Transcript WELCOME TO THE CLOSING CEREMONIES OF THE

EFFICIENCY of BANKS in SOUTH EAST ASIA:
The INFLUENCE of OWNERSHIP and RISK
Thierno Amadou Barry
Santos Jose Dacanay III
Laetitia Lepetit
Amine Tarazi
International Conference on
Safety and Efficiency of the Financial System
EDSA Shangri-La Hotel, Ortigas Center, Mandaluyong City
27 August 2007
Objectives
 Examines
cross-country differences in
efficiency among South East Asian banks
in the post-crisis period 1999-2004; and,
 Investigates
the efficiency scores in
relation to bank- and country-specific
efficiency drivers.
Methodology
 Employs
a two-stage procedure;
 First
stage involves the calculation of bank
efficiency scores using DEA;
 Second
stage entails regression of bankand country-specific variables on the
efficiency scores.
Methodology
Equation 1: DEA VRS Input-Oriented
Min
s.t.
n
 X
j 
j
ij
  X i0
i  1,2,3;
 Yr 0
r  1,2,3;
n
 Y
j 
j
rj
n

j 
j
i  0
1
j 1, 2,..., n
Methodology
Equation 2: DEA VRS Input-Oriented with Input Prices
MinWi ' Xi *
s.t.
n
 X
ij
 Xi *
 Y
 Yrj
j 
j
n
j r0
j 
n

j 
j
1
i  0
Methodology
Equation 3: Influence of Bank Characteristic Variables
Effi = ßBankChari + η
Equation 4: Influence of Environmental and Country
Dummy Variables
Effi = ßCounEnvj + γCountry Dummyj +η
Data
 Pooled



data from BankScope:
80 banks in 6 South East Asian countries;
6-year period 1999-2004;
358 observations;
 Country
environmental variables from
Asian Development Bank.
Data
Hong Kong
Indonesia
Korea
Malaysia
Philippines
Thailand
Total
No. of Banks
17
12
13
20
12
6
80
No. of Obs.
58
48
71
90
60
31
358
Results: DEA Calculations
Equation 1: DEA VRS Input-Oriented
89
49
CRTS
IRTS
DRTS
220
Results: DEA Calculations
Comparison of Efficient and Inefficient Groups
Cumulative Distribution
Score
Scale Economies
CRS vs IRS and DRS
CRS vs IRS
CRS vs DRS
D=0.9472***
D=0.9387***
D=0.9775***
Input Separability (3 Output-1Input)
(x1 and x2)
D=0.3324***
(x1 and x3)
D=0.7933***
(x2 and x3)
D=0.8184***
*** indicates significance at p<0.01.
Null Hypothesis
of No Difference
Rejected
Rejected
Rejected
Not Rejected
Rejected
Rejected
Results: DEA Calculations
Figure 1: Kolmogorov-Smirnov Test Percentile Plot
Results: DEA Calculations
Comparison Test for Equation 2 Results
Kolmogorov-Smirnov
Statistic
Technical versus Allocative
Technical versus Cost
Allocative versus Cost
*** indicates significance at p<0.01.
D=0.6554***
D=0.7712***
D=0.1864***
Spearman Rank
Order Coefficient
ρ=0.5544***
ρ=0.5500***
ρ=0.6511***
Results: DEA Calculations
Equation 2 Results: Cross-Country Comparisons
TE
x
AE
=
CE
Hong Kong
0.82
0.52
0.44
Indonesia
0.87
0.56
0.49
Korea
0.91
0.58
0.54
Malaysia
0.83
0.59
0.50
Philippines
0.83
0.60
0.51
Thailand
0.93
0.57
0.53
Results: OLS Specifications
Regression of TE score with bank
characteristic variables:
 Average
Total Assets,
 SD of ROAA, and
 Widely-held Ownership Type
are significant efficiency drivers.
Results: OLS Specifications
Regression of TE score with environmental
and country dummy variables:
 Real
GDP growth rate,
 CoV of FOREX, and
 All country dummies
are significant.
Conclusions
 There
are persistent cross-country differences
in bank efficiency in South East Asia;
 Differences can be explained by domestic
macroeconomic conditions such as Real GDP
and CoV of FOREX;
 Efficiency is positively driven by bank size and
to a lesser extent risk-taking;
 Banks which are owned by minority private
shareholders appear to be more efficient than
other banks.
Thank you
We would like to acknowledge the support of the following: