Bank levy and bank risk‐taking

• Published date: 01 September 2017
• Date: 01 September 2017
• Author: Michael Diemer
• DOI: http://doi.org/10.1016/j.rfe.2017.06.001

Review of Financial Economics 34 (2017) 10–32

Contents lists available at ScienceDirect

Review of Financial Economics

journal homepage: www.elsevier.com/locate/rfe

Bank levy and bank risk-taking

Michael Diemer

University of Leipzig, Institute for Theoretical Economics, Grimmaische Straße 12, Leipzig D-04109, Germany

ARTICLE INFO

Article history:

Received 15 February 2015

Received in revised form 6 November 2016

Accepted 4 June 2017

Available online 17 June 2017

JEL classification:

D82

G21

G28

Keywords:

Bank levy

Competition

Moral hazard

Transparency

ABSTRACT

In the aftermath of the recent financial crisis, several countries implemented a bank levy. This paper studies

the impact of different types of bank levies on the risk-taking behaviour of banks competing in the market

for secured or unsecured debt à la Hotelling. We differentiate between three types of bank levies: a levy

on secured liabilities, a levy on unsecured liabilities and a levy on risk-weighted assets. Banks collect funds

and invest in either a prudent or a gambling asset. We find that a levy on secured and unsecured liabilities

can prevent banks from investing in the gambling asset. A levy on risk-weighted assets also induces banks

to behave more prudently. Such a levy is even more effective than a levy on liabilities if banks are well-

capitalized. Finally, a guarantee on debt makes a bank levy more effective.

© 2017 Elsevier Inc. All rights reserved.

1. Introduction

During the recent financial crisis, some major banks were bailed

out in order to avoid a break-down of the financial system. Such bank

bailouts are costly for at least two reasons. Firstly, they increase fis-

cal costs (Honohan & Klingebiel, 2003). Secondly, they may change

the risk-taking incentives of banks. In response to these costs, many

countries, such as Germany, France, Sweden and United Kingdom

had introduced a bank levy which forced banks to participate in

the financing of crisis resolution measures (OECD, 2013). The design

of these charges differed significantly between these countries. In

Sweden and Germany, for instance, the levy was based on banks’

liabilities (and derivatives). In Austria, the levy was based on cap-

ital assets and on derivatives and in France, risk-weighted assets

were levied. Some countries excluded secured liabilities, such as the

Netherlands, Belgium and UK while others included these liabili-

ties (e.g. Finland, Cyprus and Hungary). Moreover, in some countries,

such as Germany, there was a minimum levy which was inde-

pendent of the banks’ financial performance. In the course of the

implementation of a new regulatory framework in the European

E-mail address: diemer@wifa.uni-leipzig.de (M. Diemer).

Union (EU), the rules specifying the bank levy were harmonized. The

annual contributions paid by European banks depend on the size and

the risk profile of these institutions. In general, the European bank

levy is based on the banks’ liabilities excluding regulatory capital and

covered deposits, whereby the risk profile of the institution shall be

taken into account.1

This paper studies and compares the impact of three different

types of bank levies on the risk-taking behaviour of banks: a levy on

secured liabilities, a levy on unsecured liabilities and a levy on risk-

weighted assets. Although, this paper does not explicitly analyse the

effects of the levy designed by the European regulators, we aim at

accounting for the two key ingredients of the European levy: the size

of the bank and its risk profile. While the first one is captured in all

three types of bank levy presented in this paper, the levy on risk-

weighted assets accounts for the risk profile of the bank. We use a

setup à la Hotelling (1929) and apply a similar framework as Repullo

(2004), where banks compete in the debt market. This setup enables

us to endogenize the costs for secured and unsecured debt and to

1See Article 103 of Directive 2014/59/EU of the European Parliament and of the

Council of 15 May 2014 establishing a framework for the recovery and resolution of

credit institutions and investment firms, OJ L 173, 12.6.2014, p. 190–348.

http://dx.doi.org/10.1016/j.rfe.2017.06.001

1058-3300/© 2017 Elsevier Inc. All rights reserved.

