Bank capital buffers around the world: cyclical patters and the effect

Bank capital buffers around the world: cyclical
patters and the effect of market power
Oscar Carvallo Valencia
CEMLA
Alberto Ortiz Bolaños
CEMLA and EGADE
XI Seminar on Risk, Financial Stability and Banking
Sao Paulo
August 10-12, 2016
The views expressed in this presentation are those of the authors, and not necessarily those of CEMLA.
Bank capital buffers around the world
•
Counter-cyclical capital buffers:
– Aiming to prevent the destabilizing cyclical impacts of capital buffers fluctuations, Basel III
requires banks to increase capital buffers during economic booms: through a "mandatory
capital conservation buffer" of 2.5% and through a "discretionary counter-cyclical buffer" of
up to another 2.5% in times of credit booms. The contra-cyclical capital buffer tool is a central
piece of Basel III.
– However, as these proposals were calibrated mostly against BIS membership information, an
interesting question remains: are buffers pro-cyclical, globally? What are other global
determinants of capital buffers? How these factors interact with the cycle?
•
Objective: understand the determinants of banks’ choice of capital buffers.
•
Methodology:
–
–
–
•
Illustrative modeling of bank’s optimal capital-to-asset ratio where cyclicality and profitability are key
determinants.
Exploit the information of 7,118 banks across 143 countries for the 2001 – 2013 period.
Control for other bank (size, competition and share of non-performing loans) and country-level (a moral
hazard index and regulatory variables as capital stringency, official supervisory power and bank accounting
standards) determinants.
Contribution: scope of banks and countries examined, the multiple dimensions factored in,
methodological improvements and robustness checks.
For advanced economies:
•
Some studies find general evidence of pro-cyclical behavior (Ayuso
et al. 2004; Lindquist 2004; Stoltz and Wedow 2011; Bikker and
Metzemakers 2004; Jokipii and Milne 2009; Shim, 2012 and Chen,
Yueh-Fang Ho and Hsu 2014).
•
However, heterogeneous cyclical patterns are detected even within
advanced economies.
– Jokipii and Milne (2009) find evidence in favor of a counter-cyclical
fluctuation of capital buffers for commercial, savings banks and large
banks in Europe.
– Capital buffers of cooperative and smaller banks are found to co-move
with the cycle.
– They also find counter-cyclical behavior by the part of banks in EU’s
recently admitted countries (RAM).
– For Australian banks, Vu and Turnell (2015) find evidence of procyclical behavior of large banks, but counter-cyclical for smaller.
For emerging economies:
•
Carvallo, Kasman and Kontbay-Busun (2015) examine the determinants of capital buffers for the
banking sectors of 13 Latin American and Caribbean countries for the 2001-2012 period.
– Negative relationship between capital buffers and GDP growth for Argentina, Colombia,
Ecuador, Peru and Uruguay.
– However, buffers in Bolivia, Brazil, Dominican Republic, Mexico, Panama and Venezuela are
found to behave pro-cyclically.
•
•
•
•
•
•
•
For Colombia, Garcia-Suaza, Gomez-Gonzalez, Pabon and Tenjo-Galarza (2012) find a negative
fluctuation of buffers with respect to the cycle, being more important the effect for large banks.
Tabak et al. (2011) also find a negative and significant relationship between business cycle and
capital buffers. Brazilian domestic private banks behave in a more pro-cyclical way than their
foreign and state-owned peers.
Indian commercial banks are also found to behave pro-cyclically, with foreign banks more intensely
so (Mahakud and Dash 2013).
Kontbay-Busun and Kasman (2015) find that in general Turkish banks behave counter-cyclically,
Gursoy and Atici (2012) observe that commercial banks included in the Turkish deposit insurance
fund scheme (Savings Deposit and Insurance Fund) behave counter-cyclically, whereas those not
included in the Fund, exhibit pro-cyclical behavior.
Regarding China, a recent study by Huang and Xiong (2015) shows that bank capital buffers have
a countercyclical relationship with business cycle fluctuations. Also,
Zheng, Xu and Liang (2012) find Chinese banks’ adjustment speed to target level to depend on the
size of capital buffer, with a faster speed of adjustment for banks with lower buffers.
Regarding more comprehensive studies:
•
Saadaoui (2014) analyzes the relationship between capital buffers, the cycle
and risk for 740 commercial banks in 50 emerging countries, over 1997-2008.
