Effectiveness of the global banking system in 2010 a data envolopment analysis approach

Chinese Business Review
Volume 10, Number 11, November 2011 (Serial Number 101)

David Publishing

David Publishing Company
www.davidpublishing.com


Publication Information:
Chinese Business Review is published monthly in hard copy (ISSN1537-1506) and online by David Publishing
Company located at 1840 Industrial Drive, Suite 160, Libertyville, Illinois 60048, USA.
Aims and Scope:
Chinese Business Review, a monthly professional academic journal, covers all sorts of researches on Economic
Research, Management Theory and Practice, Experts Forum, Macro or Micro Analysis, Economical Studies of Theory
and Practice, Finance and Finance Management, Strategic Management, and Human Resource Management, and other
latest findings and achievements from experts and scholars all over the world.
Editorial Board Members:
Moses N. Kiggundu (Canada)
Polyxeni Moira (Greece)
Iltae Kim (Korea)
Sorinel C PUŞNEANU (Romania)
ZHU Lixing (Hong Kong)
Jehovaness Aikaeli (Tanzania)
Ajetomobi, Joshua Olusegun (Nigeria)
LI Kui-Wai (Hong Kong)
Shelly SHEN (China)
Chris TIAN (China)
Ruby LI (China)
Manuscripts and correspondence are invited for publication. You can submit your papers via Web Submission, or

E-mail to , Submission guidelines and Web
Submission system are available at .
Editorial Office:
1840 Industrial Drive, Suite 160, Libertyville, Illinois 60048
Tel: 1-847-281-9862
Fax: 1-847-281-9855
E-mail:
Copyright©2011 by David Publishing Company and individual contributors. All rights reserved. David Publishing
Company holds the exclusive copyright of all the contents of this journal. In accordance with the international
convention, no part of this journal may be reproduced or transmitted by any media or publishing organs (including
various websites) without the written permission of the copyright holder. Otherwise, any conduct would be
considered as the violation of the copyright. The contents of this journal are available for any citation, however, all
the citations should be clearly indicated with the title of this journal, serial number and the name of the author.
Abstracted / Indexed in:
Database of EBSCO, Massachusetts, USA
Ulrich’s Periodicals Directory
ProQuest/CSA Social Science Collection, Public Affairs Information Service (PAIS), USA
Summon Serials Solutions
Chinese Database of CEPS, Airiti Inc. & OCLC
Chinese Scientific Journals Database, VIP Corporation, Chongqing, P. R. China
Subscription Information:
Print $450 Online $300
Print and Online $560
David Publishing Company
1840 Industrial Drive, Suite 160, Libertyville, Illinois 60048
Tel: 1-847-281-9862 Fax: 1-847-281-9855
E-mail:


Chinese

Business Review
Volume 10, Number 11, November 2011 (Serial Number 101)

Contents
Financial Forum
Effectiveness of the Global Banking System in 2010: A Data Envelopment Analysis Approach

961

Ngo Dang-Thanh
Application of Pareto Distribution in Wage Models

974

Diana Bílková
The Chaotic Monopoly Price Growth Model

985

Vesna D. Jablanovic

Marketing
Analysis of the Relationship Between Perceived Security and Customer
Trust and Loyalty in Online Shopping

990

Nihan Özgüven

Industrial Economics

Growth Potential and Profitability Analysis of Insurance Companies in the Republic of Serbia

998

Dragana Ikonić, Nina Arsić, Snežana Milošević

Regional Economics
Sustainable Consumption and Production in the Baltic Sea Region

1009

Janis Brizga, Dzintra Atstaja, Dzineta Dimante

Enterprise Management
Motifs and Impediments for the Harmonization of Accounting Regulations
for Small and Medium-Sized Companies in the EU

1021

Tamara Cirkveni
Change of Management Values in Estonian Business Life in 2007-2009
Anu Virovere, Mari Meel, Eneken Titov

1028


Public Economics
The Impact of Tax Policies on Taxpayers Budget in Terms of Risk, Sensitivity and Volatility

1043

Boloş Marcel Ioan, Otgon Cristian Ioan, Pop R zvan Valentin
Interactions Between Knowledge Sharing and Organizational Citizenship Behavior

1061

Yavuz Demirel, Zeliha Seçkin, Mehmet Faruk Özçınar

Social Economics
Enhancing Organization’s Performance Through Effective Vision and Mission

1071

Ben E. Akpoyomare Oghojafor, Olufemi O. Olayemi, Patrick S. Okonji, James. U. Okolie
Determinants of Female Employment Rate in the European Union
Irena Spasenoska, Merale Fetahu-Vehapi

1076


Chinese Business Review, ISSN 1537-1506
November 2011, Vol. 10, No. 11, 961-973

Effectiveness of the Global Banking System in 2010:
A Data Envelopment Analysis Approach∗
Ngo Dang-Thanh
University of Economics and Business (Vietnam National University), Hanoi, Vietnam
Massey University, Palmerston North, New Zealand

The current crisis has revealed the weaknesses of the global financial in general and its banking system in particular,

and puts forward a requirement for assessing the effectiveness and stability of the banking sectors across countries.
Based on available data from 64 countries over the world, the author tried to evaluate the effectiveness of the
banking sectors in those countries through the view point of the data envelopment analysis approach to define how
the global banking systems is under the effect of the current crisis. Findings from the research showed that banking
systems in advanced economies are still more effective than in developing countries. Moreover, it explained the
effect of the current financial crisis, the role of public finance (and the government), and the development of the
(privately) commercial banks to the effectiveness of the banking sectors. The research also explained some
determinants that can affect the effectiveness of the banking system, including inflation, bank concentration, and
level of economic development.
Keywords: data envelopment analysis, effectiveness, efficiency, banking, cross countries

Introduction
Because of the important role of the banking and financial system in the rapid development of new industrial
economies (NIEs) in the 1960s-1970s, there were renewed interests in the relationship between financial and
economic growth. Schumpeter (1911) argued that the role of financial intermediaries in savings mobilization,
projects evaluation and selection, risk management, entrepreneurs monitor, and facilitating transactions is
important to technological innovation and economic growth. Following this argument, many other leading
economists continuing emphasized the positively essential role of the financial sector in economic development,
including Goldsmith (1969), Shaw (1973), McKinnon (1973), King and Levine (1993a, 1993b).
Banks are the core of the financial system. They accept deposits from savers and lend them to borrowers.

∗
Acknowledgement: The author would like to offer special thanks to Professor David Tripe at Centre for Banking studies,
Massey University, New Zealand for his supports, encouragement and useful comments. The author also thanks participants at the
18th Annual Global Finance Conference in Bangkok, Thailand, April 2011 for their constructive comments and feedback to
improve the quality of the paper. The usual disclaimer applies.
Ngo Dang-Thanh, Ph.D. candidate, Lecturer, Faculty of Political Economy, University of Economics and Business (Vietnam
National University), Centre for Banking Studies, Massey University.
Correspondence concerning this article should be addressed to Ngo Dang-Thanh, Faculty of Political Economy, University of
Economics and Business (Vietnam National University). E-mail:

962

EFFECTIVENESS OF THE GLOBAL BANKING SYSTEM IN 2010

They hold liquid reserves which allowing predictable withdrawal demand. They issue liabilities which are more
liquid than the deposits. They also reduce (or some times eliminate) the need of self-finance (Bencivenga &
Smith, 1991, p. 195). Banks hold an important role within the financial system, and to some certain level,
researching the banking system therefore means researching the financial system.
Started from the bankruptcy of the Northern Rock Bank in the UK (2008, February), however, the global
financial crisis and its heavily impacts have put researchers and policy makers under the requirement of
re-assessment and re-evaluation the stability and performance of the global financial and banking system1.
A firm is effective when it reaches its target outputs. Similarly, a banking system is defined as effectiveness
if it can fulfill its missions of providing banking services and monitoring the stability of the system. Meanwhile,
if banking systems are set under similar conditions of macro- and micro-economic, the level of outcomes that a
banking system can provide (in term of services and stability) is indeed its efficiency. In this sense, the problem
of calculating effectiveness of banking systems all over the world becomes the problem of evaluating its
efficiency with a (dummy) similar and equal input. This research is trying to define the effectiveness of the global
banking system in 2010 through analysing cross-country data observed from 64 countries, using the data
envelopment analysis (DEA) approach. The remainder of this paper is organized as follows. Section 2 gives some
reviews on efficiency and effectiveness evaluation in the banking sector using DEA approach. Section 3 explains
the methodologies and technical will be applied in the research. Section 4 shows empirical results and section 5
concludes.

