ArticlePDF Available

What determines how much capital is held by UK banks and building societies

Authors:

Abstract and Figures

In this paper we review hypotheses about how decisions on capital are taken and how they affect the ‘buffer’ between actual capital and the regulatory requirement. The hypotheses come from the literature and from discussion with policy makers, supervisors and practitioners. We argue that the amount of capital held by banks and building societies depends on risk management, market discipline and regulatory environment. Using both quantitative and qualitative approaches, we provide evidence on which hypotheses hold in the UK. We find that regulatory requirements affect the amount of capital held by banks and by building societies. (2) (PDF) What determines how much capital is held by UK banks and building societies. Available from: https://www.researchgate.net/publication/248419253_What_determines_how_much_capital_is_held_by_UK_banks_and_building_societies [accessed Oct 07 2022].
Content may be subject to copyright.
F inancial
Services
Authority
Occasional Paper Series 22
July 2004
What determines how
much capital is held by
UK banks and building
societies?
Isaac Alfon
Isabel Argimon
Patricia Bascuñana-Ambrós
FSA OCCASIONAL PAPERS IN FINANCIAL REGULATION
Foreword
The FSA is committed to encouraging debate among academics, practitioners
and policy-makers in all aspects of financial regulation. To facilitate this, it is
publishing a series of occasional papers in financial regulation, extending across
economics and other disciplines.
These papers will cover such topics as the rationale for regulation, the costs and
benefits of various aspects of regulation, and the structure and development of
the financial services industry. Since their main purpose will be to stimulate
interest and debate, we welcome the opportunity to publish controversial and
challenging material and to publish papers that may have been presented or
published elsewhere.
The main criterion for acceptance of papers, which will be independently
refereed, is that they should make substantial contributions to knowledge and
understanding in the area of financial regulation. We will encourage
contributions from external authors as well as from within the FSA. In either
case, the papers will express the views of the author and not necessarily those
of the FSA. Comments on these papers are welcomed and should be addressed to
the series editors.
Authors wishing to contribute to this series should contact Kari Hale or Andrew
Sykes at:
The Financial Services Authority
25 The North Colonnade
Canary Wharf
London
E14 5HS
Telephone:(0)20 7066 1552
e-mail: kari.hale@fsa.gov.uk andrew.sykes@fsa.gov.uk
FSA Occasional Papers are available on the FSA website www.fsa.gov.uk
WHAT DETERMINES HOW
MUCH CAPITAL IS HELD BY
UK B ANKS AND BUILDING
SOCIETIES
?
I
SAAC ALFON
ISABEL AR GIMON
PATRICIA BASCUÑANA-AMBRÓS
FSA Occasional Paper
© July 2004
Biographical note
Isaac Alfon, Isabel Argimon and Patricia Bascuñana-Ambrós are economists in the
Economics of Financial Regulation team within the Finance, Strategy and Risk Division
of the Financial Services Authority.
Abstract
In this paper we review hypotheses about how decisions on capital are taken and how
they affect the ‘buffer’ between actual capital and the regulatory requirement. The
hypotheses come from the literature and from discussion with policy makers,
supervisors and practitioners. We argue that the amount of capital held by banks and
building societies depends on risk management, market discipline and regulatory
environment. Using both quantitative and qualitative approaches, we provide evidence
on which hypotheses hold in the UK. We find that regulatory requirements affect the
amount of capital held by banks and by building societies.
Acknowledgments
We are grateful: to Clive Briault, Malcolm Cook, Jonathan Fischel, Nadege Genetay,
James McGregor, Jesus Saurina, Sarah Smith and Andrew Sykes for their help and
comments at different stages of the project; to the senior staff of the banks and
building societies that took part in the survey that underpins the qualitative analysis;
to the senior staff of rating firms and equity analysts who gave us their time; and to
our colleagues in FSA supervision who helped us with the survey. The paper reflects our
personal views and not the corporate views of the Financial Services Authority.
2
Contents
1. Introduction 5
2. Capital requirements and capital holdings: Overview 7
3. Determinants of capital holdings: Overview 13
4. Determinants of capital holdings: Firms’ internal factors 14
5. Determinants of capital holdings: Market discipline 22
6. Determinants of capital holdings: Regulatory framework 28
7. Summary and conclusions 32
References 34
ANNEX 1: Summary of how changes in regulatory requirements
might affect actual capital 37
ANNEX 2: Summary of firms’ replies to the questionnaire 39
ANNEX3:Quantitative estimation of the determinants of capital ratios48
3
4
1 Introduction
Most UK banks and building societies
1
hold considerably more capital than required by
the regulator. This might lead one to assume that changes in capital requirements do
not affect the amount of capital held by firms, as the changes will be fully absorbed
by this excess or buffer.
However, the amount of capital held by firms is influenced by many different factors,
not just regulatory requirements. The risk preferences of managers, shareholders and
bondholders are also important in determining how much capital is held. In particular,
it could be argued that the 8% requirement under the current Basel Accord is meant
to act as a floor below which there is significant risk to consumers and, in some cases,
to market confidence. Actual capital, on the other hand, should be determined by
shareholders’ preferences, which reflect their desire to be properly compensated in the
form of risk adjusted investment returns and by wholesale depositors influence.
In practice, therefore, management need to weigh up different factors: the level of
capital that the market demands for the risks taken (especially the level demanded by
rating agencies to assign a given rating); the level of capital that shareholders think
is appropriate; the regulatory minimum; and managers’ own views and preferences on
risk management. So regulatory requirements need not be decisive in determining
capital holdings.
This does, however, prompt the questions of what impact capital requirements have,
given that the actual level of capital held by the firm is often significantly above the
regulatory level, and how firms respond to changes in regulatory capital. Clearly, we
cannot observe firms holding less capital than required. We also cannot assume that,
just because firms currently hold significantly more than the regulatory minimum, they
will be unaffected by changes in the minimum. A better understanding of why firms
hold excess capital will help in assessing the likely impact of changing capital requirements.
This paper analyses the factors that may affect the level of capital held by banks and
building societies. We devote special attention to the role played by regulatory capital
requirements. We review various hypotheses that explain firms’ tendency to hold more
5
1A building society is a mutual organisation whose main activity is mortgage lending for house
purchase, financed mainly but not exclusively by taking deposits from retail customers.
What determines how much capital is held by UK banks and building societies?
capital than required. We also present the results of our quantitative and qualitative
analysis to assess which hypotheses hold in the UK. The quantitative analysis uses data
from regulatory capital returns from banks and building societies. The qualitative
analysis was carried out through a questionnaire and interviews with market players.
Section 2 of the paper presents some information about individual capital requirements
set by the FSA and about firms’ desired and actual capital ratios over recent years.
Sections 3 to 6 summarise our main findings about the determinants of the amount
of firms’ capital. In particular, Section 3 outlines our framework for analysing the
determinants of capital holdings by UK banks and building societies. Section 4
analyses the factors that are internal to firms, including their attitude to risk, their
long-term strategy and the existence of adjustment costs. For each hypothesis we
present the findings of our quantitative and qualitative analysis of how the various
factors help determine the amount of capital held.
Section 5 reviews competition, the disciplinary effect of uninsured funding and how
market discipline influences firms’ decisions on capital.
Section 6 discusses how the prudential regulatory framework can affect the amount of
capital that banks and building societies hold. The role of capital requirements and of
supervisory behaviour are analysed here. The section ends with our empirical results
about how firms respond to regulation. Section 7 presents our conclusions.
The paper ends with three annexes. Annex 1 summarises the hypotheses that we
have explored as potential explanations of what determines the amount of capital that
firms hold. For each hypothesis, we show the effect of changing the regulatory
requirements. The charts in Annex 2 show the responses to the questionnaire sent to
firms. And Annex 3 details the econometric results from a model that relates actual
capital to its possible determinants.
6
What determines how much capital is held by UK banks and building societies?
2 Capital requirements and capital holdings: overview
R egulatory capital for UK banks and building societies
The FSA inherited from the Bank of England the practice of setting two separate capital
requirements for each bank: a ‘trigger ratio’, which was the minimum individual capital
ratio; and a ‘target ratio’ set above the trigger. The ‘target ratio’ was to act as a warning
light and as a cushion of capital to help prevent an accidental breach of the individual
capital requirement. For building societies, the Building Societies Commission set
‘threshold ratios’ that corresponded to banks’ trigger ratios.
Following the Financial Services and Markets Act, which came into force in 2001, the
FSA discontinued the practice of setting a target ratio for banks (FSA (2001)).
The decision was accompanied by a review of all UK banks’ individual capital
requirements to make them consistent with the new framework. A similar review took
place for building societies.
Some broad characteristics define the functioning of the UK regulatory system for
banks and building societies during the whole period of analysis. The first one is that
individual capital requirements are set at firm-specific level. The second one is that the
FSA may at any time vary a firm’s requirement. Finally, the FSA has declared that it
“will consider it to be good management practice in the financial services industry for
a UK bank to hold an appropriate capital buffer above the individual capital ratio
advised by the FSA” (FSA (2001)).
For most banks and building societies, individual capital requirements exceed the
Basel minimum of 8%. The “FSA considers that the basic 8% regulatory minimum
capital requirement is only appropriate for a well-diversified firm whose business,
management, systems and controls are strong and where the risks that it is exposed to
are captured adequately by the existing capital model” (FSA(2001)). In fact, the Basel
Committee of Banking Supervision recognised as weaknesses of the current system
7
What determines how much capital is held by UK banks and building societies?
its poor risk sensitivity, its difficulty in covering all risks and its inability in providing
the right incentives for good risk management practices (BCBS (1999)).
2
These views
are shared by many analysts (see, for example, Milne (2001) and Oliver, Wyman &
Company (2001)).
Banks and building societies also accumulate provisions against loan default. These
provisions protect against expected losses and are likely to vary over time.
3
Provisions
are therefore different from capital, which should be a buffer against unexpected
losses. Other differences between capital and provisions are that: general provisions
are classified as Tier 2 capital (i.e. regulatory capital) subject to a maximum; and the
accumulation of capital (i.e. retained earnings) and provisions attract different tax
treatments. Thus decisions about provisions and capital are unlikely to be
independent.
4
In this paper, we explore how capital requirements affect actual capital
and we therefore focus on the difference between actual capital and regulatory capital.
Information sources and method
Our empirical analysis has used both a quantitative and a qualitative approach.
Quantitative evidence on the main determinants of capital comes from data in banks’
regulatory capital returns from September 1998 to September 2002 and in building
societies’ returns from the second quarter of 1997 to the second quarter of 2002.
5
Our
quantitative analysis is derived from the econometric estimation of an equation that
8
What determines how much capital is held by UK banks and building societies?
2 It also acknowledged that it mostly deals with credit and market risk: “While the original Accord
focused mainly on credit risk, it has since been amended to address market risk. Interest rate risk in
the banking book and other risks, such as operational, liquidity, legal and reputational risks, are not
explicitly addressed. Implicitly, however, the present Accord takes account of such risks by setting a
minimum ratio that has an acknowledged buffer to cover unquantified risks” (BCBS (1999)).
3 See Pain (2003) for a survey of the provisioning experience of UK banks and a review of factors that
might help explain increases in provisions.
4For example, Laeven and Majnoni (2002) explore the relationship between capital and provisions and
find that banks tend to delay provisioning for bad loans, thereby possibly magnifying the economic
cycle’s impact on capital.
5 The frequency of reporting is quarterly for solo banks, but different banks have different reporting
dates within the quarter. (We use the term “solo banks” as shorthand for “banks that submit
unconsolidated returns”.) For the building societies the dataset is a balanced panel (i.e. it contains
the same firms throughout the period), so demutualisation does not affect the results. Initial work
carried out with data from consolidated returns seems to show that banking groups respond differently
from solo entities and building societies to some of the determinants of capital holdings.
expresses firms’ capital ratios in terms of various possible determinants. Details of the
dataset, method and results are in Annex 3.
We complement the quantitative analysis by qualitative analysis that uses the results
of a small survey, carried out between June and November 2003 as a face to face
discussion with eight banks and five building societies
6
and with equity analysts and
rating firms. The survey was based on a short questionnaire
7
, the responses to which
are summarised graphically in Annex 2.
What the regulatory returns show
Unconsolidated (solo) entities, even within a bank group, are set individual capital
requirements and submit quarterly returns.
Table 1 shows that, if weighted by total assets, the average capital ratios for banks are
nearly 50% above the average individual capital requirements set by the FSA. This rises
to over 85% for banks without trading book activity. On the other hand, building
societies hold an average of just over 31% more than individually required in weighted
terms. Building societies’ average weighted buffer (i.e. capital minus individual capital
required, as a percentage of individual capital required) is below that for banks,
even for banks of similar size (not shown in the table). The weighted averages for
actual capital ratios and percentage buffers are much lower than the unweighted
averages, indicating that larger firms tend to have lower capital ratios and lower
percentage buffers.
9
6FSA supervisors suggested a sample of firms that would provide a reasonable coverage of the market.
Those firms that agreed to take part in the survey were sent the questionnaire. They were 13 large and
medium-sized firms (8 banks and 5 building societies).
7 The questionnaire asked firms various questions about how they decide their capital. We did not try to
verify that firms’ actions were consistent with their answers. Within a firm there may be different views
about capital. Moreover, sometimes we spoke to people with a ‘risk’ background and sometimes to
people with a ‘finance’ background. We have not controlled for any differences in views within
each firm.
What determines how much capital is held by UK banks and building societies?
10
What determines how much capital is held by UK banks and building societies?
TABLE 1. Summary of data for banks (1998-2002) and building societies (1997-2002)
Average
capital
requirements
% of risk
weighted
assets
Average
actual
capital
% of risk
weighted
assets
Average
buffer
% of
capital
requirement
Proportion
of assets in
trading
book
% of risk
weighted
assets
Proportion
of tier 1
capital
% of total
adjusted
capital
Average
size
£ billion
of
assets
Number
of
firms
Number of
observations
Unweighted 12.78 41.45 234.52 7.55 87.28
Weighted by total
assets
9.42 14.16 48.06 9.83 84.78
All banks
Weighted by risk-
weigthed assets 9.42 12.92 35.03 8.61 95.14
14.54 187 2744
Unweighted 10.97 30.24 177.67 22.80 83.15
Weighted by total
assets
9.20 12.71 36.38 12.79 84.21
Banks with
trading book
activity
Weighted by risk-
weigthed assets
9.24 12.49 33.82 11.23 97.04
33.74 59 909
Unweighted 13.68 47.0 262.68 0 89.61
Weighted by total
assets
10.15 18.99 86.95 0 86.79
Banks without
trading book
activity
Weighted by risk-
weigthed assets
10.01 14.33 39.03 0 88.88
5.02 128 1835
Unweighted 9.65 15.16 56.92 n/a 95.45
Weighted by total
assets
9.45 12.40 31.13 n/a 90.32
Building
societies
Weighted by risk-
weigthed assets
9.43 12.33 30.76 n/a 90.56
2.23 65 1365
In the period considered, the FSA has slightly reduced the average individual ratio for
banks. The first signs of this came at the end of 2000 following the review of individual
capital requirements that was undertaken before the implementation of the Financial
Services and Markets Act so as to make the requirements consistent across different
firms.
8
The differences in capital ratios amongst banks have reduced in the same
period. The pattern for building societies seems to be the contrary, as the dispersion
has increased, especially since 2000.
As Figure 1 shows, over the period banks have increased their buffer (calculated as the
percentage of excess capital over individual capital requirements). Building societies
maintained theirs at a steady level. Because of the short length of the period analysed,
it is difficult to assess how different factors have influenced such developments, but
the economic cycle could be one of the contributory factors.
