Spillover effects of credit default risk in the euro area and the effects on the Euro: A GVAR approach

• Published date: 01 January 2019
• Author: Timo Bettendorf
• Date: 01 January 2019
• DOI: http://doi.org/10.1002/ijfe.1663

Received: 21 March 2018 Revised: 30 August 2018 Accepted: 9 September 2018

DOI: 10.1002/ijfe.1663

RESEARCH ARTICLE

Spillover effects of credit default risk in the euro area and

the effects on the Euro: A GVAR approach

Timo Bettendorf

DG Economics, Deutsche Bundesbank,

Frankfurt, Germany

Correspondence

Timo Bettendorf, DG Economics,

Deutsche Bundesbank,

Wilhelm-Epstein-Strasse 14, 60431

Frankfurt, Germany.

Email: Timo.Bettendorf@bundesbank.de

JEL Classification: C55; F31; H63

Abstract

During the 2008 financial crisis, increasing risk and spillovers became a main

concern for policy makers and banks. In addition, changes in sovereign and bank

risk are believed to have had strong effects on world-wide exchange rates. This

paper aims to analyse these dynamics empirically. We estimate a Global VAR

(GVAR) model for nine EMU countries plus Japan, the United Kingdom, and

the United States and identify structural risk shocks using sign restrictions. Our

results indicate that spillover effects of general risk are much stronger than those

of bailouts. Furthermore, we demonstrate that the Eurodepreciates significantly

against the Yenand U.S. dollar following general risk shocks in the euro area and

only to a small extent following bailout shocks. The PoundSterling is not affected

by any of these shocks. The Eurovariability is, from the EMU perspective, mainly

driven by shocks stemming from large countries (e.g., Germany, France, and

Italy). However, shocks from third countries also playan important role.

KEYWORDS

bailouts, credit default swaps, exchange rates, Global VAR, sign restrictions

1INTRODUCTION

When the 2008 financial crisis began, credit default risk

and contagion became a big concern for banks and policy

makers. Some financial institutions were so distressed that

they became a threat to the domestic and even the global

banking system. Subsequently, some governments bailed

out domestic banks by transferring risk from the bank

balance sheets to the public sector balance sheets. This,

however, led to so-called feedback loops (see Archarya

et al., 2014) or doom loops (see Farhi & Tirole, 2014)

as banks usually hold government bonds in their portfo-

lios, which became more risky due to the risk transfer.

These measures even intensified the crisis in countries

such as Ireland, leading to a sovereign debt crisis, which

also threatened to affect foreign economies. This had

far-reaching consequences as exchange ratesbecame more

affected by sovereign risk (see Della, Sarno, Schmeling, &

Wagner, 2014).

This study investigates the international effects of

increases in general credit default risk and bank bailouts

in the EMU as well as their effects on the Euro. The main

contribution of this research is to quantify the interna-

tional spillover effects of shocks, which have a structural

interpretation and can thus be used for policy analysis,

to credit default risk. More specifically, we identify gen-

eral risk and bailout shocks using sign restrictions in a

Global VAR (GVAR) model and quantify their effects on

domestic and foreign CDS spreads and exchange rates.Our

results demonstrate that spillover effects of general risk

shocks to sovereign and bank CDS spreads are stronger

and more significant than those of bailout shocks. More-

over, we observe heterogeneity in spillovers across differ-

ent regions, as non-EMU countries are significantly less

affected by shocks which originate in the EMU. Exchange

rate effects of risk shocks are strong and depend on the eco-

nomic size of the country in which the shock originates.

296 © 2018 John Wiley & Sons, Ltd. wileyonlinelibrary.com/journal/ijfe IntJ Fin Econ. 2019;24:296–312.

BETTENDORF 297

However, third-country effects are strong as well. Against

the background of the Great Recession, these lessons are

important for policy makers and risk management. This

research contributes to two strands of literature on credit

default risk. First, we contribute to the body of litera-

ture that is concerned with spillover effects and contagion

within the nexus of sovereign and bank credit default risk

by identifying country-specific (structural) general risk

and bailout shocks within a GVARmodel. Second, we con-

tribute to the literature that analyses the effects of credit

default risk on exchange rates, as the model provides us

with evidence on how these structural risk shocks affect

exchange rates.