Literature Review
To evaluate the efficiency of a set of firms (or banks), the most popular approaches are ratio analysis,
parametric analysis and nonparametric analysis (the latter two methods belongs to the X-efficiency approach).
While ratio analysis focuses on ratios between two variables (of inputs or outputs) to define the productivity and
efficiency, X-efficiency analysis evaluates the efficiency of a bank through a multi-variables aspect.

DEA is a popular nonparametric method applied in evaluating efficiency in finance and banking area. After
Farrell (1957) laid the foundation for a new approach in evaluating efficiency and productivity at micro-level,
Charnes, Cooper and Rhodes (1978) and then Banker, Charnes and Cooper (1984) developed the CCR and
BCC-DEA model, respectively, to evaluate the (relative) efficiencies of the researched decision making units
(DMUs). Since then, DEA was increasingly applied in efficiency evaluation, especially in social sciences2.
There are a limited number of researches using DEA to examine banking performance at cross-country level.
A study in 1997 showed that out of 130 studies on banking performance and efficiency, only six were focused on
comparing the efficiency level of banking systems across countries (Berger & Humphrey, 1997, pp. 182-184). As
shown in Table 1, all three DEA studies were using small sample data at institutional (bank) level to define the
benchmark frontier, hence, the global banking system was left untouched.
In the 2000s, further studies which used common frontier approach were developed by add in the model
1

According to Science Direct, since 2010, there are more than 2,200 journal articles regarding banking performance after the
crisis of 2007-2008 (Retrieved December 20, 2010, from ).
2
Recent study of Avkiran (2010) showed that there are more than 170 articles using DEA as a main methodology to analyse the
efficiency of banks and banks branches, including Sherman and Gold (1985), Peristiani (1997), Schaffnit, Rosen and Paradi
(1997), and Pastor, Knox Lovell and Tulkens (2006).


963

EFFECTIVENESS OF THE GLOBAL BANKING SYSTEM IN 2010

some environmental/controllable variables such as banking market conditions or market structure and regulation
(Kwan, 2003; Lozano-Vivas, Pastor, & Hasan, 2001; Maudos, Pastor, Perez, & Quesada, 2002; Sathye, 2005).
However, as they are also mainly focused on institutional level data while macro-environment is different from
country to country, they ignored that banks which are efficient in this country may not performance well if they
run as foreign-owned banks in other countries (Berger, 2007, p. 125). Hence, while trying to examine the whole

banking systems across countries, this study attempts to overcome the above problem.
Table 1
Studies on Banking Performance at Cross-Country Level (Prior to 1997)
Authors (date)
Berg, Forsund, Hjalmarsson, &
Suominen (1993)
Fecher & Pestieau (1993)
Bergendahl (1995)
Ruthenberg & Elias (1996)
Bukh, Berg, & Forsund (1995)
J. Pastor, Perez, & Quesada (1997)

Method used

Countries included

Institution

Data envelopment analysis

Norway, Sweden, Finland

Bank

Distribution free approach
Mixed optimal strategy
Thick frontier approach
Data envelopment analysis
Data envelopment analysis

11 OECD countries
Norway, Sweden, Finland, Denmark
15 developed countries
Norway, Sweden, Finland, Denmark
08 developed countries

Financial service
Bank
Bank
Bank
Bank

Note. Source: Berger and Humphrey (1997).

As DEA evaluates the efficiency of each DMU based on the optimal multipliers (or weights) of inputs and
outputs factors, it allows us to examine the effectiveness of a banking system by looking at the achievements of
the banking sector, including both quantity (assets, deposits, credits, etc.) and quality (overhead cost,
nonperforming loans, frequency of bank crises, etc.) factors of commercial banks in the economy3. They are
chosen following 122 variables represent the stability of the global financial system (WEF, 2010, Appendix A).
However, since DEA treats those factors dynamically (meaning each country can have its own preference on
them), to be understandable in evaluating and comparing the effectiveness of the banking systems between
countries, a common preference (or common set of weights) for the above analyzed factors is required. Therefore,
in this research, the DEA model will be divided into three stages, in which the first stage conducts a dynamic
DEA model (DSW model) to define the relatively efficiencies of the banking systems from these 64 countries;
the second stage examines the determinants affecting that efficiencies (Tobit model); and the third stage defines
the common set of weights for those analyzed factors (CSW model) in order to conduct the final banking
effectiveness scores.

Technical Methodologies
In the first step, DSW model is produced to calculate the maximum effectiveness scores that each country

can achieve with the observed (achievement) factors. Mahlberg and Obersteiner (2001) and Depotis (2004)
developed an input-oriented DEA-like model which treats all factors as outputs, while input is a dummy variable
(values equal to 1 for all countries). Therefore, the DSW model in this research is in fact a
constant-returns-to-scale (CRS) and input-oriented DEA model. For an evaluated country j0-th, its efficiency
score (DSWj0) can be expressed by the following non-negative linear problem:
3

It is important to notice that these factors are outcomes that a banking system is aiming for; hence, the DEA model in this paper
will use them all as output variables.


964

EFFECTIVENESS OF THE GLOBAL BANKING SYSTEM IN 2010

DSWj0 = max

∑u y
∑v x
m

mj 0

(1)

k kj 0

Subject to:

∑u

m

ymj ≤ ∑ vk xkj , 1 ≤ j ≤ n

∑v x = 1
∑u = 1
k

kj

m

um ≥ 0
xj = 1 {all original input values are assumed to be equal to 1}
where:
um: weight of m-th output factor;
vk: weight of k-th input factor;
xkj: k-th input of j-th country, k = 1;
ymj: m-th output of j-th country;
n: number of countries;
m: number of factors.
Due to the fact that some countries can have the same scores in this DSW model, a super efficiency DEA
model (Zhu, 2001) is also ran to determine the ranking order of the researched countries, makes it easier to
compare the effectiveness’s of the banking systems between countries.
In the next step, a Tobit regression (for more details, see Tobin, 1958) is used to determine the factors
affecting the country’s banking efficiencies (Tobit model). Since the CSW scores are bounded between 0 to 1,
non-censored regression models could be biased (Fethi & Pasiouras, 2010), while Tobit regression is justify as in
equation (2). Variables used in this model are ones that mainly related to the financial efficient of a banking
system at micro-level and are expressed in Table 2.

EF =

+ 1*CONC + 2*ROA + 3*ROE + 4*CIR + 5*INF
+ 6*CTA + 7*NIM + 8*CII + 9*GROUP

(2)

Table 2
Variables of the Tobit Model
Variables
EF
CONC
ROA
ROE
CIR
INF
CTA
NIM
CII
GROUP

Definition
CSW-DEA scores.
Bank concentration (assets of three largest banks as a share of assets of all commercial banks).
Bank’s average return on assets (Net income/Total assets).
Bank’s average return on equity (Net income/Total equity).
Bank’s cost to income ratio (Total costs as a share of total income of all commercial banks).
Inflation, consumer prices (annual %).
Bank’s capital to assets ratio (ratio of bank capital and reserves to total assets).
Net interest margin of banks (value of bank’s net interest revenue as a share of its interest-bearing assets).