Actual capital ratios and individually required capital ratios vary more between firms
than over time (see Annex 3). The dispersion is greater for actual capital ratios than
for required ratios, suggesting that the market differentiates more between firms than
does the regulator. Dispersion in capital ratios is much lower for building societies,
suggesting that they are a more homogeneous group of firms.
Predominantly it is large banks that report trading book activity and most of them have
a non-UK parent. For those UK-owned banks that carry out trading, on average, this
activity represents less of their business than for foreign-owned banks. On average,
trading book activity accounts for a third of risk-weighted assets in firms with a foreign
parent, compared with 16% in UK-owned banks.
11
What determines how much capital is held by UK banks and building societies?
8 The coefficient of variation (CV= standard error/mean) for the capital requirements for banks in 1998
was 0.59. It dropped to 0.42 at the end of 2000 and 0.27 in 2002. The CV for building societies has
experienced the reverse pattern: from 0.04 in 1997 to 0.08 in 2002.
Figure 1. Unweighted average buffer over
required capital (%)
0
50
100
150
200
250
300
1997 1998 1999 2000 2001 2002
Banks
Building Societies
We find that the size of the financial institution, measured by total assets, seems to
have a large influence on firms’ actual capital ratio. In particular, small banks choose
to have higher capital ratios than larger banks. Building societies seem to also show a
negative relationship between size and capital held and required. The difference in
actual capital ratios between large and small banks is much larger than the difference
in individually required capital ratios between these same firms. So the size of the
percentage buffer seems to be inversely related to the size of banks and building
societies, with large firms having a smaller percentage buffer than small firms.
Information from interviews
From our interviews with banks and building societies we concluded that firms
differentiate between their desired level of capital and their actual level of capital. As
shown in Figure 2, as at the end of 2002 most of the interviewed firms had more capital
than their desired level, which in turn was more than the individual regulatory
requirement.
9
In particular, the interviewed firms held on average 18% more capital than
their declared “desired level”
10
and 40% more than their individually required level.
All firms define their capital needs as a ratio in relation to the regulatory concept of
risk weighted assets.
11
However, while all the building societies in the sample focus on
12
What determines how much capital is held by UK banks and building societies?
9Figure 2 is based on firms’ capital as at the end of 2002 or, if that was unavailable, as at the nearest
subsequent date. Two banks responded that their actual capital tended to vary around their desired
capital. This might be related to the mergers that both banks had recently experienced.
10 However, the relationship between actual and desired capital may fluctuate over the cycle.
11 This raises some issues about how firms’ decisions on capital will be affected by the European
application of the Basel Accord.
Figure 2. Average capital of interviewed firms
0
2
4
6
8
10
12
14
16
All Banks Building
societies
% of risk weighted assets
Average
required
capital
Average
desired
capital
Average
actual
capital
the capital definition used for regulatory purposes, most banks define their capital
needs in relation to different forms of Tier 1.
Our discussions with firms showed that several factors may affect the type of capital
on which they choose to base their capital decisions:
•Equity markets focus on Tier 1 capital.
Because Tier 1 and Tier 2 capital have different costs, desired capital cannot
be specified simply in terms of total capital.
•Regulatory requirements specify certain relationships between different types
of capital. For example, the amount of Tier 2 subordinated term debt should
not exceed 50% of the amount of Tier 1 capital.
There is perceived to be a more limited supply of Tier 1 capital than Tier 2 capital.
3. DETERMINANTS OF CAPITAL HOLDINGS: OVERVIEW
After having reviewed the main features of capital ratios for UK banks and building
societies, we now consider some explanations as to what determines these holdings.
Currently there is little literature focusing on this issue, so we have mostly extracted
the relevant implications from other lines of research. Most theoretical models assume
that firms would always operate with exactly the regulatory minimum capital.
12
Our starting point is Richardson and Stephenson (2000), who briefly discuss some
possible explanations as to why banks might hold more capital than the regulator
requires. They put forward some fairly high level suggestions that we now explore further.
We group the possible explanations for holding excess capital into three categories:
explanations that focus on firms’ internal factors; explanations based on the effect
of market discipline; and explanations related to ways in which the regulator may
affect the actual level of capital. There is inevitably an overlap between these
13
What determines how much capital is held by UK banks and building societies?
12 See, for example, the literature reviews by Jackson et al. (1999) and Furfine (2000) on US banks’
response to Basel I. The alternative is to regard capital requirements as an incentive mechanism. See
Milne (2000) for a theoretical framework.
categories – for example markets may influence firms’ management – and there may
be additional explanations.
For each of these three types of explanation we summarise the relevant findings from
our qualitative and quantitative approaches and consider whether they support or
rebut the applicability of these particular explanations to the UK. The three types of
explanation are dealt with in Sections 4, 5 and 6.
4. DETERMINANTS OF CAPITAL HOLDINGS: FIRMS’ INTERNAL FACTORS
This section reviews the explanations for capital decisions that are directly linked to
management’s behaviour. They include the management’s attitude towards risk, the
approaches to risk management, the firm’s business strategy, the opportunity cost of
capital and the existence of adjustment costs. The applicability of these explanations
is affected by the market in which the firm operates, but for the sake of simplicity we
try to analyse them in isolation.
Risk assessment in individual firms
The level of capital set by the regulator might not take full account of the firm’s risks. As
stated by the Basel Committee on Banking Supervision (1999) to justify the proposal for
the new Accord, “a bank’s capital ratio, calculated using the current Accord, may not
always be a good indicator of its financial condition”. In particular it pointed at the
current risk weighting of assets resulting in a crude measure of economic risk, as there
is no adequate differentiation between borrowers’ differing default risks.
It could also be that when firms decide on capital they take into account risks that the
regulator is not concerned with, such as financial distress caused by loss of franchise
value (i.e. the present value of the bank’s ability to earn above-market rates of return
because of, for example, the strength of its brand). For example, firms with a high
franchise value may be more likely to hold capital in excess of the regulatory minimum,
to limit the negative consequences of their exposure to high-risk borrowers. Demsetz et
al. (1996) explore the relationship between franchise value and risk, analysing the 1986-
94 period for 100 US banks with publicly traded stock. The paper finds that banks with
14
What determines how much capital is held by UK banks and building societies?
higher franchise value tend to hold more capital and have less risky assets than banks
with lower franchise value. Although the proportions of risky loans held by banks do not
vary hugely, banks with high franchise value have better-diversified loan portfolios.
Our evidence on the relevance of risk assessment for capital decisions seems to support
the view that risk is a determinant of the capital ratio held by firms. In particular, three
of the largest four banks in our sample said that they form their views about desired
capital by assessing the risks rather than by deciding on a margin over regulatory
requirements. One of the risks they consider is that of breaching the regulatory
requirement. The following paragraphs provide some details on the evidence gathered.
In our quantitative approach, to test how risk affects capital ratios we have included in our
equation a variable representing the risk profile of each firm. The different weights that
assets receive to determine the amount of capital that needs to be held against the risk
that they represent try to ensure that riskier portfolios are accompanied by higher capital.
We might expect that well-managed firms hold more capital than required in relation to
their risks. If so, the higher the proportion of high-risk assets in their portfolio the higher
they would weigh them in order to determine the amount of capital to hold. However,
theoretical models also show that an insured bank can exploit a risk-independent
compensation scheme by increasing, for example, its gearing. If this effect is important,
we would expect to find a negative relationship between risk and capital ratios.
In our model, we use an ex-post measure of risk as a proxy for risk appetite: the
proportion of the total firm’s assets that attracts a weight of 100%. This is a crude
measure of risk that is also used in other papers such as Ayuso et al. (2004). We find
a negative (and statistically significant) relationship between capital ratios and our
measure of ex-post risk. In other words, the higher the risk appetite of a firm, the less
capital it holds.
13
This could be interpreted as evidence that firms assess risk
differently from the regulator, maybe only because of the variety of assets in the
category of 100% risk weighted assets.
14
15
What determines how much capital is held by UK banks and building societies?
13 We have also assessed the role of risk using other proxies, such as the ratio of highest risk assets to
risk weighted assets or including the proportion of assets weighted at 50% or the proportion of risk
weighted assets to total assets. In every case the coefficients are negative and statistically significant.
A similar result is obtained in other papers – see Annex 3.
14 It could be argued, however, that riskier firms also have stricter systems and controls allowing them
to operate prudently with less capital for a given risk. However, firms’ responses seem to indicate that
they do not fully trade off one with the other. The negative relationship could also support the
hypothesis that there is moral hazard in firms’ behaviour. However, the results relating to the role
played by insured deposits and commented below suggest that this is not the case
In our discussions some firms considered that regulatory capital overestimates the
risks run by firms on specific assets. But no firm claimed that their decisions on
desired capital were affected by risks different from those considered by the regulator.
Risk management in individual firms
High levels of capital may benefit managers because they can manage all types of risks
less actively, with increased job security.
15
In large institutions, with widely dispersed
ownership, there may be less incentive for individual shareholders to monitor the
behaviour of management and to ensure good risk management practices. There may
also be no effective mechanisms for individual shareholders to express their views. If
so, there is more scope for managers to seek a ‘quiet life’. This agency problem could
manifest itself in weak risk management practices and high levels of capital.
16
However, the market for corporate control, especially the threat of a take-over, may
mitigate this agency problem and may affect the trade-off between capital and active
risk management. Listed banks, because of their greater susceptibility to unwelcome
take-overs, may therefore be under greater pressure than their unlisted counterparts
to use and hold the most efficient level of capital.
The agency problem is likely to be greater when the market is highly concentrated, as
management can then charge sufficient for its products to remunerate all the capital
adequately.
17
For example, Berger and Hannan (1998) shows that high levels of market
concentration allow banks to charge prices in excess of competitive levels. The paper
finds strong evidence that banks in more concentrated markets are less cost efficient,
which includes choosing to hold a larger amount of capital than an efficient allocation
would suggest.
Of the 13 firms in our qualitative sample, two regarded risk management as very
important and seven as important in deciding how much capital to hold. They
considered that capital could be an imperfect substitute for good risk management. So
firms’ views seem to indicate that the negative relationship between risk profile and
16
What determines how much capital is held by UK banks and building societies?
15 Markets could also require firms to hold high levels of capital.
16 See BCBS (2001) for a cross sector comparison of risk management practices.
17 This is also relevant to deposit taking, where charging a high “price” means having a large margin
between deposit rates and lending rates.
capital is because firms with higher levels of risk tend to have stricter systems and
controls and therefore need less capital. However, management also considered that
markets reward firms for being well capitalised even when they also have good risk
management. As a result, firms may use capital to complement risk management and
internal systems and controls and thus hold too much capital. In other words, there is
only a limited trade-off between capital and active risk management.
18
Business strategy and inertia
Senior management may decide to hold capital to enable them to exploit future
business opportunities, such as mergers and acquisitions (M&A).
19
The market for
corporate control will put downward pressure on the amount of capital a listed firm
holds. A firm can become a take-over target if, for example, senior management
strategy is not credible or if the market concludes that the bank has been holding an
excessive amount of capital for too long.
Our qualitative analysis found that excess capital may arise from the firm’s need to
finance its long term strategy. Firms regarded this as the second most important
determinant of desired capital, after the desire to avoid the consequences of breaching
regulatory requirements.
20
The importance arises from firms’ perception that the
market prefers any extra capital needed for growth to be financed from retained
earnings.
21
Other reasons for the importance of long-term strategy when deciding
capital may be the desire to maintain a degree of operational flexibility and the extent
to which the firm wishes to pre-fund future acquisitions. Interestingly, although banks
17
What determines how much capital is held by UK banks and building societies?
18 The role of systems and controls in prudential regulation is partly about ensuring that management
makes adequate provision for expected losses that are not covered by capital requirements. For
example, McDonnell (2003) suggests that inadequate underwriting and reserving were the main causes
of the failures and near misses identified by a group of EU insurance regulators.
19 The causality link may go in the inverse direction.
20 The one firm for which financing its long-term strategy was not an important factor had a deliberate
policy of not accumulating capital to finance future acquisitions.
21 The evidence found in the quantitative analysis of a negative relationship between capital ratio and
size can support the hypothesis that small firms need excess capital to finance their long-term
strategies. It could also support the hypothesis that small firms have larger adjustment costs and thus
choose to hold more capital. The finding would also be consistent with the existence of diversification
benefits that can arise with a large portfolio. Our quantitative analysis cannot distinguish between the
contributions from these different hypotheses.
have access to the equity market to raise capital their views are very similar to those
of building societies, which have limited access.
High levels of capital may simply be a historical legacy where firms have not had
opportunities that they could profitably exploit. Maximisation of shareholder value
might dictate that the bank gives back surplus capital to its shareholders.
22
But
managers may consider that they are employed by the shareholders to invest their
capital, not return it to them, so that the instances when the return of capital takes
place are not very frequent. Returning capital may also have negative market signalling
implications. So under these circumstances the firm will have some undesired excess
capital. Building societies, as mutuals, may seek to return excess capital directly to
members, or benefit them through more favourable pricing or better service.
In our qualitative analysis we explored the relevance of ‘inertia’ as a possible
determinant of the difference between actual and desired capital.
23
We found that four
out of eight banks rated historical legacy as either not important or not relevant. It
was nevertheless important for two banks. Two of the five building societies in the
sample thought that historical legacy was an important determinant of their excess
capital. The responses seem to support the view that banks have stronger incentives
not to hold a large amount of excess capital over a protracted period of time. This is
possibly because banks have to remunerate capital at the going market rate and if they
carry a large amount of excess capital that they cannot remunerate adequately they
will become a takeover target. Building societies differ in that they are less exposed to
becoming a take-over target, although they might certainly be.
24
In our quantitative analysis, the estimated equations for the different types of firm all
imply the existence of some inertia. This is indicated by the statistical significance of
the coefficient of the lagged dependent variable. For building societies the coefficient
is marginally larger than for banks, so the quantitative results are consistent with
those from the qualitative analysis.
18
What determines how much capital is held by UK banks and building societies?
22 “Surplus capital” does not mean all capital above the regulatory minimum, but above what the
management of the bank thinks it really needs.
23 Our evidence supporting the relevance of inertia could equally support the agency problem hypothesis.
24 A building society can be taken over by another building society subject to the FSA’s approval – see
http://www.fsa.gov.uk/pubs/press/2003/119.html for a recent example. It can only be taken over by
a firm that is not a building society after it decides to relinquish its mutual status.
Opportunity cost of capital
If firms optimise their behaviour they should balance the overall gain of holding excess
capital against the cost of holding it. The opportunity cost of capital should discourage
banks and building societies from holding too much capital. Empirical literature
traditionally uses return on equity (in particular, profit as a percentage of reserves or
of capital) as a proxy for the opportunity cost of capital.
Our quantitative results for building societies point weakly at a negative relationship
25
between return on equity and capital if no account is taken of adjustment costs.
26
However, when we allow for adjustment costs, by including the lagged value of capital
as an additional determinant of capital, the coefficient of the return on equity (which
is our proxy for the opportunity cost of capital) becomes positive and statistically
significant. A possible explanation of this result is that profits are the main source of
capital for building societies,
27
so that the proxy variable for cost used in the
estimation is not appropriate.
For the banks that provide in their reporting returns data on their profit and loss
account, the return on equity variable shows a positive coefficient that is not
statistically significant. The role of the cost of capital and the need to find a
meaningful proxy for it need to be explored further.
Adjustment costs and the economic cycle
It could be costly for banks to adjust their level of capital smoothly in response to
unexpected changes in market conditions. Adjustment costs could arise either when
the bank needs to issue new equity or when it needs to repurchase existing equity.
