the impact of the 1988 basel accord on banks’ capital ratios and … · the context of the basel...
TRANSCRIPT
1
The impact of the 1988 Basel Accord on banks’ capital ratios
and credit risk-taking: an international study 1
Patrick Van Roy (∗)
European Center for Advanced Research in Economics and Statistics (ECARES),
Université Libre de Bruxelles
Av. F.D. Roosevelt 50, CP 114 1050 Brussels, Belgium
This draft: December 2003
Abstract
The purpose of this paper is to see whether and how G-10 banks have complied with the 1988 Basel Capital Accord. The interest of this study lies in the fact that the standardized approach of the New Basel Accord is similar to the 1988 agreement. However, very little is known about the reaction of non-US banks to the imposition of fixed minimum capital requirements. Building on previous studies, this paper uses a simultaneous equations model to analyze adjustments in capital and portfolio risk at banks from G-10 countries over the period 1988-95. The results suggest that banks close to the Basel standards have generally increased their capital adequacy ratios without any offsetting increases in portfolio risk. In addition, a negative correlation is found between changes in capital and risk ratios suggesting that, everything else being equal, the capital build-up of the early 1990s was associated with a decrease in the credit risk-taking of G-10 banks. Keywords: Capital requirements; Credit risk; Bank regulation JEL classification: G21; G28
1 I would like to thank participants at the 10th ENTER Jamboree (Tilburg University, The Netherlands), the 8th Belgian Financial Research Forum (Limburgs Universitair Centrum, Belgium), the 4th Conference of the International Economics and Finance Society – UK Chapter (City University, UK) as well as many colleagues from ECARES for helpful comments and suggestions on earlier drafts. The assistance of Bureau Van Dijk in providing the data used in this study is also gratefully acknowledged. (∗) Research Fellow, Fonds National de la Recherche Scientifique. Tel: + 322 650 34 69; Fax: + 322 650 33 69; E-mail address: [email protected].
2
1. Introduction
One of the major developments undergone by the banking industry over recent years has been
the worldwide implementation of the so-called Basel Accord that sets minimum capital
standards for internationally active banks. The Basel guidelines were originally adopted by
the central banking authorities from 12 countries (all G-10 countries plus Luxembourg and
Switzerland) in July 1988. Their implementation started in 1989 and was completed 4 years
later, in 1993. The purpose of the Accord was twofold. First, it aimed at creating a level
playing field among banks by raising capital ratios, which were generally perceived as too
low in many countries. Second, and linked to this, it aimed at promoting financial stability by
adopting a relatively simple approach to credit risk with the potential to distort incentives for
bank risk-taking.
Fifteen years after the adoption of the Basel Accord and whilst new regulatory guidelines are
being designed, it is fair to say that empirical research has failed to answer the following
questions: was the 1988 agreement effective in raising capital ratios among banking
institutions falling below the minimum requirements? How did banks respond to the capital
adequacy rules, i.e. did they increase their level of capital, forgo risky projects or sell off
assets? Did the new guidelines induce banks to modify the credit risk of their portfolio and if
so, to which risk categories did they reallocate their assets? Analysis of how G-10 banks have
responded to the 1988 risk-based standards is of course crucial if one wants to gain insight
into the likely implications of the New Basel Capital Accord.
The lack of answers to the questions raised above is largely due to the limited amount of data
on capital levels and risky assets of G-10 banks. Indeed, these data are mainly confidential or
hard to obtain on a standardized cross-country basis. Most studies focus on the US, while
evidence has remained scarce for other countries that were part of the Accord.
Therefore, an important contribution of this paper is to shed further light on the impact of the
1988 Basel Accord using data from 7 different countries: Canada, France, Italy, Japan,
Sweden, the UK and the US. More precisely, I extend the simultaneous equations model
developed by Shrieves and Dahl (1992) to a multi-country setting in order to analyze the
relationship between changes in capital ratios and credit risk at the G-10 level. The model
also allows for cross-country comparisons of undercapitalized banks’ behavior towards
capital and risk.
The results show that changes in the capital and credit risk ratios of G-10 banks were
negatively related over the 1988-95 period. In addition, the findings indicate that banks close
to the Basel minimum requirements increased their capital ratios (except in France and Italy),
whereas there is only weak evidence for a rise in their credit risk-taking. All in all, this
3
suggests that the beginning of the 1990s was a period of increased stability for the G-10
banking industry.
The remainder of the paper is organized as follows. Section 2 reviews the theoretical and
empirical literature dealing with the effects of capital requirements on banks’ portfolio risk, in
the context of the Basel Accord. Section 3 presents the data used in this study, while section 4
outlines the empirical methodology. Results are discussed in section 5 and some conclusions
are drawn in section 6.
2. Bank capital regulation and its impact on banks’ risk-taking in theory and in
practice
2.1. Theoretical review
The main justification for regulating bank capital is the need to avoid the risk-shifting
incentive generated by improperly priced deposit insurance. Indeed, although it may promote
financial stability in the short-run, risk-insensitive deposit insurance tends to reduce banks’
incentives to maintain adequate capital and may thus endanger stability in the long-run. The
ability of capital standards to successfully eliminate this “moral hazard” problem, thereby
reinforcing the effect of a higher capital ratio on default risk, has been at the heart of a
theoretical debate for more than 20 years.
A first strand of the literature focuses on utility-maximizing banks using the portfolio
approach of Pyle (1971) and Hart and Jaffee (1974). In this framework, Koehn and
Santomero (1980) show that the introduction of higher leverage ratios will lead banks to shift
their portfolio to riskier assets and that the reshuffling effect will be larger for institutions
which initially held relatively more risky assets per unit of capital. This effect occurs because
flat requirements restrict the banks’ risk-return frontier, which leads them to compensate the
loss in utility from the upper limit on leverage with the choice of a riskier portfolio. One way
to eliminate the risk-shifting incentive is to require banks to meet risk-related capital ratios, as
suggested by Kim and Santomero (1988).
These conclusions have been questioned on several grounds. Using an option model, Furlong
and Keeley (1989) and Keeley and Furlong (1990) find that higher capital requirements
reduce the incentives for a value-maximizing bank to increase asset risk. They contend that
the utility-maximization framework, which reaches opposite conclusions, is inappropriate
because it does not adequately describe the bank’s investment opportunity set by neglecting
the option value of deposit insurance and the possibility of bank failure. Within the same
4
modeling framework, Gennotte and Pyle (1991) relax the assumption that banks invest in zero
net present value assets and find that there are now plausible situations in which an increase
in capital requirements results in an increase of asset risk.
The portfolio approach is used again by Rochet (1992) who shows that when the limited
liability of banks is taken into account, insufficiently capitalized banks may exhibit risk-
loving behavior even if regulation makes use of a risk-related capital ratio. As a consequence,
it may be necessary to impose an additional requirement, in the form of a minimum capital
level.
More recently, Blum (1999) also finds that capital regulation may increase banks’ riskiness
but in a dynamic framework. Using a two-period model, he shows that an intertemporal effect
has to be considered in addition to the standard negative effect of capital regulation on credit
risk. If banks find it too costly to raise additional equity to meet new capital requirements
tomorrow or are unable to do so, they will increase risk today. This second effect will
reinforce the well-known risk-shifting incentive due to the reduction in profits.
In short, economic theory is unclear on whether imposing harsher capital requirements leads
banks to increase or decrease the risk structure of their asset portfolio. Answering this issue is
important as the probability of default of a bank depends positively on its leverage and on its
portfolio risk. If capital requirements increase asset risk, their net effect on the probability of
bankruptcy, hence on financial distress, will be ambiguous.
The next subsection attempts to clarify the debate about the risk effects of capital regulation
by briefly restating the key rules of the 1988 Basel Accord and observing how banks can
comply with them.
2.2. Capital requirements in practice: The 1988 Basel Accord
The 1988 Basel standards are almost entirely focussed on credit risk,2 the risk of loss due to
borrower or counterparty default. An amendment to incorporate market risk has been issued
in 1996, while the Basel Committee on Banking Supervision has submitted a proposal for a
New Basel Accord in 1999. This proposal is based on three mutually reinforcing pillars
(minimum capital requirements – supervisory review – market discipline) that allow banks
and supervisors to evaluate additional types of risks, like operational risk and interest rate
risk, thereby avoiding treating portfolio risk in isolation. Implementation of the New Accord
(though probably not by all adopting countries) is expected by year-end 2006.
2 The terms “credit risk” and “portfolio risk” are used interchangeably throughout the paper.
5
The key to the 1988 Basel Accord is the obligation for internationally active banks to
continually meet two capital adequacy ratios, the so-called tier 1 and total capital ratios. Both
ratios have the same denominator, which is a risk-weighted sum of banks’ on-balance and off-
balance sheet activities. A simplified formula of the risk-weighted assets (RWA) of a bank is
given by:3
RWA=0*(bucket 1) + 0.2*(bucket 2) + 0.5*(bucket 3) + 1.0*(bucket 4) [1]
where bucket 1 consists of assets with zero default risk (e.g. cash, government bonds /
securities), bucket 2 of assets with a low rate of default (e.g. loans to OECD banks), bucket 3
of medium-risk assets (essentially residential mortgage loans) and the remaining assets (in
particular loans to non-banks) fall into bucket 4. The denominator of both capital adequacy
ratios thus represents the accounting value of banks’ assets adjusted for their individual risk.
