. Recovery plans are required to be

prepared by the governing boards of banks to help their recovery from financial distress, with a minimum disruption

of critical services. In addition, resolution authorities are required to prepare resolution plans for each bank reflecting

concrete strategies for resolving them while safeguarding financial stability. Supervisory and resolution authorities

are also required to undertake annual resolvability assessments for banks, and to require changes in the legal and

operational structures of a firm if that is necessary to ensure continuity of critical functions in resolution. RRPs must

reflect cross-border issues, including ex ante agreement on operational resolution strategies to be used for resolving

such firms (see below).

Loss-absorbing capacity (LAC): In a systemic crisis, failure of any bank (regardless of size) may pose risks to financial

stability, and without sufficient LAC (both quality and quantity), resolvability will not be achieved. Adequate levels of

quality LAC in banks helps support their orderly resolution without recourse to public funds and with little or no

potential systemic risk. The Financial Stability Board has called for global systemically important banks (G-SIBs) to

maintain total loss-absorbing capacity (TLAC) in accord with their risk-weighted assets and leverage ratios. TLAC

implementation by various national authorities is ongoing; banks in key jurisdictions (for example, the European

Union, Japan, and others) are working toward complying with these requirements by 2019. The EU version, known

as “minimum requirement of own funds and eligible liabilities (MREL)” applies to all EU banks, regardless of size, at

both the individual and group consolidated levels.

Cross-border context: The Financial Stability Board’s Key Attributes call for cooperative arrangements among

national resolution authorities for the resolution of global systemically important financial institutions (G-SIFIs).

Home and key host authorities are expected to maintain crisis management groups (CMGs, made of members of

financial safety nets in each relevant jurisdiction) to prepare for and facilitate resolution of cross-border banks. They

are also expected to enter into institution-specific cooperation agreements (CoAgs) for each G-SIFI

. These

agreements enable execution of an agreed resolution strategy. National authorities have adopted one of two

operational strategies for resolving cross-border banks—namely, the single point of entry (SPE) and multiple points

of entry (MPE). Under the SPE approach, the home resolution authority of the apex holding company of the cross-

border financial group resolves the holding company—typically using the bail-in power under its legal framework.

Shareholders and eligible creditors of the holding company absorb losses of the entire group through a write-down

or restructuring of their equity and/or debt claims against the apex entity. Capital freed up from this exercise is

passed down to subsidiaries operating at a loss and used for their recapitalization and liquidity provision. Host

jurisdictions should be able to recognize and enforce resolution measures taken by the home authorities under an

SPE strategy, using statutory powers of recognition and enforcement or through contractual provisions recognizing

bail-in by foreign resolution actions. The SPE strategy requires an effective ex ante cooperation arrangement among

the authorities of relevant jurisdictions. This strategy must be

underpinned by aligned incentives that create

confidence that their respective national interest for promoting financial stability will be protected. In the absence

of such mutual trust, host authorities tend to undertake parallel resolution actions or ring-fence branches or

subsidiaries in their jurisdictions, with suboptimal outcomes for all creditors of the institution/group. Under the MPE

approach, home and relevant host authorities resolve nonviable parts of the financial group in separate proceedings

using a range of resolution tools available to them in their respective jurisdictions. In the case of cross-border banks,

resolution planning is expected to include ex ante agreement among home and key host jurisdictions on resolution

strategies and the amount and location of LAC, in the context of CMGs and CoAgs. These measures (if well

implemented) will help to promote trust and effective cooperation among relevant jurisdictions to give meaning to

orderly resolution strategies.

TRADE-OFFS IN BANK RESOLUTION

32 INTERNATIONAL MONETARY FUND

Box 5. Good International Practice in Public Solvency Support

Public solvency support for problem banks typically involves significant cost, risk, and moral hazard. Private solutions

and, if unattainable, orderly resolution without recourse to public funds, are distinctly preferable. The limits of public

support should be recognized. Debt sustainability concerns may cast doubt on the feasibility of recapitalization

strategies and contribute to further negative market reaction (the so-called sovereign-bank nexus). However, public

solvency support may be unavoidable in exceptional circumstances—for example, if spillovers are high and/or an

effective resolution regime is not in place. In such circumstances, it should be provided only under strict conditions

that maximize burden sharing, minimize moral hazard, and protect taxpayers.

• Systemic stability: Public support should be reserved for institutions whose failure would destabilize the financial

system and/or jeopardize the continuity of essential payment, clearing, and settlement functions.

• Burden sharing: Unrecognized losses must be identified, ideally via a comprehensive asset quality review if time

allows, and the bank’s equity must be written down for the losses before provision of public funds. To the extent

that it is compatible with financial stability and permissible under the legal framework, loss allocation should

continue in accordance with the creditor hierarchy, ultimately affecting claims of uninsured senior unsecured

creditors.