M. Diemer / Review of Financial Economics 34 (2017) 10–32 11

examine who effectively pays the levy. Banks compete in the market

for secured or unsecured debt and invest their funds in either a pru-

dent or a gambling asset.2The key question examined in this paper is

whether the different types of bank levies induce banks to invest in

the prudent or in the gambling asset. The revenue-generation motive

is beyond the scope of this paper.

We obtain the following results: A levy on secured liabilities can

prevent banks from investing in the gambling asset if this levy does

not depend on the banks’ financial performance. Otherwise, their

risk decision is not affected by the levy. Both types of levies, i.e.

performance-related and performance-unrelated levy, induce banks

to lower deposit rates in order to compensate for the charge. How-

ever, the size of the compensations differs. If the levy depends on

the financial performance, the compensation is not affected by the

banks’ risk decision. By contrast, if the levy does not depend on the

banks’ financial performance banks risk to pay the charge effectively

as they cannot pass the burden to the depositors in case the projects

are not successful. This risk of being charged effectively is trans-

ferred to the depositors by a lower deposit rate. As the risk of being

charged effectively is higher for gambling banks, they decrease their

deposit rates more than prudent banks which makes gambling banks

less attractive for depositors. A levy on unsecured debt also increases

risk-taking incentives if banks are opaque. Transparent banks always

coordinate towards the prudent equilibrium and a bank levy does

not change risk incentives. However, a bank levy has an impact if

the government guarantees on debt. On the one hand, such a guar-

antee induces banks to gamble (direct effect). On the other hand it

makes the bank levy more effective by increasing the marginal effect

of the levy on banks’ risk-taking (indirect effect). A levy on banks’

risk-weighted assets can also decrease risk-taking. It is even more

effective in inducing prudent behaviour than a levy on liabilities if

banks are well-capitalized.

Thus, the present paper can contribute to the discussion about

the effects of a bank levy and its optimal design. A key contribution

of this paper is to analyse the interaction of different types of bank

levy with other regulatory instruments in the context of risk-taking

behaviour. Different types of bank levy have been discussed in the

literature. Weder di Mauro (2010) and Acharya, Pedersen, Philippon,

and Richardson (2013) propose a systemic risk charge, which only

addresses systemic banks in either form, thereby internalizing the

externality of too-big-to-fail institutions. Acharya et al. (2013) argue

that a tax on systemic risk could be based on the expected loss a bank

may incur in a financial crisis. Shin (2010) suggests a tax on non-core

liabilities which would lower banks’ risk-taking incentives because

non-core liabilities are more expensive than core liabilities. Perotti

and Suarez (2009) propose liquidity risk charges depending on the

maturity of fundings because short-term uninsured liabilities induce

fire sales in a crisis and thus increase financial distress. Gorton and

Huang (2004) suggest a tax on high-value, i.e. projects with a high

payoff, but illiquid projects.

However, risk charges may not only affect banks, but also their

customers and investors. Therefore, we also examine who pays effec-

tively the charge. For example, Albertazzi and Gambacorta (2010)

find that a part of the income tax is shifted to the banks’ customers.

Similarly, Demirgüç-Kunt and Huizinga (1999) provide empirical

evidence that corporate taxes (direct taxes on banks) are passed on

to banks’ customers and thus do not negatively impact banks’ prof-

its. However, the burden shift may not only take place in the case of

direct taxes, such as income taxes, but also if indirect taxes, such as

2The differentiation between secured and unsecured debt is important because

unsecured creditors take the banks’ risk-taking into account and therefore may

influence the impact of a levy on banks’ risk behaviour.

reserves on liabilities, are required (Fama, 1985; James, 1987). Buch,

Hilberg, and Tonzer (2016) provide evidence for the German bank

levy that such a levy has an impact on the deposit rates paid by banks.

