Global link between corruption and wealth

Worldwide clustering of the corruption perception
Michal Paulus, Ladislav Kristoufek
Institute of Economic Studies, Faculty of Social Sciences, Charles University, Opletalova 26, 110 00,
Prague 1, Czech Republic
Abstract
We inspect a possible clustering structure of the corruption perception among 134 coun-
tries. Using the average linkage clustering, we uncover a well-defined hierarchy in the
relationships among countries. Four main clusters are identified and they suggest that
countries worldwide can be quite well separated according to their perception of corrup-
tion. Moreover, we find a strong connection between corruption levels and a stage of
development inside the clusters. The ranking of countries according to their corruption
perfectly copies the ranking according to the economic performance measured by the gross
domestic product per capita of the member states. To the best of our knowledge, this
study is the first one to present an application of hierarchical and clustering methods to
the specific case of corruption.
Keywords: clustering, corruption perception
PACS codes: 89.20.-a, 89.65.s, 89.65.Gh
1. Introduction
Corruption has been understood as an important aspect forming societies as well as
economic connections for a long time. Recently, it has also become more evident that
grasping corruption quantitatively can yield important additions to models in social sci-
ences and especially economics. Various studies examine corruption in fields of microeco-
nomics, macroeconomics, institutional economics or international trade and its influence
on economic performance, efficiency and decision making. Corruption is thus in the centre
of interest of many empirical and theoretical works [1, 2, 3].
The most prominent topic regarding corruption is its impact on economic growth show-
ing a negative influence of corruption on the growth [4, 5]. Another attractive domain of
corruption effects is in the inequality studies, which show that corruption increases inequal-
ity [6]. The increased corruption also tends to boost public investments but these yield
lower returns [7]. The higher corruption not only lowers the public investment returns but
Email address: kristoufek@icloud.com (Ladislav Kristoufek)
Preprint submitted to Physica A February 3, 2015
arXiv:1502.00104v1[q-fin.EC]31Jan2015

also the level of private investments with a higher public-to-private investment ratio as a
result [4, 8, 9].
In general, the empirical studies reveal negative effects of corruption on economic per-
formance in terms of lower growth, private investments, higher inequality or wasting of
talents. Except for the studies concerning the macroeconomic performance, corruption
studies are becoming more popular in the international trade area as well. The theoretical
effect of corruption on mutual trade is not straightforward as it can have both negative
and positive effects. In the case of an adverse impact, corruption can be viewed as an
additional tax because of bribes, a complicated enforcement of mutual agreements or an
institutional safety [10]. The other perspective is completely opposite. Corruption helps
to facilitate mutual business and to guarantee basic rights when the institutions are of a
low quality.
When we abandon theoretical considerations and explore empirical literature, we can-
not unanimously conclude that corruption has only a negative role in FDI (foreign direct
investments) or trade, although it seems to be mostly the case [11, 12, 13]. Studies dis-
rupting the prevailing negative effects of corruption are present as well. Egger & Winner
[14] show that corruption can stimulate FDI in the short as well as the long term. They
argue that a corrupt country attracts investments from other corrupt countries. In other
words, a country attracts investments from countries with a similar institutional quality
[15]. Brada et al. [16] go further and they develop a theoretical model showing that
the most beneficial position in the FDI engagement is to be a “medium corrupt” coun-
try. Such a country has advanced technologies, good institutions but also knowledge of
the corruption culture. Hence they can be engaged in an FDI mediation with corrupt as
well as non-corrupt countries. If a country is highly corrupt, it has a problem attracting
investments from low corrupt countries and vice versa.
Such an ambiguous role of corruption is found in the international trade analyses also.
Horsewood & Voicu [10] report adverse effects of corruption on mutual trade. On the other
hand, de Jong & Bogmans [17] find that corruption can stimulate imports under special
circumstances. Some further results can be found in Refs. [18, 19, 20].
Analysis of corruption has been of interest in the interdisciplinary research as well.
Shao et al. [21] demonstrate a power-law functional dependence between corruption level
and economic factors, such as country wealth and foreign direct investments per capita.
Podobnik et al. [22] report that an increase of the corruption perception index leads
to an increase of the annual GDP per capita growth rate between 1999 and 2004. In
addition, the authors introduce a new measure to quantify the relative corruption between
countries based on their respective wealth as measure by GDP per capita. Podobnik et
al. [23] further propose a production function (divided into the private and public sectors)
where GDP depends on market capitalization, the public or private section workforce
and competitiveness level to quantify the public sector efficiency. It is demonstrated that
the less corrupt countries receive more investments that the more corrupt ones. Finally
Podobnik et al. [24] argue that it is improbable that a country would rapidly improve its
corruption rank.
Even though the effects of corruption on various economic indicators and measures
2