• Restructuring: Solvency support must

be paired with a comprehensive restructuring plan that addresses

structural weaknesses and helps restore long-term viability, including via cost cutting and a stronger risk

management framework, capital and liquidity planning, and so forth. Public solvenc

y support should be

remunerated to help mitigate moral hazard. Plans should provide for recovery

of public support within a

reasonable time frame, if necessary via divestiture of selected assets and business lines.

• Governance: Managers responsible for the failure of the bank should be replaced, executive compensation

capped, and any bonuses paid to senior management before the failure clawed back (if possible). To fuel internal

capital generation, dividend payments (if any original shareholders remain) need to be suspended until solvency

support is repaid. The authorities should establish a high-level intra-

agency committee to coordinate the

nationalization, ensure timely and consistent information is released publicly, and manage the public sector’s

interest on an arm’s length basis. Central banks and supervisors ideally should not contribute to recapitalization

to avoid potential conflicts of interest. The central bank, however, may need to provide liquidity to viable banks

that have been recapitalized in or outside of the resolution regime.

• Strict oversight: Recipients of public support must be subjected to strict supervision and enhanced reporting to

prevent excessive risk taking, foster robust governance and safe and sound practices, and ensure consistent

implementation of the restructuring plan. Supervisors should establish measures to prevent asset stripping,

monitor, and—if needed—block intragroup and insider transactions.

• Exit planning: Solvency support should be structured to incentivize timely repayment—for example, via interest

step-ups. Divestment strategies should be carefully analyzed and initiated as soon as market conditions allow.

While exit scenarios should aim to ensure a reasonable return on the financial aid—for example, via market-

based remuneration and/or the issuance of equity warrants to the government, the overarching objective should

be to return the bank to private ownership within a reasonable amount of time—even if that entails losses on

the original investment.

• Review: A review, conducted with independent expertise, should focus on the events that led up to the bank’s

failure and identify the structural weaknesses in its business model, governance, and risk controls—with the aim

of learning how to prevent recurrence and determining the culpability of senior management. The supervisor

should also closely evaluate its own role in order to identify potential improvements to supervisory procedures,

the reporting framework, and instruments for early intervention.

• Transparency: Ongoing disclosure on the actual and estimated cost of the public solvency support and any

recovery realized (updated periodically) is crucial for accountability to taxpayers.

INTERNATIONAL MONETARY FUND 33

Appendix I. A Simple Model of Bail-ins and Bail-outs

Consider a continuum of banks, each endowed with an exogenous level of capital . Capital can

include common equity as well as state-contingent debt that is automatically converted into equity

or written off in case of financial distress. Banks are heterogeneous in size, with capital  being

distributed over the support [0, 



]. Banks leverage up to the regulatory limit by issuing debt , which

carries the interest rate 



On the asset side, banks provide loans equal to  =  + , with a lending

rate equal to 



.  denotes the proportion of lending financed with bank capital, so that  = /.

Bank loans are used by competitive firms to produce output according to a simple linear production

function  =   . The level of productivity  is subject to the shock , which is distributed as follows:

The random variable  is uniformly distributed over the [0,1] interval and is given by the sum of an

idiosyncratic and aggregate shock  = 



+ 



. Note that  is thus bank specific. We refer to the case

in which  = 1 as the “good state” and the case in which  =  as the “distress state.” Each bank can

control the probability  by exercising a monitoring effort  that entails a pecuniary cost equal to





/2, where  > 0. Specifically, we assume that  = .

Without loss of generality, we assume that banks fully appropriate output production, setting 



= .

In the good state, bank loans are fully repaid, while in the distress case banks recover only 



. Banks

operate under limited liability so that they fully repay creditors only in the good state or in the distress

state if loan repayments are high enough:

 >









= 

(1)

If instead  < , the funding gap between the bank assets and liabilities must be absorbed through a

combination of public bail-outs or the bail-in of private creditors. Bail-outs involve a government

transfer of size  and entail a fixed cost 



. We assume that a fraction  of bail-outs is appropriated

by bank owners, while the remaining part 1   is used to repay bank creditors.

Bail-ins involve

instead the write-down of bank creditors’ claims that see repayments reduced to 



 , where 

denotes the size of the bail-in. The sum of bail-ins and bail-outs must fully cover the bank’s unfunded

liabilities, so that when  <  :

In the context of the model, debt includes both deposits and bonds. For the sake of simplicity, we assume that all

deposits can in principle be bailed-in. We could easily extend the model to incorporate insured deposits, which carry

a bail-out guarantee by the government. In this case, the questions of whether to use bail-ins or bail-outs would be

relevant only for the portion of deposits that is uninsured.

This assumption is quite standard in the literature and often corresponds to the observation that banks are likely to

exploit their informational advantage to maximize the transfer they receive. In addition to being politically and socially

costly, the fact that the bank can seize part of the bail-out transfers generates moral hazard since it reduces the bank’s

incentives to monitor borrowers.