Whether charges are effective in reducing the banks’ risk-taking

also depends on how these charges interact with other regula-

tory instruments. In this paper, we consider three instruments:

bank competition, subordinated debt and government guarantee.

The impact of competition on bank risk-taking has been exten-

sively assessed in the recent years (Boyd & De Nicoló, 2005; Hell-

mann, Murdock, & Stiglitz, 2000; Keeley, 1990; Martinez-Miera &

Repullo, 2010). This relationship is ambiguous since competition in

the deposit market has effects on banks’ risk-taking behaviour which

are different from the effects of competition in the loan market. This

ambiguity is also shown in empirical studies (Beck, Demirgüç-Kunt,

& Levine, 2006; Berger, Klapper, & Turk-Ariss, 2009; Jiménez, Lopez,

& Saurina, 2007).

Subordinated debt can also have a disciplinary effect as depos-

itors may penalize excessive risk-taking through requiring higher

deposit rates for riskier banks or through restrictive covenants in

bank contracts (Goyal, 2005; Peria & Schmukler, 2001). However,

debt holders may only discipline banks if they are able to monitor

changes in bank risk-taking and if banks are able to commit to a cer-

tain risk-taking (Blum, 2002; Flannery & Sorescu, 1996). Contingent

debt instruments which convert into equity in case of a trigger event

may also have a disciplinary effect.3Hilscher and Raviv (2014) show

that appropriate choice of contingent capital parameters can elimi-

nate stockholders’ incentives to increase risk. Comparing contingent

capital with debt instruments, they find that the latter one induces

higher risk-taking than properly designed contingent capital instru-

ments. Pennacchi et al. (2014) analyse the impact of a call option

enhanced reverse convertible which convert to new equity shares if

an issuing bank’s market value of total capital to deposits falls below

a pre-determined threshold. In their setup, initial shareholders have

the option to purchase these new shares. They find that contingent

capital is more efficient in reducing moral hazard than coco bonds

and non-convertible subordinated debt. Berg and Kaserer (2015) find

that coco bonds may even increase risk-taking incentives.

However, the disciplining effects may weaken or even disappear

if debt holders are secured by either a deposit insurance or govern-

ment guarantees (Acharya, Anginer, & Warburton, 2016; Anginera,

Demirguc-Kunt, & Zhu, 2014; Berger & Turk-Ariss, 2015; Nier & Bau-

mann, 2006; Sironi, 2003). This may be a key concern of policymak-

ers when implementing and labelling new regulatory instruments

as they may make explicit that a guarantee provided by the gov-

ernment or a central bank exists what currently is at most implicit.

While new resolution tools may weaken market participants’ expec-

tations that banks or creditors are covered by an implicit guarantee,

other regulatory reforms such as the implementation of a bank levy

may boost expectations that credit institutions may receive financial

support in times of distress even though it is not (explicitly) intended

by the regulators. In the presented paper, we also analyse the impact

of a bank levy on banks’ risk-taking in the presence of an implicit or

explicit guarantee on debt and we thereby aim at contributing to the

(political) discussion about the interaction between new regulatory

reforms and (implicit) guarantees (Financial Stability Board, 2013;

Lambert, Ueda, Deb, Gray, & Grippa, 2014; Schich & Aydin, 2014).

The paper proceeds as follows: Section 2 introduces the model

and Section 3 analyses the effects of a levy on secured debt on

3Different types of contingent capital instruments have been discussed which

mainly differ in the trigger event. For example, these trigger events may be the issuer’s

capital ratio or market triggers such as the issuer’s stock price and financial institu-

tions indices (Flannery, 2016; Hilscher & Raviv, 2014; McDonald, 2013; Pennacchi,

Vermaelen, & Wolff,2014; Sundaresan &Wang, 2015), whereby the current regulatory

framework considers contingent capital instruments as debt securities that convert

into equity if equity or regulatory capital falls below a certain threshold (coco bonds).