have been often investigated, an important message remains hidden beneath the results.
The reported findings suggest that there are at least several groups of countries which
are differentiated according to their corruption level and these groups tend to trade with
each other more and they invest into similar countries as well. However, there is no
study examining the actual clustering of countries according to their corruption or their
corruption perception. Here, we aim to fill this gap in the literature by bridging an economic
topic with tools of complex systems and interdisciplinary physics – clustering analysis. In
the following section, we describe the utilized methodology in some detail. The next section
describes the analyzed dataset followed by results presentation and interpretation. The last
section concludes. We find that countries indeed cluster according to their corruption levels
and we identify four well-divided clusters. Moreover, these clusters nicely correspond to a
stage of development of the member countries. The level of corruption is thus found to be
tightly connected to country’s development.
2. Methodology
The clustering analysis has now quite a long history in economic and financial ap-
plications mainly due a pioneering work of Mantegna [25] utilizing minimum spanning
trees and hierarchical trees to uncover a hierarchical structure in the US stocks. The
methodology has then been applied to various markets such as stocks and stock indices
[26, 27, 28, 29, 30, 31, 32, 33, 34, 35], foreign exchange rates [36, 37, 38, 39], import/export
networks [40], interest rates [41], and commodities [42, 43, 44]. Most of the studies focus
on correlation analysis, using the correlation matrix as a starting point. The dissimilarity
(distance) matrix is constructed using the correlations which then allows for the spanning
and hierarchical trees construction. However, our analysis cannot use such an approach
and we must apply an alternative one due to a specific nature of the corruption data and
a different aim of the analysis.
We analyze a corruption clustering of various countries. For each country, we get
a time series of the Freedom from Corruption (COR) index of the Heritage Foundation
which ranges between 0 and 100 with higher values meaning lower corruption (more details
are given in the next section). We are primarily interested in a common level of the index
as well as in a co-movement between indices of different countries. Therefore, we cannot
use a standard correlation procedure which demeans the series and thus the information
about the corruption level is lost. Instead of a correlation coefficient, we utilize the simple
Euclidean distance dxy between time series (or in general vectors) {xt} and {yt} defined as
dxy =
T
t=1
(xt − yt)2 (1)
where T is the time series length. Even though there are various possibilities for defining
distances [45, 46], we opt for the most basic Euclidean one1
. We thus have a distance
1
We have applied other distance metrics as well and the results remain qualitatively practically intact.
3