– Bank capital buffers and loan default risk are found to be negatively correlated with
the business cycle, supporting the implementation of countercyclical buffers in
these countries.
– Yet remarkably, higher market power, is found to be critical mediating factor, as it
attenuates this pro-cyclical behavior.
•
Fonseca and Gonzales (2010) look at global determinants of capital buffers
for a sample of 1,337 banks from 70 countries, for the period 1995-2002.
– Their model is a partial adjustment one, paying special attention to the effect of
market power and the role of regulation and institutions across.
– With respect to bank-specific variables, the study includes factors such as the cost
of funds, profitability and size.
– Buffers are not found to be systematically related to the business cycle. They are
found to be related positively with market power and the cost of deposits.
– However, these effects are sensitive to country variation in regulation, notably,
better accounting disclosure rules and the stringency of regulations on the scope of
banking activities.
Regarding more comprehensive studies:
•
Chen, Yueh-Fang Ho and Hsu (2015) examine the relations between total
capital and common equity buffers with the business cycle with a sample of
banks from 171 countries, over 1995-2009.
– Capital buffers are pro-cyclical at a global scale, independently of organizational
type and size.
– Their results also suggest that banks manipulate Tier 2 capital in order to signal
stronger capital adequacy ratios.
– This seems to suggest that the stringency of accounting disclosure regulations
could play an important role.
•
However, contrary to the two previous studies, only bank-specific variables
are included in the regression analysis, which poses the question of
comparing banking ratios under unequal competitive and regulatory
landscapes.
•
The present study is in line with the the Fonseca and Gonzales (2019) and
Saadaoui (2014) articles, incorporating both bank and country variables,
while trying to improve on some of the underlying technical issues, as
further explained below.
Capital buffers and market power:
•
Theoretically, the relationship between competition and bank risk attitudes is
complex and multifaceted (Allen and Gale 2004).
– The “competition - fragility” view, competition increases risk - taking as it
undermines the “charter value” of banks. As increased risk-taking potentially
endanger that value, banks have less incentive to engage in risky activities
(Keeley, 1990).
– Under the “competition - stability” view, the “too big to fail” doctrine
implies increased risk-taking by largest banks, due to implicit public
guarantees (Mishkin, 1999).
– Empirically, while recent evidence support the “competition - fragility” view
(Beck et al., 2006), other studies provide evidence supporting the
“competition - stability” hypothesis (De Nicoló et al., 2004; Schaeck et al.,
2006; Uhde and Heimeshoff, 2009).
Capital buffers and market power:
•
This debate translate to the relation between competition and
capital ratios.
•
Schaeck and Cihák (2012) use data from 10 European countries testing
the hypothesis that competition creates incentives to hold higher capital
holdings.
– Find that market competition increase bank’s capital holdings.
•
But, market power can be associated with lower capital holdings if large
banks enjoy lower adjustment and capital costs and larger margins and
profitability, which act as substitutes for capital (Elizalde and Repullo
2007).
•
There may also be the case that banks with market power are more
vulnerable to systemic risk during large downturns, as the experience of
the recent global financial crisis has shown (Demsetz and Strahan
1997).
Bank capital buffers and GDP growth
by sub-regions (in percentages)
• Bank capital buffers, which are holdings of banks’ capital-to-asset ratio in excess
of the regulatory minimum, are persistent both across countries and over time.
Developed
Developing
4,934 banks
37 countries
average:8.6%
2,184 banks
106 countries
average:10.4%
20
8
20
6
10
6
10
4
0
4
0
2
-10
2
-10
0
-20
0
-20
Average Capital Buffer
Q75/Q25 range
Capital Buffer (right axis)
80
70
2013
2012
2011
60
2010
2003
2002
2001
Q75/Q25 range
Capital Buffer (right axis)
2009
8
2008
30
2007
10
2006
30
2005
10
2004
40
2003
12
2002
40
2001
12
2013
50
2012
14
2011
50
2010
14
2009
16
2008
60
16
2007
18
2006
70
18
2005
20
2004
80
20
Average Capital Buffer
• What are the determinants of the observed levels of banks’ capital buffers?
Following Jokipii and Milne (2008) we define the capital buffer as the bank capital ratio less the Minimum Capital Ratio (MCR).
The bank capital ratio is approximated by the Total Capital Ratio (TCR), which measures the actual regulatory capital ratio in
each jurisdiction. TCR data comes from Bankscope, MCR data from WB database “The Regulation of Banks around the World”,
surveys I, II, III and IV. Barth et al. (2001, 2004, 2006, 2012).