Depth of credit information index (measures rules and practices affecting the coverage, scope and accessibility of
credit information).
Dummy variable of income group (equals to 0 if country belongs to lower income, 1 if middle income, and 2 if high
income group).


EFFECTIVENESS OF THE GLOBAL BANKING SYSTEM IN 2010

965

The last step is to define the optimal common set of weights which should be used for compare and ranking
countries based on their banking systems’ effectiveness. It is done by applying the CSW model. It is believed that
the efficient frontier found in the DSW model in the first step is the “best practice frontier” (Grosskopf &
Valdmanis, 1987; Schaffnit, Rosen, & Paradi, 1997); hence, the optimal common weight set will be the one that
get every countries’ performances closest to that frontier. There are several ways to define that common set of
weights is based on this idea. While imposing bounds for factor weights, Roll and Golany (1993) found out that
the common set of weights can be defined by maximizing the average efficiency of all DMUs or maximizing the
number of efficient DMUs. Kao and Hung (2005) applied a compromise solution approach to minimize the total
squared distances between the optimal objective values (found by DEA) and the common weighted values (found
by using common set of weights). Jahanshahloo, Memariani, Lotfi and Rezai (2005) applied the multiple
objective programming approach to simultaneously maximize the performance scores to get it closes to the “best
practice frontier”. Liu and Peng (2008) applied the common weights analysis to minimize the vertical and
horizontal virtual gaps between the benchmark line (slope equals to 1.0, or performance scores equal to 1.0) and
the coordinate of common weighted DMUs. In this paper, we modified the model of Kao and Hung (2005) into a
minimum distance efficiencies model, in which the common set of weights can be defined as the one minimizing
the total distances between optimal efficiencies (DSW scores) and common weighted scores (CSW scores) of all
DMUs, under the condition that each DMU’s efficiency cannot exceed its DSW efficiency4. To understand the
role of each factor in CSW scores, another condition was added where the total sum of weights is equal to 1 (or
100%). The country’s banking effectiveness scores will be constructed based on that CSW scores and findings
from the super efficiency DEA results in the previous step. This CSW model can be expressed as a

non-negatively linear problem as follows:

(

min ∑ e*j − e j

)

(3)

Subject to:

e*j = DSWj
ej =

∑u y
∑v x
m

mj

k

kj

,1≤j≤n

e j ≤ e*j

∑v x = 1

∑u = 1
k

kj

m

um ≥ 0.015
xj = 1 {all original input values are assumed to be equal to 1}
where:
um: weight of m-th factor;
ymj: m-th factor of j-th country;
4

This constrain makes these distances non-negative, hence, they can be used directly rather than the squared distances.
Mahlberg and Obersteiner (2001) found that restriction weights with lower bound of 0.01 steered a middle course between too
strong predetermination and too large flexibility.
5


966

EFFECTIVENESS OF THE GLOBAL BANKING SYSTEM IN 2010
n: number of countries;
m: number of factors.
The final effectiveness scores will then be calculated following this equation:

ES j =

∑u

CSW
m

ymj

(4)

where:
ESj: Effectiveness score of country j-th;
umCSW: Common weight of factor m-th;
ymj: Value of factor m-th of country j-th.

Empirical Results
In the first stage, countries and factors are collected from the database of Beck, Demirgüç-Kunt and Levine
(2000), Laeven and Valencia (2010), the World Bank (World Development Indicator, Global Development
Financial, and Doing Business databases), the International Monetary Fund (IMF, 2010), the Consultative Group
to Assist the Poor (CGAP, 2010) and Annual Reports from Central Banks of such researched countries. Ten
factors6 are included in this research, covering both quantitative (the first 5 factors) and qualitative (the last 5
factors) aspect of the banking sectors (see Table 3). It is important to notice that the last 3 factors are undesirable
factors (as they have negative effect to the banking effectiveness), hence, they was transformed into desirable
ones through the linear monotone decreasing transformation method7.
Table 3
Descriptive Statistics of Factors
Factors

Mean

Commercial banks’ assets/GDP
0.74

Domestic credit provided by banking sector (% of GDP)
80.21
Commercial banks' deposits/GDP
0.60
Number of ATMs per 100,000 people
28.27
Number of branches per 100,000 people
11.47
Private credit bureau coverage (% of adults)
36.72
Public credit bureau coverage (% of adults)
8.24
Banks' overhead costs/Total assets
0.22
Nonperforming loans ratios of commercial banks
17.39
Frequency of banking crises
2.92
Note. Data of the last three variables are already transformed.

Standard
error
0.06
8.74
0.04
4.87
1.23
4.38
1.60
0.01

0.78
0.09

Standard
deviation
0.48
69.92
0.36
38.96
9.86
35.03
12.76
0.05
6.23
0.72

Minimum

Maximum

0.09
-11.17
0.12
0.06
0.53
0
0
0
0
0

2.42
379.30
1.80
236.07
45.60
100
48.50
0.26
22.80
4.00

As mentioned in section 3, those factors will be treated as output variables, while a dummy-input (equals to
1) will be set for the whole 64 countries. The DSW model then produces an effective frontier built from 25
countries, while the other 39 are ineffective (see Appendix A Table A2).
Within the ineffective ones, none of them is developed countries, suggesting that the banking systems in
6

According to Dyson et al. (2001, p. 248) and Avkiran (2001, p. 68), one rule of thumb in using DEA is that the sample size has
to be at least 3 times bigger than the number of total inputs and outputs to overcome the discrimination problem. As we have 64
samples over 10 variables, hence, this research is justified.
7
In this method, the transformed values will be calculated by the difference between a proper translation vector w with the
original values of those undesirable factors. For more details, see Seiford and Zhu (2002) and Fare and Grosskopf (2004).


EFFECTIVENESS OF THE GLOBAL BANKING SYSTEM IN 2010

967

advanced economies still run better than in developing countries although they had to bear stronger effect from
the current crisis. This can be explained by the difference between projected values and original values of these
factors (in percentage of original values), in which the biggest differences are mainly for quantity factors, except
for the case of private credit bureau coverage. The results show that, major weaknesses of ineffective countries in
banking system development are the ATM network, bank deposits to GDP, private credit coverage, bank assets,
and bank’s domestic credits. Those are the disadvantage of developing countries as they are still on their way
developing their financial and banking systems (see Table 4).
Table 4
Differences Between Projected and Original Values for Inefficient Countries
Total differences

Factors
Commercial banks’ assets/GDP
Domestic credit provided by banking sector (% of GDP)
Commercial banks’ deposits/GDP
Number of ATMs per 100,000 people
Number of branches per 100,000 people
Private credit bureau coverage (% of adults)
Public credit bureau coverage (% of adults)
Banks’ overhead costs/Total assets
Nonperforming loans ratios of commercial banks
Frequency of banking crises
Average

In value
21.72
2,338
21.67
1,373
379.4

1,230
56.46
0.741
80.16
21.68
552.3

In percentage of original value
45.56
45.55
56.44
75.88
51.7
52.34
10.71
5.376
7.201
11.59
36.24

In the second stage, the results from Tobit model show the relation between the banking systems’
effectiveness and various variables such as inflation level of the economy, income group that the country belongs
to, concentration of the banking system, etc., as summarized in Figure 1. It is obvious that higher inflation,
banking concentration, and bank’s cost-income ratio can reduce the effectiveness of the banking sector
(respectively significant at 1, 5 and 10 percent), while the high level of economic development (improving to
higher income group) can help increase the effectiveness of the banking system (5% significant level).

Figure 1. Determinants of the global banking effectiveness.

968

EFFECTIVENESS OF THE GLOBAL BANKING SYSTEM IN 2010
In the last stage, solving the non-linear problem of the CSW model (equation (3)) helped us defining a

common set weight for the ten factors of every country in the research (see Table 5). Noticeably, important
factors which strongly affect the performance of the banking sector in those countries are non-performing loans
ratio (79.49%), public credit bureau coverage (10.47%), and number of branches per 100,000 people (3.03%).
The other factors only keep minimum role (1% weight) in the final results. It shows that the effectiveness of the
banking sector is mainly affected by the damage of the global crisis, the (financial) public policy of the
government, and the development of the commercial bank system of each country respectively. It also suggests
that the quality of the banking sector is now becoming more important than the quantity aspect, not only for
countries with developed banking systems but for developing countries as well. Thus, country which focuses on
improving the quality of its banking sector can have higher effectiveness and is more stable.
Table 5
Common Set of Weights for the Effectiveness Scores
Factors
Commercial banks’ assets/GDP
Domestic credit provided by banking sector (% of GDP)
Commercial banks’ deposits/GDP
Number of ATMs per 100,000 people
Number of branches per 100,000 people
Private credit bureau coverage (% of adults)
Public credit bureau coverage (% of adults)
Banks’ overhead costs/Total assets
Nonperforming loans ratios of commercial banks
Frequency of banking crises

Weight
1.00

1.00
1.00
1.00
3.03
1.00
10.47
1.00
79.49
1.00

By applying this common set of weights, the effectiveness scores of country’s banking systems can be
calculated and countries can be ranked as in Table 6. Since non-performing loans ratio became the most
important factor, countries having problems with NPLs became less efficient and ranked bottom in the list,
including even Denmark and New Zealand.
Table 6
The Global Banking Effectiveness in 2010
Rank
1
2
3
4
5
6
7
8
9
10
11
12

Country
Japan
Canada
Chile
Malaysia
Australia
Switzerland
United States
Bulgaria
Argentina
Ecuador
Costa Rica
United Kingdom

Effectiveness score
23.231
23.231
23.231
22.275
22.177
22.079
22.037
21.755
21.671
21.461
21.421
21.415

Rank
33

34
35
36
37
38
39
40
41
42
43
44

Country
Kuwait
Venezuela, RB
Moldova
Lithuania
Bolivia
Croatia
Uganda
Jordan
Mozambique
Poland
Colombia
Armenia

Effectiveness score
17.606
17.556
17.504

17.394
17.333
17.307
16.947
16.891
16.853
16.771
16.770
16.276


969

EFFECTIVENESS OF THE GLOBAL BANKING SYSTEM IN 2010
(Table 6 continued)
Rank
Country
13
Korea, Rep.
14
Sweden
15
Brazil
16
El Salvador
17
Dominican Republic
18
Peru
19

Israel
20
Guatemala
21
Singapore
22
Estonia
23
Panama
24
Indonesia
25
Turkey
26
South Africa
27
Czech Republic
28
Hungary
29
Saudi Arabia
30
India
31
Macedonia, FYR
32
Slovak Republic

Effectiveness score
21.066

21.060
20.968
20.232
20.070
19.907
19.735
19.626
19.326
19.276
19.085
18.993
18.749
18.538
18.302
18.233
18.045
17.921
17.842
17.750

Rank
45
46
47
48
49
50
51
52
53

54
55
56
57
58
59
60
61
62
63
64

Country
Thailand
Russian Federation
Georgia
Morocco
Kazakhstan
Albania
Yemen, Rep.
Nigeria
Kenya
Bangladesh
Tunisia
Romania
Egypt, Arab Rep.
Mauritius
Denmark
New Zealand
Vietnam

Angola
Botswana
Sierra Leone

Effectiveness score
16.203
16.066
15.859
15.475
15.288
15.116
14.566
14.202
11.871
10.486
9.696
9.442
8.051
7.601
6.519
5.338
4.841
4.761
0.662
0.203

Conclusions
Using data from 64 countries in the world, this research applied the data envelopment analysis (DEA) to
evaluate the effectiveness of banking systems in the World in 2010. The research was divided into three steps, in
which the first stage applied data envelopment analysis method to build a common frontier for these 64 countries;

the second step detected the determinants of the banking sector’s effective; and the last step defined a common set
of weights for analyzing factors helping in ranking the effectiveness of the global banking system in 2010.
The research evaluated the effectiveness of the global banking systems using a dummy input and ten outputs
to create a common frontier for the whole banking systems of 64 countries (while previous studies used
institutional level data of smaller sample size); and after that building a common set of weights to calculate the
effectiveness scores of the global banking system, applied to the DEA method. This proposes an interesting
function for using DEA in examining the effectiveness (and efficiency) in the banking sector.
Findings from the research showed that banking systems in advanced economies are still more effective than
in developing countries. Reasons seem to be related to the development of the banking sector in quantity (number
of bank branches) and more importantly in quality aspects (including the NPL ratio, public credit bureau coverage,
bank concentration, bank’s capital, and cost-income ratio). It is also included the effect of economic development,
expresses through level of income (group) and inflation rates. These results partly explained the effect of the
current financial crisis to the banking sector, the role of public finance (and the government) in this kind of
situation, and the important role of developing commercial banking system to its efficiency and effectiveness.

References
Avkiran, N. K. (2001). Investigating technical and scale efficiencies of Australian universities through data envelopment analysis.


970

EFFECTIVENESS OF THE GLOBAL BANKING SYSTEM IN 2010

Socio-Economic Planing Sciences, 35, 57-80.
Avkiran, N. K. (2010). Association of DEA super-efficiency estimates with financial ratios: Investigating the case for Chinese banks.
Omega, 39, 323-334.
Banker, R. D., Charnes, A., & Cooper, W. W. (1984). Some models for estimating technical and scale inefficiencies in data
envelopment analysis. Management Science, 30(9), 1078-1092.
Beck, T., Demirgüç-Kunt, A., & Levine, R. (2000). A new database on financial development and structure. World Bank Economic
Review, 14, 597-605.

Bencivenga, V. R., & Smith, B. D. (1991). Financial intermediation and endogenous growth. Review of Economic Studies, 58,
195-209.
Berg, S. A., Forsund, F. R., Hjalmarsson, L., & Suominen, M. (1993). Banking efficiency in the Nordic countries. Journal of
Banking and Finance, 17, 371-388.
Bergendahl, G. (1995). DEA and benchmarks for Nordic banks. Working paper. Gothenburg University, Gothenberg, Sweden.
Berger, A. N. (2007). International comparison of banking efficiency. Financial Markets, Institutions and Instruments, 16(3),
119-144.
Berger, A. N., & Humphrey, D. B. (1997). Efficiency of financial institutions: International survey and directions for future research.
European Journal of Operational Research, 98, 175-212.
Bukh, P. N. D., Berg, S. A., & Forsund, F. R. (1995). Banking efficiency in the Nordic countries: A four-country Malmquist index
analysis. University of Aarhus, Denmark.
CGAP. (2010). Financial access 2010: The state of financial inclusion through the crisis. Washington, D.C.: Consultative Group to
Assist the Poor.
Charnes, A., Cooper, W. W., & Rhodes, E. (1978). Measuring the efficiency of decision making units. European Journal of
Operational Research, 2, 429-444.
Depotis, D. (2004). A reassessment of the human development index via data envelopment analysis. Journal of Operational
Research Society, 1-12.
Dyson, R. G., Allen, R., Camanho, A. S., Podinovski, V. V., Sarrico, C. S., & Shale, E. A. (2001). Pitfalls and protocols in DEA.
European Journal of Operational Research, 132, 245-259.
Fare, R., & Grosskopf, S. (2004). Modeling undesirable factors in efficiency evaluation: Comment. European Journal of
Operational Research, 157, 242-245.
Farrell, M. J. (1957). The measurement of productive efficiency. Journal of the Royal Statistical Society, 120(3), 253-281.
Fecher, F., & Pestieau, P. (1993). Efficiency and competition in O.E.C.D. financial services. In H. O. Fried, C. A. K. Lovell, & S. S.
Schmidt (Eds.), The measurement of productive efficiency: Techniques and applications (pp. 374-385). Oxford University
Press, UK.
Fethi, M. D., & Pasiouras, F. (2010). Assessing bank efficiency and performance with operational research and artificial intelligence
techniques: A survey. European Journal of Operational Research, 204, 189-198.
Goldsmith, R. W. (1969). Financial structure and development. New Haven, C.T.: Yale University Press.
Grosskopf, S., & Valdmanis, V. (1987). Measuring hospital performance: A non-parametric approach. Journal of Health Economics,
6, 89-107.