between time series as a measure of dissimilarity between corruption levels in pairs of
countries forming a distance matrix D, which is symmetric with zeros on the diagonal.
For the hierarchical structure construction, we utilize the average linkage clustering
approach (also known as UPGMA). The algorithm starts with nodes in a network treated
as a separate cluster each. The closest pair is identified based on the lowest distance in
the distance matrix D (except for the diagonal elements). This closest pair forms a new
cluster and it is substituted into the distance matrix instead of the original two elements.
The new cluster is assigned the average of the distances of all elements of the newly formed
cluster. This is repeated until we have a single cluster [47, 48, 49]. Such construction can
be seen as a cautious one as it uses the average distance for the new cluster instead of
minima or maxima of the alternative approaches2
.
The resulting hierarchical structure is standardly reported in the form of a dendrogram
– a tree diagram (from Greek dendron for a tree and gramma for drawing). The diagram
shows the most important connections as an output of the clustering analysis and a pos-
sible hierarchical structure in the analyzed dataset. The most standard way of drawing a
dendrogram is in a form of roots of a tree which is mostly of a rectangular shape. For large
datasets and mainly a large number of time series, alternative methods can be utilized for
a better graphical representation3
.
3. Data
We examine the corruption clustering using the Freedom from Corruption (COR) index
of the Heritage Foundation between years 1996 and 2014. Only the countries with statistics
for each year within the analyzed period are included in the final dataset. This accounts
for 134 countries. The Freedom from Corruption is based primarily on the Corruption
Perception Index (CPI) constructed by the Transparency International. The CPI is a 0-10
range index with 0 meaning the most corrupt country and 10 the opposite. The COR is
constructed by simply multiplying the CPI by 10. Hence the COR is on 0-100 scale. If the
CPI index is unavailable for a country, the Heritage Foundation estimates the index using
expert information from other reliable sources 4
. However, it needs to be noted that the
number of countries not covered by the CPI is very low.
The CPI is not an index of an objective actual state of corruption but of the perception
of corruption in the public sector of a country. The Transparency International uses several
sources to derive the index5
. The institutions ask country experts to assess the corruption
2
In the Results section, we also provide two alternative clustering linkage methods – complete [47] and
Ward’s [50] algorithms.
3
The analysis has been performed in RStudio 0.98.1028 via R 3.1.1 using the ape and stats packages.
4
U.S. Department of Commerce, Economist Intelligence Unit, Office of the U.S. Trade Represen-
tative and official government publications of each country. Additional information can be found at
http://www.heritage.org/index/freedom-from-corruption.
5
E.g. African Development Bank Governance Ratings, Economist Intelligence Unit Country Risk Rat-
ings, IMD World Competitiveness Yearbook, Political Risk Services International Country Risk Guide or
World Justice Project Rule of Law Index.
4

level in the country in many aspects. The data is then gathered and the CPI is constructed
to reflect the level of corruption in a country perceived by its experts.
Of course, the CPI has some natural weaknesses. The most important one is that the
CPI is derived from the experts’ judgement. The results can be obviously biased by the
“elite” position of the respondent. The perception of corruption could be different if the
“ordinary ones” were asked. Another problem lays in the comparison of the CPI scores
between years. Temporal changes in ranking can be caused by changes in the country CPI
sample or methodology. To overcome these shortcomings, several alternative corruption
indices have been developed (e.g. the Global Corruption Barometer, the Bribe Payers
Index or the WB’s Worldwide Governance Indicators – Control of Corruption index).
Even though all the criticism points to the relevant CPI shortcomings, the index is still
the key measure of corruption level across countries. The discussion about CPI validity
shows difficulties concerning the corruption measurement which can never be objective and
must somehow rely on perceptions. The CPI still offers the largest database of corruption
indicator with sufficient historical time series and it remains the most suitable resource for
cross-country corruption studies.
4. Results
We apply the average linkage clustering technique based on the Euclidean distances to
find a possible hierarchical structure between corruption levels of 134 studied countries.
In Fig. 1, we present the final depiction of clustering between analyzed states6
. Due to
a large number of the examined countries, we opt for a spherical representation of the
dendrogram. In the centre of the figure, the distance between clusters is the highest and
the further we get from the centre of the circle, the closer the clusters or countries are.
We observe that the network of countries breaks down into four well-defined clusters7
.
To help reader with orientation in Fig. 1, we separate the clusters by additional dashed
lines. Furthermore, we list the countries according to their cluster membership in Table 1
as well. A very interesting clustering structure emerges.
Cluster #1 is mainly formed of developed countries, i.e. the USA, Japan, the Western
states of the EU, Hong Kong, Singapore, Australia and New Zealand, and it forms a solid
piece of network structure with the average COR of 82.92 ± 1.82. Members of this cluster
thus have very low corruption levels. Compared to the values of other clusters, which
6
In the Appendix, we also present Figs. 2 and 3 which are based on the complete and Ward’s linkage
algorithms, respectively. Even though the details of the hierarchical structure are not identical, the inter-
pretation of the results is qualitatively very close to what we present for the average linkage clustering in
the main text. The results are thus quite robust to the linkage algorithm selection.
7
The definition of clusters is quite arbitrary here and we label the main clusters based on a visual
inspection. If desirable, the number of clusters could be reduced to two – Clusters #1 and #2 together,
and Clusters #3 and #4 together. The main message of the results would remain very similar. Either
way, the reported results, their structure and implications presented in this section show that such cluster
selection is reasonable.
5