1 with probability

0 1 with probability 1

≤≤ −







TRADE-OFFS IN BANK RESOLUTION

34 INTERNATIONAL MONETARY FUND

(

1  

)

 +  = 



 





(2)

The expected net return for an individual bank is thus given by





= 

 





 + 











(

1 

)





















 +

















where  is the gross risk-free rate.

Regarding bank funding costs, we assume that bank creditors are risk-neutral, act competitively, and

can observe the monitoring level chosen by the bank.

Creditors require an interest rate that allows

them to break even in expectation by providing compensation against the risk of bail-ins:





=  +

(

1 

)















By substituting out 



, the bank’s expected returns can be rewritten as





= D 

 





 +









(



)















(

1  

)















(3)

Note that banks appropriate in expectation all benefits from public bail-outs, not only the share  that

is directly seized by bank owners. This is because banks also benefit from the bail-out share 1   that

is used to repay creditors, since, by reducing the expectation of bail-ins, bail-outs lower bank funding

costs 



. Nonetheless, the higher the share , the greater the risk of moral hazard, since a larger bail-

out is needed to achieve a given reduction in bail-ins, as shown in equation (2).

We now turn to the costs imposed on the rest of society when a bank becomes insolvent; that is,  <

. The model incorporates two sources of losses. First, in the case of bail-outs, the government faces

the associated fiscal cost  and the fixed costs 



. Second, following Sandri 2015, we allow for the

possibility that bail-ins may involve spillovers. Besides imposing losses on bank creditors equal to ,

we assume that bail-ins involve negative externalities for the rest of society equal to , where the

parameter   0 controls the severity of spillovers. The cost for the rest of society  from resolving an

individual bank is then

 =

(

1 

)

























where 1



is an indicator function taking the value of 1 if bail-outs are positive and zero otherwise. The

model takes the intensity of spillovers  as an exogenous parameter and can thus encompass different

mechanisms through which spillovers can emerge. One plausible assumption is that spillovers are

proportional to the overall capital shortfall in the banking sector so that

If bank monitoring is not observable, the provision of bail-outs could actually increase monitoring as explained in

note 2.

INTERNATIONAL MONETARY FUND 35

 









(

  1

)





 





(



)







,

where 

(



)

is the probability density function of  and 1





is an indicator taking a value of 1 if the

bank is insolvent; that is,   . In this case, spillovers become severe if a large financial institution fails

or if an aggregate shock pushes many banks into insolvency.

By subtracting from the bank’s returns the costs faced by the rest of the society in case of insolvency,

we can then define aggregate social welfare as





= 



  = D 

 





 +









(



)

















(

1  

)

























Consider now the problem of a social planner that aims to maximize social welfare 



by choosing

which portion  of a bank’s unfunded liabilities 



 



 should be subject to a bail-in in case  <

. The remaining share 1   is then covered with public bail-outs. Importantly, the planner cannot

control the level of monitoring that is chosen by the bank to maximize its own profit. Formally, the

planner chooses the bail-in’s share  by solving max







subject to  = arg max







From an ex post perspective; that is, once the bank faces a capital shortfall, bail-outs are preferable to

bail-ins whenever the spillover costs associated with bail-ins exceed the fixed cost of bail-outs.

However, the choice between bail-outs and bail-ins should consider not only the ex post costs, but

also the ex ante implications for bank monitoring. In this regard, note that equation (3) shows that the

provision of bail-outs reduces monitoring incentives for the bank since it provides compensation even

if the bank faces a capital shortfall; that is,  < . This is how bail-outs generate moral hazard.

The optimal choice between bail-ins and bail-outs should thus be based on a careful trade-off

between the ex post costs associated with bail-ins and the ex ante moral hazard effects arising from

bail-outs. The solution is illustrated in Figure 1, showing that bail-outs should play an important role

in bank resolution only if spillovers are particularly large.

Appendix II. Sample Coverage and Empirical Specifications

The empirical analysis is conducted for a global sample of banks. These banks are selected in two

steps. The first step criterion is bond issuance activity and size (total assets): a set of active issuers is

identified based on available information from Bloomberg Finance L.P. and Thomson Reuters

Datastream on issuances of senior unsecured, subordinated, and contingent convertible bonds. In the

second step, a set of banks that are not selected based on this criterion but rank in the top 500 banks

globally are added to the sample. Daily data on equity prices, credit default swap spreads, and bond

prices are merged with quarterly data on balance sheet components (the source in this case is SNL

Financial). The process delivers a global sample that consists of 841 banks across 75 countries

encompassing both advanced and emerging market economies with coverage from 2008 to 2016.

The countries covered are Argentina, Australia, Austria, Bahrain, Bangladesh, Barbados, Belgium,

Note that we can neglect bank creditors since they always break even in expectation.