The tier 1 ratio and the total capital ratio differ by their numerator. The numerator of the
former ratio consists only of tier 1 capital while the numerator of the latter ratio includes both
tier 1 and tier 2 capital. Tier 1 capital, also called “core capital”, consists mainly of
stockholder equity capital and disclosed reserves whereas tier 2 capital or “supplementary
capital” includes elements like undisclosed reserves and subordinated term debt instruments
(provided that their original fixed term to maturity does exceed five years). The difference
between tier 1 and tier 2 capital thus reflects the degree to which capital is explicit or
permanent. Total capital is the sum of tier 1 and tier 2 capital.
The 1988 capital adequacy guidelines require banks to have a tier 1 ratio of at least 4% and a
total capital ratio of at least 8% with the contribution of tier 2 capital to total capital not
exceeding 50%, i.e., the following inequalities must hold:
Tier 1 ratio = Tier 1 capital / RWA ≥ 0.04 [2]
Total capital ratio = Total capital / RWA = (Tier 1 capital + Tier 2 capital) / RWA ≥ 0.08 [3]
Tier 1 capital ≥ Tier 2 capital [4]
The regulation also limits general loan-loss reserves and subordinated debt that can be eligible
for inclusion in tier 2 capital (see Appendix 1). The implementation of the Basel guidelines in
G-10 countries occurred in two steps. Interim standards of 7.25% for the total capital ratio and
3 Strictly speaking, formula [1] is only valid for on-balance sheet assets. Off-balance sheet items are also assigned to four risk buckets but they involve additional weights reflecting the nature of the operation. See Dewatripont and Tirole (1994, pp. 52-53) for the precise regulatory definition of RWA.
6
3.25% for the tier 1 ratio had to be met by the end of 1990, whereas full compliance with the
definitive standards was expected by year-end 1992.
Banks that wish to raise their capital adequacy ratio (either to obey the minimum
requirements or for other non-regulatory reasons) can use three types of balance-sheet
adjustments. They can increase their capital level (depending on the regulatory ratio
concerned, this can be done in several ways), decrease their risk-weighted assets as a
proportion of total assets and / or decrease their total assets. This is summarized in equation
[5], which decomposes the growth rate of the capital adequacy ratio into three terms: the
growth rate of capital, the growth rate of the credit risk ratio and the growth rate of assets - a
proof is given in Appendix 2:
t t t t
t t t t
CAR K RISK A= - -
CAR K RISK A∆ ∆ ∆ ∆
[5]
where CAR = K / RWA = capital adequacy ratio (tier 1 ratio or total capital ratio)
K = capital (tier 1 capital or total capital)
RISK = RWA / A = credit risk ratio
A = total assets
t denotes time
From equation [5], it can be seen that a (mandatory) increase in the capital adequacy ratio
does not prevent banks from raising the credit risk of their portfolio provided that the growth
rate of the credit risk ratio is lower than the growth rate of capital.4 Thus, the Basel Accord,
which aimed at imposing a higher capital buffer against insolvency, may well have
encouraged banks to take on more credit risk, thereby having an ambiguous effect on their
probability of failure.
In the remainder of the paper, I focus on the relationship between changes in capital and risk
ratios.5 Furthermore, I analyze the behavior of banks which are close to the minimum capital
requirements, as they can be reasonably expected to have had a stronger and faster response
to the Basel Accord than better capitalized institutions.
4 Holding total assets constant. 5 Other strands of the empirical literature have looked at the effect of capital requirements on capital levels “K” and on total assets “A”. For a review of both types of studies, see Basel Committee on Banking Supervision (1999).
7
2.3. Review of the related empirical literature
The main papers that investigated the impact of the Basel capital requirements on banks’
credit risk-taking are listed in Table 1. With the exception of Ediz et al. (1998), all use the
simultaneous equations approach that is described in section 4. This modeling framework
allows to compare the behavior of undercapitalized and adequately capitalized banks with
respect to changes in capital and risk ratios and to see whether these changes are related.
(insert Table 1 here)
The studies surveyed in Table 1 generally support the idea that undercapitalized banks (i.e.,
those failing to achieve the Basel requirements) increased their capital adequacy ratios in the
first half of the 1990s. A similar phenomenon, although to a lesser extent, is observed for
adequately capitalized institutions. The latter result is consistent with an idea that will be used
when modeling regulatory pressure, namely that banks tend to be well above the minimum
requirements for precautionary and / or reputational reasons. Furthermore, there is little
consensus among the papers reviewed on whether banks – adequately capitalized or not –
engaged in riskier activities. Finally, changes in capital ratios and credit risk appear to be
mostly unrelated.
Results of US studies are difficult to interpret as the implementation of the second stage of the
Basel Accord, between end-1990 and end-1992, coincide with the passage of the Federal
Deposit Insurance Corporation Improvement Act (FDICIA) in December 1991. Section 131
of FDICIA, Prompt Corrective Action (PCA), goes one step further than the Basel Accord by
defining three regulatory ratios (the Basel capital standards plus a leverage requirement) and
five categories into which banks are classified according to their compliance with the three
ratios. Thus, it is hard to ascribe the findings of the two papers by Aggarwal and Jacques
(1997, 2001) to the Basel Accord as opposed to FDICIA, as US banks’ behavior is likely to
have been affected by both regulations over the period that they consider. However, this is not
a too big problem to the extent that one is interested in the effects of capital regulation in a
broad sense. The study by Jacques and Nigro (1997) deals exclusively with the consequences
of the Basel Accord, as it concentrates on the years 1990-91, i.e. the period before FDICIA
was passed. However, the very small number of undercapitalized institutions in Jacques and
Nigro’s sample - less than 2% of the total number of banks - may reduce the reliability of
some of their estimates.
8
Two papers present some non-US evidence regarding the relationship between capital ratios
and credit risk. Ediz et al. (1998) base their study on confidential UK data whereas Rime
(2001) uses Swiss data. Like other studies listed in Table 1, the first of these two papers uses
a partial adjustment framework but, unlike them, treats changes in capital and risk-taking as
two separate decisions. Unfortunately, Ediz et al.’s model leads to the puzzling result that
banks adjust their capital levels each year by more than the difference between the current
level and the target they have in mind, which means that banks overshoot the target (and by a
higher amount each year). The study by Rime (2001) is interesting because it provides the
first application of the simultaneous-equations model reviewed in section 4 to non-US banks.
However, Rime adopts the PCA regulatory classification to measure regulatory pressure on
Swiss banks, which might be inappropriate given that the additional requirements set by PCA
have not been adopted formally by any other country besides the US.6
Finally, a paper by Sheldon (1996) – not listed in Table 1 – looks at the risk effects of capital
adequacy on eleven G-10 countries using an option-pricing framework. Sheldon’s main result
is that the Basel Accord did not have a risk-increasing impact on banks’ portfolio but is a bit
difficult to interpret, as he is does not control for regulatory and non-regulatory influences.
Moreover, his sample is not always representative of the banking industry of each G-10
country, as some countries are only represented by very small banks.
Therefore, the main contribution of this study is to extend the empirical literature on the
effects of the Basel Accord using a simultaneous-equations model for seven different
countries along with a representative data set, the construction of which is detailed in the next
section.
3. Data sources and description
3.1. Sources
Data were obtained from Bankscope, a database of bank account figures which is a joint
product of Fitch Ratings and Bureau Van Dijk, a major rating agency and a publisher of
financial databases on CD-ROM, respectively. Consistent with most studies on the impact of
6 Rime also uses a regulatory pressure variable relatively similar to mine, which does not alter his results.
9
the Basel requirements, I chose to restrict the sample to commercial banks over the 1988-95
period.7
3.2. Methodology
All the variables used in this study were available on Bankscope, except the credit risk ratio
of banks. Therefore, capital adequacy ratios (CAR), capital levels (K) and total assets (A)
were extracted from the database in order to compute the credit risk ratio (RISK) using the
following formulas:
K K= = RWA
KCARRWA
[6]
RWA= RISK
A [7]
Equation [6] can be computed in two different ways, either using tier 1 data (i.e., K = tier 1
capital and CAR = tier 1 ratio) or using total capital data (i.e., K = total capital and CAR =
total capital ratio). Obviously, both methods yield the same value for risk-weighted assets. In
order to check for consistency of the results, I computed the credit risk ratio using both
methods except for Japanese banks, which generally do not report their total capital.