Bank capital buffers and GDP growth
by sub-regions diverge after the crisis
Developed
Developing
We separate the data also in two groups: Developed: OECD non-emerging, OHI (other high income,
non-OECD); Developing: emerging and other developing, following recent classification in Claessens,
Stijn and Neeltje van Horen, 2015. "OECD" only includes the core OECD countries and "OHI" includes
all countries classified as high-income by the World Bank in 2000, not belonging to OECD.
A benchmark model of bank’s optimal
capital-to-asset ratio
• Banks use equity 𝐸𝐸𝑡𝑡 and deposits (𝐷𝐷𝑡𝑡 ) to fund loans
(𝐿𝐿𝑡𝑡 ) to firms and households.
• The objective of the banks is to maximize the net present
value of the stream of dividends 𝑑𝑑𝑑𝑑𝑑𝑑𝑡𝑡 :
subject to
𝑚𝑚𝑚𝑚𝑚𝑚
𝑑𝑑𝑑𝑑𝑑𝑑𝑡𝑡 ,𝐸𝐸𝑡𝑡,𝐷𝐷𝑡𝑡 ,𝐿𝐿𝑡𝑡
∞
𝔼𝔼𝑡𝑡 � 𝛽𝛽𝑡𝑡 𝛬𝛬𝑡𝑡 𝑑𝑑𝑑𝑑𝑑𝑑𝑡𝑡
𝑡𝑡=0
𝐿𝐿𝑡𝑡 ≤ 𝐷𝐷𝑡𝑡 + 𝐸𝐸𝑡𝑡
𝐷𝐷 𝐷𝐷
𝑑𝑑𝑑𝑑𝑑𝑑𝑡𝑡 + 𝐸𝐸𝑡𝑡 − 𝐸𝐸𝑡𝑡−1 ≤ 𝑅𝑅𝑡𝑡𝐿𝐿 𝐿𝐿𝑡𝑡−1 − 𝑅𝑅𝑡𝑡−1
𝑡𝑡−1 −
𝜓𝜓𝑡𝑡
𝐸𝐸𝑡𝑡
𝐸𝐸
𝐿𝐿𝑡𝑡 − 𝐿𝐿
2
𝑟𝑟𝑟𝑟𝑟𝑟 2
𝑡𝑡
Optimal level of capital-to-asset (loan) ratio
•
The first order necessary condition for an interior solution of the optimal
choice of equity, 𝐸𝐸𝑡𝑡 , is:
𝛬𝛬𝑡𝑡
+
�
foregone dividends (utility)
𝛬𝛬𝑡𝑡 𝜓𝜓𝑡𝑡
𝐸𝐸𝑡𝑡
𝐸𝐸
−
𝐷𝐷𝑡𝑡 + 𝐸𝐸𝑡𝑡
𝐷𝐷 + 𝐸𝐸
𝑟𝑟𝑟𝑟𝑟𝑟
𝑡𝑡
𝐷𝐷𝑡𝑡
𝐷𝐷𝑡𝑡 + 𝐸𝐸𝑡𝑡
2
opportunity cost of excess capital accumulation
marginal cost of increasing equity
•
=
𝔼𝔼𝑡𝑡 𝛽𝛽𝛬𝛬𝑡𝑡+1 𝑅𝑅𝐿𝐿𝑡𝑡+1 + 1
returns on the lent equity
marginal benefit increasing equity
The first order necessary condition for an interior solution of the optimal
choice of deposits, 𝐷𝐷𝑡𝑡 , is:
𝔼𝔼𝑡𝑡 𝛽𝛽𝛬𝛬𝑡𝑡+1 𝑅𝑅𝐷𝐷
𝑡𝑡
interest payment on deposits
marginal cost of increasing deposits
=
𝔼𝔼𝑡𝑡 𝛽𝛽𝛬𝛬𝑡𝑡+1 𝑅𝑅𝐿𝐿𝑡𝑡+1
expected return on the loans financed with these deposits
+
𝛬𝛬𝑡𝑡 𝜓𝜓𝑡𝑡
𝐸𝐸𝑡𝑡
𝐸𝐸
−
𝐷𝐷𝑡𝑡 + 𝐸𝐸𝑡𝑡
𝐷𝐷 + 𝐸𝐸
𝑟𝑟𝑟𝑟𝑟𝑟
𝑡𝑡
𝐸𝐸𝑡𝑡
𝐷𝐷𝑡𝑡 + 𝐸𝐸𝑡𝑡
2
reduction of the opportunity cost of over accumulating equity
marginal benefit of increasing deposits
Benchmark determinants of bank’s optimal
capital-to-asset (loan) ratio
• According to this simple model of bank’s capital
optimization, it is optimal for a 𝐿𝐿 bank to increase
its
𝐷𝐷
𝔼𝔼𝑡𝑡 𝛽𝛽𝛬𝛬𝑡𝑡+1 𝑅𝑅𝑡𝑡+1 −𝔼𝔼𝑡𝑡 𝛽𝛽𝛬𝛬𝑡𝑡+1 𝑅𝑅𝑡𝑡
∗
capital-to-loan ratio, 𝜙𝜙𝑡𝑡 = 𝛬𝛬 −𝔼𝔼 𝛽𝛽𝛬𝛬 −𝔼𝔼 𝛽𝛽𝛬𝛬 𝑅𝑅𝐷𝐷 , when:
𝑡𝑡
𝑡𝑡
𝑡𝑡+1
𝑡𝑡
𝑡𝑡+1 𝑡𝑡
– GDP decreases, under the assumptions that the utility function
exhibits decreasing marginal utility of consumption and
𝛬𝛬
1
, which represents a pro-cyclical behavior.
𝔼𝔼𝑡𝑡 𝑡𝑡+1 >
𝐷𝐷
𝛬𝛬𝑡𝑡
1+𝑅𝑅𝑡𝑡
1+𝛽𝛽
𝐿𝐿
– The marginal return from lending 𝑅𝑅𝑡𝑡+1