TRADE-OFFS IN BANK RESOLUTION

36 INTERNATIONAL MONETARY FUND

Bermuda, Brazil, Bulgaria, Canada, Chile, China, Colombia, Croatia, Cyprus, Czech Republic, Denmark,

Egypt, Estonia, Faroe Islands, Finland, France, Germany, Greece, Hong Kong SAR, Hungary, India,

Indonesia, Iran, Ireland, Israel, Italy, Japan, Kuwait, Lebanon, Liechtenstein, Luxembourg, Malaysia,

Malta, Mexico, Morocco, Netherlands, New Zealand, Nigeria, Norway, Oman, Pakistan, Panama, Peru,

Philippines, Poland, Portugal, Qatar, Romania, Russia, Saudi Arabia, Singapore, Slovak Republic, South

Africa, South Korea, Spain, Sri Lanka, Sweden, Switzerland, Taiwan Province of China, Thailand, Togo,

Trinidad and Tobago, Turkey, United Arab Emirates, United Kingdom, United States, Venezuela, and

Vietnam.

For Figure 2, daily changes in CDS spreads and equity prices around three events—September 15,

2008 (Lehman bankruptcy), October 30, 2008 (TARP passage), and September 30, 2008 (Dexia capital

injection)—are plotted against bank size.

For Tables 1, 4, and 5, cumulative abnormal changes in CDS spreads and equity prices are computed

in three steps. First, we estimate the relationship between the bank-level variable and its market

equivalent over 60 trading days prior to relevant event. Market equivalents are benchmark regional

indices collected from Thomson Reuters Datastream and Bloomberg Finance L.P. In the second step,

abnormal changes are calculated as the difference between the actual change and the change

predicted based on the regression estimated in the first step. Finally, these abnormal changes are

summed over an event window to arrive at the cumulative abnormal changes. We use several event

windows ranging from 1 to 10 days in length (that is, from t to t + 1, …, t + 10).

The regression specification in Table 1 is





=  + 



+ 



+ 



where  denotes the bank and  denotes the country of residence of the bank. The regressions are run

separately for each of the three bail-out events featured in Figure 2, with and without country fixed

effects, 



. Standard errors are clustered at the country level. A positive (negative) beta for CDS

spreads (equity prices) can be interpreted as larger banks being affected adversely by the event in

question.

The regression specification in Table 4 is





=  + _



+ _



+ 



where _ is a dummy that takes the value of 1 if the bank’s residence is in the country where

the bail-in event happened and _ is a dummy that takes the value of 1 if the bank’s residence

is in a country that is an EU member but is not the country where the bail-in event happened. The

regressions are run separately for each of the 16 bail-in events. The list of events is in Appendix Table

1. For the Cyprus event, an additional dummy for banks whose residence is Russia is included. Results

are robust to clustering standard errors at the country level.

A similar specification is used in Table 5:





=  + _



+ _



+ 



INTERNATIONAL MONETARY FUND 37

where _ is a dummy that takes the value of 1 if a bank is headquartered in the same state as

the bank resolved by the Federal Deposit Insurance Corporation (“in-state”), and _ is a dummy

that takes the value of 1 if a bank is headquartered in a different US state (“out-of-state”). The

regressions are run separately for each of the 32 bail-in events. The list of events is in Appendix Table

2. The results are robust to clustering at the state level.

The coefficients obtained on the dummy variables in Tables 4 and 5 can be interpreted as the marginal

impact of being “closer” to a bail-in event, conditional on a bail-in event. Given that non-bail-in events

are not included in the estimation sample, the coefficients do not represent the average overall impact

of the bail-in event. Note that there is no restriction on the samples used for these tables; that is, the

samples include not only European/US banks but also banks from the other regions.

For Table 2, summary statistics on funding costs (CDS spread, senior unsecured bond spread,

subordinated bond spread, contingent convertible bond (CoCo) spread), risk measures (Z-score, Tier

1 ratio, leverage ratio, NPL ratio), and profitability metrics (ROA, ROE) are displayed separately for

large and small banks in the sample. Large banks are defined as those with assets in the top percentile

of the full sample distribution. Z-score is computed as equity capital ratio plus return on assets divided

by the standard deviation of return on assets (calculated over a rolling window of 10 quarters). Tier 1

ratio is defined as the ratio of Tier 1 capital to risk-weighted assets, and leverage ratio is defined as

the ratio of Tier 1 capital to total tangible assets. NPL ratio is the ratio of nonperforming loans to total

loans. ROA is return on average assets and ROE is return on average equity.

For Table 3, the risk measures and profitability metrics displayed in Table 2 are treated as the

dependent variable in the following regression specifications:





=  + 



+ 







+ 







+ 



+ 



+ 



+ 







=  + 



+ 







+ 







+ 







+ 



+ 



+ 



+ 



where  refers to gross loans in percent of total assets and  refers to total deposits in

percent of total assets. Z-score is omitted in the specification where the dependent variable is a risk

measure given that it is also a risk measure, but it is included in the profitability regression to assess

returns controlling for risk. Of course, the set of fixed effects included in each specification is either

bank and time (



, 



) or country and time (



, 



). Results are robust to clustering standard errors at

the country or bank level.