3.3. Sample Statistics
Using the data from Bankscope, I constructed an unbalanced panel containing 586
commercial banks from seven G-10 countries with assets of more than $ 100 million during
the period 1988-95. Banks that did not report their total capital ratio, their tier 1 ratio or their
credit risk ratio for at least 2 consecutive years were omitted from the data set. Also, banks
with a capital ratio above 50% or a credit risk ratio above 200% were treated as outliers and
excluded from the sample.
(insert Table 2 here)
7 Data on capital adequacy are not available for years prior to 1988, preventing any comparison with the pre-Basel period. The choice of 1995 is somewhat arbitrary but quite standard given that most studies on the impact of the Basel Accord focus on the first half of the 1990s. In the case of the US, Flannery and Rangan (2002) have shown that none of the 100 largest banking firms appears to have been constrained by regulatory capital requirements since 1995.
10
The second column of Table 2 shows the distribution of banks by country. Unfortunately,
Bankscope does not contain data on the capital levels of Belgian, Dutch, Luxembourg,
German and Swiss commercial banks over the period of interest, which prevents the
computation of their risk-weighted assets using equation [6]. This explains why the analysis is
limited to the remaining seven G-10 countries. Unsurprisingly, Japanese and US banking
institutions constitute the vast majority of the sample.
The remainder of Table 2 indicates that the sample is in fact quite representative of each
national banking sector. With the exception of the UK, the data set includes at least 7 of the
10 biggest commercial banks (in terms of assets) of each country. Moreover, the sum of
sample banks’ assets almost always exceeds half of the total national banking assets.
(insert Table 3 here)
Table 3 lists the average total capital ratio, tier 1 ratio and credit risk ratio of each country
between 1988 and 1995. The comparability of the figures displayed in the table is not
guaranteed since the number of observations is increasing over time.8 Nevertheless, some
tentative remarks can be made.
First, the tier 1 ratio and the total capital ratio of all countries are far above their minimum
required levels of 4% and 8%, respectively. Moreover, both series are upward trending across
the years 1988-93, with no significant increase afterwards. Two groups of countries coexist in
the sample: those with relatively low capital adequacy ratios throughout the whole period
(Canada, France and Japan) and those that exhibit higher figures for both regulatory ratios
(Italy, Sweden, the UK and the US).
Second, with respect to credit risk, some countries (Canada, the UK and possibly the US)
appear to have experienced a decrease in their banking portfolio risk whereas others (France,
Italy, Japan and Sweden) have seen their credit risk ratio remaining relatively constant. Once
again, the sample can be divided into two groups of countries: those with an average credit
risk ratio varying between 50% and 60% (France, Italy, Sweden and perhaps the UK) and
those with a credit risk ratio equal or higher than 70% (Canada, Japan and the US). At first
glance, it might be tempting to attribute the higher credit risk ratio of these 3 countries to their
pre-1988 capital adequacy rules. Indeed, until the Basel guidelines were adopted, simple
gearing ratios were in force in Canada, Japan and the US (Pecchioli, 1987), which may
account for their historically higher level of credit risk. But then, the 1988 Accord should
have led to a decrease in the credit risk ratio of these countries as risk-based standards take
8 G-10 banks started to implement the Basel standards only gradually, which explains the low number of observations at the end of the 1980s. The slightly lower number of banks in 1994 and 1995 is due to a few mergers, which were not motivated by the level of capitalization of merging banks.
11
into account the composition of a bank’s portfolio when assessing capital charges. However,
Table 3 (Panel C) provides mixed support for this hypothesis: Canadian banks did indeed
experience a decrease in their portfolio risk whereas the level of risk-taking of Japanese
institutions remained fairly constant. The increasing number of observations for the US
prevents any reliable comparison.
On the whole, Table 3 suggests that the seven G-10 countries have raised their capital
adequacy ratios during the 1988-93 period whereas no specific trend could be found in the
credit risk series. Table 4 presents further descriptive statistics on the relationship between
capital and risk, by decomposing the average annual growth rate of the total capital ratio and
of the tier 1 ratio into three terms, as in equation [5].
(insert Table 4 here)
As can be seen from Table 4, the growth rate of both ratios over the sample years is roughly
similar and is mainly driven by a rise in capital levels, which offsets the rise in total assets.
The growth rate of the credit risk ratio is close to zero for all countries except Canada,
Sweden and the UK. However, Tables 3 and 4 do not tell us whether ∆CAR and ∆RISK are
related, nor whether the increase in capital adequacy ratios that took place between 1988 and
1993 was due to the introduction of capital adequacy rules. Indeed, it could be the case that
banks that were not part of the Basel Accord also decreased their leverage.
(insert Figure 1 and Figure 2 here)
Figure 1 shows that a rise in the equity to asset ratio9 was not only experienced by G-10 banks
at the beginning of the 1990s, but also by non G-10 banks and even by banks from countries
where capital adequacy rules were not put in place before 1993.10 In a similar way, Figure 2
indicates that the reliance of these three groups of banks on subordinated debt11 – a key
component of tier 2 capital – was roughly similar throughout the period surveyed.
9 The equity to asset ratio is a very rough approximation of the tier 1 ratio, as its numerator does not include disclosed reserves and its denominator does not assign different weights to the different asset categories. 10 The use of this third group of banks is motivated by the fact that a large number of non G-10 countries did adopt Basel-like rules between 1988 and 1993. I consider that a country has not implemented the Basel Accord before 1993 if its banks did not report a capital adequacy ratio before that year. 11 Expressed as a percentage of total assets. Note that Figure 2 does not distinguish between subordinated debt with a term to maturity of less or more than five years. Only the latter category is allowed to count as tier 2 capital, along with other elements (cf. section 2.2).
12
Thus, determining whether the Basel agreement caused changes in capital and risk ratios of
G-10 banks and whether these changes were related requires a more complex econometric
analysis than just looking at descriptive statistics. The following section sets up a model that
aims at assessing the empirical determinants of observed changes in capital and risk ratios
with a particular emphasis on the role played by capital adequacy rules.
4. The model
In order to acknowledge that capital and risk decisions are determined together, I extend the
simultaneous equation model developed by Shrieves and Dahl (1992) to a multi-country
setting. In this model, observed changes in banks’ capital and credit risk ratios consist of two
components, a discretionary adjustment and a change caused by factors exogenous to the
bank:
d
i,t i,t i,tCAR = CAR +E∆ ∆ [8]
di,t i,t i,tRISK = RISK +S∆ ∆ [9]
where ∆CARi,t and ∆RISKi,t are the observed changes in capital and risk ratios, respectively,
for bank i in period t. The d di,t i,tCAR and RISK∆ ∆ variables represent discretionary
adjustments in capital and risk, and Ei,t and Si,t are exogenously-determined factors.
Following Shrieves and Dahl, I model the discretionary changes in capital and risk using a
partial adjustment framework such that:
d *
i,t i,t i,t-1CAR = (CAR -CAR )α∆ [10]
d *i,t i,t i,t-1RISK = (RISK -RISK )β∆ [11]
where * *i,t i,tCAR and RISK are the target capital and risk ratios, respectively. Thus, the
discretionary changes in capital and risk for bank i are proportional to the difference between
the target ratio and the (observed) ratio in period t-1.
Substituting equations [10] and [11] into equations [8] and [9], the changes in capital and risk
can now be written as:
13
*i,t i,t i,t-1 i,tCAR = (CAR -CAR ) +Eα∆ [12]
*i,t i,t i,t-1 i,tRISK = (RISK -RISK ) +Sβ∆ [13]
This means that observed changes in capital and risk in period t are a function of the target
capital and risk ratios, the lagged capital and risk ratios, and any random shocks. Due to the
low number of observations for most of the sample countries, I decided to pool cross-country
data to estimate this system of equations.
Although the target capital and risk ratios of a bank are not observable, they are assumed to
depend on some set of observable variables describing the bank’s financial condition and the
state of the economy in each country. The variables that I use to approximate the target capital
ratio (CAR*) are the size of the bank (SIZE), a measure of its liquidity (LOANS), a
profitability indicator (ROA), changes in the risk ratio (∆RISK), country dummies, the degree
of regulatory pressure (REG) interacted with country dummies, and year dummies (YEAR).
The variables used to proxy the target credit risk ratio (RISK*) are the size of the bank
(SIZE), a measure of its liquidity (LOANS), a measure of asset quality (LLOSS), changes in
the capital ratio (∆CAR), country dummies, the degree of regulatory pressure (REG)
interacted with country dummies, and year dummies (YEAR).
The explanatory variables can thus be divided into bank-specific and country-specific factors
plus year dummies.
4.1. Bank-specific variables
All the variables presented here have been used in the studies listed Table 1 with the
exception of the LOANS variable.