IMF. (2010). Global financial stability report: Sovereigns, funding, and systemic liquidity. Washington, D.C.: International
Monetary Fund.
Jahanshahloo, G. R., Memariani, A., Lotfi, F. H., & Rezai, H. Z. (2005). A note on some of DEA models and finding efficiency and
complete ranking using common set of weights. Applied Mathematics and Computation, 166, 265-281.
Kao, C., & Hung, H. T. (2005). Data envelopment analysis with common weights: the compromise solution approach. Journal of
Operational Research Society, 56, 1196-1203.
King, R. G., & Levine, R. (1993a). Finance and growth: Schumpeter might be right. Quarterly Journal of Economics, 108, 717-737.
King, R. G., & Levine, R. (1993b). Finance, entrepreneurship, and growth: Theory and evidence. Journal of Monetary Economics,
32, 513-542.
Kwan, S. H. (2003). Operating performance of banks among Asean economies: An international and time series comparison.
Journal of Banking and Finance, 27, 471-489.
Laeven, L., & Valencia, F. (2010). Resolution of banking crises: The good, the bad, and the ugly. IMF Working paper 10/146.
Liu, F. H. F., & Peng, H. H. (2008). Ranking units of the DEA frontier with common weights. Computers and Operations Research,
35, 1624-1637.


971

EFFECTIVENESS OF THE GLOBAL BANKING SYSTEM IN 2010

Lozano-Vivas, A., Pastor, J. T., & Hasan, I. (2001). European bank performance beyond country borders: What really matters.
European Finance Review, 5, 141-165.
Mahlberg, B., & Obersteiner, M. (2001). Remeasuring the HDI by data envelopment analysis. Laxenburg, Austria: International
Institute for Applied System Analysis (IIASA).
Maudos, J., Pastor, J. M., Perez, F., & Quesada, J. (2002). Cost and profit efficiency in European banks. Journal of International
Financial Markets, Institutions and Money, 12, 33-58.
McKinnon, R. I. (1973). Money and capital in economic development. Washington D.C.: Brookings Institution.
Pastor, J., Perez, F., & Quesada, J. (1997). Efficiency analysis in banking firms: An international comparisons. European Journal of
Operational Research, 98, 396-408.
Pastor, J. T., Knox Lovell, C., & Tulkens, H. (2006). Evaluating the financial performance of bank branches. Annals of Operations

Research, 145(1/July), 321-337.
Peristiani, S. (1997). Do mergers improve the x-efficiency and scale efficiency of U.S. banks? Evidence from the 1980s. Journal of
Money, Credit, and Banking, 29(3), 326-337.
Roll, Y., & Golany, B. (1993). Alternate methods of treating factor weights in DEA. OMEGA, 21(1), 99-109.
Ruthenberg, D., & Elias, R. (1996). Cost economies and interest rate margins in a unified European banking market. Journal of
Economics and Business, 48, 231-249.
Sathye, M. (2005). Technical efficiency of large bank production in Asia and the Pacific. Multinational Finance Journal, 9(1-2), 1-22.
Schaffnit, C., Rosen, D., & Paradi, J. C. (1997). Best practice analysis of bank branches: An application of DEA in a large Canadian
bank. European Journal of Operational Research, 98, 269-289.
Schumpeter, J. A. (1911). The theory of economic development. Cambridge, M.A.: Havard University Press.
Seiford, L. M., & Zhu, J. (2002). Modeling undesirable factors in efficiency evaluation. European Journal of Operational Research,
142, 16-20.
Shaw, E. S. (1973). Financial deepening in economic development. New York: Oxford University Press.
Sherman, D. H., & Gold, F. (1985). Bank branches operating efficiency: Evaluation with Data Envelopment Analysis. Journal of
Banking and Finance, 9, 297-315.
Tobin, J. (1958). Estimation of relationships for limited dependent variables. Econometrica, 26(1), 24-36.
WEF. (2010). The financial development report 2010. Geneva, Switzerland: World Economic Forum.
Zhu, J. (2001). Super-efficiency and DEA sensitivity analysis. European Journal of Operational Research, 129, 443-455.
Appendix A
Table A1
Countries’ Data
Country
Albania
Angola
Argentina
Armenia
Australia
Bangladesh
Bolivia
Botswana

Brazil
Bulgaria
Canada
Chile
Colombia
Costa Rica
Croatia
Czech Republic
Denmark

y1
0.77
0.24
0.18
0.20
1.29
0.54
0.32
0.19
0.91
0.85
1.40
0.78
0.51
0.49
0.90
0.67
2.42

y2

66.88
9.34
24.47
16.66
143.75
59.38
55.24
-11.17
117.85
66.74
178.07
115.92
43.26
53.90
75.09
57.98
211.45

y3
0.74
0.24
0.20
0.12
1.14
0.51
0.38
0.58
0.66
0.77
1.04

0.55
0.22
0.25
0.77
0.62
0.72

y4
2.37
9.58
14.91
1.37
64.18
0.06
4.80
9.00
17.82
29.79
135.23
24.03
9.60
12.83
40.10
19.57
52.39

y5
2.11
0.60
10.01

7.59
29.86
4.47
1.53
3.77
14.59
13.87
45.60
9.39
8.74
9.59
23.36
11.15
37.63

y6
0.00
0.00
100.00
34.50
100.00
0.00
33.90
51.90
59.20
6.20
100.00
33.90
60.50
56.00

77.00
73.10
5.20

y7
9.90
2.50
34.30
4.40
0.00
0.90
11.60
0.00
23.70
34.80
0.00
32.90
0.00
24.30
0.00
4.90
0.00

y8
0.24
0.23
0.18
0.22
0.24
0.24

0.21
0.22
0.14
0.25
0.24
0.23
0.21
0.15
0.24
0.24
0.23

y9
16.70
5.34
20.60
18.90
22.80
12.10
19.00
0.00
20.20
20.90
22.20
22.30
19.30
21.80
18.40
20.00
3.30

y10
3.00
4.00
0.00
3.00
4.00
3.00
2.00
4.00
2.00
3.00
4.00
2.00
2.00
2.00
3.00
3.00
3.00


972

EFFECTIVENESS OF THE GLOBAL BANKING SYSTEM IN 2010

(Table A1 continued)
Country
y1
y2
y3

y4
y5
Dominican Republic
0.22
39.06 0.21
15.08
6.00
Ecuador
0.28
19.76 0.28
6.32
9.30
Egypt, Arab Rep.
0.56
77.70 0.75
1.78
3.62
El Salvador
0.42
49.94 0.42
11.07
4.62
Estonia
1.18
97.26 0.48
57.70 15.19
Georgia
0.40
32.87 0.22
1.17

3.14
Guatemala
0.44
40.11 0.37
20.20 10.12
Hungary
0.90
80.70 0.50
29.40 28.25
India
0.69
68.35 0.70
7.29 10.64
Indonesia
0.29
36.75 0.33
4.84
8.44
Israel
0.95
82.16 0.87
18.81 14.74
Japan
1.48
379.30 1.80
113.75
9.98
Jordan
1.29
114.92 1.09