28
19
What determines how much capital is held by UK banks and building societies?
25 The estimated coefficient is negative and statistically significant at the 20% level of confidence. The
results obtained using a fixed effect estimator are not shown.
26 Adjustment costs are the firm’s costs associated with either raising extra capital or reducing its capital.
27 They can also raise Tier 2 capital.
28 The costs of issuing equity are likely to be higher than the costs of repurchasing equity and therefore
asymmetrical.
In particular, there is inevitably a time lag between the decision to raise capital and the
capital being available in the business, as – amongst other factors – the raising of
capital needs legal and regulatory work. By the same token, the disposal of capital also
requires time and gives rise to procedural costs. Moreover, the bank will incur
transaction costs, for instance fees to investment banks and lawyers. Finally, there are
indirect costs arising from information asymmetries between management and
investors. For example, the investor may see issuing (or repurchasing) equity as a signal
that the firm considers market prices to be above (or below) the intrinsic value of the
firm. If so, the share price may go down when a new issue is announced, thereby
increasing the cost of the adjustment. For example, Cornett and Tehranian (1994)
reports statistically significant negative share price reactions to announcements of
equity issues in the banking industry. Increasing equity may be less expensive if the
shares are overvalued.
Our empirical findings support the hypothesis that accessing the desired amount of
capital gives rise to adjustment costs and that these costs could be a determinant of
the observed capital buffers.
29
Nine of the thirteen firms in our sample considered that the cost of raising extra
capital is the main reason for holding a buffer over desired capital. Banks pointed out
that adjustment costs included the indirect costs arising from movement in share price
when capital is issued. Building societies, especially small ones, also considered their
costs of raising extra capital at short notice to be extremely high, though it could be
because some of their costs are independent of the amount raised.
Five of the eight banks in our sample rated adjustment costs due to unexpected
developments within the firm as the second most important determinant of the
difference between desired and actual capital. Four of the five building societies rated
this as not important. Their different response could be partly because banks tend to
be engaged in a wider range of activities than building societies.
20
What determines how much capital is held by UK banks and building societies?
29 The adjustment costs of reducing capital are likely to be higher for banks than for building societies.
This is because building societies, being mutual organisations, can reduce their capital by narrowing
the interest margin between their lending rate and deposit rate. Banks, on the other hand, may find
it difficult to adopt a business strategy that reduces operating profits. Thus, in practice, banks can
reduce their capital only by increasing their dividend or repurchasing their stock – either of which
sends a message to the market, with the associated costs discussed in the main text. This could be one
reason why banks have, on average, larger buffers than building societies.
Our econometric analysis produced results consistent with the qualitative analysis of
adjustment costs. However, the econometric analysis investigated the existence of
adjustment costs by introducing the lagged capital ratio into the equation. As
explained in the section on inertia, it is difficult to distinguish between the inertia
hypothesis and the adjustment costs hypothesis, especially when return on equity does
not seem to be a good proxy of the opportunity cost of capital.
The economic cycle may play a role in the determination of capital holdings, through
its effect on the decisions of both current and past management. Firms may wish to
hold extra capital to reduce adjustment costs, especially the cost of going to the
capital markets when conditions for raising extra capital are less than ideal. The
likelihood of needing extra capital in a downturn is much higher than in an upturn, so
making it more costly for firms to raise the required amount. Firms might hold extra
capital to guard against such an eventuality.
In a downturn, when risks are more likely to materialise, capital decreases because of
write-offs and increases in specific provisions. Moreover, the default probabilities of
loans and the value of collateral could be highly correlated, in which case many market
participants might want to sell at the same time when the downturn hits the market.
A macroeconomic downturn will lead to a decline in ratings and hence to additional
demands on capital for those banks using ratings to assess their loans’ risk and to
decide on their capital. In an upturn, risk reduces and firms can safely hold less capital
than in a downturn. The related empirical evidence is scant, although Ayuso et al.
(2004) find a significant negative relationship between business cycles and capital
buffers for Spanish banks.
Our capital ratio equation includes a GDP growth variable. The estimated coefficient
for the variable is negative and statistically significant for banks. However, the data
does not cover a full economic cycle as the period analysed is too short.
30
The
statistically non-significant negative coefficient for building societies could be due to
their larger weight of household-based retail business, which may be less affected by
the economic cycle. Therefore our quantitative evidence could support the existence of
a negative relationship in the UK between the economic cycle and bank capital.
21
What determines how much capital is held by UK banks and building societies?
30 Moreover, half the banks have a foreign parent, which may be linked to a more internationally
diversified portfolio which could isolate them from changes in UK GDP.
One could then expect that, during a downturn, banks would try to hold larger capital
ratios than during an upturn. Our qualitative analysis supports this, as eleven out of
thirteen firms regard maintaining capital as a cushion against the effect of an
economic downturn either important or very important.
5. DETERMINANTS OF CAPITAL HOLDINGS: MARKET DISCIPLINE
This section reviews the role that market discipline can play in the determination of
capital holdings. The main focus is on the relevance of uninsured funding as a
discipline mechanism and the role of information disclosure. Uninsured funding is
much more significant for banks than building societies.
Banks’ shareholders, depositors, and creditors face risks that increase with the risks
borne by banks and may demand higher interest rates or withdraw their funds if the bank
assumes more risk (Berger (1991)). In anticipation of this, the bank may therefore
choose to hold high levels of capital. In the absence of regulation, market discipline
could cause banks to hold even more capital than that set by the regulator. Market
discipline may be weakened if stakeholders perceive themselves to be insulated from
losses, either because of compensation arrangements or because they expect a
government bail-out. Competition policy and market structure in the banking sector may
also condition the effect of market discipline on the determination of capital holdings.
31
Uninsured funding
Bondholders and investors in banks’ liabilities, such as subordinated debt or uninsured
deposits may feel that the probability of default resulting from holding an amount of
capital very close to the regulatory minimum is insufficient to cover the risks they
face. They will therefore seek a higher rate of interest. In order to reduce the costs of
22
What determines how much capital is held by UK banks and building societies?
31 See Carletti and Hartmann (2002) for a review of the empirical and theoretical literature on the
relationship between competition and stability in banking.
this form of financing, banks may choose to hold more capital.
32
This is supported by
Nier and Baumann (2002), who find that a higher share of uninsured funding
33
has a
disciplining effect, leading banks to choose a larger capital buffer for a given risk.
The discipline exerted by uninsured investors and depositors may be stronger if the
bank is rated. Rating agencies act as intermediaries in the disclosure process. They
gain access to information that it is not publicly available to investors and feed this
information into the rating. Therefore, in effect, investors have more information
about a rated bank than about a non-rated bank. Nier and Bauman (2002) find, for
banks from a variety of countries, that rated banks have capital ratios 0.5 percentage
points higher than banks without a rating. Overall they find evidence in favour of the
idea that banks – rated or not – which disclose more information, limit their
probability of default by choosing higher capital ratios. They find, however, that all of
these effects are weaker when looking at the sub-sample of banks for which the market
believes the government would bail out.
Banks may also hold high levels of capital to get a rating that gives them access to
specific capital markets. This is likely to be more relevant to OTC markets, where the
absence of a centralised market means that there is a material amount of counterparty
risk. Jackson et al. (2002) examines the way in which banks’ solvency standards (as
shown by their ratings) appear to influence their access to the swap market. It finds
evidence that banks rated below ‘A’ have very much smaller than average swap
liabilities. The paper interprets this as indicating that banks with a rating below this
level have much reduced access to the market. A bank moving from a rating of ‘A’ to
‘BBB’ would have to change its strategy to limit its trading in swaps, because of costs
and access problems. So if market practice is that counterparties are only accepted if
they have a rating above some agreed level, banks might build and maintain their
capital ratios to achieve better ratings, enabling them to trade in these instruments
and also reduce their funding costs. The greater use of collateralisation in the swap
market over recent years makes it easier for lower-rated counterparties to have some
access to the market, but they might need to limit their use of swaps because of cost.
23
What determines how much capital is held by UK banks and building societies?
32 This is not the only option. Banks will accept the imposition of covenants (e.g. restrictions on issuing
new debt, restrictions on dividend payments, etc.) designed to protect bondholders from various
actions that can diminish their security. However, no set of covenants can eliminate all risks, so a case
for higher capital remains.
33 They use certificates of deposit as a measure of uninsured funding. The rationale is that the lending
bank is likely to be subject to the same kinds of shocks to risk and profitability as the borrowing bank.
As a result, certificates of deposit are likely to be sensitive to the risk that the borrowing bank is taking.
We might therefore expect a dependency between capital levels and ratings. The
evidence in support of the hypothesis that banks raise their capital ratios to obtain a
particular rating is inconclusive. The illustrative evidence in Matten (1998) suggests
that high capital ratios do not guarantee a good credit rating but that low capital
ratios seem to be associated with a low credit rating.
Overall, the evidence summarised above seems to support the hypothesis that uninsured
funding is an effective mechanism of market discipline and that part of the excess capital
held by banks arises from this. However, the impact of market discipline would show not
necessarily in capital ratios but through its potential to curb the incentive that banks
may have to take excessive risk, by making risk-taking more costly for banks.
Our quantitative analysis tested the strength of uninsured funding as a market
discipline mechanism by including in the model a variable calculated as the proportion
of (partially) insured deposits (non-interbank deposits) over total deposits.
34
The
rationale is that insured funding, even if payments from the compensation scheme are
capped, should weaken market discipline. We find that the higher a bank’s proportion
of partially insured deposits, the more capital it holds for a given level of risk. Therefore
it does not seem that weaker market discipline results in lower levels of capital.
The qualitative analysis found that attracting and maintaining uninsured funding –
wholesale deposits or access to money markets or both – was regarded as an important
determinant of desired capital, not only by banks but also by building societies. Of the
13 firms in the sample, five regarded this as very important and seven as important.
They said that in the short term relative costs will dominate the decisions on how firms
can fund themselves. But in the medium to long term the objective is to develop and
maintain a range of funding options that will support the firm’s strategy. It is in this
context that a rating and peer comparisons become important.
Information and peer pressure
The regulator might feel confident about the level of capital it sets for a bank on the
basis of inspections and private information. The market does not have access to such
information and, given that in practical terms there are too many limitations of
24
What determines how much capital is held by UK banks and building societies?
34 Commercial deposits, which are not covered by the Compensation scheme, are also included, thus
affecting the interpretation of the results.
accounting, auditing and disclosure requirements (Oliver Wyman (2001)) – at least for
financial conglomerates – it may force banks to hold a different amount of capital than
that set by the regulator.
The quality and quantity of disclosure depends on where the bank is listed, as different
regulators set different disclosure and accounting rules. Empirical research largely
supports the view that financial statements complying with US GAAP are more useful
than those in alternative disclosure regimes (Nier and Bauman (2002)).
35
In addition,
a US listing may also entail more disclosure than required under the bank’s national
accounting rules. So whether a bank is listed in the US might affect the amount of
capital held by the firm.
As a result of incomplete information, the views of the market and rating agencies
about the bank may be affected by where its capital ratio stands in relation to others
– i.e. by peer group pressure. The peer group may include banks from other countries
and banks with similar capital requirements, if that is known. If so, changes in
regulatory capital may affect not only the bank whose capital requirement has changed
but also those within its peer group.
In conditions of imperfect information, banks may hold high levels of capital as a
signalling mechanism to differentiate themselves from their peers. High capital ratios
can be regarded as a substitute for information that firms cannot easily or convincingly
disclose directly, e.g. adequate systems and controls, or good earnings prospects. For
example, small or privately owned UK banks active in private banking may want to
show potential depositors that they are well capitalised and thus that funds deposited
with them are as safe as those deposited with a larger competitor.
If the signal is effective, the increase in actual capital should lead to an increase in
earnings. However, the hypothesis that management signal private information about
‘good’ future prospects by increasing capital is not supported by the empirical work in
Berger (1995).
25
What determines how much capital is held by UK banks and building societies?
35 It seems that the wave of accounting frauds in the US did not meet the US accounting standards. It is
not clear whether these cases should change the belief that accounts apparently complying with US
GAAP are more informative than accounts complying with other standards.
Our qualitative evidence suggests that ratings affect firms’ behaviour. However, we
cannot conclude that rated firms hold higher capital ratios than their non-rated
counterparties,
We found that banks in the sample regarded the market’s most likely reaction to an
unexpected drop in capital as being a review of their rating (with a possible increase
in funding costs) followed by their shares’ trading at a lower multiple of earnings.
36
Withdrawal of wholesale funding was seen as a likely reaction only by four out of the
thirteen firms. Firms’ answers reflected their funding structure: those with significant
dependency on wholesale funding rated this reaction as likely. Therefore firms perceive
there to be a certain degree of market discipline, exercised through ratings and the
wholesale markets.
In our qualitative analysis we also explored the role that rating can play in determining
capital. The results indicate that desired capital is affected by management’s intention
to maintain the firm’s credit rating. Of the 13 firms in the sample, five regarded this
as very important in deciding desired capital and seven as important. The subsequent
discussions with firms and rating agencies suggested that this is especially so where
the firm relies heavily on non-retail deposits for its funds or where the firm’s credit
rating is an important factor in its strategy.
37
To gather further evidence on the role of ratings, we also asked firms to rank the
likelihood of various reactions to a rating downgrade. We focused on the likelihood in
the short and medium term of changes in desired capital, actual capital, and risk-
weighted assets (as a proxy for changes in the business).
38
Most banks answered that
they would change their desired capital, actual capital and distribution of risk
26
What determines how much capital is held by UK banks and building societies?
36 Most respondents said that the extent of the reaction would depend on the cause and materiality of the
unexpected reduction in actual capital.
37 Firms recognised that some strategies demand particular ratings and hence that when a firm changes
its strategy it must take account of the rating implications. More generally, rating agencies emphasise
that they take account of more than just the firm’s level of capital – a point also acknowledged by other
stakeholders.
38 We want to distinguish between a decision to change the terms of existing business (e.g. the interest
rates charged) and a decision to withdraw from some business. The first (if competitive pressures allow)
may lead to higher profits and an increase in capital. We treat this as a change in actual capital – but
it is by no means the only way of increasing the firm’s capital. The second one may reduce the firm’s
risk-weighted assets but we treat it as a change in the business or in the firm’s portfolio of activity.
weighted assets in the medium term only. Most building societies were not sensitive to
changes in the rating, given that most of their funding comes from retail savings.
Peer pressure also seems to play a role in capital determination in the UK. In our
quantitative analysis we test for it, by including in our regression, for each firm and
for each period t, the average capital ratio for all other firms of similar size reporting
at time t. We find evidence that peer pressure affects the level of capital. The result
suggests either that firms are using their capital ratio as a signal or that the capital
ratio of similar firms is a focal point for competition in capital markets.
The interviewed firms regarded peer pressure as at least an “important” factor in capital
decisions. The firms gave various explanations and qualifications about the importance
of peer pressure as a determinant of desired capital. Broadly speaking, firms responded
to the questionnaire in terms of how they perceive that other firms take their decisions
rather than how they do it themselves. Firms may aim to appear well capitalised in
relation to their peers. Nevertheless, firms consider that peer comparisons can
sometimes be overly simplistic and can ignore material issues such as the quality of
capital – for example, by not taking sufficient account of the differences between Tier
1 and Tier 2 capital. For building societies, the absence of shareholders may limit the
extent to which they use peer comparisons to form their views about desired capital. In
spite of this, our quantitative results show evidence of such behaviour.