SIZE is measured as the natural log of total assets. It is included as a control variable because
large banks have an easier access to equity capital markets and are thus expected to have
lower capital ratios than smaller banks. In addition, large banks carry out a wider range of
activities, which should increase their ability to diversify their portfolio hence to decrease
their credit risk. The variable LOANS, defined as the percentage of total assets tied up in
loans, is included in the system of equations as a measure of the riskiness of the bank. As
higher LOANS values correspond to lower investment in non earning assets, they should lead
to higher portfolio risk and a greater need for capital. Therefore, the impact of LOANS on
capital ratios will be positive if capital levels increase by more than what is necessary to
compensate the increase in the credit risk ratio (cf. equation [5]). Consistent with previous
14
studies, the return on assets, ROA, is included in equation [12] as profitable banks may prefer
to increase capital through retained earnings rather than through equity issues in the presence
of asymmetric information in capital markets. Loan loss provisions as a percentage of total
assets, LLOSS, are included in equation [13] as they represent funds that banks set aside to
cover bad loans. Loan loss provisions are deducted from outstanding loans and should
therefore lead to a decrease in risk-weighted assets. As a consequence, they are expected to
have a negative impact on the credit risk ratio.
Finally, the analysis in section 2 indicates that banks’ capital and risk choices are
interdependent, which suggests the inclusion of ∆CAR in equation [13] and of ∆RISK in
equation [12]. The sign of the relationship between these two variables is not clear a priori. A
positive sign would mean that G-10 banks increased their capital buffer and their credit risk
simultaneously, a result consistent with the unintended effects of more stringent capital
requirements suggested by the literature reviewed in section 2.1. This would also imply that
banks rely on an increase in capital levels and / or a decrease in total assets to increase their
capital adequacy ratios. On the other hand, a negative relationship between changes in capital
and risk ratios would indicate that higher capital ratios did not give banks greater incentives to
increase credit risk, leading to an unambiguous decrease in the probability of bankruptcy at
the G-10 level.
4.2. Country-specific variables
Country fixed-effects are included in the model in order to account for factors that are not
reflected by the set of bank-specific variables, such as national differences in capital
preferences and risk-aversion. The country dummies are also assumed to capture the extent to
which national variants of the Basel Accord had an impact on capital and risk. Some countries
that were part of the 1988 agreement have indeed supplemented the original guidelines with
additional requirements such as higher capital thresholds or new regulatory ratios over the
period studied.12
Of greater interest here is the regulatory pressure variable (REG), which is interacted with a
second set of country dummies. This variable is meant to capture the impact of the Basel
Accord across the G-10 as it describes the behavior of the banks that fell short of the
regulatory standards. More precisely, the regulatory pressure variable used in this study
12 As already mentioned, FDICIA was adopted in the US but there is also evidence that national regulators in other countries have set additional constraints on banks. See Ediz et al. (1998) for the UK and Rime (2001) for Switzerland.
15
emphasizes two aspects. First, the level below which a bank should be regarded as
undercapitalized and hence influenced by capital adequacy rules. Second, the size of the gap
between a bank’s capital ratio and this level, hence the magnitude of regulatory pressure
experienced by the bank. With the exception of Jacques and Nigro (1997), the papers listed in
Table 1 focus on the first aspect and neglect the second one. That is, they create a simple
dummy variable, which is equal to one when capital adequacy ratios are below some
threshold level and zero otherwise. The threshold level is the minimum regulatory ratio (cf.
equations [2] and [3]) or the minimum plus some bank-specific quantity.
The need to take into account the second characteristic of supervisory pressure – i.e., a
measure of how far banks are from the threshold – suggests the use of the following
regulatory pressure variable:
REG=THR -CAR if CAR < THR0 otherwise���
[14]
where THR represents the threshold level.
Thus, supervisory pressure is positive whenever CAR < THR, but decreasing as CAR
approaches THR from below. Banks with a CAR above THR are considered to be unaffected
by capital adequacy regulation. Of course, a proper threshold value still needs to be chosen.
For this purpose, I use Table 5, which lists the number of banks’ observations experiencing
positive regulatory pressure for different values of THR (from now on, I use the term
“undercapitalized” to describe these observations).
(insert Table 5 here)
As shown in Panels A and B, the percentage of banks’ observations that can be regarded as
undercapitalized is very low when the threshold is first set to the interim standards and then to
the definitive standards (Panel A) or directly to the latter standards (Panel B). Although this
result seems to support the widespread idea that banks were not affected by the Basel
standards, it could be the case that lower requirements would have induced banks to hold less
capital and that banks actually constructed a buffer above the regulatory minimum for
precautionary and / or reputational reasons. Panels C to E illustrate this possibility by setting
the threshold value for the total capital ratio and for the tier 1 ratio to the regulatory minimum
augmented by one standard deviation of the bank’s own capital ratio (Panel C), to 9% and
16
4.5% (Panel D) and finally to 10% and 5% (Panel E), respectively. Consequently, the
percentage of undercapitalized observations based on the total capital ratio increases to
roughly 18% in Panels C and D and to 34% in Panel E whereas the percentage of
observations that do not meet the different threshold values for the tier 1 ratio remains
systematically below 8%. Thus, a first finding that emerges from Table 5 is that G-10 banks
were more constrained by their total capital ratio than by their tier 1 ratio over the 1988-94
period. A second result is that, whatever the value taken by THR, the UK and the US always
have the lowest percentages of banks with CAR < THR. This is probably due to the fact that
most observations for both countries are from the years 1992-1995, when banks were strictly
obliged to obey the regulatory requirements.
Looking at the percentages of undercapitalized observations and their dispersion across
countries, it seems appropriate to consider two different threshold levels: the one set in Panel
B and the one shown in Panel C. The choice of the first threshold is motivated by the fact that
it corresponds to the binding regulatory minimum even though it generates an insufficient
number of undercapitalized observations for the purpose of estimation.13 The second
threshold solves this problem by producing sensible percentages of banks with CAR < THR
in each country. On the contrary, Panel A does not have enough undercapitalized observations
whereas Panels D and E classify too many Japanese banks and / or too few American banks in
the undercapitalized category.
To sum up, I measure regulatory pressure using equation [14], where THR alternatively takes
two values: the minimum requirement and the minimum requirement plus one bank-specific
standard deviation. Note that the use of the regulatory pressure variable to capture the impact
of the Basel Accord faces one important criticism, namely that the behavior of banks below
the threshold is likely to be influenced by other factors than capital requirements, including
market pressure coming from peer banks, private investors and even rating agencies. Thus, it
may be hard to disentangle the effects of the Basel Accord from increased market discipline
(Basel Committee on Banking Supervision, 1999). However, one could reasonably argue that
market pressure is ultimately driven by the prevailing capital adequacy rules, and hence, that
the REG variable is appropriate to capture the impact of the Basel standards on banks’
behavior.
13 I have looked at several ways to increase this number. For instance, by taking into account the additional requirements regarding tier 2 capital (cf. equation [4]) or the amount of general loan-loss reserves and subordinated debt eligible for inclusion in tier 2 capital (cf. Appendix 1). Unfortunately, none of the sample banks violates equation [4] whereas the composition of tier 2 capital is not disclosed in Bankscope.
17
4.3. Year dummy variables (YEAR)
Dummy variables for each year of the reference period – except 1989 in order to avoid perfect
collinearity – are added to the specification to take account of macroeconomic shocks (e.g.
changes in the volume or in the structure of loans demand) that can affect banks’ capital and
credit risk ratios.
4.4. Specification
On the basis of the variables selected to explain target capital and risk ratios, the model
defined by equations [12] and [13] is written as follows:
i,t 0 i,t 1 i,t 2 i,t 3 i,t-1 4 i,t j j i,t-1j j
CAR =a SIZE +a LOANS +a ROA +a CAR +a RISK + b + c REG∆ ∆ � �
t i,tt
+ YEAR +ε� [15]
i,t 0 i,t 1 i,t 2 i,t 3 i,t-1 4 i,t j j i,t-1j j
RISK =f SIZE +f LOANS +f LLOSS +f RISK +f CAR + g + h REG∆ ∆ � �
t i,tt
+ YEAR +ν� [16]
where b, c, g and h are the country dummy variables and j is a country index
εi,t and νi,t are disturbance terms
The coefficients of particular interest in this system of equations are cj and hj, which
represent, for country j, the impact of regulatory pressure on observed changes in capital and
risk, respectively, and a4 and f4 that test the relationship between changes in capital and risk
ratios at the G-10 level.
5. Results
Since the right-hand side of both ∆CAR and ∆RISK includes an endogenous variable, the
estimation of the system formed by equations [15] and [16] is carried out by three stage least
squares (3SLS). Unlike ordinary least squares (OLS), estimation by 3SLS provides consistent
estimates of the parameters. In addition, 3SLS estimation is more efficient than two stage
18
least squares (2SLS) as it exploits error correlation across equations.14 Tables 6 and 7 present
the results for the system based on the total capital ratio and on the tier 1 ratio, respectively.