9.38 10.02
Kazakhstan
0.89
33.51 0.39
7.01
2.47
Kenya
0.29
40.09 0.29
0.99
1.38
Korea, Rep.
1.21
112.32 0.59
90.03 13.40
Kuwait
0.81
74.92 0.71
19.69
8.27
Lithuania
0.73
64.37 0.36
28.78
3.39
Macedonia, FYR
0.55
42.70 0.56
49.97 26.79
Malaysia

0.99
115.54 1.09
16.44
9.80
Mauritius
0.88
111.78 0.86
22.04 11.92
Moldova
0.49
39.76 0.45
236.07 10.07
Morocco
0.91
95.54 0.94
9.68 15.80
Mozambique
0.22
14.14 0.29
4.90
2.92
New Zealand
1.55
156.45 0.96
50.36 28.04
Nigeria
0.45
26.73 0.26
18.63
6.42

Panama
0.86
85.41 0.88
16.19 12.87
Peru
0.21
18.51 0.26
5.85
4.17
Poland
0.55
60.06 0.42
17.31
8.17
Romania
0.52
40.91 0.32
12.47 13.76
Russian Federation
0.49
26.03 0.36
6.28
2.24
Saudi Arabia
0.55
9.42 0.53
14.70
5.36
Sierra Leone
0.09

7.35 0.15
1.14
2.76
Singapore
1.10
79.17 1.18
37.93
9.13
Slovak Republic
0.55
53.80 0.49
29.21 10.28
South Africa
0.95
215.47 0.67
17.50
5.99
Sweden
1.40
133.43 0.57
29.56 21.80
Switzerland
1.89
180.59 1.31
70.60 37.99
Thailand
0.84
145.65 0.79
17.05
7.18

Tunisia
0.62
72.04 0.52
17.69 15.51
Turkey
0.51
52.54 0.42
18.00
8.50
Uganda
0.22
11.45 0.20
0.70
0.53
United Kingdom
2.08
211.35 1.71
42.45 18.35
United States
0.73
271.64 0.83
120.94 30.86
Venezuela, RB
0.38
20.49 0.39
16.60
4.41
Vietnam
1.24
94.99 0.93

15.36
3.42
Yemen, Rep.
0.13
11.29 0.21
2.75
1.97
Note. y1, y2,..., y10 are respectively referred to ten factors in Table 3.

y6
46.10
46.00
8.20
94.60
20.60
12.20
28.40
10.30
10.20
0.00
89.80
76.20
0.00
29.50
2.30
93.80
30.40
18.40
0.00
82.00

0.00
0.00
14.00
0.00
100.00
0.00
45.90
31.80
68.30
30.20
14.30
17.90
0.00
40.30
44.00
54.70
100.00
22.50
32.90
0.00
42.90
0.00
100.00
100.00
0.00
0.00
0.00

y7
29.70

37.20
2.50
21.00
0.00
0.00
16.90
0.00
0.00
22.00
0.00
0.00
1.00
0.00
0.00
0.00
0.00
12.10
28.10
48.50
36.80
0.00
0.00
2.30
0.00
0.00
0.00
23.00
0.00
5.70
0.00

0.00
0.00
0.00
1.40
0.00
0.00
0.00
0.00
19.90
15.90
0.00
0.00
0.00
0.00
19.00
0.20

y8
0.13
0.22
0.22
0.23
0.17
0.18
0.01
0.00
0.24
0.23
0.24
0.25

0.24
0.23
0.21
0.25
0.23
0.24
0.22
0.24
0.24
0.21
0.25
0.20
0.25
0.23
0.19
0.23
0.24
0.18
0.18
0.25
0.16
0.26
0.24
0.22
0.25
0.23
0.24
0.24
0.22
0.20

0.25
0.22
0.21
0.25
0.25

y9
19.80
20.80
8.50
20.50
21.40
19.20
20.90
20.30
21.00
20.10
21.80
21.60
19.10
18.20
14.30
22.20
20.20
18.70
16.50
18.50
2.50
18.10
17.30

20.50
1.70
17.00
21.60
21.10
18.90
9.50
19.50
21.90
0.00
21.90
20.10
19.40
22.30
22.80
17.60
7.80
19.70
21.10
21.70
20.30
21.40
2.00
18.00

y10
3.00
2.00
3.00
3.00

3.00
3.00
4.00
2.00
3.00
3.00
3.00
3.00
3.00
3.00
2.00
3.00
3.00
3.00
3.00
3.00
4.00
4.00
3.00
3.00
4.00
3.00
3.00
3.00
3.00
3.00
2.00
4.00
3.00
4.00

3.00
4.00
2.00
3.00
2.00
3.00
2.00
3.00
3.00
2.00
3.00
3.00
3.00


973

EFFECTIVENESS OF THE GLOBAL BANKING SYSTEM IN 2010
Table A2
Dynamic DEA Efficiencies
Rank
1
2
3
4
5
6
7
8
9

10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32

Country
Moldova
Malaysia
Japan
Canada
United Kingdom

Denmark
Mauritius
Argentina
Switzerland
United States
Chile
Guatemala
Singapore
Macedonia, FYR
South Africa
New Zealand
Australia
Bulgaria
Vietnam
Sweden
Korea, Rep.
El Salvador
Botswana
Saudi Arabia
Angola
Ecuador
Yemen, Rep.
Costa Rica
Morocco
Tunisia
Peru
Israel

DSW score
1.000

1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
0.985
0.984
0.980
0.972
0.970
0.969

0.965

Rank
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59

60
61
62
63
64

Note. First 25 countries are ranked based on super-efficiency DEA results.

Country
Thailand
India
Dominican Republic
Croatia
Panama
Czech Republic
Lithuania
Estonia
Venezuela, RB
Poland
Indonesia
Jordan
Albania
Brazil
Slovak Republic
Uganda
Bangladesh
Kuwait
Turkey
Mozambique
Kazakhstan

Nigeria
Hungary
Armenia
Bolivia
Egypt, Arab Rep.
Russian Federation
Colombia
Georgia
Kenya
Romania
Sierra Leone

DSW score
0.961
0.957
0.955
0.951
0.947
0.947
0.944
0.939
0.939
0.938
0.937
0.935
0.931
0.930
0.929
0.925
0.920

0.912
0.904
0.901
0.893
0.893
0.890
0.870
0.867
0.863
0.855
0.846
0.842
0.813
0.750
0.750


Chinese Business Review, ISSN 1537-1506
November 2011, Vol. 10, No. 11, 974-984

Application of Pareto Distribution in Wage Models∗
Diana Bílková
University of Economics in Prague, Prague, Czech Republic

This paper deals with the use of Pareto distribution in models of wage distribution. Pareto distribution cannot
generally be used as a model of the whole wage distribution, but only as a model for the distribution of higher or of
the highest wages. It is usually about wages higher than the median. The parameter b is called the Pareto coefficient
and it is often used as a characteristic of differentiation of fifty percent of the highest wages. Pareto distribution is
so much the more applicable model of a specific wage distribution, the more specific differentiation of fifty percent
of the highest wages will resemble to differentiation that is expected by Pareto distribution. Pareto distribution

assumes a differentiation of wages, in which the following ratios are the same: ratio of the upper quartile to the
median; ratio of the eighth decile to the sixth decile; ratio of the ninth decile to the eighth decile. This finding may
serve as one of the empirical criterions for assessing, whether Pareto distribution is a suitable or less suitable model
of a particular wage distribution. If we find only small differences between the ratios of these quantiles in a specific
wage distribution, Pareto distribution is a good model of a specific wage distribution. Approximation of a specific
wage distribution by Pareto distribution will be less suitable or even unsuitable when more expressive differences
of mentioned ratios. If we choose Pareto distribution as a model of a specific wage distribution, we must reckon
with the fact that the model is always only an approximation. It will describe only approximately the actual wage
distribution and the relationships in the model will only partially reflect the relationships in a specific wage
distribution.
Keywords: Pareto distribution, Pareto coefficient, estimation methods for parameters, least squares method, wage
distributions