The negative relationship between volume of assets and capital ratios found in our
quantitative analysis could support the hypothesis that small firms use
overcapitalisation as a signal to the market. As poorly capitalised banks could lose
their stakeholders’ confidence, small banks could be using their capital holdings as a
mechanism to show the market their high prudential standards. In the qualitative
analysis, ten firms regarded inaccurate market understanding of their risks as either
not important or not relevant for determining their desired capital.
39
This probably
implies that these information asymmetries can be more effectively corrected by means
such as appropriate disclosures and covenants rather than by extra capital.
27
What determines how much capital is held by UK banks and building societies?
39 The interviews suggested that some firms may have thought the question about inaccurate market
understanding of the firm’s risks referred to the firm’s misunderstanding of the risk of operating in
certain product markets. We clarified this orally and were given no indication that this would change
any firm’s answer. Furthermore, if a firm had misinterpreted the question and if it had considered
inaccurate market understanding of the firm’s risks to be material, no doubt the firm would have
mentioned it under “other”. However, none of the firms added other reasons for holding capital.
6. DETERMINANTS OF CAPITAL HOLDINGS:
REGULATORY FRAMEWORK
This section reviews the regulatory environment’s role in determining capital holdings.
In particular, decisions on capital may be affected by how capital requirements are
perceived and set by the regulator, by the degree of intervention through regulation
and supervision and by the fines imposed for regulatory breaches.
Capital requirements as a minimum
The regulator may set a capital requirement with the explicit intention that banks
always operate with a higher capital ratio. In fact, this is the interpretation of the 8%
ratio of the current Basel Accord (BCBS (1999)). In the UK, the regulator sets
individual capital requirements as minima with the expectation that firms will always
exceed this. Ediz et al. (1998) explores UK banks’ reaction to regulatory pressure. It
finds evidence that banks boost their capital ratios as soon as they fall to a certain
level above the regulatory minimum.
40
Holding capital above the minimum is thus
consistent with the aims of prudential regulation
41
and reflects an alignment between
banks’ behaviour and the regulator’s prudent approach.
Our quantitative analysis for UK firms indicates that individual capital requirements
are a significant factor in capital decisions.
42
While Ediz et al. (1998) assessed the
regulatory pressure created by target ratios, we have looked at how specific individual
capital requirements and their changes affect capital holdings. We provide details of
both quantitative and qualitative analysis below.
28
What determines how much capital is held by UK banks and building societies?
40 As explained earlier, historically the FSA has set a ‘target’ ratio above the individual capital ratios
(‘trigger ratios’) to act both as a ‘warning light’ and as a cushion of capital to help prevent an
accidental breach of the individual capital ratios.
41 This contrasts with Danielsson (2002), which suggests that excess capital is evidence of the
impossibility of risk-based regulation.
42 We have run the regression with both the trigger ratio and the target ratio as explanatory variables.
The results reported by Ediz et al. (1998) assess the relevance of the trigger ratio. Using the trigger
ratio or the target ratio makes no major difference to our results.
It could be argued that the regulator and the firms might have similar views about the
implications of shifts in the characteristics of business or in risk levels. If that were so,
regulatory and actual (or desired) capital would tend to move together, but there
would be no causality link.
43
Our quantitative analysis allows for such a dependency by
including in the regression the other factors that are commonly suspected to
determine firms’ decisions about capital.
For both building societies and banks we obtain a clearly significant positive
relationship between regulatory requirements and actual capital ratios, indicating that
the higher the required individual capital ratio the higher the actual capital ratio.
44
The short term coefficients range from 0.28 for all banks to 0.43 for banks engaged in
trading activities. The long term coefficients are 0.41 and 0.61 respectively. For
building societies the short term coefficient is 0.18 and the long term one is 0.29. So
there is never a one-to-one response. These figures suggest that, in general, less than
50% of changes in individual capital requirements is translated into changes in the
capital ratio in the short term, and a little bit more in the long term. In other words,
the buffer only partially absorbs changes in individual capital requirements.
We find that the estimated coefficient for the required capital ratio is higher for banks
and building societies with low buffers than for those with larger buffers. In other
words, firms with large buffers react less to changes in capital requirements. In fact,
in many cases, firms with a large buffer seem not to react at all to changes in capital
requirements: the coefficient is not statistically significant, whether or not the firm
has experienced a change in the required capital ratio.
In order to assess the robustness of these results we test for asymmetric responses to
changes in capital requirements. When the sample is split between firms that have
experienced an increase in their required capital ratio during the period of our data
and firms that have experienced a decrease the results show different responses to
changes.
45
In particular, banks that have experienced an increase in their requirements
raise their actual capital ratio by 50% of the increase in the requirement in the short
29
What determines how much capital is held by UK banks and building societies?
43 See note 54 in Annex 3 for further details on our causality test results.
44 Most of the empirical work for other countries has been carried out using the buffer as the explanatory
variable (Ayuso et al. (2004) and Lindquist (2004)). The reason is that, in many cases, the regulator
in these other countries does not set individual capital requirements.
45 In the period considered, our dataset has 58 instances of the capital requirement being raised and 18
instances of it being lowered.
term and by nearly 71% in the long term. For banks that have experienced a decrease,
the adjustment is around 20%. This result seems to suggest that firms are more
concerned with the possibility of regulatory breach than with the additional costs
associated with holding excess capital.
46
Building societies do not seem to react very
differently to increases or to reductions in regulatory capital requirements.
We also used the qualitative analysis to explore whether firms react to changes to their
regulatory requirements. As stated above, nine of the 13 firms in the survey said that
they form their views about desired capital by adding a margin to the required
minimum. Some of the nine firms emphasised that they still assess specific risks from
a bottom-up perspective – for example, to price new business. The other four firms said
that they form their views about desired capital by assessing the risks. Three of these
four were the largest banks in the sample.
We asked firms directly about their potential reaction to changes in their own
individual capital requirements. In relation to desired capital, banks answered either
that they would not change the amount or that they would change it in the same
direction in the medium term. Building societies, on the other hand, said they would
probably change their desired capital in the same proportion as the change in
regulatory requirements in the short term. Building societies would change their
actual capital, while banks would be less likely to do so.
47
Both banks and building
societies said they were unlikely to change their business or portfolio composition.
All the firms that forecast a reduction in their capital requirements under Pillar 1 of
the new Basel Accord said that the effect of this overall reduction would be to reduce
both their desired and actual capital in the medium term. Most also said that this
reduction in capital requirements would also affect their business or portfolio
composition in the medium term.
30
What determines how much capital is held by UK banks and building societies?
46 The lack of statistical significance of the return on equity variable (ROE) in the estimation, mentioned
above, could support such an interpretation. However, as already pointed out, the results on ROE must
be treated with caution.
47 Four out of the five building societies said they would change actual capital, while only two out of the
eight banks said they would.
R egulatory rules and supervisory behaviour
The regulator affects the actual level of capital in various ways, as well as by just
setting the capital requirement. First, a breach of the capital requirement will usually
attract serious enough regulatory intervention that firms will want to hold more
capital than required, in order to avoid or reduce the likelihood of a breach. Milne
(2002) suggests that capital requirements act as an incentive mechanism in which a
breach gives rise to a penalty. It is then shown that banks would want to hold more
capital than the regulatory minimum.
Goodhart (1995) has suggested that regulators commit themselves to specific types of
action that are triggered when thresholds above the minimum margin are breached.
This recognises that the underlying requirements must be somewhat arbitrary. For
example, what is the difference between a minimum requirement of 8% or 7.9%?
Goodhart suggests that the inevitable degree of arbitrariness in any set of minimum
requirements will encourage regulators to retain some discretion in responding to
breaches of the requirements. Goodhart further suggests that in reality the regulator’s
discretion means that the regulator will react with forbearance, partly because of a
concern that its intervention will cause consumers to lose confidence in banks. If so,
regulated banks will soon factor the regulator’s behaviour into their decisions and hold
less capital than they would otherwise do. The results presented in Aggarwal and
Jacques (1998), for instance, show that regulatory commitment to action linked to the
Federal Deposit Insurance Corporation Improvement Act was effective in getting US
banks simultaneously to increase their capital ratios and reduce their portfolio risk
Regulators can also affect capital by indicating their willingness to intervene when
banks find themselves in financial distress. For instance, the FSA’s statement about
non-zero failure (FSA (2003)) is an example of how the regulator can inform the market
of its intentions. As with any regulatory policy the effect will depend on its credibility.
Our qualitative analysis found that avoiding “the consequences of a potential breach of
regulatory capital” was regarded as “very important” by all firms in the sample.
PricewaterhouseCoopers (2003) reports a similar result from a sample of internationally
active European banks.
48
Our subsequent discussions with firms indicated that they
regard capital requirements as the absolute minimum for capital rather than a sort of
31
What determines how much capital is held by UK banks and building societies?
48 The survey covers amongst other things the determinants of capital in excess of the minimum required
amount. About 80% of respondents identified avoiding a breach of capital requirements as very
important or important.
target. Managers said that a regulatory breach might be regarded as comparable to
“deceiving customers” and might therefore affect their position in the firm.
The statistically significant negative effect of size on capital holdings could reflect that
larger banks tend to have better diversified portfolios. Also larger banks may have better
risk controls than smaller banks, as there could be a significant fixed cost of
implementing a control system – so capital ratio could be negatively associated with size.
7. SUMMARY AND CONCLUSIONS
In this paper we have argued that the amount of capital held by banks and building
societies depends on risk management, market discipline and regulatory environment.
Using both quantitative and qualitative approaches, we provide evidence on which
hypotheses hold in the UK. In particular, we analysed prudential returns for UK banks
and building societies and the responses to a questionnaire sent to a sample of firms that
we later interviewed. Our findings are in line with the results obtained with data from
other countries (Ayuso et al. (2004) and Lindquist (2004)).
Even though all firms have a buffer over individual capital requirements, our analysis
indicates that changes in these individual capital requirements are very likely to be
accompanied by some response in the capital ratio. For example, if a bank (building
society) which is holding capital at 15% of risk weighted assets has its individual required
capital ratio increased from 10% to 11%, it would on average increase its actual capital
ratio to 15.6% (15.4%).
Our evidence indicates that the dependency of capital ratios on capital requirements is
somewhat greater for firms operating close to their regulatory requirements than for
those that hold a large amount of excess capital. As the firms with smaller capital buffers
are generally the larger banks, it could be argued that capital policy changes introduced
by the regulator will affect large banks more than smaller ones. The firm’s degree of risk
aversion will determine the final impact.
Adjustment costs affect the amount that firms hold. They seem to be marginally larger
for building societies than for banks, maybe because of the formers’ limited access to
32
What determines how much capital is held by UK banks and building societies?
capital markets. Firms say that the difference between actual and desired capital is
mainly determined by the costs of raising further capital and by provision for unexpected
events in the economy and in the firm. We find that the economic cycle is negatively
associated with capital ratios, at least for banks. Firms also say that their desired capital
is mainly determined by the need to finance their long term business strategy.
Risk appetite and risk management help determine capital holdings. Perhaps
surprisingly, portfolios with a higher proportion of assets falling into the high risk group
category (in our case, 100% weighted assets) are associated with lower capital ratios, a
result also obtained by Lindquist (2004). Some firms consider systems and controls to
be a partial substitute for capital.
Market discipline seems to affect how firms assess their capital needs and seems to play
a complementary role to regulation in this assessment. Our analysis indicates that firms
adjust their level of capital in response to changes in their peers’ level. So changes in
the regulatory capital requirement for a subset of firms could produce a similar change
in actual capital for firms that have not experienced the change but are competing in the
same market. Any regulatory change that enhances market discipline, such as
requirements for firms to disclose new information, will not necessarily result in more
capital in the system.
The implementation of the new Basel Accord will certainly change the relationship
between regulatory capital requirements and capital held by firms. The fact that
regulatory capital will more heavily rely on internal models devised by firms to set their
desired capital suggests that the link may be reinforced. However, it is an area that
requires further research.
Some of our evidence is consistent with a number of different hypotheses about firms’
behaviour and the determinants of capital. Further evidence should also be gathered on
how banks calculate the size of the margin of capital that they aim to hold over the
regulatory minimum. The finding that nine of 13 interviewed firms said that they form
their views about desired capital by adding a margin to the required minimum could
provide a starting point. Milne’s (2002) model could provide the framework.
33
What determines how much capital is held by UK banks and building societies?
REFERENCES
Aggarwal R. and K.T. Jacques, “Assessing the impact of prompt corrective action on bank
capital and risk”, FRBNY Economic Policy Review, October 1998
Ayuso, J., D. Perez and J. Saurina, “Are capital buffers pro-cyclical? Evidence from Spanish
panel data”, Journal of Financial Intermediation (forthcoming), 2004
Basel Committee on Banking Supervision “A new capital adequacy framework”,
Consultative paper issued by the BCBS, Bank for International Settlements, June 1999
Basel Committee on Banking Supervision, “Risk management practices and regulatory
capital”, Bank for International Settlements, November 2001
Berger, A.N., “Market discipline in Banking”, Proceedings of a conference on bank
structure and competition, Federal Reserve Bank of Chicago, pp. 419-437, May 1991
Berger, A.N., “The relationship between capital and earnings in banking”, Journal of
Money, Credit, and Banking, Vol. 27, No. 2, May 1995
Berger A.N., R.J. Herring and G.P. Szego, “The role of capital in financial institutions”,
Journal of Banking and Finance, Vol. 19, pp. 383-430, 1995
Berger A.N. and T.H. Hannan, “The efficiency cost of market power in the banking industry:
a test of the ‘quiet life’ and related hypothesis”, The Review of Economics and Statistics,
Vol. 80, pp. 454-465, August 1998
Cornett M. and H. Tehranian An examination of voluntary versus involuntary security
issuance by commercial banks: the impact of capital regulations on common stock returns”,
Journal of Financial Economics, Vol. 35, pp. 99-122, 1994
Carletti E. and P. Hartmann, “Competition and stability: what’s special about banking?”
LSE Financial Markets Group, Special Paper 140, April 2002
34
What determines how much capital is held by UK banks and building societies?
Danielsson J., “On the feasibility of risk-based regulation”, London School of Economics,
2002
Demsetz R.S, M.R. Saidenberg, and P.E.Strahan, “Banks with something to lose: the
disciplinary role of franchise value”, FRBNY Economic Policy Review, October 1996
Ediz T., I. Michael and W. Perraudin, “The impact of capital requirements on UK bank
behaviour”, FRBNY Economic Policy Review, October 1998
FSA, “New regulator for the new millennium”, February 2000
FSA, “Individual capital ratios for banks”, Policy Statement, July 2001
FSA, “Reasonable expectation: regulation in a non-zero failure world”, September 2003
Furfine C., “Evidence on the response of US banks to changes in capital requirements”, BIS
Working Paper, Bank for International Settlement, No. 88, June 2000
Goodhart, C. A. E., “Some regulatory concerns”, LSE Financial Markets Group, 1995
Jackson, P., W. Perraudin and V. Saporta, “Regulatory and ‘economic’ solvency standards
for internationally active banks”, Bank of England Working Paper, 2002
Jackson, P., “Capital requirements and bank behaviour: the impact of the Basel Accord”,
Basel Committee on Banking Supervision Working Paper, Bank for International
Settlements, April 1999
Laeven, L. and G. Majnoni, “Loan loss provisioning and economic slowdowns: too much,
too late?”, World Bank, Working Paper 2749, 2002
Lindquist, K., “Banks’ buffer capital: How important is risk”, Journal of International
Money and Finance, Vol. 23, pp. 493-513, 2004
McDonnell, W., “Managing risk: practical lessons from recent ‘failures’ of EU insurers”, FSA
Occasional Paper 20, January 2003
Matten C., “Managing bank capital: capital allocation and performance measurement”,
WILEY, 2nd Edition, 1998
35
What determines how much capital is held by UK banks and building societies?
