This item was submitted to Loughborough’s Institutional Repository
(https://dspace.lboro.ac.uk/) by the author and is made available under the
following Creative Commons Licence conditions.
For the full text of this licence, please go to:
http://creativecommons.org/licenses/by-nc-nd/2.5/
The Cyclical Behaviour of European Bank
Capital Buffers
♠
by
Terhi Jokipii
♣
and
Alistair Milne
♦
October 2006
Abstract
We examine the cyclical behaviour of European bank capital buffers using an
unbalanced panel data set comprising eight years (1997-2004) of bank balance sheet
data for commercial, savings and co-operative banks. Controlling for other potential
determinants of bank capital, we find that capital buffers of the banks in the acces-
sion countries (RAM10) have a significant positive relationship with the cycle,
while for banks operative in the EU15 and the EA and the combined EU25 the rela-
tionship is significantly negative. We also find fairly slow speeds of adjustment,
with around two-thirds of the correction towards desired capital buffers taking place
each year. We further distinguish by type and size of bank, finding that capital buff-
ers of commercial and savings banks, and also of a sub-sample of large banks, ex-
hibit negative co-movement. Co-operative banks and smaller banks on the other
hand, tend to exhibit positive cyclical co-movement.
Key words: Bank Capital, Bank Regulation, Business Cycle Fluctuations
JEL classification numbers: G21, G28
♠
We would like to thank Antonello D'Agostino, Ari Hyytinen, Giuliano Iannota, Esa Jokivuolle, Juha Kilponen,
Jukka Vauhkonen and Matti Virén as well as the audiences of the XXVIII Kansantaloustieteen Päivät 2006, The
2006 European Banking Symposium held at Bocconi University, The Journal of Banking and Finance 30th An-
niversary Conference as well as internal Bank of Finland workshops for their very useful comments and sugges-
tions.
♣
Bank of Finland and Stockholm School of Economics. Terhi.Jokipii@hhs.se; Department of Finance, SE-113
83 Stockholm, Sweden. +46-8-728-5124.
♦
Bank of Finland and Cass Business School. A.K.L.Milne@city.ac.uk; Cass Business School London, Faculty
of Finance, London EC2Y 8HB; +44-(0)20-7477-8738.
1
1. Introduction
Much debate surrounding the new Accord (Basel II) on bank capital requirements,
due to come into force in 2007, has centred on its potential ‘pro-cyclicality’. One of
the primary aims of Basel II is to create a closer link between capital requirements
and risks, so it is clear that these requirements will become more dependent on the
business cycle. In a cyclical downturn, when counterparties are more likely to be
downgraded than upgraded, the resultant effect could be a significant increase in the
capital requirements to account for increased counterparty risk. Similarly, during an
economic upturn, the amount of capital required would be reduced. Since raising
capital is costly, especially during an economic recession when profits are decreas-
ing, banks might be forced to reduce their loan portfolio in a recession, so as to meet
rising capital requirements. Thus many have argued that the new Accord will make
it much harder for policy makers to maintain macroeconomic stability.
The growing literature on the potential pro-cyclicality of Basel II has largely fo-
cussed on quantifying the likely range of variation in ‘Pillar 1’ capital requirements
through the business cycle.
1
In practice, well-functioning banks hold capital well in
excess of the minimum requirements, which will reduce the impact of Pillar 1 regu-
latory capital requirements on loan portfolios. Moreover, the supervisory review
powers granted to regulators under Pillar 2, allowing them to demand a buffer of ad-
ditional capital during a business cycle expansion, provide policy makers with a tool
to counter the potential pro-cyclicality effect of the new accord. Al this implies that
the management of bank capital buffers over the course of the business cycle will be
as or even more important than the ‘Pillar 1’ requirements as a determinant of the
cyclical impact of the new capital regulations.
With this policy concern in mind, we investigate the behaviour of bank capital
buffers of European banks, under the old Basel 1988 accord on capital regulation.
By ‘capital buffer’ we mean the amount of capital banks hold in excess of that re-
quired of them by national regulators. The main objective of our paper is to establish
the extent of co-movement between this buffer and the cycle, and to determine
whether such co-movement is country, bank type or bank size specific. We also ana-
lyse the impact of various cost and revenue variables on the behaviour of bank capi-
tal buffers.
Our estimation results reveal substantial differences in the cyclical behaviour of
capital buffers between the various sub-groups. We find that capital buffers of RAM
(10 countries that joined the EU in May 2004) banks appear to move together with
1
Basel II is based on three complementary pillars. Pillar 1 consists of the regulatory calculations of
capital requirements for market, credit, and operational risk. Pillar 2 is the supervisory review pro-
cess, where supervisors assess both the bank’s total capital adequacy for the full range of risks includ-
ing those not covered by Pillar 1 and the bank’s management of capital. Pillar 3 is market discipline.
In order to improve transparency of banks to counterparties and investors, banks will be required to
disclose detailed information on their risk profile and capital adequacy.