Pareto Distribution
The question of income and wage distributions and their models is quite extensively treated in the statistical
literature (Bartošová, 2006; Bartošová & Bína, 2009; Bílková, 2007; Dutta, Sefton, & Weale, 2001; Majumder &
Chakravarty, 1990; McDonald & Snooks, 1985; McDonald, 1984; McDonald & Butler, 1987).
∗

Acknowledgment: The paper was supported by grant project IGS 24/2010 called “Analysis of the Development of Income
Distribution in the Czech Republic Since 1990 to the Financial Crisis and Comparison of This Development With the
Development of the Income Distribution in Times of Financial Crisis, According to Sociological Groups, Gender, Age, Education,
Profession Field and Region” from the University of Economics in Prague.
Diana Bílková, Ing./Dr., Department of Statistics and Probability, Faculty of Informatics and Statistics, University of Economics
in Prague.
Correspondence concerning this article should be addressed to Diana Bílková, Department of Statistics and Probability, Faculty
of Informatics and Statistics, University of Economics in Prague, Sq. W. Churchill 1938/4, Prague 3, Czech Republic, post code:
130 67. E-mail:

APPLICATION OF PARETO DISTRIBUTION IN WAGE MODELS

975

Pareto distribution is usually used as a model of the distribution of the largest wages, not for the whole wage
distribution. In this article, we will consider using the Pareto distribution to model wages higher than median.
The 100·P% quantile of the wage distribution will be denoted by xP, 0 < P < 1. This value represents the upper
bound of 100·P% lowest wages and also the lower bound of 100(1 – P) % highest wages. A particular quantile
(denoted as xP0) which will be the lower bound of some small number of the highest wages is usually set to be the
maximum wage. If the following formula (1) holds for any quantile xP, the wage distribution is Pareto distribution.

x P0
xP

⎛ 1− P ⎞
=⎜
⎟
⎝ 1 − P0 ⎠

b

(1)

The parameter b of the Pareto distribution (1) is called the Pareto coefficient. It can be used as a
characteristic of differentiation of 50% highest wages.
We will now consider a pair of quantiles xP1 and xP2, P1 < P2. It follows from equation (1) that:

x P0
x P1

b

⎛ 1 − P1 ⎞
⎟
= ⎜⎜
⎟
P
−
1
0⎠
⎝

(2)

and

x P0
x P2

b

⎛ 1 − P2 ⎞
⎟
= ⎜⎜
⎟
⎝ 1 − P0 ⎠

(3)

From what we can derive for the rate of xP2 to xP1 that:

x P2
x P1

⎛ 1 − P1 ⎞
=⎜
⎟
⎝ 1 − P2 ⎠

b

(4)

The rate xx P 2 is an increasing function of the Pareto coefficient b. If the rate of quantiles increases, the
P1

relative differentiation of wages increases too. If only absolute differences between quantiles increase, only the
absolute differentiation of wages increases.
It follows from the equation (1) that once the values xP0 and b are chosen, we can determine the quantile xP
for any chosen P or the other way around for any value xP we can find the corresponding value of P. In the first
case, it is advantageous to write the equation (1) as:

xP =

x P0
b

⎛ 1− P ⎞
⎜
⎟

⎜1 − P ⎟
0⎠
⎝

(5)

or after logarithmic transformation as:

log xP = log xP0 − b [log (1 - P) - log (1 − P0)]

(6)

in the second case:
1 − P = (1 − P0) b

x P0
xP

(7)


976

APPLICATION OF PARETO DISTRIBUTION IN WAGE MODELS

or after logarithmic transformation as:

log (1 − P) = log (1 − P 0) +

1

(log x P − log x P)
0
b

(8)

The equations (2)-(4) will after logarithmic transformation have the following forms:
log
b=
log

x P1
1 − P1

(9)

1 − P0

log
b=

x P0

x P2

x P1
1 − P1
log
1 − P2

(10)

It follows from the equation (9) that instead of the Pareto coefficient b we can use any other quantile xP1 of
the Pareto distribution and it follows from the equation (10) that the Pareto coefficient b can be calculated using
any known quantiles xP1 and xP2. Then we can also determine the value xP0 using the formulas:
b

⎛ 1 − P1 ⎞
⎟,
x P 0 = x P1 ⎜⎜
⎟
⎝ 1 − P0 ⎠

(11)

b

⎛ 1 − P2 ⎞
⎟
x P 0 = x P 2 ⎜⎜
⎟
P
−
1
0
⎠
⎝

(12)

The model characterized with the relationship (1) will be practically applicable if the following is known:
 The value of the quantile that characterizes the assumed wage maximum and the value of the Pareto

coefficient b;
 The value of the quantile that characterizes the assumed wage maximum and the value of any other quantile;
 The values of any two quantiles of the Pareto distribution.

Any two quantiles can be written as xP and xP+k, where 0 < k < 1 – P . Using the equation (4), we can derive
for the rate of these two quantiles:
b

x P + k ⎛⎜ 1 − P ⎞⎟
=⎜
⎟
xP
⎝1 − P − k ⎠

The rate (13) will be equal for such pairs of quantiles for which the following formula holds:
1− P
= c,
1− P − k

(13)

(14)

where c is a constant, i.e., the rate will be the same for all pairs of quantiles for which:

k=

c −1
(1 − P)
c

(15)

We will use the constant c = 2 in equation (15) and we will choose gradually P = 0.5; 0.6; 0.8. Then using the
equation (13) we can show the equality of rates of some frequently used quantiles:


APPLICATION OF PARETO DISTRIBUTION IN WAGE MODELS
x 0.75 = x 0.8 = x 0.9
x 0.5
x 0.6 x 0.8

977

(16)

From the relationship (16) we can conclude that Pareto distribution assumes such a wage differentiation for
which the rate of the upper quartile to median is the same as:
 The rate of the 8th to the 6th decile;
 And as the rate of the 9th to the 8th decile.

If in a particular case, the observed differences of the rates of the above mentioned quantiles are negligible,
Pareto distribution will be an appropriate model of the considered wage distribution. In the case, the differences
are quite material, the approximation of the considered wage distribution with Pareto distribution will be more or
less inappropriate. More about the theory of Pareto distribution is described in statistical literature (Forbes, Evans,
Hastings, & Peacock, 2011; Johnson, Kotz, & Balakrishnan, 1994; Kleiber & Kotz, 2003; Krishnamoorthy,
2006).

Parameter Estimates
If the Pareto distribution is chosen as a model for a particular distribution we have to keep in mind that this
model is only an approximation. The wage distribution will be only approximated and the relations derived from
the model will also hold for the “true distribution” only approximately. Which relations will hold more precisely
and for which the precision will be lower will be mostly dependent on the method of parameter estimates.
There are many possibilities to choose from. In the following text the quantiles of Pareto distribution will be
denoted as xP and the quantiles of the observed wage distribution will be denoted as yP.
First we need to decide which quantile to choose as xP0. It this article we will assume that xP0 = x0.99. From
the equation (1) we can see that the considered Pareto distribution will be defined by the equation:
b

x 0.99 = ⎛ 1 − P ⎞
(17)
⎜
⎟
x P ⎝ 0.01 ⎠
Then we need to determine the value x0.99 and the value of the Pareto coefficient b. Because it is necessary to
estimate the values of two parameters we need to choose two equations to estimate from.
A natural choice is the equation xP0 = yP0; that is in our case x0.99 = y0.99. As the other equation we set a
quantile xP1 equal to the corresponding observed quantile, i.e., xP1 = yP1. In this case, the parameters of the model
will be:

x P0 = y P0

(18)

and using equation (9):
log

b=

y P0

y P1
1 − P1
log
1 − P0

(19)