Milne A. and A.E. Whalley, “Bank capital regulation and incentives for risk-taking”, Journal
of Economic Literature, March 2000
Milne A. “Minimum capital requirements and the design of the new Basel Accord: a
constructive critique”, Journal of Financial Regulation and Compliance, vol. 9, no. 4,
November 2001
Milne, A, “Bank capital regulation as an incentive mechanism: implications for portfolio
choice”, Journal of Banking and Finance, January 2002
Nier E. and U Baumann , “Market discipline, disclosure and moral hazard in banking”, Bank
of England (Draft), September 2002
Oliver, Wyman & Company, “Study on the risk profile and capital adequacy of financial
conglomerates”, February 2001
Pain, D., “The provisioning experience of the major UK banks: a small panel investigation”,
Bank of England, Working Paper no. 177, 2003
PricewaterhouseCoopers, “Basel … hopes and fears. A European banking view of the
practical application of Pillar 2”, 2003
Richardson J. and M. Stephenson, “Some aspects of regulatory capital”, FSA Occasional
Paper 7, March 2000
36
What determines how much capital is held by UK banks and building societies?
37
What determines how much capital is held by UK banks and building societies?
WW
WW
hh
hh
yy
yy
ff
ff
ii
ii
rr
rr
mm
mm
ss
ss
mm
mm
ii
ii
gg
gg
hh
hh
tt
tt
hh
hh
oo
oo
ll
ll
dd
dd
mm
mm
oo
oo
rr
rr
ee
ee
cc
cc
aa
aa
pp
pp
ii
ii
tt
tt
aa
aa
ll
ll
tt
tt
hh
hh
aa
aa
nn
nn
tt
tt
hh
hh
ee
ee
rr
rr
ee
ee
gg
gg
uu
uu
ll
ll
aa
aa
tt
tt
oo
oo
rr
rr
yy
yy
mm
mm
ii
ii
nn
nn
ii
ii
mm
mm
uu
uu
mm
mm
PP
PP
oo
oo
tt
tt
ee
ee
nn
nn
tt
tt
ii
ii
aa
aa
ll
ll
ii
ii
mm
mm
pp
pp
aa
aa
cc
cc
tt
tt
oo
oo
ff
ff
ii
ii
nn
nn
cc
cc
rr
rr
ee
ee
aa
aa
ss
ss
ii
ii
nn
nn
gg
gg
rr
rr
ee
ee
gg
gg
uu
uu
ll
ll
aa
aa
tt
tt
oo
oo
rr
rr
yy
yy
cc
cc
aa
aa
pp
pp
ii
ii
tt
tt
aa
aa
ll
ll
EE
EE
xx
xx
cc
cc
ee
ee
ss
ss
ss
ss
cc
cc
aa
aa
pp
pp
ii
ii
tt
tt
aa
aa
ll
ll
aa
aa
ss
ss
aa
aa
rr
rr
ee
ee
ss
ss
uu
uu
ll
ll
tt
tt
oo
oo
ff
ff
hh
hh
oo
oo
ww
ww
dd
dd
ee
ee
cc
cc
ii
ii
ss
ss
ii
ii
oo
oo
nn
nn
ss
ss
oo
oo
nn
nn
cc
cc
aa
aa
pp
pp
ii
ii
tt
tt
aa
aa
ll
ll
aa
aa
rr
rr
ee
ee
tt
tt
aa
aa
kk
kk
ee
ee
nn
nn
ww
ww
ii
ii
tt
tt
hh
hh
ii
ii
nn
nn
ee
ee
aa
aa
cc
cc
hh
hh
ff
ff
ii
ii
rr
rr
mm
mm
Excess capital as a result of
firms taking account of risks not
relevant to the regulator
Excess capital as a result of
firms assessing risks differently
from the regulator
The firm’s response to a change in the regulator’s requirements may
depend on what drove the change. If it is to cover risks not covered
under the current requirement, there may be no response; otherwise,
the firm may wish to maintain the same buffer to mitigate those risks
not taken into account by the regulator.
Excess capital as a result of
agency problems
If management have little pressure from shareholders for capital
remuneration and other market discipline mechanism are weak, an
increase in capital requirement will lead them to increase capital to
maintain a “quiet life”.
Excess capital as a result of
weak risk management practices
For a given a business risk profile, prudent firms with poor risk
management will tend to hold more capital and will be likely to raise
their capital in line with regulatory increases
49
. However, changes in
capital requirements that give adequate incentives for risk management
should lead these firms to hold less excess capital.
Excess capital as financial slack
to finance long term strategy
If firms have a long-term strategy about their growth, an increase in
capital requirements is likely to cause them either to raise capital to
maintain their ability to deliver their long-term strategy or to change
their strategy.
Excess capital as a historical
legacy
In this case, a change in capital requirement would not lead to a
change in actual capital.
Excess capital as a result of
adjustment costs from issuing
equity
If adjusting levels of capital is costly, firms may respond differently to
increases and reductions in the regulatory minimum – possibly reacting
to increases in the regulatory minimum but not significantly responding
to reductions. Much may depend on the size of the change and on how
the firm views its longer-term outlook and any possible future changes
to its regulatory capital.
Excess capital as a result of
firms’ optimising their capital
structure
This may change the overall balance between the benefits and costs of
holding excess capital. However, because of the opportunity cost of
capital and if other things are equal, the higher the cost of capital the
lower the amount of excess capital
Excess capital to cushion the
effect of an economic downturn
We would expect firms to raise their capital ratio
Annex 1: Summary of how changes in regulatory requirements
might affect actual capital
38
What determines how much capital is held by UK banks and building societies?
WW
WW
hh
hh
yy
yy
ff
ff
ii
ii
rr
rr
mm
mm
ss
ss
mm
mm
ii
ii
gg
gg
hh
hh
tt
tt
hh
hh
oo
oo
ll
ll
dd
dd
mm
mm
oo
oo
rr
rr
ee
ee
cc
cc
aa
aa
pp
pp
ii
ii
tt
tt
aa
aa
ll
ll
tt
tt
hh
hh
aa
aa
nn
nn
tt
tt
hh
hh
ee
ee
rr
rr
ee
ee
gg
gg
uu
uu
ll
ll
aa
aa
tt
tt
oo
oo
rr
rr
yy
yy
mm
mm
ii
ii
nn
nn
ii
ii
mm
mm
uu
uu
mm
mm
PP
PP
oo
oo
tt
tt
ee
ee
nn
nn
tt
tt
ii
ii
aa
aa
ll
ll
ii
ii
mm
mm
pp
pp
aa
aa
cc
cc
tt
tt
oo
oo
ff
ff
ii
ii
nn
nn
cc
cc
rr
rr
ee
ee
aa
aa
ss
ss
ii
ii
nn
nn
gg
gg
rr
rr
ee
ee
gg
gg
uu
uu
ll
ll
aa
aa
tt
tt
oo
oo
rr
rr
yy
yy
cc
cc
aa
aa
pp
pp
ii
ii
tt
tt
aa
aa
ll
ll
Excess capital as a result of market discipline
Excess capital to reduce the risks
faced by uninsured investors
Excess capital to obtain access to
certain capital markets
The extent to which firms react to an increase in the capital
requirement will then be conditioned by the possible effect on
their funding costs. Firm are likely to react if:
the increase is observed by uninsured investors (whether
or not through a change in rating) and this reveals new
information that would lead them to demand a higher
yield unless capital is increased; or
the increase leads to a lower rating that reduces the firms’
ability to access certain capital markets, thereby
increasing their funding costs.
Excess capital as a result of
inaccurate information in the
market
Changes in regulatory capital may affect not only the firm
whose capital requirement has increased but also those
within its peer group. In this case, an increase in the
capital ratio set by the regulator may have knock-on
effects amongst a peer group and tend to drive capital
levels upwards. This would imply that the market
indirectly
2
observes the change in regulatory capital
and alters its judgement as a result.
An increase in regulatory capital will make the firm react if
the market interprets the change as deterioration in
the firm’s relative capital position. In this case a firm
will have to choose between increasing its capital,
taking other credible remedial action or accepting the
consequences of the deterioration.
EE
EE
xx
xx
cc
cc
ee
ee
ss
ss
ss
ss
cc
cc
aa
aa
pp
pp
ii
ii
tt
tt
aa
aa
ll
ll
aa
aa
ss
ss
aa
aa
rr
rr
ee
ee
ss
ss
uu
uu
ll
ll
tt
tt
oo
oo
ff
ff
tt
tt
hh
hh
ee
ee
rr
rr
ee
ee
gg
gg
uu
uu
ll
ll
aa
aa
tt
tt
oo
oo
rr
rr
yy
yy
ff
ff
rr
rr
aa
aa
mm
mm
ee
ee
ww
ww
oo
oo
rr
rr
kk
kk
Capital requirements set only the
minimum amount of capital
We would expect banks to either not respond or raise their
actual capital ratio.
Excess capital to avoid regulatory
intervention when faced with a
breach or likely breach of the
capital requirement
To avoid the costs of regulatory intervention, banks will want to
hold a buffer above the level of capital set by the regulator, and
will increase their actual capital in line with any increase in the
likelihood or severity of intervention.
Excess capital as a result of the
market’s perception of a non-
zero failure regime
An increase in the credibility of a non zero failure regime would
lead to more excess capital and vice versa.
Annex 2: Summary of firms’ replies to the questionnaire
Q1: How is desired capital specified?
Q2: What are the determinants of the firm’s desired
capital and regulatory requirements?
39
What determines how much capital is held by UK banks and building societies?
024681012
14
A level of equity
A ratio (of equity
to risk-weighted
assets)
Banks Building Societies
02468
We assess the risks, then decide
how much capital is required to run
the business and finally verify that
this is more than the regulatory
requirement.
Given the regulatory capital
requirement, we assess how much
additional capital is required to run
the business.
Banks Building Societies
Number of firms
Number of firms
Q3: What does the desired capital represent?
Q4: What are the determinants of the firm’s desired
capital?
40
What determines how much capital is held by UK banks and building societies?
012345
6
78
The minimum
capital that we
wish to hold.
A range (or a point
in a range) within
which we aim to
stay at all times.
Banks Building Societies
02468
Avoid the consequences of a potential breach of regulatory capital
Maintaining an existing credit rating
Capital held by your peers
Accessing markets with high reliance on counterparties credit risk
Inaccurate understanding of the markets about your risks
Financing your long-term strategy
Securing wholesale deposits / access to money markets
Cushion against the effect of an economic downturn
Risks to the business
Complement to risk management and internal systems and controls
Regulatory capital underestimates the risks that it captures
Activities attracting no capital requirements and requiring capital
Very Important Important Not Important Not Relevant
Number of firms
Number of banks
Q5: How does desired capital differ from actual capital?
41
What determines how much capital is held by UK banks and building societies?
0 12345
Avoid the consequences of a potential breach of regulatory capital
Maintaining an existing credit rating
Capital held by your peers
Accessing markets with high reliance on counterparties credit risk
Inaccurate understanding of the markets about your risks
Financing your long-term strategy
Securing wholesale deposits / access to money markets
Risks to the business
Complement to risk management and internal systems and controls
Regulatory capital underestimates the risks that it captures
Activities attracting no capital requirements and requiring capital
Very Important Important Not Important Not Relevant
Cushion against the effect of an economic downturn
Number of building societies
0246810
Desired capital is
always very close
to actual capital
Desired capital
usually exceeds
capital held
Capital held
usually exceeds
desired capital
Capital held varies
around desired
capital
Banks
Building Societies
Number of firms
Q6: Why does actual capital differ from desired capital?
42
What determines how much capital is held by UK banks and building societies?
0123456
Cost of raising extra capital
Historical legacy
Unexpected events in the economy
affecting all firms in the sector
Unexpected developments within the
firm
Very important Important Not important Not relevant
Number of banks
0123 45
Cost of raising extra capital
Historical legacy
Unexpected events in the economy
affecting all firms in the sector
Unexpected developments within the
firm
Very important
Important Not important Not relevant
Number of building societies
Q7:What would be the reaction to an unexpected
reduction in the actual capital arising from, say,
a reduction in profits
43
What determines how much capital is held by UK banks and building societies?
01234567
Retail depositors
withdraw funds
Wholesale
depositors
withdraw funds
Ratings will be
reviewed
Shares will trade
at a lower multiple
of earnings
Very likely Likely Unlikely Not relevant
Number of banks
012345
Retail depositors
withdraw funds
Wholesale
depositors
withdraw funds
Ratings will be
reviewed
Shares will trade
at a lower multiple
of earnings
Very likely
Likely Unlikely Not relevant
Number of building societies
Note: One bank did not answer this question and another did not
say whether it would affect the share price
Q8: Suppose you had experienced a downgrade in your
rating as a result of an event affecting you such as an
unexpected reduction in profits. How would you have
reacted to the downgrade?
44
What determines how much capital is held by UK banks and building societies?
012345
The desired capital would change in
the short term only
The desired capital would change in
the medium term only
The desired capital would change in
the short and medium term
The desired capital would not change
in the short or medium term
Banks
Building Societies
Change in desired capital: number of firms
012345
The actual capital would change in
the short term only
The actual capital would change in
the medium term only
The actual capital would change in
the short and medium term
The actual capital would not change
in the short and medium term
Banks Building Societies
Change in actual capital: number of firms
Q9: If your individual capital requirement [increased
by 1 percentage point from 9.0% to 10.0%].
How would you react to it?
45
What determines how much capital is held by UK banks and building societies?
01234
The distribution of RWA would
The distribution of RWA would
The distribution of RWA would
The distribution of RWA would not
change in the short or medium term
Number of responses
Banks Building Societies
change in the short and medium term
change in the medium term only
change in the short term only
Change in the distribution of risk weighted
assets (RWA): number of firms
01234
The distribution of RWA would
The distribution of RWA would
The distribution of RWA would
The distribution of RWA would not
change in the short or medium term
Number of responses
Banks Building Societies
change in the short and medium term
change in the medium term only
change in the short term only
Change in desired capital: number of firms
46
What determines how much capital is held by UK banks and building societies?
01234567
No change in the short and medium
term
No change in the short term and
different changes in the medium
term
Changed by the same proportion or
less in the short term and remain at
that level
Banks Building Societies
Change in actual capital: number of firms
02468
10 12
Changed by the
same proportion
No change in the
short and
medium term
Banks Building Societies
Change in distribution of RWA: number of firms
Q10: How would the implentation of Basel II requirements
for regulatory capital (Pillar I) affect your capital
requirement? How would you react to it?
Of the 13 firms interviewed, six firms took the view that Basel II’s new Pillar
1 would mean that they need less regulatory capital to support the current
business. Two other firms said that they expected the effect of Pillar 1 to be
neutral. The remaining firms did not express a view.
The six firms that expected a reduction in regulatory capital said that it would
lead to a reduction of actual and desired capital in the medium term. Five of
them also said that it would also affect the distribution of assets in the
medium term.
47
What determines how much capital is held by UK banks and building societies?
What determines how much capital is held by UK banks and building societies?
48
ANNEX 3: QUANTITATIVE ESTIMATION OF THE DETERMINANTS OF
CAPITAL RATIOS
The model
This annex presents the empirical results from our quantitative approach. The
starting point of the analysis is Ayuso et al. (2004) with the extension
incorporated by Lindquist (2004). These two papers analyse the behaviour of the
buffer capital of Spanish and Norwegian banks respectively.
We specify the model in terms of capital ratios.