Note that the number of banks in the regression tables is fewer than 841 because of missing data

over the estimation and/or event windows.

TRADE-OFFS IN BANK RESOLUTION

38 INTERNATIONAL MONETARY FUND

Event Date Description

Austria 11-Apr-16 Resolution of HETA; first test of covered bonds under resolution; BRRD regime

Cyprus 25-Mar-13

Resolution and restructuring of Laiki and Bank of Cyprus, with Laiki's insured deposits, emergency liquidity assistance, and enough assets

to meet regulatory requirements transferred to the Bank of Cyprus and uninsured deposits and other assets left in a run-off unit, and

Bank of Cyprus recapitalized with participation of creditors including uninsured depositors to attain regulatory limits; an unexpected bail-

in following ad hoc adoption of a resolution law inspired by the BRRD (which was then still under negotiation so a pre-BRRD experience)

Denmark (1) 7-Feb-11

Bankruptcy of Amagerbanken; first time haircut imposed on senior debt in Europe; first test of the Danish bail-in law passed six months

prior to the event; pre-BRRD

Denmark (2) 5-Oct-15 Resolution of Andelskassen; hybrid application of the bridge bank and bail-in tools; BRRD regime

Greece (1) 10-Oct-11 Resolution of Proton Bank via a bridge bank; pre-BRRD

Greece (2) 18-Jan-13 Resolution of Hellenic Post Bank via a bridge bank; pre-BRRD

Greece (3) 11-Dec-15

Precautionary recapitalization of National Bank of Greece and Piraeus Bank; resolution not triggered but subordinated debt and senior

bonds converted to equity; BRRD regime

Italy (1) 17-Jul-15

Liquidation of Banca Romagna Coop; burden sharing of equity and subordinated debt (later paid in full by cooperative bank’s voluntary

fund); under national insolvency law; pre-BRRD

Italy (2) 23-Nov-15

Resolution of Banca Marche, Cassa di risparmio di Ferrara, Popolare Eturia, Carichieti; equity and subordinated debt bailed in; BRRD

regime

Netherlands 1-Feb-13 Nationalization of SNS Reaal; junior debt wiped out; pre-BRRD regime and in the run-up to the Cypriot bail-in

Portugal (1) 4-Aug-14

Resolution of Banco Espirito Santo (BES) via a bridge bank (Novo Banco); equity and subordinated debt become shares and bonds of the

bad bank; BRRD regime (directive adopted in 15 May and entered into force on 2 July, although compliance and implementation

deadline at the end of the year)

Portugal (2) 29-Dec-15

Transfer of some non-subordinated bonds back from Novo Banco to BES; BRRD regime (and eve of BRRD bail-in rule kick-in on 1

January 2016)

Slovenia 12-Dec-13

Public recapitalization of NLB, NKBM, Abanka, Probanka, and Factor Banka; bail-out accompanied by private loss absorption where

shares and subordinated bonds are wiped out; pre-BRRD

Spain (1) 10-Jul-12

Spanish bank rescue plan involving subordinated liability exercises in addition to public recapitalization; preference shares and

subordinated debt affected by haircuts and conversion; pre-BRRD

Spain (2) 22-Mar-13 Resolution of Bankia; haircuts on subordinated bonds; pre-BRRD

United Kingdom 17-Jun-13 Recapitalization of Co-operative Bank; negotiated/consensual bail-in or "liability management exercise"; outside the BRRD regime

Appendix Table 1. Bail-in Events in Europe

Sources: European Parliament reports and news articles; Philippon and Salord 2017; Schafer, Schnabel, and Weder di Mauro 2016; World Bank 2016.

Note: The term bail-in is used in a generic sense to capture any resolution that imposed losses on private stakeholders. Bank Recovery and Resolution Directive (BRRD)

2014/59/EU of the European Parliament and of the Council of 15 May 2014 established a framework for the recovery and resolution of credit institutions and investment

firms. The BRRD entered into force on 2 July 2014. EU member states were required under Article 130 of the BRRD to adopt and publish the laws, regulations, and

administrative provisions necessary to comply with the BRRD by 31 December 2014 and to apply those with effect from 1 January 2015, except in relation to the bail-in

provisions, which were to apply from 1 January 2016 at the latest.