We can get different modifications using different choice of the maximum wage and the second quantile. If we
use equation x0.99 = y0.99 and we use the median in the second equation, i.e., x0.5 = y0.5 we get a model with


978

APPLICATION OF PARETO DISTRIBUTION IN WAGE MODELS

parameters:

x 0.99 = y 0.99

(20)

y 0.99
y 0.5
b=
0.5
log

0.01

(21)

log

Another possibility is setting any two quantiles of the model equal to the quantiles of the observed distribution:

x P1 = y P1

(22)

x P2 = y P2

(23)

Using the formula (10), we get the following parameter estimates:
y
log P 2
y P1
b=
1 − P1
log
1 − P2

(24)

and from equations (11) and (12) we get:
b

⎛ 1 − P1 ⎞
⎛ 1 − P2 ⎞
x P 0 = y P1 ⎜
⎟ = y P2 ⎜
⎟
⎝ 1 − P0 ⎠
⎝ 1 − P0 ⎠

b

(25)

With this alternative we can also get numerous modifications depending on the choice of quantiles yP1 and
yP2 that are used.
The third possibility is based on the request that xP0 = yP0 and that the rate of some other two quantiles of the
Pareto distribution xP2/xP1 is equal to the rate yP2/yP1 of correspoding quantiles of the wage distribution observed.
In this case we will estimate the parameters using equation (10):
x P0 = y P0
log

b=

(26)

y P2

y P1
1 − P1
log
1 − P2

(27)

In this case, notwithstanding that xP2/xP1 = yP2/yP1 holds, the equality of quantiles itself, xP1 ≠ yP1 and xP2 ≠ yP2,
does not hold. In this case, we can also arrive to numerous modifications depending on what maximum wage is
chosen and what quantiles yP1 and yP2 are chosen.
For all of the above methods the equality of two characteristics of the model and the observed distribution
was required. There are also different approaches to the parameter estimates.
The least squares method is frequently used for the Pareto distribution parameter estimates as well. We will
consider the following quantiles of the observed wage distribution yP1, yP2, …, yPk and corresponding quantiles of the


APPLICATION OF PARETO DISTRIBUTION IN WAGE MODELS

979

Pareto distribution xP1, xP2, …, xPk. The model distribution will be most precise when the sum of squared differences:
k
∑ ( yP
i
i =1

− x P i) 2

(28)

is minimized. In this case closed formula solution does not exist. Therefore sum of squared differences of
logarithms of quantiles is often considered:
k
∑ (log y P

i
i =1

− log x Pi)

2

(29)

Minimizing the objective function (29), it is possible to derive the following estimates:

k
b=

k
∑ log y
Pi
i =1

log

1 − P0 k
− ∑ log y P
i
1 − P i i =1

k
1 − P0
∑ log
i =1

1 − Pi
2

⎛k
1 − P0 ⎞
− ⎜ ∑ log
⎟
⎝ i =1 1 − P i ⎠
k
1 − P0
k
∑ log
∑ log y
1
=
i
1 − Pi
Pi
−b
log x P0 = i =1
k
k
k
1 − P0
k ∑ log 2
i =1
1 − Pi

(30)

(31)

In the case we use this estimating method, it is needed to keep in mind that the equality of model quantiles
and observed quantiles is not guaranteed for any P. Again we can arrive to different results depending on what
quantiles yP1, yP2, …, yPk are used for the calculations. Furthermore the parameter estimates also depend on the
choice of the maximum wage.

Characteristics of the Appropriateness of the Pareto Distribution
For the application of Pareto distribution as a model of the wage distribution, it is crucial that the model fits
the observed distribution as close as possible. It is important that the observed relative frequencies in particular
wage intervals are as close to the theoretical probabilities assigned to these intervals by the model as possible.
It is needed to note that the same parameter estimation method does not always lead to the best results. It is
of particular importance in “what direction” is the observed wage distribution different from Pareto distribution.
Pareto distribution assumes such wage differentiation that the relations (16) hold. With real data we can
encounter many different situations:
y 0.75 y 0.8 y 0.9
<
<
y 0.5 y 0.6 y 0.8

(32)

y 0.75 y 0.8 y 0.9
>
>
y 0.5 y 0.6 y 0.8

(33)

y 0.75 y 0.9 y 0.8

<
<
y 0.5 y 0.8 y 0.6

(34)

y 0.75 y 0.9
>
>
y 0.5 y 0.8
y 0.8 y 0.75
<
<
y 0.6 y 0.5

y 0.8
y 0.6
y 0.9
y 0.8

(35)
(36)


980

APPLICATION OF PARETO DISTRIBUTION IN WAGE MODELS

y 0.8 y 0.75 y 0.9
>

>
y 0.6 y 0.5 y 0.8

(37)

It follows from equations (32)-(37) that the observed distributions will more or less systematically differ
from the Pareto distribution. In the case of equation (32) the differentiation of the observed wage distribution is
higher; in the case of equation (33) the differentiation will be lower than in the case of Pareto distribution. Some
bias occurs in cases equations (34)-(37) as well (but cannot be so specified). Systematical bias should be a signal
for potential adjustment of the model which could be based for example on adding one or more parameters into
the model. These adjustments usually lead to more complicated models. Therefore, the above mentioned bias is
often neglected and simple models are preferred even though they lead to some bias.

Wage Distribution of Males and Females in the Czech Republic in 2001-2008
The data used in this article is the gross monthly wage of male and female in CZK in the Czech Republic in
the years 2001-2008. Data were sorted in the table of interval distribution with opened lower and upper bound in
the lowest and highest interval respectively. The source is the web page of the Czech statistical office. The
following quantiles were calculated (see Table 1).
Table 1
Median y0.50 (in CZK), 6th Decile y0.60 (in CZK), Upper Quartile y0.75 (in CZK), 8th Decile y0.80 (in CZK), 9th
Decile y0.90 (in CZK) a 99th Percentile y0.99 (in CZK) of Gross Monthly Wages in the Czech Republic in the Years
2001-2008 (Total and Split up to Male and Female Separated)
Total

Males

Females

Year
2001

2002
2003
2004
2005
2006
2007
2008
2001
2002
2003
2004
2005
2006
2007
2008
2001
2002
2003
2004
2005
2006
2007
2008

y0.50
12,502
15,545
16,735
17,709
18,597

19,514
20,987
22,310
14,152
16,985
18,240
19,344
20,281
21,199
22,933
24,498
10,770
13,746
14,831
15,642
16,454
17,311
18,390
19,399

y0.60
14,042
17,125
18,458
19,557
20,566
21,564
23,227
24,696
15,781

18,667
20,116
21,321
22,446
23,460
25,366
27,115
12,187
15,181
16,453
17,303
18,211
19,202
20,392
21,600

y0.75
16,987
20,215
22,224
23,077
24,470
25,675
27,590
29,553
19,037
22,604
24,145
25,306
26,822

28,090
30,284
32,343
14,655
17,727
19,281
20,293
21,426
22,530
24,024
25,558

y0.80
18,254
22,193
23,797
24,849
26,328
27,693
29,900
31,769
20,697
24,199
26,041
27,286
28,989
30,525
32,663
35,105
15,700

18,903
20,628
21,560
22,804
23,966
25,924
27,215

y0.90
23,319
27,754
29,590
31,082
33,292
35,230
37,892
40,541
26,264
31,101
34,564
34,819
37,211
39,381
42,815
46,375
18,904
23,291
24,637
25,776
27,503

29,082
31,338
33,405

y0.99
44,921
47,172
47,719
56,369
56,852
57,326
66,395
68,828
46,781
48,047
48,417
57,514
57,808
58,104
70,522
72,338
37,526
43,339
44,883
50,776
52,508
54,054
58,649
63,628