2
the business cycle while the buffers of banks operative in the EU25, EU15, Den-
mark, Sweden and the United Kingdom (DK,SE,UK) and EA samples rather exhibit
negative co-movement. The latter finding is broadly in line with most of the individ-
ual country studies that analyse the determinants of excess capital and their relation-
ship to the cycle (see among others Ediz et al., 1998; Rime, 2001; Ayuso et al.,
2004; Bikker and Metzemakers, 2004; Lindqvist, 2004; Stoltz and Wedow, 2005).
Breaking the sample down further by size and type of bank, we find additional
distinctions. Capital buffers of commercial and savings banks, as well as those of
larger banks have a negative relationship with the cycle while those of co-operative
banks and of smaller banks move together with the cycle. With regard to the associ-
ated costs, in almost all cases we find a fairly slow speed of adjustment towards de-
sired capital buffers. These results provide a benchmark from which inferences relat-
ing to the introduction of Basel II and its effect on capital buffer management can be
made. In particular, they shed some light on how capital management decisiosn may
need to be adjusted through Pillar II and III of Basel II in order to offset the potential
cyclical effects of the new accord.
The paper is organized as follows. Section 2 discusses the motivation for hold-
ing excess capital, sets out the hypotheses we test, and describes our data including
the various controls we introduce for the non-cyclical determinants of bank capital.
We can only test hypotheses about the reduced form cyclical behaviour of capital
buffers. Section 3 presents our specification and empirical results together with
some robustness checks. Finally, section 4 concludes.
2. Hypotheses and Data Description
Our data, for the years 1997-2004, reported in Table 1, indicates that banks hold far
more prudential capital than that required by the regulators.
2
Tier 1 capital buffers of
banks within the EU15 vary from 1.87 percent of risk-weighted assets in Portugal to
4.79 percent in Finland with an average across the EU15 of 2.93 percent. Buffers are
also substantial in the accession countries, ranging between 2.64 percent in Cyprus
and 6.99 in Malta. The average buffer for the RAM10 is around 5.14 percent which
is considerably larger than in the EU15.
Several reasons have been put forward to explain why banks hold excess capital
(see amongst other studies Marcus, 1984; Berger et al., 1995; Jackson et. al., 1999;
Milne and Whalley, 2001; Estrella, 2004; Milne, 2004). Banks generally will tend to
assess their risks differently than regulators, for instance using their own internal
economic capital models. Appropriate bank-specific capital levels will therefore be
set according to their own assumptions and risk appetites. Banks may also need to
hold excess capital in order to signal soundness to the market and satisfy the expec-
2
Similarly large capital buffers are also held by US and Asian banks. See for example Peura and Jok-
ivuolle (2004) for a tabulation of US capital buffers.
3
tations of rating agencies (Jackson et. al., 1999). These ‘market disciplines’ may
lead banks to holding more capital required by regulators.
Banks may also hold a buffer of capital as a protection against the violation of
the regulatory minimum requirements (Marcus, 1984; Milne and Whalley, 2001;
Milne, 2004). By holding capital as a buffer, banks insure themselves against costs
arising from a supervisory intervention in response to a violation of the require-
ments.
A further reason for holding a capital buffer is to take advantage of future
'growth opportunities', putting banks in a position to obtain wholesale funds quickly
and at a competitive rate of interest in the event of unexpected profitable investment
opportunities. In the event of a substantial increase in loan demand, banks with rela-
tively little capital may lose market share to those that are well capitalised.
It is difficult to empirically distinguish these different underlying determinants
of bank capital buffers: for example higher portfolio volatility can be expected to
increase capital buffers, whether these are the result of market disciplines or of a de-
sire to avoid supervisory interventions. Our paper has a more limited objective, to
investigate how capital buffers of European banks behave over the business cycle,
and in particular whether capital buffers are higher in business upturns and lower in
business downturns (positive co-movement) or the reverse (negative co-movement),
controlling as far as we can for various bank specific determinants of capital buff-
ers..
We thus test the following null hypothesis:
0
H
Under the Basel I Accord,
business cycle fluctuations do not have an impact on the capital buffers of Euro-
pean banks; against two alternatives:
)(
1
aH
Capital buffers co-move positively with
the business cycle i.e. banks tend to increase capital in business cycle expansions
and reduce capital in recessions; and
)(
1
bH
Capital buffers co-move negatively
with the business cycle i.e. banks tend to reduce capital in business cycle expan-
sions and increase capital in recessions.
These descriptive hypotheses are consistent with a number of different underly-
ing structural models of bank capital dynamics. Estrella (2004) examines the rela-
tionship between optimal forward looking capital buffers and deterministic cycles of
loan losses. He finds that banks, subject to costs of capital adjustment, will build up
capital buffers in anticipation of loan losses. Since loan losses themselves tend to lag
the business cycle, this suggests that actual capital buffers will rise during cyclical
downturns, i.e. negative cyclical co-movement.
It is also argued (see amongst others Rajan, 1994; Borio et al., 2001; Crockett,
2001) that portfolio risks actually increase during an economic upturn. During an
economic boom, lenders provide large amounts of credit while imbalances that will
become responsible for the following recession continue to build up, increasing the
possibility of unusually large losses during a cyclical downturn. Under this interpre-
tation rational forward looking banks may build up capital buffers during cyclical
upturns, i.e. positive co-movement.