increases.
– The marginal cost of borrowing 𝑅𝑅𝑡𝑡𝐷𝐷 increases.
𝐿𝐿
– The profitability of lending over deposits 𝔼𝔼𝑡𝑡 𝛽𝛽𝛬𝛬𝑡𝑡+1𝑅𝑅𝑡𝑡+1
−
𝐷𝐷
𝔼𝔼𝑡𝑡 𝛽𝛽𝛬𝛬𝑡𝑡+1𝑅𝑅𝑡𝑡 increases.
Capital buffers: research hypothesis
•
Persistence: capital buffers exhibit persistence which could reflect the
presence of adjustment costs of capital buffers to targets.
•
Cyclicality: according to the model, we expect to observe a negative
relationship between capital buffers and GDP growth, a pro-cyclical
behavior.
𝛬𝛬
– The necessary condition for this relationship in the model is that 𝔼𝔼𝑡𝑡 𝑡𝑡+1 >
𝛬𝛬𝑡𝑡
1
. It is likely to be the case that for a growing economy, marginal utility of
1+𝑅𝑅𝐷𝐷
𝑡𝑡 1+𝛽𝛽
income is relatively high in the present compared to the future and therefore
𝛬𝛬
𝔼𝔼𝑡𝑡 𝑡𝑡+1 < 1 .This condition is more easily satisfied for an economy where
𝛬𝛬𝑡𝑡
consumption is relatively constant such that 𝔼𝔼𝑡𝑡
happen in a mature/high income economy.
𝛬𝛬𝑡𝑡+1
𝛬𝛬𝑡𝑡
→ 1, which is more likely to
•
Profitability: according to the model, profitability is positively associated
with the level of capital buffers.
•
Cost of capital: according to the model, the cost of capital is positively
associated with the level of capital buffers.
Capital buffers: research hypothesis (cont.)
•
We also study the determinants of:
– Size: moral hazard motives could imply a negative relation with capital buffers,
charter value a positive one.
– Risk: the higher the ratio of non-performing loans as a percentage of total bank
assets, the larger the capital buffers, a positive relation.
– Organizational type of banks: we will investigate the different behavior of
commercial, cooperative and saving banks.
– Competition: higher competition could lead to lower levels of buffers due to lower
profitability and charter value, but if pool of borrowers is affected by competition,
then could lead to higher levels to cover losses.
– Deposit insurance: due to moral hazard motives, there could be a negative
relationship with capital buffers.
– Regulation: considering the stringency of capital regulation, the level of intervening
power by authorities, and the enforcement of accounting standards, which could be
expected to have a positive relationship.
– Financial depth: measures of financial depth associated with easiness of access
to capital markets are expected to have a positive relationship with capital buffers.