are reported in Table 1, the first cluster is very well separated from the others and a
considerable distance to other clusters is evident.
The cluster is clearly separated from the others not only in terms of the corruption
level but also in terms of the economic development. The average GDP per capita8
in
the group is 52,138 current USD9
. There is no country from the African continent. Both
most developed countries from the North and Central America (the USA and Canada) are
members of the cluster as well as is the only one and also the most developed country
from the Latin America (Chile), the most developed states from Oceania (New Zealand
and Australia), except for Kuwait the most developed nations on the Asian continent
(Singapore, Japan, Hong Kong and UAE) and 13 most developed countries of Europe.
The lowest GDP per capita is less than 7 times lower than the highest one. In other
words, the cluster of the least corrupt countries is composed of the most developed nations
across all continents. Hence we see that low corruption is clearly connected with economic
development.
Cluster #2 is more heterogeneous as it contains various EU states such as Malta,
Cyprus, Spain, Portugal and Slovenia (i.e. somewhere between the most developed Western
states and the post-communist East) as well as a mix of Taiwan, Israel, Barbados, Botswana
and others for which the connection can be seen usually only for small groups (such as a
mini-cluster of oil exporters Oman, Bahrain and Kuwait). The average COR of 59.21±1.47
still suggests a decent level of corruption.
This cluster has a lower average GDP per capita of about half the GDP of the most
developed cluster (23,521 USD). It consists of the fourth most developed African state
(Botswana), the third and fifth most developed countries of the Central and North America,
and the second most developed country of the Latin America. Also five (including the first
one) from the eleven most developed states in Asia are members of the cluster. The ratio
between the lowest and the highest GDP per capita is less than 8. The cluster hence
contains states which can be labeled as the upper-middle class of economic development.
They exhibit significantly lower economic performance in terms of per capita GDP but still
much higher than the other states. These countries are the “runner-up” leaders on their
continents and they are trying to catch up the leading group. Hence we can regard the
cluster as a “successfully transitioning” or a “catching up” group. The cluster is not only
the second richest one but it also has the second lowest corruption level.
Cluster #3 is then the biggest one with 60 members and the average COR of 24.15 ±
0.78, i.e. the lowest value among clusters suggesting the highest level of corruption. The
basket of countries is quite diverse here ranging from the new EU members (Bulgaria
and Romania) to Russia and countries in between (Belarus, Moldova, the Ukraine) to
China and India as well many African, Asian and a few Latin American countries. Several
interesting implications arise from this structure. First, Bulgaria and Romania seem to
8
Gross domestic product (GDP) for year 2012 is reported. GDP is a widely used measure of overall
economic performance of a given country.
9
Data source: World Development Indicators, WB, year 2012.
6