The first two columns of each table show the results when THR is set to the minimum
regulatory requirement, while the remaining two columns present the results when THR is
defined as the minimum requirement plus one standard deviation of the bank’s own capital
ratio. A quick examination of both tables reveals that most of the explanatory variables are
statistically significant15 in at least one of the regressions.
The time dummies are all positive and significant in the ∆CAR equations of Table 6, which
indicates that target total capital ratios were higher over the period 1990-1995 than in 1989.
The magnitude of the year dummies implies that the biggest increase in total capital ratios
across the seven countries took place in 1992 (dummy 1992 - dummy 1991 ≅ +1.2%). This
result is logical since the definitive Basel standards had to be implemented by the end of
December 1992. On the contrary, the time dummies included in the ∆CAR equations of
Table 7 are mostly insignificant, which suggests that target tier 1 ratios were relatively
constant across years. These results echo Table 5, which showed that banks had more
difficulties in meeting the minimum requirement for the total capital ratio than the minimum
requirement for the tier 1 ratio. The time dummies included in the ∆RISK equations are only
negatively significant in Table 7, providing mixed support for the hypothesis that target risk
ratios were lower in the first half of the 1990s than in 1989.
Next, the country dummy variables, used alone, capture the extent to which target capital and
risk ratios differed across countries over 1989-95. The country dummies are essentially
significant in the ∆CAR equations, where their magnitude is roughly identical for each
country. This result suggests that most banks across the G-10 had the same target capital
ratios in mind and that the national variants of the Basel Accord did not bring major changes
to it. Indeed, Canada is probably the first country to have seriously departed from the 1988
agreement by requesting its banks to hold a minimum tier 1 ratio of 7% and a minimum total
capital ratio of 10% but only recently, in 1999. In the regressions where THR is set to the
minimum regulatory requirement, an F-test rejected the null hypothesis that the target capital
ratio of Swedish banks was of the same order of magnitude as that of other G-10 banks. The
higher magnitude of the dummy variable for Sweden is more than probably due to the
14 However, 3SLS may be sensitive to misspecification or measurement error. This suggests comparison with 2SLS estimates as a specification check. Estimation of equations [15] and [16] using 2SLS produces essentially the same results. 15 Throughout the paper, “significance” refers to a level of confidence of 95% and “marginal significance” to a level of confidence of 90%.
19
aftermath of the banking crisis experienced by this country, with banks forced to recapitalize
heavily, leading them to overshoot the target induced by Basel.
Before analyzing the regulatory pressure brought about by the Basel Accord, I briefly discuss
the sign of some bank-specific variables (lower panel of Tables 6 and 7). The parameter
estimates on lagged capital and risk ratios are negative and significant in both tables, with
values between –0.214 and –0.152. This result indicates that banks were only slowly
adjusting their capital and risk ratios to desired levels in the first half of the 1990s. Bank size
(SIZE) has a negative impact on bank capital ratios and a positive impact on the credit risk
ratio. Possible interpretations are that large banks have easier access to capital markets and
can therefore operate with lower amounts of capital or that they feel less pressure to increase
their capital ratio because of a “too-big-to-fail” effect. A larger size also allows a greater
diversification to mitigate the credit risk exposure. As hypothesized, loans as a percentage of
total assets (LOANS) are a good proxy of the target risk profile of a bank as they always
increase the credit risk ratio significantly. Their impact on capital ratios is negative, which
suggests that capital levels rise by less than what is necessary to compensate the increase in
risk. The return on assets (ROA) was found to have a positive effect on banks’ capital ratios, a
result consistent with the hypothesis that banks with higher earnings can retain more capital.
Finally, loan loss provisions as a percentage of total assets (LLOSS) had no effect on target
risk ratios overall.
More importantly, Tables 6 and 7 provide some insights on the behavior of banks
experiencing regulatory pressure and on the overall relationship between ∆CAR and ∆RISK.
Hereunder, I focus on the results obtained by setting THR to the minimum regulatory
requirement plus one bank-specific standard deviation, i.e. on the last two columns of each
table. Regressions with THR equal to the regulatory minimum include too few
undercapitalized observations and may therefore display unreliable estimates or no estimates
at all for the regulatory pressure variables (see the first two columns of Table 7).
Impact of the Basel Accord on bank capital ratios
In Table 6, banks put under regulatory pressure because they were below the 8% requirement
plus one standard deviation increased their total capital ratio significantly in four countries:
Canada, Japan, Sweden and the US.16 It is well-known that one of the hidden goals of the
Basel Accord was to create a level playing field by eliminating the funding-cost advantage
16 Interpreting the magnitude of the regulatory pressure variable (REG) interacted with country dummies is not possible due to its definition (see equation [14]).
20
enjoyed by Japanese banks which operated with significantly lower capital to asset ratio
compared to their competitors in other G-10 countries (Wagster, 1996). Results in Table 6
show that the Basel Accord was quite effective in raising the total capital ratio of
undercapitalized Japanese institutions, which tends to indicate that part of this competitive
advantage was reduced. In France and in the UK, some banks (e.g. Banque Paribas or Lloyds
Bank plc) did not raise their total capital ratio in the first years of the Basel implementation
process, which probably accounts for the fact that the regulatory pressure variable is found to
be insignificant for both countries.17 The fact that French banks were slow in meeting the
Basel requirements is not surprising given that they had previously only been subject to a
minimum total capital ratio of 5%. Surprisingly, I found that Italian banks within one standard
deviation of the regulatory minimum actually decreased their total capital ratio even though
only marginally significantly. Two things might explain this result. First, Italian banks
belonging to a banking group are allowed to have a capital ratio as low as 7% provided that
the 8% requirement is respected for the group as a whole. Second, some troubled banking
institutions not only fell below the regulatory threshold, but also saw their capital position
deteriorate dramatically over the period studied. For instance, the total capital ratio of Banco
di Napoli, the seventh biggest Italian bank in terms of assets, was down from 8.46% in 1992
to 1.74% in 1995. These banks were eventually recapitalized.
Examining the coefficients on the regulatory pressure variable in the capital equation of
Table 7 indicates that undercapitalized banks showed the same response to the Basel Accord
in their tier 1 ratio than in their total capital ratio, except in Canada and in Japan for which the
coefficients are non significant. The result for Japan is consistent with several papers18 which
show that undercapitalized Japanese banks tended to issue more subordinated debts (an
increase in tier 2 capital, hence in total capital), while leaving their tier 1 capital relatively
unchanged. Finally, it should be mentioned that the regulatory pressure variable has a positive
and significant impact on both capital adequacy ratios when it is not interacted with country
dummies.19
Impact of the Basel Accord on bank portfolio risk
Regarding the impact of the Basel standards on portfolio risk, the parameter estimates for the
regulatory pressure variable show little significance, except in Table 6 where they are positive
17 Regulatory pressure has a positive and (marginally) significant impact on UK banks when REG is replaced by a simple dummy variable, a result consistent with what Ediz et al. (1998) found using a much larger data set of UK banks. 18 See for instance Ito and Sasaki (1998). 19 This result is not reported in any table.
21
and significant for Canada, Sweden and Japan, but only marginally in the latter case. Thus,
the findings weakly support the hypothesis that undercapitalized banks in these three
countries were increasing risk in response to the Basel agreement between 1988 and 1995.
Moreover, the regulatory pressure variable is found to be insignificant when not interacted
with country dummies,19 suggesting that capital requirements did not increase credit risk-
taking at the G-10 level. This result contradicts earlier findings by Koehn and Santomero
(1980) and others, who argue that more stringent bank capital regulation will cause a utility
maximizing bank to increase asset risk and may, as a result, increase the probability of bank
failure.
Overall relationship between changes in capital and risk ratios
Finally, a robust finding emerges from Tables 6 and 7, namely the negative relationship
between changes in capital and risk ratios over 1988-95. Conditioning on other variables, an
increase in the credit risk ratio of 1% decreases the total capital ratio by only 0.07% and the
tier 1 ratio by a mere 0.05%, suggesting that G-10 banks almost offset any increase in risk
with an increase in capital and / or a decrease in total assets. Looking at the “risk” equations,
a rise of 1% in both capital ratios is associated with a similar decrease in the credit risk ratio.
This means that, conditioning on other variables, higher capital ratios led to a decrease in
portfolio risk at the G-10 level during the 1988-95 period. The existence of a negative
relationship between changes in capital ratios and credit risk contrasts with the empirical
studies reviewed in section 2.3, which generally find that these changes are either somewhat
positively related or not related at all (see the last column of Table 1). This result also
strongly rejects various theories providing a rationale for a positive relationship between
changes in capital and risk, including the “bankruptcy cost avoidance” theory and the
“managerial risk aversion” theory (see Shrieves and Dahl, 1992).
6. Conclusion
This paper documents the behavior of banks from seven G-10 countries toward capital and
risk between 1988 and 1995 by using a modified version of the model developed by Shrieves
and Dahl (1992). Prior research, both at the theoretical and empirical levels, indicates that
banks may well respond to an increase in capital requirements by a corresponding increase in
the credit risk of their portfolio.