Spillover effects of credit default risk have been exten-

sively discussed in the literature. They occur not only

between sovereigns and between banks but also between

sovereigns and banks. This can happen via different chan-

nels (see BIS, 2011; Prisker,2001; De Bruyckere, Gerhardt,

Schepens, & Vennet, 2013). Using the canonical model of

contagion by Pesaran and Pick (2007), Metiu (2013) finds

evidence for contagion in EMU bond yields. De Bruyckere

et al. (2013) define contagion as excess correlation and

find empirical evidence for contagion between banks and

sovereigns in 2012. Caceres, Guzzo, and Segoviano (2010)

find that the source of contagion shifted from countries

such as Austria, the Netherlands, and Ireland, which were

severely affected by the financial crisis, to countries with

high short-term refinancing risk and uncertain long-term

fiscal sustainability.Fratzscher and Rieth (2015) show that

there was a strong bank-sovereign nexus during the crisis

and that policies undertaken to reduce risk led to nega-

tive feedback loops as suggested by Acharya et al. (2014).

Studies more closely related to our research are those of

Alter and Beyer (2014); Gray, Gross, Paredes, and Sydow

(2013); and Gross and Kok (2013). Alter and Beyer (2014)

measure credit risk spillovers across sovereign and bank

CDSs between October 2009 and July 2012 using a method-

ology based on Diebold and Yilmaz (2011).1Their results

show increasing interdependencies between banks and

sovereigns and reveal mitigating effects of policy mea-

sures such as the EFSF and LTROs. Gross and Kok (2013)

confirm these findings in a mixed-cross-section GVAR

and show furthermore that spillover potential was high

for banks in 2008 and high for sovereigns in 2012. To

investigate the effects of sovereign and bank risk on the

real economy, Gray et al. (2013) employ a different risk

measure, namely, expected loss, and add a macro sphere

to the mixed-cross-section GVAR. Their results indicate

a strong contraction in real activity following sovereign

risk shocks. These studies rely on generalized impulse

responses, which have the advantage that they are invari-

ant to the ordering of the variables and can easily be

implemented into a framework such as that of Diebold and

Yilmaz (2011). However, they do not allow for a structural

interpretation. Our identification strategy is based on an

economic model and enables us to present an economic

interpretation of the identified shocks.

The empirical literature on linkages between credit

default swaps and exchange rates is scarce. Only Della

et al. (2014) provide an extensive analysis of the relation-

ship between changes in sovereign CDS and exchange

rates. Their findings suggest that an increase in a country's

sovereign credit default risk translates into a deprecia-

tion of the domestic currency. They explainthis behaviour

using a model in which agents ask for a risk premium

for holding a specific currency. As sovereign credit risk is

priced into this risk premium, the value of the currency

falls with an increase in sovereign credit default risk et

vice versa. However,this relationship can also be explained

by the standard monetary model of exchange rate deter-

mination. If, for example, a negative business cycle shock

reduces output, then the exchange rate will depreciate.

At the same time, the negative business cycle shock is

expected to increase the sovereign (and bank) CDS spread

(see Stanga, 2014) as economic conditions worsen.

Other studies focus on FX options and volatility, but

not on currency returns. Hui and Fong (2015) show that

in the long run currency options of the U.S. dollar, Yen,

Swiss Franc, and the Euro are driven by corresponding

CDSs. Carr and Wu (2007) demonstrate, using data for

Mexico and Brazil, that CDS and currency option-implied

volatilities are positively related.

Overall, the research by Stanga (2014) uses a simi-

lar identification strategy and is thus the closest to our

approach. The main difference is that the present paper

uses a GVAR model to quantify international effects,

whereas Stanga (2014) uses single-country VAR models

to analyse effects of idiosyncratic shocks on CDS spreads

and the business cycle. The multi-country setting of our

approach enables us to focus on the international trans-

mission of shocks and the behaviour of exchange rates.

Within this context, the literature often fails to differen-

tiate between contagion and spillover effects. In line with

Alter and Beyer (2014), we follow a definition by Allen and

Gale (2000), who interpret contagion as excess spillover

effects. Spillover effects can thus be seen as a necessary

condition for contagion. It is therefore preferable to focus

on spillover effects of credit default risk rather than con-

tagion, which is difficult to capture without additional

assumptions regarding excess.

The remainder of this paper is structured as follows. In

Section 2, we present the data and methodology that we

used in our analysis. The results of this study are discussed

in Section 3. Finally, Section 4 concludes.