51
The most general specification
of our model can be represented as:
itiitit
it
ittititititit
KZPeerDepGSizeRiskKrAK
εη
µςλτγδβα
1
= (1)
where the subscript i denotes the firm and t the period. K is the capital as a
proportion of risk weighted assets
52
and Z are other explanatory variables that are
relevant to the specific dataset (building societies and banks) and that will be
described below.
Kr is the individual capital requirements as a proportion of risk weighted assets,
using the trigger ratio (i.e. threshold ratio) as the regulatory requirement.
53
The
51
The empirical work for Spain (Ayuso et al. (2004)) and Norway (Lindquist (2004)) has used the
buffer as the explanatory variable. The reason is that, in many cases, the regulators in these two
countries do not set individual capital requirements. We have also run the regressions specifying
the buffer as a dependent variable and not including the capital requirement as a regressor. The
results show that a) there is inertia b) risk, size, quality of capital and growth have a negative
impact on the buffer, while c) non-wholesale deposits and peer pressure have a positive effect
(although the latter is not statistically significant). Also foreign firms have a larger buffer.
52
We have also run the regression with the ratio of tier 1 capital to risk weighted assets as the
explained variable. This produces no major differences in the estimated coefficients.
53
As already mentioned, we also ran the regression using the target ratio for banks, with not
significantly different results.
What determines how much capital is held by UK banks and building societies?
49
coefficient α is expected to be positive and statistically significant,
54
as we
expect that firms react to individual regulatory requirements.
55
The variable Risk is a proxy for risk: it is an ex-post measure of risk, calculated as
the proportion of a firm’s highest risk assets (i.e. those with a 100% risk weight)
over its total assets.
56
A statistically significant β could be interpreted as
evidence that firms assess risk differently from regulators. A positive β would
suggest that increases in the riskiness of the portfolio are associated with capital
increases beyond those required by the heavier regulatory weight given to those
assets. This would suggest that firms assign an even larger risk to these assets
than the regulator; a negative β would be compatible with either the presence of
moral hazard behaviour in firms or the possibility that riskier firms also have
better risk management mechanisms.
57
54
We carry out a Granger causality test to assess the hypothesis that the changes in the trigger
ratio precede the changes in the capital ratio and not the other way round. We carry out the test
by running the regression of both variables on the lagged values of the other and the variable
itself, using a random-effects (within) estimator. The p-value of the lagged capital ratio in the
trigger equation is 0.33 (0.64 in the within estimation) for banks. The p-value for the lagged
trigger in the capital specification is 0.03 (0.43 in the within estimation). Therefore, there is
some evidence that the changes in the trigger ratio precede the capital changes.
55
We have tested for the relevance of the quality of capital by running the regression including the
ratio of Tier 1 capital to total adjusted capital as an additional endogenous regressor. The
coefficient is negative but not statistically significant. So there seems to be a limited degree of
trade-off between quantity and quality.
56
As shown in Tables A1 and A2, the average of such a proxy of risk is around 7% for building
societies and 30% for banks. The pattern that emerges for the relationship between banks’ size and
riskiness shows that the largest firms seem to be holding the lowest proportion of risky assets
(25.6% against 38.5% on average for the smallest firms). On the other hand, the relationship
between growth and this measure of risk is negative but not statistically significant.
57
If an index of governance were available it could be used to distinguish between the two
possibilities. We have also used the proportion of risk weighted assets to total assets as an
alternative measure of risk appetite. The coefficient is negative and statistically significant.
What determines how much capital is held by UK banks and building societies?
50
Size is measured here by total assets.
58
Most hypotheses suggest a negative
relationship between size and capital held by firms, so that δ is expected to be
negative.
59
G is the real annual GDP growth rate by quarter. We would expect a negative sign
if buffers are pro-cyclical.
For each period t, the variable Peer is the average capital ratio of all firms,
except firm i, that report at time t and that are of similar size to firm i.
60
We
would expect the sign of λ to be positive if peer pressure is relevant.
Dep is the proportion of non-inter-bank deposits (commercial and retail deposits,
the latter being insured) over total deposits, so that it partially captures the
relevance of insurance.
61
The larger the amount of insured funding the weaker
market discipline can be, so that a negative τ could be regarded as evidence of
moral hazard behaviour by firms.
Other variables (Z) have been also included in the estimation depending on their
availability and relevance within the different databases. In particular, for
building societies, return on equity (ROE) – measured as the proportion of profits
over capital
62
is included as an additional regressor. This variable is also
58
Building societies data provides information on number of employees. When that is tried as a
measure of size the estimated coefficient is negative but not statistically significant.
59
We are therefore not using risk-weighted assets as an explanatory variable in our equation
because it conflates the effect of size and risk. Our approach is based on separating size and risk
and is more in line with FSA’s ARROW approach (FSA (2000)) in that this distinguishes between
impact (proxied by an indicator of size) and probability (proxied by risks).
60
For each firm, the variable is calculated as the mean of the capital ratios of all firms, except the
firm itself, that submit returns in the same period and that are in the same quartile. Lindquist
(2004) proposes a similar but simpler measure, as he does not group firms by size.
61
On the liability side of a bank’s balance sheet.
62
An alternative to ROE for building societies is the ratio of profit to reserves. The results are very
similar to those reported here with the ratio defined relative to capital.
What determines how much capital is held by UK banks and building societies?
51
constructed for those banks that report a profit in their financial returns (BSD3).
ROE is a proxy for the opportunity cost of capital, so the sign of the coefficient
is expected to be negative. However, for building societies, profits are one of the
main sources of capital, so changes in capital are closely equivalent to profits.
For banks, the variable trade, measured as the proportion of the trading book’s
notional risk weighted assets over total risk weighted assets, is a proxy for the
amount of trading book activity. It is included as it could be argued that the
assessment of risk associated with these activities could have different
implications for capital than credit risk. The variable could capture business
differences
63
associated with investment banking activities in the portfolio.
The dummy variable foreign is a trend dummy that takes that value of 0 if the
country of origin of the firm’s parent is the UK and 1 otherwise.
64
The hypothesis that there are adjustment costs in attaining the desired level of
capital justifies the inclusion of the dependent variable lagged one period.
Finally,
η
i
is an unobservable variable that captures the idiosyncratic features of
each firm that are constant over time but vary from firm to firm. These could
cover management’s aversion to regulatory risk, management’s strategy for new
business opportunities or management’s freedom from shareholders.
ε
it
is a
random shock.
Data
We estimate the model with two different sets of data. First, we use building
societies’ FSA quarterly capital returns covering the period 1997 Q2 - 2002 Q2,
that constitute a balanced panel dataset. These returns include both capital
statements and the income and expenditure account. Second, we use quarterly
63
The average value of the variable trade for banks is 2.3% for UK-owned banks, 13.3% for other
EU-owned banks, 17.1% for US-owned banks and 12.2% for other non-UK banks.
64
We have tested not only whether foreign-owned banks have different capital ratios but also
whether they respond differently to changes. The coefficient of the slope variable is not
statistically significant.
What determines how much capital is held by UK banks and building societies?
52
capital returns submitted to the FSA by UK incorporated banks, covering the
period 1998 Q3 – 2002 Q3.
65
Banks report at different months, in some cases
their reporting pattern is not regular and some have changed it during the period
analysed. It is an unbalanced panel dataset that in most cases does not contain
information on income and expenditure but only on capital accounts.
Tables A1 and A2 summarise the main statistics of the variables used in the
estimations for building societies and UK banks respectively.
66
The tables show
three measures of standard deviation of the relevant variables: “total”, “over
time” and “within firms”. The first measures the overall dispersion of the dataset;
the second measures the deviation in firms’ average over time; the third
measures the deviation within each firm over time.
All variables in the tables, except for ROE, show more dispersion across firms
than within a given firm over time.
67
In many instances, the differences among
firms are four times larger than the differences across time. These variations
suggest that the different behaviour among firms needs to be properly analysed.
68
65
In particular we used BSD3 returns.
66
We have dropped for bank returns observations where:
the value of the capital ratio or the capital requirement was missing;
the capital ratio exceeded 500%;
the variable risk or dep was larger than 100.
We have also excluded two observations with capital ratios below or at the required minimum
level, as these banks would have been subject to intervention by the supervisory authorities.
67
The mean of ROE for banks is 6.88 and the dispersion is higher across firms than within a given
firm over time.
68
Average buffers (and their dispersion) for UK banks are higher than those reported for Spanish
(Ayuso et al. (2004)) and Norwegian (Lindquist (2004)) banks. The average buffer for building
societies is below or close to the ones reported there. In particular, the variable buffer2 for
Norwegian banks is 9.4% with a 6.3 standard deviation and the variable buffer for Spanish banks
is 40.3% with 40.4 of standard deviation.
What determines how much capital is held by UK banks and building societies?
53
A comparison of the tables shows that much higher average capital ratios are
observed for banks than for building societies. Regulatory capital requirements
are also set at a higher level for banks than for building societies, but the
dispersion of the former is much higher than the one for the latter.
TABLE A1: SUMMARY STATISTICS OF BUILDING SOCIETIES
(1)
Variable
(2)
Mean Standard
deviation
Standard
deviation
over time
Standard
deviation
within firms
K
15.16 4.11 4.05 0.88
Kr
9.65 0.50 0.40 0.30
Tier 1
(3)
95.45 6.89 6.52 2.37
Buffer
(4)
56.92 40.51 39.70 9.39
Buffer2
(5)
5.50 4.02 3.96 0.87
Size
2229.0 7947.4 7859.9 1513.1
Risk
6.88 4.18 3.75 1.89
Peer
15.16 1.31 1.25 0.43
ROE
6.83 4.52 2.55 3.75
Observations 1365
Firms 65
(1) Variable definitions are in the text.
(2) Size in millions of £; K, Kr, Buffer, Risk, Peer and ROE in % terms.
(3) Tier 1 capital/capital in % terms.
(4) Buffer is defined as (K-Kr)/Kr in % terms.
(5) Buffer2 is defined as K-Kr
.
What determines how much capital is held by UK banks and building societies?
54
TABLE A2: SUMMARY STATISTICS OF BANKS
(1)
Variable
(2)
Mean Standard
deviation
Standard
deviation
over time
Standard
deviation
within firms
K
41.45 61.01 61.44 22.60
Kr
12.78 5.90 5.21 3.33
Tier 1
(3)
87.28 14.89 13.89 5.87
Buffer
(4)
234.52 558.38 546.15 223.14
Buffer2
(5)
28.67 59.67 60.02 22.81
Size
14536.48 46452.24 42617.7 11559.41
Risk
30.82 26.22 24.69 9.29
Dep
63.04 35.00 32.72 13.14
Peer
41.45 18.07 15.49 9.38
Trade
7.55 20.08 18.62 6.04
Observations 2744
Firms (foreign) 187 (92)
(1) Variable definitions are in the text.
(2) Size in millions of £; K, Kr, Buffer, Risk, Dep, Peer and Trade in % terms.
(3) Tier 1 capital/adjusted capital in % terms.
(4) If target ratios are used the buffer mean is 213.99.
(5) If target ratios are used the buffer2 mean is 27.66.
What determines how much capital is held by UK banks and building societies?
55
Empirical results
In order to carry out the estimation, we re-specify equation (1) in logs, so that it
becomes:
itiititit
it
ittititititit
kzdeppeer
depgsizeriskkrak
εηµξτλ
τ
γ
β
α
++++++
+
+
+
+
+=
1
(2)
where the lower case names indicate variables in logs.
We treat the risk variable (risk), the deposits variable (dep), the trading book
variable (trade) and the opportunity costs variable (roe) as endogenous. As the
lagged endogenous variable is included among the regressors, we proceed to
transform the above equation into first differences, as is traditionally done with
panel data.
69
We estimate equation (2) using instrumental variables.
70
We report
the coefficients and their robust standard errors from the robust one-step
estimators. We report the Sargan test of over-identifying restrictions and
autocorrelation tests of first and second order. When we use differenced data, we
should observe first order autocorrelation and no second order autocorrelation.
We present the results from the two different types of return. Table A3 has the
results for building societies and Table A4 the results for banks.
In columns 1 to 4 in Table A3 a first difference estimate is reported, using a GMM
estimator with lagged levels of the dependent variable and the endogenous
variables (second, third and fourth lags) and the lagged differences of the
exogenous as instruments. The variables risk and roe are treated as endogenous.
69
For the static estimation with building society data to assess the role of ROE we use a fixed effect
estimate The fixed effects estimator leads to consistent estimates even if the time-invariant
component of the error term (η
i
) is correlated with the regressors.
70
A GMM estimator is used for the coefficients using lagged levels of the dependent variable and
the endogenous variables (second or more lagged periods) and differences of the strictly
exogenous variables.
What determines how much capital is held by UK banks and building societies?
56
(1) First difference regression. First-step robust standard errors in brackets. All equations include
quarterly dummies.
(2) Observations with an excess capital lower than 30.5%, which corresponds to the lowest quarter of
the population.
(3) Prob >
χ
2 (degrees of freedom).
(*) Statistically significant at 10%.
(**) Statistically significant at 5%.
TABLE A3: DETERMINANTS OF BUILDING SOCIETIES CAPITAL RATIOS (1997 Q2-2002 Q2)
(1)
Variable All sample Firms that
experienced an
increase in kr
Firms that
experienced a
reduction in kr
Firms with a low
buffer
(2)
kr
0.18 (0.07)** 0.22 (0.10)** 0.21 (0.10)** 0.29 (0.14)**
size
-0.02 (0.02) -0.05 (0.03)** 0.01 (0.01) -0.08 (0.04)**
risk
-0.04 (0.01)** -0.05 (0.01)** -0.04 (0.01)** -0.01 (0.02)
g
-0.002 (0.005) -0.003 (0.007) -0.004 (0.006) -0.01 (0.009)
peer
0.16 (0.06)** 0.17 (0.09)** 0.04 (0.08) 0.17 (0.13)
roe
0.01 (0.004)**
k
t-1
0.48 (0.06)** 0.39 (0.06)** 0.63 (0.12)** 0.34 (0.09)**
k
t-2
-0.10 (0.04)** -0.10 (0.04)**
Instrumented
variable
risk, roe, k
t-1
, k
t-2
risk, k
t-1
, k
t-2
risk, k
t-1
, k
t-2
risk, k
t-1
, k
t-2
Instruments (all
exogenous
and…)
2
nd,
3
rd
and 4
th
lag
of risk, k
t-1
and
roe
2
nd
3
rd
and 4
th
lag of risk, k
t-1
2
nd,
3
rd
and 4
th
lag
of risk, k
t-1
2
nd,
3
rd
and 4
th
lag
of risk, k
t-1
Sargan test
(3)
1.00 (291) 1.00 (239) 1.00 (242) 1.00 (242)
H
0
= No 1
st
order
autocorrelation
Prob > z
0.000 0.0004 0.13 0.03
H
0
= No 2
nd
order
autocorrelation
Prob > z
0.87 0.93 0.80 0.55
Observations 1162 954 304 305
Firms 65 53 16 33
What determines how much capital is held by UK banks and building societies?
57
Robust standard errors and the estimated coefficient from first step estimates are
reported. Column 1 reports the results obtained for the whole sample. Column 2
and 3 report the results obtained when firms that have experienced an increase
or a decline in requirements are considered, respectively. The last column shows
the results obtained with the observations corresponding to those building
societies whose buffer (difference between capital held and capital required) is in
the lowest quartile.
Table A4 shows the results obtained with the banks’ quarterly returns. As already
mentioned, some of them have changed their reporting pattern, some new firms have
appeared and some have stopped reporting. It is therefore an unbalanced panel.