The evidence presented here shows that G-10 banks within one standard deviation of the
minimum regulatory capital requirement tended to increase their capital adequacy ratios,
22
except in France and in Italy. This finding indicates that regulatory pressure, i.e., the degree to
which banks are urged to raise their capital ratios by the prudential authorities, was generally
effective in raising the capital buffer of banks across the G-10. Moreover, regulatory pressure
only marginally raised the credit risk of banks, a result suggesting that banks that had to raise
their capital ratios drastically did not in general engage in riskier activities. In addition, the
tendency toward higher capital ratios, which characterizes G-10 banks during the 1988-95
period, was shown to have gone hand in hand with a decrease in credit risk. All in all, the
evidence presented here indicates that the 1988 Basel Accord did not lead banks to engage in
riskier activities, while providing them with a higher capital buffer against insolvency. These
findings have important policy implications for regulators as they suggest that the use of risk
buckets to assess and limit credit risk-taking is likely to produce the desired effect. This
approach to portfolio risk is currently being refined under the “standardized approach” of the
New Basel Accord, which is likely to be adopted by many small banks that do not have the
resources to use one of the more advanced “internal ratings-based approaches”.20
Two caveats are in order, however. First, as pointed out by Jones (2000), banks may attempt
to arbitrage between their economic assessment of risk and regulatory requirements. This can
be done either by boosting capital ratios through “cosmetic arrangements” or by exploiting
shortcomings in the measure of risk through “regulatory capital arbitrage”, both methods
allowing banks to misreport their effective capital ratios and / or credit risk. Although the lack
of data prevents measuring the extent to which these techniques were used by banks (Basel
Committee on Banking Supervision, 1999), this points out that risk-weighted assets as
defined by the 1988 Basel standards may not fully reflect the actual risk of a bank’s portfolio.
Second, the paper analyzes portfolio risk in isolation from other types of risks, like market
risk or interest rate risk. It could be the case that banks chose not to modify their credit risk to
comply with the Basel guidelines, while increasing their interest or market exposure.
However, evidence on the existence of such a trade-off is scarce, even if it is due to a lack of
data.21 Moreover, the fact that credit risk is still the biggest risk faced by banks strengthens
confidence in the conclusion that G-10 banks did not become riskier in the first half of the
1990s.
20 Under the “standardized approach”, corporate exposures are now assigned to five risk buckets (0%, 20%, 50%, 100% and 150%) according to the nature of the claim and the assessment of external agencies, while retail exposures receive a 75% risk weight. The “internal ratings-based approaches” allow banks to determine their own risk weights through the combination of their quantitative inputs and formulas specified by the Basel Committee on Banking Supervision. 21 Typically, estimating interest-risk exposure requires data on the duration of banks’ assets. This information is almost impossible to obtain on a standardized cross-country basis. In the case of the US, Allen et al. (1996) provide some evidence that substitution of unpriced interest rate risk for priced credit risk did occur. However, they do not allow for other features of banks’ books, which makes their results difficult to interpret.
23
References Aggarwal, R. and K. Jacques (1997): “A Simultaneous Equations Estimation of the Impact of
Prompt Corrective Action on Bank Capital and Risk,” in Financial Services at the Crossroads: Capital Regulation in the 21st Century, Conference, 26-27 February 1998, Federal Reserve Bank of New York.
Aggarwal, R. and K. Jacques (2001): “The impact of FDICIA and Prompt Corrective Action
on bank capital and risk: estimates using a simultaneous equation model”, Journal of Banking and Finance, vol.25 (6), pp. 1139-1160.
Allen, L., J. Jagtiani and Y. Landskroner (1996): “Interest Rate Risk Subsidization in
International Capital Standards”, Journal of Economics and Business, vol.48 (3), pp. 251-267.
Basel Committee on Banking Supervision (1988): “International convergence of capital
measurement and capital standards”, Bank for International Settlements, Basel. Basel Committee on Banking Supervision (1999): “Capital requirements and bank behavior:
the impact of the Basel Accord”, working paper #1, Bank for International Settlements, Basel.
Blum, J. (1999): “Do capital adequacy requirements reduce risks in banking?”, Journal of
Banking and Finance, vol.23 (5), pp. 755-771. Dewatripont, M. and J. Tirole (1994): “The prudential regulation of banks”, MIT Press. Ediz, T., I. Michael and W. Perraudin (1998): “Capital Regulation and UK Banks’ behavior”,
Financial Stability Review, Bank of England. Flannery, M.J. and K.P. Rangan (2002): “Market Forces at Work in the Banking Industry:
Evidence from the Capital Buildup of the 1990s”, working paper, Department of Finance, Insurance and Real Estate, Graduate School of Business, University of Florida.
Furlong, F.T. and M.C. Keeley (1989): “Capital regulation and bank risk-taking: a note”,
Journal of Banking and Finance, vol.13 (6), pp. 883-891. Gennotte, G. and D. Pyle (1991): “Capital controls and bank risk”, Journal of Banking and
Finance, vol.15 (4-5), pp. 805-824. Hart, O.D. and D.M. Jaffee (1974): “On the application of portfolio theory to depository
financial intermediaries”, Review of Economic Studies, vol.41 (1), pp. 129-147. Ito, T. and Y.T. Sasaki (1998): “Impact of the Basel Standard on Japanese Banks Behavior”,
working paper #6730, National Bureau of Economic Research, Cambridge, MA. Jacques, K. and P. Nigro (1997): “Risk-Based Capital, Portfolio Risk and Bank Capital: A
Simultaneous Equations Approach,” Journal of Economics and Business, vol.49 (6), pp. 533-547.
Jones, D. (2000): “Emerging problems with the Basel Capital Accord: Regulatory capital
arbitrage and related issues”, Journal of Banking and Finance, vol.24 (1-2), pp. 35-58.
24
Keeley, M.C. and F.T. Furlong (1990): “A re-examination of the mean-variance analysis of bank capital regulation”, Journal of Banking and Finance, vol.14 (1), pp. 69-84.
Kim, D. and A.M. Santomero (1988): “Risk in banking capital regulation”, Journal of
Finance, vol.43 (5), pp. 1219-1223. Koehn, M. and A.M. Santomero (1980): “Regulation of bank capital and portfolio risk”,
Journal of Finance, vol.35 (5), pp. 1235-1244. Pecchioli, R. (1987): “Prudential Supervision in Banking”, OECD, Paris. Pyle, D.H. (1971): “On the theory of Financial Intermediation”, Journal of Finance,
vol.26 (3), pp. 737-747. Rime, B. (2001): “Capital Requirements and Bank Behavior: Empirical Evidence for
Switzerland”, Journal of Banking and Finance, vol.25 (4), pp. 789-805. Rochet, J.C. (1992): “Capital requirements and the behavior of commercial banks”, European
Economic Review, vol.36 (5), pp. 1137-1170. Sheldon, G. (1996): “Capital Adequacy Rules and the Risk-Seeking Behavior of Banks: A
Firm-Level Analysis,” Swiss Journal of Economics and Statistics, vol.132, pp. 709-734.
Shrieves, R.E. and D. Dahl (1992): “The Relationship Between Risk and Capital in
Commercial Banks,” Journal of Banking and Finance, vol.16 (2), pp. 439-457. Wagster, J.D. (1996): “Impact of the 1988 Basel Accord on International Banks,” Journal of
Finance, vol.51 (4), pp. 1321-1346.
25
Table 1: Previous studies on the impact of the 1988 Basel Accord on capitalization and credit risk-taking.
Author(s) (Year of
Publication) Sample and Period
Impact of regulatory pressure
on ∆CAR
Impact of regulatory pressure
on ∆RISK
Relationship between
∆CAR and ∆RISK
Jacques and Nigro (1997)
2,570 US commercial banks with assets > $100 million over 2 years (1990-91) – First step in the implementation process.
+ for A 0 / - for U
– for A 0 for U
mostly 0
Aggarwal and Jacques (1997)
2,849 US commercial banks with assets > $100 million over 3 years (1991-93) – Second step in the implementation process
+ for A in 93 + for U
+ in 91 / – in 92-93 for A and U
– in 91-92 / + in 93
Ediz et al. (1998)
94 UK banks over 25 quarters (4th quarter 1989 - 4th quarter 1995)
+ for U 0 for U not studied
Rime (2001) 154 Swiss banks over 7 years (1989-95)
0 for A + for U
0 for A and U
0 / +
Aggarwal and Jacques (2001)
1,685 US commercial banks with assets > $100 million over 6 years (1991-96)
+ for A and U in 91
+ in 91 / 0 in 92 / – in 93-96 for A and U
+ and – in 91-92 / + in 93-96
Note: +: significantly positive; –: significantly negative; 0: insignificant. A: adequately capitalized banks; U: undercapitalized banks. Table 2: Representativeness of the sample
Number of banks Number of banks from the national
top-10 *
Sample bank assets / Total national banking assets *
Canada 7 7 0.92 France 9 7 0.49 Italy 16 9 0.74 Japan 76 9 0.84 Sweden 10 8 0.97 United Kingdom 9 6 0.70 United States 459 10 0.91 * As of December 1995.