be institutionally very different from the rest of the EU as no other EU country falls into
the most corrupt cluster. Even the newest EU member state (Croatia) is a member of the
last cluster which possesses lower levels of corruption. Second, all BRIC countries (Brazil,
Russia, India and China), which historically form a group of large, fast-growing economies,
belong to the cluster. These countries are thus still well-behind the developed countries
with respect to corruption environment and behavior. And third, various authoritarian
regimes fall into this cluster as well – Azerbaijan, Iran, Laos, North Korea, Syria, Turkey,
Venezuela and Vietnam. The corruption levels are thus tightly connected to the form of
reign in a country. Note that most of the countries in the least corrupt cluster are standard,
western-type democracies.
From the GDP per capita point of view, the third cluster is again quite diverse. The
average GDP per capita is the lowest one with 3,888 USD. However, a range between the
richest and the poorest country in the cluster is immense with a ratio of 53 – between
Malawi (287 USD) and Russia (14,090 USD). Nonetheless, we can regard the cluster as a
group of countries with the lowest economic performance. The cluster contains not only the
poor countries or countries in civil war situations (e.g. Libya or Syria) but also transition
states with a high future economic potential (e.g. Brazil or Turkey). It is likely the highest
corruption can be a relevant obstacle for the development of these countries and hence the
corruption-economic development correlation seems to be valid even in this case.
Cluster #4 is somewhere between the previous two clusters with the average COR of
41.28 ± 0.98. This is again quite a heterogeneous cluster (the ratio between the the richest
and the poorest country is 45) with several interesting subgroups. First, the whole Visegrad
group (the Czech Republic, Hungary, Poland and Slovakia), post-communist countries in
the Central Europe, as well as the Baltic post-communist states (Latvia and Lithuania)
are members of this cluster10
. Second, the newest member state of the EU – Croatia – is
also part of the cluster and it thus readily joins the post-communist countries of the region,
overtaking Bulgaria and Romania. Third, some of developed countries such as Italy and
Greece fall into this cluster as well which might give a notion about economic problems
and its causes in these states. And fourth, practically the whole region of the North
Africa (Algeria, Egypt, Morocco and Tunisia) is inside the cluster. From the economic
perspective, the last cluster falls within the second and the third one with the average
GDP of 9,751 USD. The connection between corruption and economic development is thus
supported even by the last detected cluster.
5. Conclusions
We have inspected a possible clustering structure of the corruption perception among
134 countries. Using the average linkage clustering, we have uncovered a well-defined hi-
erarchy in the relationships among countries. Four main clusters have been identified and
they suggest that countries worldwide can be quite well separated according to their per-
ception of corruption. Moreover, we have found a strong connection between corruption
10
Estonia is not in the dataset due to incomplete time series.
7

levels and a stage of development inside the clusters. The ranking of countries according
to their corruption perfectly copies the ranking according to the economic performance
measured by the gross domestic product per capita of the member states. Even though
our analysis does not (as it cannot) discuss potential causality between the two, the re-
ported results remain promising. We believe that our analysis, which is the first one to
the best of our knowledge, can be used as a starting point of various analyses in social sci-
ences which struggle to find an appropriate mechanism of separating countries according
to such a specific criterion as corruption. We are convinced that the appropriate clustering
can be beneficial especially for studies concerning international trade where the corrup-
tion is gaining increasing attention and the studies lack rigorous differentiation of corrupt
countries.
Acknowledgements
This research has been supported by the CERGE-EI Foundation under a program
of the Global Development Network with a grant number RRC15+55. All opinions ex-
pressed are those of the authors and have not been endorsed by CERGE-EI or the GDN.
The research has also received funding from the Czech Science Foundation under project
No. P402/12/0982 and the SVV project 260 113 “Strengthening Doctoral Research in
Economics and Finance”. Michal Paulus acknowledges the support of the Deutscher
Akademischer Austausch Dienst (DAAD) sponsoring the long-term research visit (grant
No. 50015537) at the Humboldt Unviersitt zu Berlin and of the Grant Agency of the
Charles University in Prague.
References
References
[1] S.J. Wei. How taxing is corruption on international investors. The Review of Eco-
nomics and Statistics, 82:1–11, 2000.
[2] J. Svensson. Eight questions about corruption. Journal of Economic Perspectives,
19:19–42, 2005.
[3] A. Shleifer and R. Vishny. Corruption. Quarterly Journal of Economics, 109:599–617,
1993.
[4] P. Mauro. Corruption and growth. The Quarterly Journal of Economics, 110:681–712,
1995.
[5] P. Mauro. Corruption and the global economy, chapter The Effect of Corruption
on Growth, Investment, and Government Expenditure, pages 83–107. Institute for
International Economics, 1997.
[6] S. Gupta, H. Davoodi, and R. Alonso-Terme. Does corruption affect income inequality
and poverty? Economics of Governance, 3:23–45, 2002.
8