•
The differential patterns of these effects in developed and developing
banking systems are analyzed.
Partial adjustment model
The presence of highly persistent capital buffers could indicate that
banks approach their optimum target with a partial adjustment.
0
50
100
0
Capital Buffer
50
100
•
Capital Buffer(t-1)
Developed
Linear regression
𝐵𝐵𝐵𝐵𝐵𝐵𝑖𝑖,𝑡𝑡 = 𝛼𝛼
•
𝐵𝐵𝐵𝐵𝐵𝐵𝑖𝑖∗,𝑡𝑡
not observable
Developing
Fitted values
+ 1 − 𝛼𝛼 𝐵𝐵𝐵𝐵𝐵𝐵𝑖𝑖,𝑡𝑡−1 +
The empirical model will be specified as:
𝑋𝑋�
𝐵𝐵𝐵𝐵𝐵𝐵𝑖𝑖,𝑡𝑡 = 𝛼𝛼0 + 𝛼𝛼1 𝐵𝐵𝐵𝐵𝐵𝐵𝑖𝑖,𝑡𝑡−1 + 𝛿𝛿 𝑖𝑖
𝑖𝑖,𝑡𝑡
𝜂𝜂⏟𝑖𝑖
bank−specific determinants
∗
vector of variables that determine 𝐵𝐵𝐵𝐵𝐵𝐵𝑖𝑖,𝑡𝑡
+ 𝜂𝜂𝑖𝑖 +
𝑢𝑢�
𝑖𝑖,𝑡𝑡
i.i.d error term
Estimation: Baseline specifications
Equation
Main Explanatory Variables
Controls
(1)
ROAA, LLRGL, BOONE, OCS, OSP, BACC
(2)
ROAA, LLRGL, BOONE, OCS, OSP, BACC, MCM
(3)
BUFFERt-1
(4)
Other variables of interest (Size,
SizeCo, GGDP, Developing
Dummy, Cooperative Banks, Saving
Banks, Interactions with GGDP and
Developing Dummy)
(5)
(6)
ROAE, LLRGL, BOONE, OCS, OSP
ROAE, LLRGL, BOONE, OCS, OSP, BACC
ROAE, LLRGL, BOONE, OCS, OSP, MCM
ROAE, LLRGL, BOONE, OCS, OSP, BACC, MCM
(7)
ROAA, LLRGL, BOONE, OCS, OSP, CF
(8)
ROAE, LLRGL, BOONE, OCS, OSP, BACC, CF
Estimation results: Baseline regressions
Estimation results: First Lag of Capital Buffer
Buffer (T-1)
1
2
3
4
5
6
7
8
0.565***
0.570***
0.560***
0.569***
0.597***
0.600***
0.579***
0.585***
0.595***
0.601***
0.577***
0.586***
0.491***
0.490***
0.485***
0.483***
0.563***
0.563***
0.596***
0.578***
0.597***
0.580***
0.488***
0.482***
0.549***
0.549***
0.580***
0.565***
0.579***
0.567***
0.480***
0.472***
0.565***
0.563***
0.597***
0.580***
0.596***
0.581***
0.477***
0.469***
0.485***
0.488***
0.507***
0.486***
0.505***
0.490***
0.331***
0.324***
0.486***
0.490***
0.508***
0.488***
0.507***
0.492***
0.326***
0.320***
0.574***
0.570***
0.570***
0.600***
0.584***
0.602***
0.586***
0.491***
0.579***
0.567***
0.567***
0.598***
0.583***
0.599***
0.584***
0.484***
0.572***
0.559***
0.561***
0.590***
0.575***
0.592***
0.577***
0.486***
Excluding US
World no-US
Developed no-US
Developing
2001-2006
0.567***
0.592***
0.577***
0.589***
0.719***
0.715***
0.717***
0.577***
0.5754***
0.5455***
0.6110***
0.713***
0.467***
0.715***
0.566***
0.5648***
0.5917***
0.5993***
0.717***
0.568***
0.462***
0.4677***
0.4952***
0.5172***
0.568***
Laeven and Valencia (2013)
0.586***
0.587***
0.605***
0.600***
0.606***
0.602***
0.563***
0.557***
Total sample
Including Size
Including SizeCo
Including SizeCo,
Cooperative and Savings
Including SizeCo,
Cooperative*GGDP and
Savings*GGDP
Including SizeCo and
SizeCo*GGDP
Including SizeCo and FID
Including SizeCo and
FID*Developing
Including SizeCo and
Boone*Developing
Including SizeCo and
SizeCo*Developing
Including SizeCo and
GGDP*Developing
Standard errors in parentheses * p < 0.10, ** p < 0.05, *** p < 0.01.