INTERNATIONAL MONETARY FUND 39

Event

Date Headquarter Location

Washington Mutual (WaMu)

25-Sep-08 Nevada

Silverton Bank

1-May-09 Georgia

Independent Bankers Bank

18-Dec-09 Illinois

Waterfield Bank

5-Mar-10

Maryland

Midwest Bank and Trust Company

12-May-10

Illinois

TierOne Bank

4-Jun-10

Nebraska

Enterprise Banking Company

21-Jan-11 Georgia

First Community Bank

28-Jan-11 New Mexico

BankEast

27-Jan-12 Tennessee

Bank of the Eastern Shore

27-Apr-12 Maryland

DuPage National Bank

17-Jan-14 Illinois

The Bank of Union

24-Jan-14 Oklahoma

Syringa Bank

31-Jan-14 Idaho

Millennium Bank NA

28-Feb-14 Virginia

Vantage Point Bank

28-Feb-14 Pennsylvania

Allendale County Bank

25-Apr-14 South Carolina

AztecAmerica Bank

16-May-14 Illinois

Columbia Savings Bank

23-May-14 Ohio

Slavie Federal Savings Bank

30-May-14 Maryland

The Freedom State Bank

27-Jun-14 Oklahoma

Eastside Commercial Bank

18-Jul-14

Georgia

GreenChoice Bank, fsb

25-Jul-14

Illinois

NBRS Financial

17-Oct-14 Maryland

The National Republic Bank of Chicago

24-Oct-14 Illinois

Frontier Bank, FSB D/B/A El Paseo Bank

7-Nov-14

California

Northern Star Bank

19-Dec-14 Minnesota

First National Bank of Crestview

16-Jan-15 Florida

Highland Community Bank

23-Jan-15 Illinois

Capitol City Bank & Trust Company

13-Feb-15 Georgia

Doral Bank

27-Feb-15 Puerto Rico

Edgebrook Bank

8-May-15 Illinois

Premier Bank

10-Jul-15 Colorado

The Bank of Georgia

2-Oct-15 Georgia

Hometown National Bank

2-Oct-15 Washington

Appendix Table 2. Bail-in Events in the United States

Source: Federal Deposit Insurance Corporation (FDIC).

Note: Compiled based on FDIC-executed resolution of failed banks as reported on the

agency's website. The cases were selected from more than 500 bank failures the agency

handled since 2000, subject to the condition that the approach used was a purchase &

assumption (P&A) and the failed bank's total assets ranked among the top 30. The term

bail-in is used in a generic sense to capture any resolution that imposed losses on private

stakeholders.

TRADE-OFFS IN BANK RESOLUTION

40 INTERNATIONAL MONETARY FUND

References

Acharya, Viral, Deniz Anginer, and A. J. Warburton. 2014. “The End of Market Discipline? Investor

Expectations of Implicit State Guarantees.” Unpublished.

Adrian, Tobias, and Hyun Song Shin. 2014. “Procyclical Leverage and Value-at-Risk.” Review of

Financial Studies 27(2): 373–403.

Allen, Franklin, Elena Carletti, and Robert Marquez. 2011. “Credit Market Competition and Capital

Regulation.” Review of Financial Studies 24(4): 983–1018.

Anginer, Deniz, Asli Demirguc-Kunt, and Min Zhu. 2014. “How Does Deposit Insurance Affect Bank

Risk? Evidence from the Recent Crisis.” Journal of Banking and Finance 48: 312–21.

Avgouleas, Emilios, and Charles Goodhart. 2015. “A Critical Evaluation of Bail-in as a Bank

Recapitalization Mechanism.” In The New International Financial System: Analyzing the Cumulative

Impact of Regulatory Reform, edited by Douglas D. Evanoff, Andrew Haldane, and George G.

Kaufman. World Scientific Studies in International Economics 48: 267–305.

Barth, J. R., Gerard Caprio, and Ross Levine. 2004. “Bank Regulation and Supervision: What Works

Best?” Journal of Financial Intermediation 13: 205–48.

Beck, Thorsten, Asli Demirguc-Kunt, and Ross Levine. 2006. “Bank Concentration, Competition, and

Crises: First Results.” Journal of Banking and Finance 30: 1581–1603.

Berger, Allen N., Christa H. S. Bouwman, Thomas Kick, and Klaus Schaeck. 2016. “Bank Liquidity

Creation Following Regulatory Interventions and Capital Support.” Journal of Financial

Intermediation 26: 115–41.

Berger, Allen N., Raluca A. Roman, and John Sedunov. 2016. “Do Bank Bailouts Reduce or Increase

Systemic Risk? The Effects of TARP on Financial System Stability.” Federal Reserve Bank of Kansas

City Research Working Paper RWP 16–08.

Bhagat, Sanjai, Brian Bolton, and Jun Lu. 2015. “Size, Leverage, and Risk-Taking of Financial

Institutions.” Journal of Banking and Finance 59: 520–37.

Black, Lamont, and Lieu Hazelwood. 2012. “The Effect of TARP on Bank Risk-Taking.” Board of

Governors of the Federal Reserve System International Finance Discussion Paper IFDP 1043.

Blancher, Nicolas, Srobona Mitra, Hanan Morsy, Akira Otani, Tiago Severo, and Laura Valderrama.