[7] M.E. Haque and R. Kneller. Public investment and growth: The role of corruption.
CGBCR Discussion Papers, 98:1–21, 2008.
[8] H. Davoodi and V. Tanzi. Corruption, public investment, and growth. IMF Working
Papers, 97:139, 1997.
[9] S. Tarhan. Public investment and corruption in an endogenous growth models. MPRA
Paper, 21319:1–35, 2008.
[10] N. Horsewood and A.M. Voicu. Does corruption hinder trade for the new eu members?
Economics: The Open-Access, Open-Assessment E-Journal, 6(47):1–28, 2012.
[11] P. Egger and W. Hannes. How corruption influences foreign direct investment: A
panel data study. Economic Development and Cultural Change, 54:459–486, 2006.
[12] M. Busse and C. Hefeker. Political risk, institutions and foreign direct investment.
European Journal of Political Economy, 23:397–415, 2007.
[13] M. Habib and L. Zurawicki. Country-level investments and the effect of corruption
some empirical evidence. International Business Review, 10:687–700, 2001.
[14] P. Egger and H. Winner. Evidence on corruption as an incentive for foreign direct
investment. European Journal of Political Economy, 21:932–952, 2005.
[15] A. Cuervo-Cazura. Who cares about corruption? Journal of International Business
Studies, 37:807–822, 2006.
[16] J.C. Brada, Z. Drabek, and F. Perez. The effect of home-country and host-country
corruption on foreign direct investment. Review of Development Economics, 16:640–
663, 2012.
[17] E. De Jong and C. Bogmans. Does corruption discourage international trade? Euro-
pean Journal of Political Economy, 27:385–398, 2011.
[18] P. Dutt and D. Traca. Corruption and bilateral trade flows: Extortion or evasion?
The Review of Economics and Statistics, 92:843–860, 2010.
[19] R.K. Goel and I. Korhonen. Exports and cross-national corruption: A disaggregated
examination. Economic Systems, 35:109–124, 2011.
[20] S. Thede and N.-A. Gustafson. The multifaceted impact of corruption on international
trade. The World Economy, 35:651–666, 2012.
[21] J. Shao, P.Ch. Ivanov, B. Podobnik, and H.E. Stanley. Quantitative relations between
corruption and economics factors. European Physical Journal B, 56:157–166, 2007.
9