Estimation results: Cycle Indicator (GGDP)
GGDP
1
2
3
4
5
6
7
8
-0.108***
-0.113***
Total sample
-0.128***
-0.109*** -0.0921*** -0.0970*** -0.0786*** -0.0827***
Excluding US
-0.103***
-0.0738** -0.0797***
World no-US
2001-2006
Laeven and Valencia (2013)
-0.0509*
-0.046 -0.0991*** -0.0806***
-0.0738**
-0.046
0.0721
0.0415
0.077
0.0721
-0.1047***
-0.225*** -0.224***
-0.247***
-0.260***
-0.245***
-0.0785**
-0.260***
-0.230**
-0.0662
-0.232**
-0.163***
-0.149***
-0.141***
-0.135***
-0.128***
-0.160***
-0.152***
Developed no-US
Developing
-0.0682**
-0.149***
Standard errors in parentheses * p < 0.10, ** p < 0.05, *** p < 0.01.
•
Procyclical behavior of capital buffers, specially in the US and developing
Estimation results: Relative Size (SizeCo)
SizeCo (=1 if the bank belongs to the top quartile in their national size distribution)
Total sample
Excluding US
World no-US
Developed no-US
Developing
2001-2006
Laeven and Valencia (2013)
1
-0.929***
-1.188***
-0.573
-0.870**
2
-0.911***
-1.177***
-1.1772***
-0.7503*
-1.0257**
-0.562
-0.910**
3
-0.769***
-1.125***
4
-0.868***
-1.141***
5
-0.698**
-1.073***
-0.745
-0.869**
-0.693
-0.828**
-0.712
-0.817**
6
-0.853***
-1.133***
-1.1332***
-0.8998**
-0.9429**
-0.658
-0.861**
7
-1.147***
-1.660***
-1.080*
-0.859*
8
-1.143***
-1.561***
-0.9410**
-0.9410**
-1.1893**
-1.109*
-0.765*
GGDP*SizeCo
1
2
3
4
5
6
7
8
Total sample
-0.126***
-0.127***
-0.0984**
-0.105***
-0.0942**
-0.107***
-0.161***
-0.155***
Excluding US
-0.135***
-0.136***
-0.115**
-0.117***
-0.113**
-0.119***
-0.187***
-0.178***
World no-US
-0.1363***
-0.1192***
-0.1445***
Developed no-US
-0.2035***
-0.2151***
-0.1498**
Developing
-0.1052**
-0.0917*
-0.1183
Standard errors in parentheses * p < 0.10, ** p < 0.05, *** p < 0.01.
•
•
The coefficient on this variable is negative, significant and robust, indicating that the
average global bank has lower capital buffers the larger its size.
We find large banks in global and domestic markets to behave substantially more procyclically that their smaller peers.
Estimation results: Cooperative and Saving
Banks with their interaction with the cycle
Cooperative
1
2
3
4
5
6
7
8
Total sample
0.677***
0.557***
0.272
0.592***
0.216
0.538***
0.208
0.397
Excluding US
0.678***
0.450**
0.443**
0.568***
0.328
0.425**
0.448*
0.387
Savings
1
2
3
4
5
6
7
8
Total sample
1.372***
1.337***
1.085***
1.277***
1.129***
1.260***
1.163***
1.233***
Excluding US
0.745***
0.571**
0.733***
0.668***
0.498**
0.558**
0.504*
-0.457
6
0.211***
0.228***
7
0.261***
0.299***
8
0.269***
0.294***
6
0.245***
0.0385
7
0.207***
-0.0162
8
0.177**
-0.134
Cooperative*GGDP
Total sample
Excluding US
1
0.261***
0.296***
2
0.238***
0.264***
3
0.215***
0.248***
4
0.230***
0.254***
5
0.206***
0.238***
Savings*GGDP
Total sample
Excluding US
1
0.266***
0.0441
2
0.273***
0.0435
3
0.245***
0.0785
4
0.241***
0.0395
5
0.246***
0.0376
Standard errors in parentheses * p < 0.10, ** p < 0.05, *** p < 0.01.
•
Cooperative and saving banks tend to hold higher capital buffers in developed
economies. Saving banks tend to behave counter-cyclically in the US.