2013. “Systemic Risk Monitoring (“SysMo”) Toolkit—A User Guide.” IMF Working Paper 13/168,

International Monetary Fund, Washington, DC.

Boyd, John H., and Mark Gertler. 1994. “The Role of Large Banks in the Recent U.S. Banking Crisis.”

Federal Reserve Bank of Minneapolis Quarterly Review (Winter): 2–21.

Caprio, Gerard, and Maria Soledad Martinez Peria. 2002. “Avoiding Disaster: Policies to Reduce the

Risk of Banking Crises.” In Monetary Policy and Exchange Rate Regimes: Options for the Middle East,

edited by E. Cardoso and A. Galal. Cairo: Egyptian Center for Economic Studies, 193–230.

Cordella, Tito, Giovanni Dell’Ariccia, and Robert Marquez. 2016. “Government Guarantees,

Transparency, and Bank Risk-Taking.” Unpublished.

Dam, Lammertjan, and Michael Koetter. 2012. “Bank Bailouts and Moral Hazard: Evidence from

Germany.” Review of Financial Studies 25 (8): 2343–80.

Damar, Evren, Reint Gropp, and Adi Mordel. 2012. “The Ex-Ante Versus Ex-Post Effect of Public

Guarantees.” Bank of Canada Working Paper 2012-22.

Demirguc-Kunt, Asli, and Enrica Detragiache. 2002. “Does Deposit Insurance Increase Banking System

Stability? An Empirical Investigation.” Journal of Monetary Economics 49: 1373–406.

Demsetz, Rebecca S., Marc R. Saidenberg, and Philip E. Strahan. 1996. “Banks with Something to Lose:

The Disciplinary Role of Franchise Value.” FRBNY Economic Policy Review (October): 1–14.

Demsetz, Rebecca S., and Philip E. Strahan. 1997. “Diversification, Size, and Risk at Bank Holding

Companies.” Journal of Money, Credit, and Banking 29: 300–13.

INTERNATIONAL MONETARY FUND 41

Dobler, Marc, Simon Gray, Diarmuid Murphy, and Bozena Radzewicz-Bak. 2016. “The Lender of Last

Resort Function after the Global Financial Crisis.” IMF Working Paper 16/10, International

Monetary Fund, Washington, DC.

Duchin, Ran, and Deniz Sosyura. 2014. “Safer Ratios, Riskier Portfolios: Banks’ Response to

Government Aid.” Journal of Financial Economics 113: 1–28.

Elliott, Douglas J. 2014. “Implicit Subsidies for Very Large Banks: A Primer.” Economic Studies at

Brookings Institution, July.

Farhi, Emmanuel, and Jean Tirole. 2012. “Collective Moral Hazard, Maturity Mismatch, and Systemic

Bailouts.” American Economic Review 102 (1): 60–93.

Financial Stability Board (FSB). 2016a. “Peer Review of Japan.”

http://www.fsb.org/wp-

content/uploads/Japan-peer-review-report.pdf.

———. 2016b. “Second Thematic Review on Resolution Regimes.” http://www.fsb.org/wp-

content/uploads/Second-peer-review-report-on-resolution-regimes.pdf.

———. 2016c. “Implementation and Effects of the G20 Financial Regulatory Reforms 2nd Annual

Report.”

http://www.fsb.org/wp-content/uploads/Report-on-implementation-and-effects-of-

reforms.pdf.

Fratianni, Michele, and Francesco Marchionne. 2013. “The Fading Stock Market Response to

Announcement of Bank Bailouts.” Journal of Financial Stability 9: 69–89.

Furlong, Frederick T., and Simon H. Kwan. 2006. “Sources of Bank Charter Value.” Unpublished.

Geanakoplos, John. 2010. “The Leverage Cycle.” In NBER Macroeconomics Annual, edited by D.

Acemoglu, K/ Rogoff, and M. Woodford. Chicago: University of Chicago Press.

Geithner, Timothy F. 2014. Stress Test: Reflections on Financial Crises. Crown Publishing Group.

Goddard, John, Phil Molyneux, and John O. S. Wilson. 2004. “The Profitability of European Banks: A

Cross-Sectional and Dynamic Panel Analysis.” The Manchester School 72: 363–81.

Government Accountability Office (GAO). 2014. “Large Bank Holding Companies: Expectations of

Government Support.” GAO report GAO-14-621, Washington, DC.

Gropp, Reint, Christian Gruendl, and Andre Guettler. 2014. “The Impact of Public Guarantees on Bank

Risk-Taking: Evidence from a Natural Experiment.” Review of Finance 18: 457–88.

Gropp, Reint, and Jukka Vesala. 2004. “Deposit Insurance, Moral Hazard, and Market Monitoring.”

Review of Finance 8: 571–602.

Hovakimian, Armen, and Edward J. Kane. 2000. “Effectiveness of Capital Regulation at U.S. Commercial

Banks, 1985 to 1994.” Journal of Finance 55 (1): 451–68.