All columns in Table A4 show first difference estimates, obtained using a GMM
estimator with lagged levels of the dependent variable and the endogenous
variables (second, third, fourth lags) and the lagged differences of the exogenous
variables as instruments. Column 1 show the results obtained for the whole
sample. Column 2 shows the results only for the firms that carry out trading book
activity. Columns 3 and 4 show the results for firms that have experienced an
increase and a decline, respectively in their regulatory capital requirements. 33
firms have experienced both an increase and a decrease in their requirements
during the period considered and so are included in both columns. Column 5
reflects the results obtained for banks that have the lowest buffer (those in the
lowest quartile).
None of the estimated equations presented in Tables A3 and A4 can reject the
Sargan test of over-identifying restrictions. They all show good properties in the
face of autocorrelation: as expected, they show first order autocorrelation, but
the hypothesis of no second order autocorrelation can be rejected in all cases.
What determines how much capital is held by UK banks and building societies?
58
TABLE A4: DETERMINANTS OF BANKS CAPITAL RATIOS (1998 Q3-2002 Q3)
(1)
Variable All sample Firms with
trading book
Firms that
experienced an
increase in kr
(2)
Firms that
experienced a
reduction in
kr
(3)
Firms with a
low buffer
(4)
kr
0.28 (0.14)** 0.43 (0.23)** 0.49 (0.23)** 0.19 (0.10)** 0.41 (0.15)**
size
-0.23 (0.06)** -0.38 (0.07)** -0.25 (0.04)** -0.26 (0.06)** -0.13 (0.06)**
risk
-0.22 (0.07)** -0.23 (0.05)** -0.17 (0.04)** -0.25 (0.07)** -0.12 (0.06)**
g
-0.04 (0.02)** -0.11 (0.04)** -0.08 (0.04)** -0.01 (0.02) -0.04 (0.02)**
peer
0.07 (0.02)** 0.06 (0.03)* 0.08 (0.03)** 0.06 (0.03)** 0.01 (0.01)
dep
0.02 (0.03) 0.02 (0.01)** 0.02 (0.01)* -0.03 (0.03) -0.02 (0.06)
trade
- -0.06 (0.03)** - - -
foreign
0.005 (0.005) 0.008 (0.007) -0.002 (0.007) 0.01 (0.007)* 0.0005 (0.006)
k
t-1
0.32 (0.09)** 0.30 (0.06)** 0.31 (0.07)** 0.24 (0.05)** 0.22 (0.07)**
Const 0.007
(0.003)**
0.006 (0.006) 0.01 (0.005)* 0.006 (0.004) -0.02
(0.004)**
Sargan test
(5)
0.93 (201) 1.00 (242) 1.00 (201) 1.00 (201) 1.00 (190)
H
0
= No 1
st
order
autocorrelation
Prob > z
0.000 0.005 0.009 0.0004 0.05
H
0
= No 2
nd
order auto-
correlation
Prob > z
0.92 0.99 0.75 0.78 0.47
Observations 2052 646 821 1180 403
Firms 182 59 72 97 76
(1) First difference regression. First-step robust standard errors in brackets. Risk, dep, trade, k
t-1
and k
t-2
have
been instrumented with their 2
nd
, 3
rd
and 4
th
lags and the lagged differences of the exogenous variable.
(2) Includes only firms that have experienced an increase in regulatory capital requirements. 35 of them
carry out trading book activity.
(3) Includes only firms that have experienced a decrease in regulatory capital requirement. 32 of them carry
out trade book activity.
(4) Includes only observations corresponding to an excess capital lower than 27.78 %, which corresponds to
the lowest quartile of the population.
(5) Prob >
2 (degrees of freedom).
(**) Statistically significant at 5%.
(*) Statistically significant at 10%.
What determines how much capital is held by UK banks and building societies?
59
We observe a clearly significant positive relationship between capital
requirements and observed capital ratios in the building societies and banks’
returns datasets, suggesting that the higher the required capital ratio the higher
the actual capital ratio. The value of the coefficient for the short term ranges
from 0.28 for all banks to 0.43 for banks engaged in trading activities. The long
term coefficient for these same groups of firms is estimated to be 0.41 and 0.61
respectively. For the whole building societies sample the short term coefficient is
0.18 and the long term 0.29. The figures in columns 3 and 4 of Table A4 suggest
that, when faced with an increase in the regulatory requirements, banks transfer
nearly 50% of the change into changes in the capital ratio in the short term. In
the long term it could reach 71%. When faced with a regulatory reduction, they
reduce their holdings just below 20% of the change in the short term (25% in
the long run). Building societies do not seem to have such an asymmetric
response.
We might also expect that firms whose actual capital ratio is closer to their
regulatory individual requirement would react more strongly to changes in capital
requirements, because of the regulatory pressure. To test this hypothesis, we
estimate the equation for various groups of firms defined according to their
buffer. However, only the results for the lowest quartile are reported (last column
of Tables A3 and A4). Both building societies and banks whose buffer is in the
lowest quartile, show an estimated coefficient for the regulatory capital much
higher than for the whole sample. Firms with a large buffer do not seem in
general to react to changes in the required capital: the coefficient of kr is either
not statistically significant or much lower than for firms whose capital is closer
to the regulator’s requirement. So capital requirements seem not to be a binding
constraint for those firms with a large enough buffer.
We find a statistically significant negative effect of size on capital holdings, so
the larger the firm the lower its capital ratio. This negative relationship could
reflect any of the various hypotheses that have been discussed in the main text.
The current specification does not allow for a unique interpretation.
The relationship between capital ratios and the proposed measure of ex-post risk
is estimated to be negative, so the higher the risk appetite of a firm, the less
What determines how much capital is held by UK banks and building societies?
60
capital it holds.
71
Evidence of a negative risk effect is also obtained in Ayuso et
al. (2004) and Lindquist (2004). This result could support the hypothesis that
there is moral hazard in firms’ behaviour. However, this interpretation is not
consistent with the positive relationship estimated between the ratio of partially
insured deposits over total deposits and capital holdings. It could be argued,
moreover, that riskier firms also have stricter systems and controls allowing them
to be comfortable with less capital for a given risk.
72
We tried including a multiplicative variable composed of risk and size and either
it is not statistically significant or it is positive, reflecting the compound effect
of the two variables.
We find a negative relationship between the economic cycle and the capital
ratio, the coefficient being statistically significant for banks. Not surprisingly, for
building societies the results are more mixed. So we can tentatively conclude
that the amount of capital held by firms depends negatively on GDP growth. We
can therefore expect that, during a contraction, banks would try to hold more
capital than in a boom. Thus we cannot find evidence of the buffer being pro-
cyclical.
We find evidence of peer pressure through the capital ratio. This suggests that
firms may be using the capital ratio as a signalling device and that the capital
ratio of similar firms is a relevant area for competition. Peer pressure seems to be
relevant for all the types of firm considered, although the coefficient is not very
high. Lindquist (2004) also finds evidence that competitors’ buffer capital weakly
affects the size of the firm’s buffer.
We estimate a negative relationship between the amount of assets in the trading
book and capital holdings. Again it could suggest that firms engaging in these
activities value risks differently, at least as regards the risks that could be
71
We have tried other proxies for risk such as the ratio of highest risk assets over risk weighted
assets or including the proportion of assets weighted at 50%. The coefficients are always
negative and statistically significant.
72
Rather similar results are obtained when we proxied risk by the ratio of Tier one capital over total
eligible capital and when we used other risk buckets.
What determines how much capital is held by UK banks and building societies?
61
reflected in capital. When we consider only firms engaging in trading activities,
the coefficient of capital requirements rises. It seems that this type of firm is
more responsive to regulatory pressure.
The coefficient of the opportunity cost of capital for building societies is
positive, when the dynamic specification is used. Profits are the main source for
building society capital and this dependency may explain the positive
relationship. For banks we have not been able to find a meaningful proxy.
73
For banks, the parents’ origin (i.e. whether it is UK or foreign) seems to have no
influence neither on the average capital ratio held nor on the reaction to
changes in capital requirements.
73
Following Lindquist (2004), we have tried including a market value of beta obtained from 8 large
banks. The coefficient is positive and not statistically significant.
FSA Occasional Papers in Financial Regulation
1 The Economic Rationale for Financial Regulation April 1999
David Llewellyn
2 The Rationale for a Single National Financial Services Regulator May 1999
Clive Briault
3 Cost-Benefit Analysis in Financial Regulation September 1999
Isaac Alfon and Peter Andrews
4 Plumbers and Architects: a supervisory perspective
on international financial architecture January 2000
Huw Evans
5 Household Sector Saving and Wealth Accumulation:
Evidence from balance sheet and flow of funds data February 2000
Iftikhar Hussain
6 The Price of Retail Investing in the UK February 2000
Kevin R James
7 Some Aspects of Regulatory Capital March 2000
Jeremy Richardson and Michael Stephenson
8Saving for Retirement May 2000
Malcolm Cook and Paul Johnson
9Past Imperfect? August 2000
The performance of UK equity managed funds
Mark Rhodes
10 A More Market Based Approach to Maintaining August 2000
Systemic Stability
David Mayes
11 CAT standards and Stakeholders September 2000
Paul Johnson
12 Some cost-benefit issues in financial regulation October 2000
David Simpson, Geoff Meeks, Paul Klumpes
and Peter Andrews (Editor)
13 Paying for pensions November 2000
Edward Whitehouse
63
What determines how much capital is held by UK banks and building societies?
14 Low inflation April 2001
Ed Harley and Stephen Davies
15 The Regulation of Funded Pensions - December 2001
A Case Study of the United Kingdom
E Philip Davis
16 Revisiting the rationale for a single national February 2002
financial services regulator
Clive Briault
17 The impact of fees and levies on non-networked February 2002
Independent Financial Adviser (IFA) firms
David O’Neill
18 To switch or not to switch, that’s the question September 2002
An analysis of the potential gains from switching
pension provider
Isaac Alfon
19 Losing interest: How much can consumers save by October 2002
shopping around for financial products?
Malcolm Cook, Fionnuala Earley, Jody Ketteringham
and Sarah Smith
20 Why some Insurers fail: Practical lessons from recent December 2002
cases in Europe
William McDonnell
21 Stopping short: Why do so many consumers stop January 2004
contributing to long-term saving policies?
Sarah Smith
Available from
FSA Publications
Financial Services Authority
25 The North Colonnade
Canary Wharf
London E14 5HS
Tel 0845 608 2372
and on the FSA website www.fsa.gov.uk
64
What determines how much capital is held by UK banks and building societies?
The Financial Services Authority
25 The North Colonnade Canary Wharf London E14 5HS
Telephone: +44 (0)20 7066 1000 Fax: +44 (0)20 7066 1099
Website: http://www.fsa.gov.uk
Registered as a Limited Company in England and Wales No. 1920623.
Registered Office as above.
ISBN 1-84518-198-0
... The approach, adopted in our study, complements the previous researches on the effect of capital buffers on banks' risk incentive (Alaeddin et al., 2017;Alfon et al., 2004;Aysan et al., 2015;Ayuso et al., 2004;Flannery & Rangan, 2006;Ghosh, 2017;Lindquist, 2004;Nier & Baumann, 2006). Using a cross-country panel data set-up from the MENA region, we examine and compare empirically whether or not and; to what extent market discipline exerted by Investment account holders (for Islamic banks) and banks depositors (for conventional banks) can be an effective incentive for banks to maintain capital buffers to limit the risks associated to their portfolios arising from adverse outcomes. ...
... Previous studies also show that banks' size and capital level are negatively correlated (Alfon et al., 2004;Ayuso et al., 2004;Flannery & Rangan, 2006). Indeed, larger banks can expand their lending across business lines and regions, which helps decrease the idiosyncratic shocks for these banks and reduce the need for a capital buffer. ...
Article
This paper examines the effectiveness of market discipline in motivating banks to mitigate their default risk by using capital buffers against adverse effects in portfolio risk. Using a large sample of 126 banks from 9 MENA countries from 2009 -2014, we study whether support, funding, and risk disclosure motivate banks to hold larger capital buffers to cover potential portfolio risk or not; and whether it affects differently Islamic and conventional banking sectors. Our paper findings provide evidence of the contradictory effects of these factors on Islamic banks compared to conventional banks. While government support results in higher capital buffers, stronger market discipline resulting from uninsured liabilities and disclosure results in larger capital buffers. Most factors look effective only for the conventional banking industry and ineffective for the Islamic banking industry due to its risk-sharing characteristic. Our paper findings provide evidence of the effectiveness of market discipline mechanisms, especially for the conventional banking industry. Moreover, high government support reinforces the disclosure and uninsured funding effects, especially for conventional banks. Finally, competition minimizes bank's risk incentives.
... Anginer and Demirgüç-Kunt (2014) report a negative association between capital measures and several indicators of a bank's risk based on a large sample of banks in 65 countries. Examining the UK banks, Alfon et al. (2004) also find a negative relationship between capital and risk-taking from 1998 to 2003. Similar results are reported by Klomp and de Haan (2012) for a sample of banks in 70 developing and emerging countries and Lee and Hsieh (2013) studying banks in 42 Asian countries. ...
Article
Full-text available
This article investigates the impact of capital requirements and market competition on the stability of financial institutions in the Middle East and North African (MENA) region. We test the hypothesis that capital requirements significantly affect the risk behaviour of both Islamic and conventional banks in the MENA region. We also investigate the moderating effect of market power and concentration on the relationship between capital regulation and bank risk. We find that capital ratio has a strong positive impact on conventional banks’ credit risk, whereas this effect is insignificant in the sample of Islamic banks. Our analysis indicates that, for the conventional banking sector, the increase in the capitalization level is negatively linked to bank credit risk only when banks’ level of market power is high. Regarding the Islamic banks’ behaviour, we find that the relationship between capital and credit risk is weakly moderated by banking competition. This means that Islamic banks are less sensitive to the market’s competitive conditions in the MENA countries, as they still apply their theoretical models, based on prohibition of interest. Our findings inform regulatory authorities concerned with improving the banking sector’s financial stability in the MENA region to strengthen their policies and force banks to better align with regulatory capital requirements during the COVID-19 pandemic.
... Regulatory variables; capital requirements -CAPRQ, under the regulatory paradigm, high capital provides more confidence to customers leading to an increase in spending and investments. These are examined beside the risk analysis and the capital shortfall measure to investigate the procyclicality effect of capital requirements (Alfon et al., 2004) for UK market and Ayuso et al., 2004, Lindquist, 2004, Fonseca and Gonz alez, 2010, Stolz and Wedow, 2011, for rest of Europe and the world). Supervisory power -SPOWER examines the strength of the regulator's supervision power and their influence in the decision of mergers approvals. ...
Article
Full-text available
Purpose The purpose of this paper is to examine the effects of bank mergers on systemic and systematic risks on the relative merits of product and market diversification strategies. It also observes determinants of M&A deals criteria, product and market diversification positioning, crisis threshold and other regulatory and market factors. Design/methodology/approach This research examines the impact and association between merger announcements and regulatory reforms at bank and system levels by investigating the impact of various bank consolidation strategies on firms’ risks. We estimate beta(s) as an index of financial institutions’ systematic risk. We then develop an index of the estimated equity value loss as the long-rum marginal expected shortfall (LRMES). LRMES contributes to compute systemic risk (SRISK) contribution of these firms, which is the capital that a firm is expected to need if we have another financial crisis. Findings Large acquiring banks decrease systemic risk contribution in cross-border M&As with a non-bank financial institution, and witness profitability (ROA) gains, supporting geographic diversification stability. Capital requirements, activity restrictions and bank concentration increase systemic risk contribution in national mergers. Bank mergers with investment FIs targets enhance productivity but impair technical efficiency, contrary to bank-real estate deals where technical efficiency change accompanied lower systemic risk contribution. Practical implications Financial institutions are recommended to avoid trapped capital and liquidity by efficiently using local balance sheet and strengthening them via implementing models that clearly set diversification and netting benefits to determine capital reserves and to drive capital efficiency through the clarity on product–activity–geography diversification and focus. This contributes to successful ringfencing, decreases compliance costs and maximises returns and minimises several risks including systemic risk. Social implications Policy implications: the adversative properties of bank mergers in respect of systemic risk require strict and innovative monitoring of bank mergers from the bidding level by both acquirers and targets and regulators and competition supervisory bodies. Moreover, emphasis on regulators/governments intervention and role, as it provides a stabilising factor of the markets and consecutively lower systemic risk even if the systematic idiosyncratic risk contribution was significant. However, such roles have to be well planned and scaled to avoid providing motives for banks to seek too-big-too-fail or too-big-to-discipline status. Originality/value This research contributes to the renewing regulatory debate on banks sustainable structures by examining the risk effect of bank diversification versus focus. The authors aim to address the multidimensional impacts and risks inherent to M&A deals, by examining the extent of the interconnectedness of M&A and its implications within and beyond the banking sector.