Estimation results: Bank competition (Boone
Indicator)
BOONE
1
2
3
4
5
6
7
8
Total sample
2.374***
2.394***
1.962***
1.633**
1.934***
1.589**
2.628***
2.740***
Excluding US
2.292***
2.302***
1.977***
1.491**
1.903***
1.438**
2.830***
2.618***
World no-US
2.3020***
1.4376**
-3.0782
Developed no-US
-3.8714***
-3.6022***
-3.0782
2.1394***
2.971
3.046
2.3703***
5.489
5.371
Developing
2001-2006
Laeven and Valencia (2013)
2.434***
2.497***
1.515
2.622
1.295
1.5009**
2.414
2.194***
1.859***
2.189***
1.853***
2.582***
2.483***
BOONE*Developing
1
2
3
4
5
6
7
8
Total sample
9.463***
9.669***
7.844***
7.225***
8.030***
7.516***
11.33***
11.03***
Exluding US
6.595***
5.977***
5.115***
4.505**
4.361**
4.064**
7.106***
6.303***
Standard errors in parentheses * p < 0.10, ** p < 0.05, *** p < 0.01.
•
Our results show that more competition leads to higher buffers in developed countries
but to lower in developing.
Estimation results: Moral hazard indicator
(insured deposits)
FID
1
2
3
4
5
6
7
8
Total sample
0.00968**
0.00970**
0.00859*
0.00940*
0.0100**
0.00941*
0.0154**
0.0173**
Excluding US
-0.00417
-0.00349
-0.00409
-0.00478
-0.0051
-0.00442
0.00127
-0.00191
FID*Developing
1
2
3
4
5
6
7
8
Total sample
-0.0276***
-0.0255***
-0.0231***
-0.0232***
-0.0199***
-0.0217***
-0.0294***
-0.0280***
Excluding US
-0.0207***
-0.0143**
-0.0204***
-0.0175***
-0.0121**
-0.0132*
-0.0242***
-0.0138*
Standard errors in parentheses * p < 0.10, ** p < 0.05, *** p < 0.01.
•
Higher moral hazard index is related to lower capital buffers in developing countries.
Estimation results: Regulatory indicators
Bank Accounting
Total sample
Excluding US
World no-US
Developed no-US
Developing
2001-2006
Laeven and Valencia (2013)
1
0.310**
-0.219
0.58
0.714***
2
0.426***
-0.0918
-0.0918
-0.6847*
0.0386
0.539
0.719***
3
4
0.399***
-0.0294
5
0.835*
0.690***
6
0.470***
0.06
0.06
-0.6704*
0.1266
0.880*
0.692***
7
8
0.294
-0.365
-1.3934***
-1.3934***
-0.1528
0.255
0.488*
6
0.245***
0.0951**
0.0951**
0.0083
0.1036
-0.0868
0.280***
7
0.314***
0.104*
8
0.287***
0.06
-0.2113
-0.2113
0.1500*
-0.165
0.363***
6
-0.0477
-0.053
-0.053
-0.0989
-0.0036
-0.0162
-0.0698
7
-0.011
-0.0935*
Overall Capital Stringency
Total sample
Excluding US
World no-US
Developed no-US
Developing
2001-2006
Laeven and Valencia (2013)
1
0.258***
0.0953**
-0.025
0.295***
2
0.275***
0.104**
0.1040**
0.0153
0.1243*
-0.0227
0.304***
3
0.244***
0.105**
4
0.233***
0.0877*
5
0.253***
0.100**
-0.132
0.306***
-0.105
0.274***
-0.116
0.306***
-0.19
0.394***
Overall Supervisory Power
Total sample
Excluding US
World no-US
Developed no-US
Developing
2001-2006
Laeven and Valencia (2013)
•
1
-0.0548*
-0.0797**
0.0278
-0.114***
2
-0.0222
-0.0439
-0.0439
-0.1130*
0.0151
0.0426
-0.0544
3
-0.0451
-0.0864**
4
-0.0678**
-0.0767*
5
-0.0167
-0.0423
0.104*
-0.0591*
0.0113
-0.117***
0.0857
-0.00562
-0.0648
0.00185
Banks account standards regulation, overall capital regulatory stringency, bank risk and
financial depth rise the capital buffers.