———, and Luc Laeven. 2003. “How Country and Safety-Net Characteristics Affect Bank Risk-Shifting.”

Journal of Financial Services Research 23 (3): 177–204.

Hughes, Joseph P., and Loretta J. Mester. 2013. “Who Said Large Banks Don’t Experience Scale

Economies? Evidence from a Risk-Return-Driven Cost Function.” Journal of Financial

Intermediation 22: 559–85.

Huser, Anne-Caroline, Grzegorz Halaj, Christoffer Kok, Cristian Perales, and Anton van der Kraaij. 2017.

“The Systemic Implications of Bail-in: A Multi-layered Network Approach.” ECB Working Paper

Series 2010.

International Monetary Fund (IMF). 2013. “Changes in Bank Funding Patterns and Financial Stability

Risks.” Global Financial Stability Report, Chapter 3, October, Washington, DC.

———. 2014a. “How Big Is the Implicit Subsidy for Banks Considered Too Important to Fail?” Global

Financial Stability Report, Chapter 3, April, Washington, DC.

———. 2014b. “Cross-Border Bank Resolution: Recent Developments.” IMF Policy Paper, June,

Washington, DC.

International Monetary Fund, Bank for International Settlements, and Financial Stability Board (IMF-

BIS-FSB). 2009. “Guidance to Assess the Systemic Importance of Financial Institutions, Markets,

TRADE-OFFS IN BANK RESOLUTION

42 INTERNATIONAL MONETARY FUND

and Instruments: Initial Considerations,” Report to the G20 Finance Ministers and Central Bank

Governors, October.

Kroszner, Randall. 2016. “A Review of Bank Funding Cost Differentials.” Journal of Financial Services

Research 29: 151–74.

Laeven, Luc, and Ross Levine. 2009. “Bank Governance, Regulation, and Risk Taking.” Journal of

Financial Economics 93: 259–75.

Laeven, Luc, Lev Ratnovski, and Hui Tong. 2016. “Bank Size, Capital, and Systemic Risk: Some

International Evidence.” Journal of Banking and Finance 69: S25–S34.

Laeven, Luc, and Fabian Valencia. 2010. “Resolution of Banking Crises: The Good, the Bad, and the

Ugly.” IMF Working Paper 10/146, International Monetary Fund, Washington, DC.

Mehran, Hamid, and Anjan V. Thakor. 2011. “Bank Capital and Value in the Cross-Section.” Review of

Financial Studies 24 (4): 1019–67.

Morrison, Alan D., and Lucy White. 2005. “Crises and Capital Requirements in Banking.” American

Economic Review 95 (5): 1548–72.

Philippon, Thomas, and Aude Salord. 2017. “Bail-ins and Bank Resolution in Europe: A Progress

Report.” Geneva Reports on the World Economy Special Report 4.

Sandri, Damiano. 2015. “Dealing with Systemic Sovereign Debt Crises: Fiscal Consolidation, Bail-ins or

Official Transfers?” IMF Working Paper 15/223, International Monetary Fund, Washington, DC.

Santos, Joao. 2014. “Evidence from the Bond Market on Banks’ “Too-Big-to-Fail” Subsidy.” FRBNY

Economic Policy Review (December): 29–39.

Sarin, Natasha, and Lawrence H. Summers. 2016. “Understanding Bank Risk through Market

Measures.” Brookings Papers on Economic Activity (Fall).

Schafer, Alexander, Isabel Schnabel, and Beatrice Weder di Mauro. 2016. “Bail-in Expectations for

European Banks; Actions Speak Louder Than Words.” ESRB Working Paper Series 7.

Schenk, Natalya A., and John H. Thornton. 2016. “Charter Values, Bailouts, and Moral Hazard in

Banking.” Journal of Financial Regulation 49: 172–202.

Schneider, Martin, and Aaron Tornell. 2004. “Balance Sheet Effects, Bailout Guarantees and Financial

Crises.” Review of Economic Studies 71 (3): 883–913.

Schnabel, Isabel. 2009. “The Role of Liquidity and Implicit Guarantees in the German Twin Crisis of

1931.” Journal of International Money and Finance 28: 1–25.

Ueda, Kenichi, and Beatrice Weder di Mauro. 2013. “Quantifying Structural Subsidy Values for

Systemically Important Financial Institutions.” Journal of Banking and Finance 37: 3830–42.

World Bank. 2016. “Bank Resolution and “Bail-in” in the EU: Selected Case Studies Pre and Post BRRD.”

World Bank Group Financial Sector Advisory Center.

Zhou, Jianping, Virginia Rutledge, Wouter Bossu, Marc Dobler, Nadege Jassaud, and Michael Moore.

2012. “From Bail-out to Bail-in: Mandatory Debt Restructuring of Systemic Financial Institutions.”

IMF Staff Discussion Note 12/03, International Monetary Fund, Washington, DC.