[22] B. Podobnik, J. Shao, D. Njavro, P.Ch. Ivanov, and H.E. Stanley. Influence of cor-
ruption on economic growth rate and foreign investments. European Physical Journal
B, 63:547, 2008.
[23] B. Podobnik, B.E. Baaquie, S. Bishop, D. Njavro, and B. Li. Growth versus gov-
ernment management improvement during economic downturn. Scientific Reports,
3:1612, 2013.
[24] B. Podobnik, D. Horvatic, D.Y. Kenett, and H.E. Stanley. The competitiveness versus
the wealth of a country. Scientific Reports, 2:678, 2012.
[25] R.N. Mantegna. Hierarchical structure in financial markets. European Physical Journal
B, 11:193–197, 1999.
[26] J.-P. Onnela, A. Chakraborti, K. Kaski, and J. Kertesz. Dynamic asset trees and
portfolio analysis. European Physical Journal B, 30:285–288, 2002.
[27] J.-P. Onnela, A. Chakraborti, K. Kaski, J. Kertesz, and A. Kanto. Dynamics of
market correlations: Taxonomy and portfolio analysis. Physical Review E, 68:056110,
2003.
[28] J.-P. Onnela, K. Kaski, and J. Kertesz. Clustering and information in correlation
based financial networks. European Physical Journal B, 38:353–362, 2004.
[29] G. Bonanno, G. Caldarelli, F. Lillo, S. Micciche, N. Vandewalle, and R. Mantegna.
Networks of equities in financial markets. European Physical Journal B, 38:363–371,
2004.
[30] M. Tumminello, T. Di Matteo, T. Aste, and R. Mantegna. Correlation based networks
of equity returns sampled at different time horizons. European Physical Journal B,
55:209–217, 2007.
[31] J.M. Brida and W.A. Risso. Multidimensional minimal spanning tree: The dow jones
case. Physica A, 387:5205–5210, 2008.
[32] C. Eom, G. Oh, W.-S. Jung, H. Jeong, and S. Kim. Topological properties of stock
networks based on minimal spanning tree and random matrix theory in financial time
series. Physica A, 388:900–906, 2009.
[33] T. Di Matteo, F. Pozzi, and T. Aste. The use of dynamical networks to detect the
hierarchical organization of financial market sectors. European Physical Journal B,
73:3–11, 2010.
[34] C. Eom, O. Kwon, W.-S. Jung, and S. Kim. The effect of a market factor on informa-
tion flow between stocks using the minimal spanning tree. Physica A, 389:1643–1652,
2010.
10

[35] A. Nobi, S.E. Maeng, G.G. Ha, and J.W. Lee. Effects of global financial crisis on
network structure in a local stock market. Physica A, 407:135–143, 2014.
[36] T. Mizuno, H. Takayasu, and M. Takayasu. Correlation networks among currencies.
Physica A, 364:336–342, 2006.
[37] M.J. Naylor, L.C. Rose, and B.J. Moyle. Topology of foreign exchange markets using
hierarchical structure methods. Physica A, 382:199–208, 2007.
[38] W. Jang, J. Lee, and W. Chang. Currency crises and the evolution of foreign exchange
market: Evidence from minimum spanning tree. Physica A, 390:707–718, 2011.
[39] M. Keskin, B. Deviren, and Y. Kocakaplan. Topology of the correlation networks
among major currencies using hierarchical structure methods. Physica A, 390:719–
730, 2011.
[40] E. Kantar, B. Deviren, and M. Keskin. Hierarchical structure of Turkey’s foreign
trade. Physica A, 390(20):3454–3476, 2011.
[41] B. Tabak, T. Serra, and D. Cajueiro. The expectation hypothesis of interest rates and
network theory: The case of Brazil. Physica A, 388:1137–1149, 2009.
[42] B. Tabak, T. Serra, and D. Cajueiro. Topological properties of commodities networks.
European Physical Journal B, 74:243–249, 2010.
[43] L. Kristoufek, K. Janda, and D. Zilberman. Correlations between biofuels and related
commodities before and during the food crisis: A taxonomy perspective. Energy
Economics, 34:1380–1391, 2012.
[44] L. Kristoufek, K. Janda, and D. Zilberman. Regime-dependent topological properties
of biofuels networks. European Physical Journal B, 86:40, 2013.
[45] K.V. Mardia, J.T. Kent, and J.M. Bibby. Multivariate Analysis. Academic Press,
1979.
[46] I. Borg and P. Groenen. Modern multidimensional scaling. Theory and applications.
Springer, 1997.
[47] T. Sorensen. A method of establishing groups of equal amplitude in plant sociology
based on similarity of species and its application to analyses of the vegetation on
danish commons. Biologiske Skrifter, 5:1–34, 1948.
[48] B.S. Everitt, S. Landau, and M. Leese. Cluster analysis. London: Arnold, 2001.
[49] D. Defays. An efficient algorithm for a complete link method. The Computer Journal,
20:364–366, 1977.
11