26
Table 3 (Panel A): Total capital ratio (%) - number of observations in parenthesis 1988 1989 1990 1991 1992 1993 1994 1995
Canada 7.14 (6) 7.59 (7) 7.89 (7) 8.86 (7) 9.01 (7) 9.84 (7) 9.94 (7) 9.91 (7) France 6.00 (1) 7.03 (7) 7.34 (9) 7.72 (9) 8.26 (9) 8.80 (9) 9.65 (9) 9.41 (9) Italy 10.02 (1) 8.84 (3) 8.97 (6) 9.75 (8) 9.32 (11) 10.15 (14) 11.69 (16) 11.33 (16) Japan 9.24 (37) 8.61 (51) 8.93 (74) 8.49 (75) 9.23 (76) 9.60 (76) 9.24 (76) 9.58 (72) Sweden - 8.77 (1) 8.26 (4) 9.40 (7) 8.76 (8) 11.64 (9) 16.09 (10) 16.79 (10) United Kingdom 10.95 (2) 9.91 (6) 11.31 (7) 12.16 (7) 12.22 (9) 12.59 (9) 13.71 (9) 13.55 (9) United States - 9.30 (1) 8.90 (3) 10.08 (156) 12.30 (418) 13.10 (444) 12.63 (429) 13.01 (399) All countries 8.99 (47) 8.49 (76) 8.86 (110) 9.55 (269) 11.64 (538) 12.42 (568) 12.14 (556) 12.46 (522) Table 3 (Panel B): Tier 1 ratio (%) - number of observations in parenthesis 1988 1989 1990 1991 1992 1993 1994 1995 Canada 4.84 (6) 5.33 (7) 5.54 (7) 6.07 (7) 6.23 (7) 6.67 (7) 6.76 (7) 6.97 (7) France 3.20 (1) 4.34 (5) 4.41 (8) 4.58 (8) 5.11 (8) 5.41 (8) 5.95 (8) 5.89 (8) Italy 8.88 (1) 7.31 (3) 7.76 (4) 7.85 (8) 7.37 (10) 8.00 (14) 9.28 (16) 8.96 (14) Japan 4.91 (37) 5.30 (51) 5.61 (74) 5.66 (75) 5.89 (76) 6.07 (76) 6.21 (76) 6.12 (72) Sweden - 7.31 (1) 5.72 (4) 7.03 (7) 5.97 (7) 8.80 (8) 11.12 (9) 12.57 (10) United Kingdom 6.05 (2) 5.72 (6) 6.68 (7) 7.47 (7) 7.14 (8) 7.72 (9) 8.85 (9) 8.89 (9) United States - 5.50 (1) 6.47 (3) 8.37 (156) 10.56 (418) 11.34 (444) 10.79 (429) 11.13 (399) All countries 5.00 (47) 5.39 (74) 5.69 (107) 7.36 (268) 9.59 (534) 10.32 (566) 9.97 (554) 10.23 (519) Table 3 (Panel C): Credit risk ratio (%) - number of observations in parenthesis 1988 1989 1990 1991 1992 1993 1994 1995 Canada 97.65 (6) 93.28 (7) 89.40 (7) 83.59 (7) 79.78 (7) 74.58 (7) 71.64 (7) 68.37 (7) France 72.00 (1) 55.80 (5) 58.75 (8) 59.11 (9) 58.89 (9) 57.11 (9) 54.71 (7) 53.00 (7) Italy - 58.00 (1) 56.00 (2) 57.50 (6) 60.22 (9) 55.36 (14) 54.63 (16) 57.46 (13) Japan 66.76 (37) 68.47 (51) 67.76 (74) 68.28 (75) 69.75 (76) 68.78 (76) 68.33 (76) 67.97 (72) Sweden - 66.00 (1) 61.50 (4) 57.43 (7) 66.29 (7) 59.25 (8) 56.50 (10) 53.22 (9) United Kingdom 75.00 (1) 79.50 (4) 76.50 (6) 74.14 (7) 65.50 (8) 65.13 (8) 62.67 (9) 61.56 (9) United States - 92.00 (1) 86.00 (3) 78.94 (156) 69.93 (418) 69.28 (444) 70.90 (428) 72.91 (397) All countries 71.18 (45) 70.81 (70) 69.09 (104) 74.23 (267) 69.57 (534) 68.54 (566) 69.49 (553) 70.95 (514)
27
Table 4: Decomposition of the average annual growth rate of CAR for the period 1988-95 (%)
Panel A: CAR = Total capital ratio; K = Total capital
t
t
CARCAR
∆ t
t
KK
∆ t
t
RISKRISK
∆ t
t
AA
∆ Number of
observations
Canada 4.56 8.67 -4.23 8.33 48 France 4.55 7.76 -1.57 4.77 46 Italy -1.68 1.92 0.75 2.85 45 Japan 1.55 5.02 0.32 3.15 49 Sweden 11.80 13.04 2.87 -1.64 36 United Kingdom 3.16 4.96 -4.04 5.85 43 United States 3.34 13.93 0.68 9.91 1,385 All countries 3.40 12.76 0.39 8.97 1,652 Panel B: CAR = Tier 1 ratio; K = Tier 1 capital
t
t
CARCAR
∆ t
t
KK
∆ t
t
RISKRISK
∆ t
t
AA
∆ Number of
observations
Canada 4.56 8.67 -4.23 8.33 48 France 5.47 8.50 -2.00 5.03 40 Italy -5.01 -1.40 0.75 2.85 45 Japan 2.24 9.32 0.26 6.82 461 Sweden 9.22 9.39 2.59 -2.43 33 United Kingdom 3.98 5.98 -3.91 5.90 42 United States 3.00 13.59 0.68 9.91 1,385 All countries 2.86 11.86 0.36 8.65 2,054
28
Table 5: Impact of the choice of a particular threshold level on the percentage of observations for which CAR < THR (data are for the 1988-94 period).
CAR = Total capital ratio (*) CAR = Tier 1 ratio (**) Threshold level (THR) Country # of obs.
(total)
# of obs. for which
CAR < THR
% of obs. for which
CAR < THR
# of obs. (total)
# of obs. for which
CAR < THR
% of obs. for which
CAR < THR
Panel A Canada 48 10 20.83 48 0 0.00 For * : 7.25% France 53 18 33.96 46 5 10.87 until 1990, then Italy 59 8 13.56 56 1 1.79 8% Japan 465 18 3.87 465 4 0.86 For ** : 3.25% Sweden 39 2 5.13 36 0 0.00 until 1990, then UK 49 0 0.00 48 0 0.00 4% US 1,451 17 1.17 1,451 2 0.14
Total 2,164 73 3.37 2,150 12 0.56 Panel B Canada 48 14 29.17 48 1 2.08 For * : 8% France 53 19 35.85 46 6 13.04 For ** : 4% Italy 59 8 13.56 56 1 1.79
Japan 465 31 6.67 465 12 2.58 Sweden 39 2 5.13 36 0 0.00 UK 49 1 2.04 48 0 0.00 US 1,451 17 1.17 1,451 2 0.14 Total 2,164 92 4.25 2,150 28 1.30
Panel C Canada 48 30 62.50 48 8 16.67 For * : 8% + std France 53 38 71.70 46 18 39.13 For ** : 4% + std Italy 59 22 37.29 56 5 8.93 Japan 465 146 31.40 465 37 7.96
Sweden 39 22 56.41 36 10 27.78 UK 49 4 8.16 48 4 8.33 US 1,451 123 8.48 1,451 30 2.07 Total 2,164 385 17.79 2,150 112 5.21
Panel D Canada 48 29 60.42 48 3 6.25 For * : 9% France 53 41 77.36 46 15 32.61 For ** : 4.5% Italy 59 19 32.20 56 2 3.57
Japan 465 238 51.18 465 41 8.82 Sweden 39 12 30.77 36 0 0.00 UK 49 2 4.08 48 1 2.08 US 1,451 76 5.24 1,451 3 0.21 Total 2,164 417 19.27 2,150 65 3.02
Panel E Canada 48 44 91.67 48 11 22.92 For * : 10% France 53 48 90.57 46 22 47.83 For ** : 5% Italy 59 34 57.63 56 2 3.57
Japan 465 404 86.88 465 103 22.15 Sweden 39 18 46.15 36 4 11.11 UK 49 14 28.57 48 3 6.25 US 1,451 173 11.92 1,451 8 0.55 Total 2,164 735 33.96 2,150 153 7.12
Note: std = standard deviation of banks’ own capital ratios series.