... Heid et al. (2003) argue that banks hold a higher amount of capital than their regulatory capital ratio faces lower problems of normal banking activities in crisis conditions. Alfon et al. (2004) conclude a negative relationship between capital and risk in the U.K Banks. Van Roy (2005) reports an inverse relationship between bank capital ratios and risk-taking, and the finding is in line with (Lee & Hsieh, 2013). ...
Article
Full-text available
This study aims to explore how different capital ratios influence the risk-taking of large commercial banks of the USA. The study collects the data from FDIC for commercial banks from 2003 to 2019. We use a two-step GMM method to manage the endogeneity, simultaneity, heteroscedasticity, and auto-correlations issue. The findings conclude that an increase in the risk-based capital ratios decreases the banks’ risks. Empirical findings demonstrated a significant and positive association between non-risk-based capital ratios and bank risk-taking. The findings also demonstrate that an increase in capital buffer ratios decreases the banks’ risks. The impact of capital ratios on risk-taking is heterogeneous for well and under-capitalized banks. The findings suggest that State-chartered member and non-member banks are inclined to take a higher risk than nationally chartered banks. The findings have implications for regulators to consider the State-chartered member, non-member, and nationally chartered banks while formulating the new guidelines for required capital ratios.
... While considering the competitive market and apart from capital regulation, the market discipline also motivates the banks in developing countries to manage the level of capital adequacy. In addition, Barrios and Blanco (2003), Alfon et al. (2004), and Gropp and Heider (2008) reveal a positive and significant effect of market discipline on capital adequacy ratio. According to Bliss and Flannery (2002), external sanctions on banks are imposed because of market discipline. ...
Article
Full-text available
We aim to analyze the impact of capital regulation and market discipline on capital to risk-weighted assets ratio. We used the panel data of Asian developing-countries banks for the period from 2009 to 2018. We collected data from the financial statements of 73 banks of Pakistan, Jordan, Indonesia, the Philippines, Saudi Arabia, and Thailand. We used the generalized method of moment (GMM) to analyze the results. We find that capital regulation and market disciplines significantly influence the capital ratio in Asian developing countries.
Article
Full-text available
This paper examines the impact of monetary policy instruments such as discount rate, reserve requirement and liquidity requirement on bank credit to the private sector in the Kingdom of Eswatini. Monthly data sourced from the Central Bank of Eswatini and Eswatini Central Statistics Office is used for the period January 2000 to December 2017. Using the Johansen cointegration test and Vector Error Correction Model, our results show that: there is one cointegration in the model and the current levels of the discount rate, reserve requirement and liquidity requirement bear a negative and significant effect to bank credit to the private sector. This indicates that the three instruments are not supportive to economic growth and they are not optimal to stimulate credit. Applying the Quadratic approach, findings of this study reveal that the optimal monetary policy mix (thresholds) to stimulate bank credit to the private sector while maintaining inflation within reasonable levels is 5.42% for the discount rate, 4.03% for the reserve requirement, 12.48% for the liquidity requirement. The study recommends that the Central Bank of Eswatini should always take into account the existence of trade-off in monetary policy instruments results in order to stimulate bank lending to the private sector.
Article
Purpose This paper aims to investigate the impact of regulation and market competition on the risk-taking Behaviour of financial institutions in the Middle East and North Africa (MENA) region. Design/methodology/approach The empirical framework is based on panel fixed effects/random effects specification. For robustness purpose, this study also uses the generalized method of moments estimation technique. This study tests the hypothesis that regulatory capital requirements have a significant effect on financial stability of Islamic and conventional banks (CBs) in the MENA region. This study also investigates the moderating effect of market power and concentration on the relationship between capital regulation and bank risk. Findings The estimation results support the view that capital adequacy ratio (CAR) has no significant impact on credit risk of Islamic banks (IBs), whereas market competition does play a significant role in shaping the risk behavior of these institutions. This study report opposite results for CBs – an increase in the minimum capital requirements is followed by an increase in a bank’s risk level, which has a negative impact on their financial stability. Furthermore, the results support the notion of a non-linear relationship between banking concentration and bank risk. The findings inform the regulatory authorities concerned with improving the financial stability of banking sector in the MENA region to set their policy differently depending on the level of concentration in the banking market. Research limitations/implications This study contributes to the literature on the effectiveness of regulatory reforms (in this case, capital requirements) and market competition for bank performance and risk-taking. In regard to IBs, capital requirements are less effective in requiring IBs to adjust their risk level according to the Basel III methodology. This study finds that IBs’ risk behavior is strongly associated with market competition, and therefore, the interest rates. Moreover, banks operating in markets with high banking concentration (but not necessarily, low competition), will decrease their credit risk level in response to an increase in the minimum capital requirements. As a result, these banks will be more stable compared to their conventional peers. Thus, regulators and policymakers in the MENA region should restrict the risk-taking behavior of IBs through stringent capital requirements and more intense banking supervision. Practical implications The practical implications of these findings are that the regulatory authorities concerned with improving banking sector stability in the MENA region should proceed differently, depending on the level of banking market concentration. The findings inform regulators and policymakers to set capital requirements at levels that would restrict banks from taking more risk to increase their returns. They are also important for bank managers who should avoid risky strategies in response to increased regulatory pressure (e.g. increase in the minimum required capital level of 8%), as they may lead to an increase in the level of non-performing loans, and therefore, a greater probability of bank default. A future extension of this study will focus on testing the effect of bank risk-taking and market competition on the capitalization levels of banks in the MENA countries. More specifically, this study will investigates if banks raise their capitalization levels during the COVID-19 pandemic. Originality/value The analysis of previous research indicates that there is no unambiguous answer to the question of whether IBs perform differently than CBs under different competitive conditions. To fill this gap, this study examines the influence of capital regulation and market competition (both individually and interactively) on bank risk-taking behavior using a large sample of banking institutions in 18 MENA countries over 14 years (2005–2018). For the first time in this line of research, this study shows that the level of market power is positively associated with the level of a bank’ insolvency risk. In others words, IBs operating in highly competitive markets are more inclined to take a higher risk than their conventional peers. Regarding the IBs credit risk behavior, this study finds that market power has a limited impact on the relationship between CAR and risk level. This means that IBs are still applying in their operations the theoretical models based on the prohibition of interest.
Article
We examine how bank-specific capital requirements affect banks’ capital ratios and balance sheet composition over the period 1989-2013. We find that higher requirements, even when not binding in a regulatory sense, result in higher bank capital ratios, achieved by a combination of asset contraction, risk reduction and capital raising strategies. We find that after the 2007-09 financial crisis, banks place greater emphasis on raising capital, but that they continue to focus first on raising lower-quality capital. Our results have implications for evaluating the efficacy of bank-specific capital requirements as a policy tool to influence bank balance sheet behavior and resilience.
Article
Full-text available
Critical requirement imposed by regulatory authorities on financial institutions and banks is sufficient and adequate capital. State Bank of Pakistan imposed capital adequacy ratio (associated risk to its assets) requirement on microfinance banks to ensure its stability. This study focuses on determinants that influence capital adequacy in Pakistani microfinance banks for the period 2011-2015. Analysis of observations considers 12 microfinance banks operating under State Bank of Pakistan. Five factors have been considered in current study that is Share of Deposit, Return on Equity, Portfolio Risk, GDP, and Total Assets. The results indicate that total assets to capital adequacy shows insignificant and negative relationship (r=-.008, sig=.953) while the other four factors indicate positive and significant relationship. The results of multiple regression indicate total assets ( =-0.823, t=-17.200, p=0.000) and growth rate ( =-0.766, t=-5.112, p=0.000) having negative and significant impact, however, share of deposit shows insignificant and negative impact on capital adequacy ( =-0.039, t=-2.446, p=0.01), while the other factors have positive and significant impact on capital adequacy ratio.
Article
Full-text available
This study examines whether the capital requirements under Basel III are effective in enhancing the profitability and efficiency of the banking sector. Drawing on a sample of the largest commercial banks from the UK and Australia over the period from 2000 to 2019, we employ the FMOLS (Fully Modified OLS) and DOLS (Dynamics OLS) estimation approaches. The results indicate that stricter capital ratio increases operating earnings, whist it fails to improve bank profitability and bank efficiency. Our findings cast doubts on the effectiveness of tax policy in the observed banks. Further empirical testing shows an optimal capital structure in which the banks can achieve the best performance. Interestingly, these optimal ratios are broadly in line with the minimum common equity ratio required under Basel-III. The macroeconomic outlook also contributes to the performance of British and Australian banks. British banks are found to perform well under inflationary pressure and higher policy rates, contrary to the Australian banks whose performance deteriorates. Our results hold across different samples, efficiency, and profitability measures and various estimation models.
Article
Full-text available
The Basel committee proposes making regulatory capital requirements more risk sensitive. Cost-benefit assessment suggests that this yields greater efficiency in the use of regulatory capital, but has substantial enforcement and compliance costs and may well increase the severity of banking crises. Better if the new Accord sets risk-insensitive regulatory minimum capital standards and encourages banks to set risk-sensitive desired capital targets with generous reductions in required capital for healthy banks with effective systems of risk management. Ten specific suggestions for improvement of the new Accord are made.
Article
Foreword The FSA is committed to encouraging debate among academics, practitioners and policy-makers in all aspects of financial regulation. To facilitate this, it is publishing a series of occasional papers in financial regulation, extending across economics and other disciplines. These papers will cover such topics as the rationale for regulation, the costs and benefits of various aspects of regulation, and the structure and development of the financial services industry. Since their main purpose will be to stimulate interest and debate, we welcome the opportunity to publish controversial and challenging material and to publish papers that may have been presented or published elsewhere. The main criterion for acceptance of papers, which will be independently refereed, is that they should make substantial contributions to knowledge and understanding in the area of financial regulation. We will encourage contributions from external authors as well as from within the FSA. In either case, the papers will express the views of the author and not necessarily those of the FSA. Comments on these papers are welcomed and should be addressed to the series editors.
Article
This paper examines differences in stock price reactions following voluntary capital injections by commercial banks and involuntary capital injections required to meet regulatory capital requirements. Empirical tests document that stock price declines associated with voluntary common stock issues are significantly greater than those associated with involuntary common stock injections, consistent with Ross (1977). Empirical test also confirm that for both voluntary and involuntary stock issuances, the abnormal stock price reaction is negatively related to the relative size of the offering and positively related to managerial ownership prior to the security issue, although these relationships are stronger for voluntary issues.
Article
This paper addresses three linked questions, in three separate sections. The first section, "The Structure of Capital Requirements", asks what the purpose is of regulatory capital requirements for different types of financial services firms, and attempts to build a coherent theoretical framework to reflect the answer to that question. It starts from first principles on the basic rationale for regulation, and describes the various ways in which regulators can use capital requirements to help meet the objectives of regulation. It suggests a broad categorisation of financial services firms into four types, broadly aligned to the importance of preventing their failure, and the impact if failure were to occur. It then proposes different forms of capital requirement for each of those categories and discusses the appropriate mixture of capital with other tools of regulation. The second section, "Setting Differentiated Capital Ratios", aims to shed light on the question of how capital can best be used as one of the tools of regulation. It describes the FSA's current approach to setting capital requirements for UK-incorporated banks, which involves specifying 'trigger' and 'target' capital adequacy ratios tailored to individual banks' risk profiles. It summarises how banks' current actual capital ratios compare with their target ratios set by the FSA, and discusses the reasons for the pattern seen. And it draws out the implications for initiatives to revise capital requirements for banks. The third section, "Capital Requirements for Cross-Border Claims", looks at one example of the impact that capital requirements can have. From the Basel Committee's June 1999 proposals for reforming its Capital Accord, the paper takes the proposal that the risk-weighting of cross-border lending could be linked to the credit rating of the borrower's country of residence. Looking at the periods of crisis surrounding East Asia and Russia, this section estimates how the new proposals would have changed banks' capital requirements for lending to some major countries that underwent rating changes over the period. It attempts to shed some light on the question of 'procyclicality', the concern that increasing risk weights when a borrower is downgraded could add to the volatility of international capital flows. The paper finds that the amount of extra capital required by BIS-area banks in this case would have been very small compared to their total capital: it does not therefore provide strong support for concerns about procyclicality.
Article
We analyse the incentive impact of bank capital regulation in a model with endogenous capital, assuming regulators randomly audit banks and require undercapitalised banks either to bear the fixed cost of new issue or to liquidate. Forward looking banks with sufficient franchise value maintain a buffer of capital in excess of the regulatory minimum. In our dynamic setting we show, amongst other results: that incentives for risk taking depend upon this buffer of free capital, not the total level; and that the regulatory capital requirement has no long run effect on bank risk-taking.
Article
This paper contrasts the conventional interpretation of prudential capital regulation as a system of ex ante enforcement (`hard wired') with an alternative interpretation as a system of sanctions for ex post violation (`incentive based'). Under the incentive based interpretation risk-weightings affect portfolio choice only when assets are illiquid. Under both interpretations, the medium term impact on portfolio allocation depends upon the relative costs of debt and equity finance. Viewing enforcement as incentive based suggests there is relatively less need to match risk weightings accurately to portfolio risk.
Article
This paper examines empirically the hypothesis that market discipline is effective in providing incentives for banks to limit their risk of default, by holding capital buffers against adverse outcomes in portfolio risk. We have constructed a large cross-country panel data set consisting of observations on 729 individual banks from 32 different countries over the years 1993 to 2000. Theory implies that the strength of market discipline ought to be related to the extent of the government safety net, the observability of bank risk choices and to the proportion of uninsured liabilities in the bank's balance sheet. Using panel data techniques, we test whether these factors provide incentives for banks to hold larger capital buffers against adverse outcomes in portfolio risk. Our results suggest that government safety nets result in lower capital buffers and that stronger market discipline resulting from uninsured liabilities and disclosure results in larger capital buffers, all else equal. While our results therefore point to the effectiveness of market discipline mechanisms in general, we also find that the effect of disclosure and uninsured funding is reduced when banks enjoy a high degree of government support. Our results finally suggest that while competition leads to greater risk taking incentives, market discipline is more effective in curbing these incentives in countries where competition among banks is strong.
Article
Only recently the debate on bank capital regulation has devoted specific attention to the role that bank loan loss provisions can play as a part of the overall capital regulatory framework. This paper contributes to the ongoing debate by demonstrating empirically that loan loss provisioning needs to be an integral component of capital regulation. We find empirical evidence that many banks around the world delay provisioning for bad loans until too late, when cyclical downturns have already set in, thereby magnifying the impact of the economic cycle on banks' income and capital.