8
0.014
-0.0548
-0.1715*
-0.1715*
-0.016
-0.0894
-0.0118
Estimation results: Profitability indicators
(Return on average assets and equity)
ROAA
Total sample
Excluding US
World no-US
Developed no-US
Developing
2001-2006
Laeven and Valencia (2013)
1
0.528***
0.563***
0.499***
0.447***
2
0.543***
0.590***
0.5903***
0.5441**
0.5656***
0.493***
0.463***
3
4
5
6
7
0.564***
0.537***
8
0.561***
0.545***
0.494**
0.493***
0.477**
0.490***
7
8
ROAE
1
Total sample
Excluding US
World no-US
Developed no-US
Developing
2001-2006
Laeven and Valencia (2013)
2
3
0.0211***
0.0184***
4
0.0201***
0.0172***
5
0.0214***
0.0191***
0.0222*
0.0202*
0.0215*
6
0.0203***
0.0178***
0.0178***
0.0081**
0.0248**
0.0192
0.0229***
0.0226***
0.0235***
0.0231***
Standard errors in parentheses * p < 0.10, ** p < 0.05, *** p < 0.01.
Estimation results: Risk measure (Loan
reserves to gross loans)
LLRGL
1
2
3
4
5
6
7
8
Total sample
0.193***
0.202***
0.173***
0.177***
0.181***
0.184***
0.152***
0.163***
Excluding US
0.204***
0.219***
0.178***
0.188***
0.196***
0.199***
0.172***
0.187***
World no-US
Developed no-US
Developing
0.2194***
0.1990***
0.0843
0.0801
0.0724
0.0843
0.344**
0.343**
0.351**
0.1895***
0.349**
0.429
0.0697
0.453
0.220***
0.228***
0.228***
0.235***
0.191***
0.202***
2001-2006
0.371***
0.2120***
0.371***
Laeven and Valencia (2013)
0.234***
0.243***
Standard errors in parentheses * p < 0.10, ** p < 0.05, *** p < 0.01.
Summary of the results
•
The average bank in the world exhibits pro-cyclical behavior. This would call for the extensive and
effective implementation of Basel’s III proposed counter-cyclical buffer tools, at a global scale.
•
Nevertheless, according to our empirical exercise, this average result is conditional on a series of
factors, some affecting the functioning of local markets, its regulation, the level of competition and
several bank-specific characteristics.
•
Costs of adjustment are important and do not seem to differ greatly across levels of economic
development.
•
We find large banks in global and domestic markets to behave substantially more pro-cyclically
that their smaller peers.
•
Cooperative and saving banks tend to hold higher capital buffers in developed economies. Saving
banks tend to behave counter-cyclically in the US.
•
Banks account standards regulation, overall capital regulatory stringency, bank risk and financial
depth rise the capital buffers.
•
Another important result relates to competition. Our results show that more competition leads to
higher buffers in developed countries but to lower in developing. Higher moral hazard index is
related to lower capital buffers in developing countries.
Policy implications
•
We hypothesize that these differential effects regarding competition and financial
stability are related to market transparency and discipline, competition policies and
incentives, the risk-sensitiveness of deposit insurance schemes and supervisory
infrastructure.
•
Pro-cyclicality seems to be a more generalized behavior in developing, as in average,
not only their larger banks but also the rest banks, tend to fluctuate negatively with
the cycle.
•
Heterogeneity should be taken into account, but watch for political pressures!
Policy implications…again
Developed
Developing
Policy implications
•
Although in both regions capital buffers seem to become more counter-cyclical after
the global financial crisis, the road ahead regarding policy prescriptions have different
starting points.
•
For developed economies, we regard our empirical evidence as suggesting playing
more close attention to the pro-cyclical patters of its largest banks.
These banks are more systemically important because of its relative position in the
financial network and because of the potential cyclical externalities they may cause.
•
•
•
•
•
With respect to developing economies, it would seem that the timing for
introduction of new macro-prudential tools is critical.
It has been recognized the potential contractionary effect of capital regulation
(Repullo and Suarez 2013). This have to be taken in consideration.
Pave the way for more transparent, competitive and deep financial markets, stronger
and effective supervisory power by authorities, all of which implies committed
financial reform.
This will probably guarantee appropriate conditions for the future implementation of
more sophisticated macro-prudential tools, once the current factors behind the growth
volatility of such economies recede.
Bank capital buffers around the world: cyclical patters
and the effect of market power
Oscar Carvallo Valencia
CEMLA
Alberto Ortiz Bolaños
CEMLA and EGADE
Thank you