29
Table 6: Results for the system based on CAR = Total capital ratio
(THR = minimum) (THR = minimum + one std) ∆CAR ∆RISK ∆CAR ∆RISK
Year Dummies 1990 0.007 -0.003 0.007 -0.004 (2.45)** (-0.29) (2.51)** (-0.37)
1991 0.005 -0.012 0.005 -0.01 (1.73)* (-1.29) (1.92)* (-1.1)
1992 0.017 -0.015 0.016 -0.013 (5.81)*** (-1.53) (6.15)*** (-1.28)
1993 0.012 -0.016 0.014 -0.012 (4.51)*** (1.69)* (5.43)*** (-1.28)
1994 0.006 -0.01 0.008 -0.007 (2.22)** (-1.16) (3.01)*** (-0.75)
1995 0.013 -0.003 0.015 0.001 (4.56)*** (-0.3) (5.64)*** (-0.14)
Country Dummies Canada 0.05 -0.035 0.04 -0.045 (7.67)*** (2.08)** (6.22)*** (2.55)**
France 0.046 0.001 0.038 -0.005 (7.71)*** (-0.03) (6.48)*** (-0.31)
Italy 0.047 0.006 0.04 0.003 (8.05)*** (-0.36) (7.06)*** (-0.2)
Japan 0.05 -0.005 0.041 -0.01 (8.96)*** (-0.33) (7.71)*** (-0.69)
Sweden 0.067 -0.015 0.039 -0.048 (10.42)*** (-0.87) (5.94)*** (2.73)***
UK 0.052 -0.011 0.042 -0.015 (8.27)*** (-0.69) (7.09)*** (-0.94)
US 0.049 0.021 0.039 0.017 (9.36)*** (-1.64) (7.74)*** (-1.3)
Regulatory Pressure * REG * Canada 0.906 3.839 0.568 1.946 Country Dummies (1.71)* (2.27)** (2.34)** (2.35)**
REG * France -0.329 -0.083 -0.156 0.128 (-1.55) (-0.12) (-1.01) (-0.25)
REG * Italy -1.002 -5.544 -0.48 -1.316 (-0.72) (-1.25) (1.72)* (-1.4)
REG * Japan 0.862 0.467 0.669 1.052 (2.50)** (-0.41) (3.82)*** (1.70)*
REG * Sweden 2.621 15.162 1.819 3.005 (-1.28) (2.39)** (10.61)*** (4.19)***
REG * UK 1.085 -2.012 0.764 -0.322 (-0.41) (-0.24) (-1.28) (-0.16)
REG * US 0.792 0.342 0.732 0.095 (2.85)*** (-0.37) (11.41)*** (-0.32)
Control variables SIZE -0.002 0.003 -0.002 0.003 (6.78)*** (2.56)** (6.05)*** (2.71)***
LOANS -0.028 0.142 -0.027 0.141 (8.12)*** (10.96)*** (8.20)*** (10.77)***
ROA 0.255 0.319 (5.55)*** (7.23)***
CAR(-1) -0.2 -0.168 (11.98)*** (10.71)***
∆RISK -0.082 -0.069 (4.04)*** (3.51)***
LLOSS -0.01 -0.009 (-1.22) (-1.14)
RISK(-1) -0.157 -0.156 (12.89)*** (12.90)***
∆CAR -0.935 -0.949 (4.35)*** (3.91)***
R-squared 0.282 0.218 0.365 0.227 # Observations 2,017 2,017 2,017 2,017 t statistics in parentheses; * significant at 10%; ** significant at 5%; *** significant at 1%.
30
Table 7: Results for the system based on CAR = Tier 1 ratio
(THR = minimum) (THR = minimum + one std) ∆CAR ∆RISK ∆CAR ∆RISK
Year Dummies 1990 -0.001 -0.015 -0.000 -0.012 (0.31) (1.52) (0.10) (1.27)
1991 0.000 -0.019 0.001 -0.018 (0.12) (2.13)** (0.23) (2.00)**
1992 0.005 -0.032 0.005 -0.030 (1.78)* (3.69)*** (2.01)** (3.37)***
1993 0.004 -0.029 0.005 -0.027 (1.45) (3.42)*** (1.77)* (3.09)***
1994 -0.002 -0.023 -0.002 -0.021 (0.79) (2.71)*** (0.62) (2.43)**
1995 0.004 -0.016 0.005 -0.013 (1.41) (1.85)* (1.78)* (1.53)
Country Dummies Canada 0.070 -0.013 0.063 -0.021 (10.79)*** (0.80) (10.12)*** (1.24)
France 0.061 0.015 0.055 0.010 (10.33)*** (0.92) (9.49)*** (0.59)
Italy 0.065 0.018 0.060 0.014 (11.01)*** (1.12) (10.50)*** (0.89)
Japan 0.067 0.012 0.061 0.006 (12.13)*** (0.78) (11.37)*** (0.39)
Sweden 0.080 0.003 0.062 -0.022 (12.64)*** (0.17) (9.93)*** (1.30)
UK 0.066 0.003 0.060 -0.000 (10.76)*** (0.19) (10.06)*** (0.00)
US 0.067 0.037 0.060 0.032 (12.65)*** (2.81)*** (11.72)*** (2.44)**
Regulatory Pressure * REG * Canada -1.609 28.784 0.294 2.854 Country Dummies (0.11) (0.58) (0.34) (0.98)
REG * France -0.431 -1.393 -0.144 -0.464 (0.56) (0.56) (0.42) (0.40)
REG * Italy - - -2.146 -2.660 - - (2.70)*** (0.98)
REG * Japan 0.750 0.025 0.468 0.751 (0.74) (0.01) (1.09) (0.52)
REG * Sweden - - 4.371 7.383 - - (11.93)*** (4.76)***
REG * UK - - 0.455 -4.765 - - (0.30) (0.95)
REG * US 0.346 6.096 0.911 0.400 (0.22) (1.19) (8.20)*** (0.93)
Control variables SIZE -0.003 0.003 -0.003 0.003 (9.32)*** (2.40)** (8.60)*** (2.76)***
LOANS -0.035 0.134 -0.034 0.137 (9.89)*** (10.09)*** (9.92)*** (10.50)***
ROA 0.238 0.279 (5.29)*** (6.32)***
CAR(-1) -0.214 -0.199 (12.25)*** (11.97)***
∆RISK -0.056 -0.045 (2.56)** (2.16)**
LLOSS -0.007 -0.007 (0.93) (0.81)
RISK(-1) -0.152 -0.155 (12.57)*** (12.98)***
∆CAR -1.042 -1.040 (5.09)*** (4.79)***
R-squared 0.261 0.237 0.332 0.252 # Observations 2,011 2,011 2,011 2,011 t statistics in parentheses; * significant at 10%; ** significant at 5%; *** significant at 1%.
31
Figure 1: Median of the equity to asset ratio over 1988-95, by type of banks
4
5
6
7
8
9
10
11
1988 1989 1990 1991 1992 1993 1994 1995
Equ
ity /
Tot
al A
sset
s (%
)
G-10 banks Non G-10 banks Banks not reporting a CAR before 1993
Figure 2: Median of the subordinated debt to asset ratio over 1988-95, by type of banks
0
1
2
3
1988 1989 1990 1991 1992 1993 1994 1995
Subo
rdin
ated
Deb
t / T
otal
Ass
ets
(%)
G-10 banks Non G-10 banks Banks not reporting a CAR before 1993
Source: Bankscope, Update 77 (December 1996)
32
Appendix 1: The 1988 Basel Accord: Transitional and implementing arrangements
End-1990 End-1992
1. Total capital ratio
7.25 % 8 %
2. Tier 1 ratio 3.25 % 4%
3. Limit on general provision (or general loan loss reserves) in Tier 2 capital (∗)
Maximum 1.5 % or, exceptionally, up to 2% of
Tier 2 capital
Maximum 1.5 % or, exceptionally and
temporarily, up to 2% of Tier 2 capital
4. Limit on term subordinated
debt in Tier 2 capital No limit (at discretion) Maximum 50% of Tier 1
capital
5. Deduction for goodwill Deducted from Tier 1 capital (at discretion)
Deducted from Tier 1 capital
(∗) In the event that no agreement was reached on the definition of unencumbered resources
eligible for inclusion in Tier 2 capital. Source: Basel Committee on Banking Supervision (1988).
33
Appendix 2: Proof of equation [5]
t tt t
t t
K RWAAs CAR = and RISK =
RWA A
tt
t t
KWe have that CAR =
RISK .A Taking logs and differentiating with respect to time:
t t t tdlog(CAR ) dlog(K ) dlog(RISK ) dlog(A )= - +
dt dt dt dt� �� �
Using the notation
•
t t t
t t
dlog(X ) dX X1= × =
dt X dt X
We obtain easily that
• • • •
t t t t
t t t t
CAR K RISK A= - -
CAR K RISK A
Note that equation [5] uses discrete time changes rather than time derivatives and is therefore only an approximation of the correct formula for the growth rate of capital adequacy ratios.