[50] F. Murtagh and P. Legendre. Wards hierarchical agglomerative clustering method:
Which algorithms implement wards criterion? Journal of Classiﬁcation, 31:274–295,
2014.
Table 1: Countries in clusters. Clusters from Fig. 1 are listed here and the average values with standard
errors are reported as well. The countries are listed alphabetically within the clusters.
Cluster #1 – average 82.92 ± 1.82
Australia Austria Belgium Canada Chile
Denmark Finland France Germany Hong Kong
Ireland Japan Luxembourg Netherlands New Zealand
Norway Singapore Sweden Switzerland United Arab Emirates
United Kingdom USA
Bahamas Bahrain Barbados Botswana Cyprus
Estonia Israel Kuwait Malta Oman
Portugal Slovenia Spain Taiwan Uruguay
Albania Azerbaijan Bangladesh Belarus Bolivia
Brazil Bulgaria Burkina Faso Burma Cameroon
China Colombia Cuba Dominican Republic Ecuador
Ethiopia Fiji Georgia Guinea Guyana
Haiti Honduras India Indonesia Iran
Ivory Coast Kenya Laos Lebanon Lesotho
Libya Madagascar Malawi Mali Mauritania
Moldova Mozambique Nepal Nicaragua Niger
Nigeria North Korea Pakistan Paraguay Peru
Philippines Republic of Kongo Romania Russia Suriname
Swaziland Syria Tanzania Turkey Uganda
Ukraine Venezuela Vietnam Yemen Zimbabwe
Algeria Argentina Armenia Belize Benin
Cape Verde Costa Rica Croatia Czech Republic Egypt
El Salvador Gabon Ghana Greece Guatemala
Hungary Italy Jamaica Jordan Latvia
Lithuania Malaysia Mexico Mongolia Morocco
Panama Poland Saudi Arabia Senegal Slovakia
South Africa South Korea Sri Lanka Thailand Trinidad and Tobago
Tunisia Zambia
12

Figure 1: Dendrogram of corruption perception. The dendrogram is based on the average linkage clustering
procedure using the Euclidean distance as a measure of dissimilarity. The spherical depiction gives a base
of the tree in the centre and roots are growing centrifugally. The identiﬁed clusters are based on a visual
inspection and for a better orientation, these are separated by dashed lines.
13

Figure 2: Alternative dendrogram of corruption perception – complete linkage. The dendrogram is based
on the complete linkage clustering procedure using the Euclidean distance as a measure of dissimilarity.
The spherical depiction gives a base of the tree in the centre and roots are growing centrifugally. The
identiﬁed clusters are based on a visual inspection and for a better orientation, these are separated by
dashed lines. Structure of the clusters is very similar to the one based on the average linkage clustering in
Fig. 1.
15

Figure 3: Alternative dendrogram of corruption perception – Ward’s linkage. The dendrogram is based on
the Ward’s [50] linkage clustering procedure using the Euclidean distance as a measure of dissimilarity.
The spherical depiction gives a base of the tree in the centre and roots are growing centrifugally. The
identiﬁed clusters are based on a visual inspection and for a better orientation, these are separated by
dashed lines. Structure of the clusters is more complex than for the average and complete linkage methods,
such as splitting of Cluster #3 into two clusters. Nevertheless, the most important features of the network
remain intact.
16

Global link between corruption and wealth

More Related Content

Similar to Global link between corruption and wealth (20)

More from Marcos CAVALCANTI (20)

Recently uploaded (20)

Global link between corruption and wealth