1

What Is Corruption and Why Does It Matter?

Iht• Millennium Development Goals (MDGs), established in 2000 under
till' auspices of the United Nations, aimed to reduce extreme poverty to
h,,lf of its 1990 level, by 2015. This goal was achieved ahead of schedule, by
'II I 0, but as impressive as this achievement is, the gains were not distrib-
ult'd equally across the world: 94% of the reduction in the number of peo-
pl,· living in extreme poverty occurred in China. 1 In Paul Collier's (2007)
lt'.rminology, a "bottom billion'' - 1.2 billion people - still live in extreme
poverty (less than $1.25 per day) and 2.4 billion live in poverty (less than
~;) per day). 2
Poverty, poor health, low life expectancy, and an unequal distribution of
l11nune and wealth remain endemic. Many poor countries have had very
lc ,w or negative growth rates that challenge convergence models of develop-
111,·nl. 3 Others have weak economic records in spite of a well-educated labor
lorn:. Even some countries that are well endowed with natural resources
h,,vt• poor growth records, low per capita income, and massive inequality.
I ht• MDGs set specific global development targets, but fulfilling those goals
II I he country level has proven much more challenging in some countries
th.in in others.
World Bank, "Poverty Overview (Results)," http://www.worldbank.org/en/topic/poverty/
uwrview#3 (accessed June 20, 2014).
lnrnme is measured in real purchasing power parity U.S. dollars with a base year of 2005.
St·t· World Bank, "Poverty Overview (Context)," http://www.worldbank.org/en/topic/
povcrt y/ overview# 1 or http://www.un.org/millenniumgoals/pdf/Goal_l_fs.pdf (accessed
June 20, 2014).
< ·onvcrgence models argue that, as less-developed countries tend to grow faster than
more developed countries, the former catch up with the latter. Such convergence was
npccted to occur in the latter half of the twentieth century, but was not realized for many
, ountries, so that the gap between rich and poor grew rather than shrank.
 4 Introduction What Is Corruption and Why Does It Matter? 5
The world's leaders continue to debate how to move forward. The tlill1l ult. Dams, highways, and port facilities are technically straightfor-
MDGs, now called Sustainable Development Goals (SDGs), have been rd, l(cforming government and nurturing a strong private sector are
reformulated with the shortfalls of the first effort in mind. 4 Recognizi.ng ore ,ubtle and difficult tasks that cannot be reduced to an engineering
the remaining problems of low growth and development, the World hl11cprint. The United Nations' recently proposed SDGs include fighting
Bank in 2013 announced the establishment of a new mission: eliminat- on 11ption specifically to promote equity, justice, and peace, but reducing
9
ing extreme poverty by 2030. 5 Recent data suggest that this goal is overly 1 rnption will help achieve all the goals.
ambitious for a variety of reasons, including the fact that a large num- ll11til the mid-1990s, international development organizations, such as
ber of people were just below the original cutoff. 6 However, one part of the World Bank and the International Monetary Fund (IMF), mostly took
the explanation is dysfunctional public and private institutions that both '"titutions as given; since then, some (most notably the World Bank) have
hold back growth and restrict the flow of benefits to those at the bottom 111,,dl• institutional reform and good governance priorities. Bilateral lend-
of the income distribution. 7 Neither public funds nor outside assistance i11H or aid is also often conditional on staying off "black lists" that highlight
are used as effectively as they could be. Low-income countries and those , orruption, drug trafficking, and other illicit activities. 10 Several factors
with weak growth records are often in difficulty because they are unable to , on verged to contribute to this change in policy. The end of the Cold War
use their human and material resources to further development and to aid n•duced incentives for the more powerful countries to tolerate corruption
the poorest. 8 These countries need institutional reform, but such reform In their allies (Theobald 1999). Transitions from centrally planned econo-
111ics to market economies opened up new opportunities for both licit and
illicit profit (Rose-Ackerman 1998b). Accelerated globalization and a 1977
• United Nations, "Sustainable Development G,pal~' http://www.un.org/sustainabledevelopment/ U.S. law criminalizing overseas bribery11 pressured governments to reduce
sustainable-development-goals/ (accessed July 22, 2015).
5 unfair dealing and firms to reex~mine their overseas practices. The found-
The exact goal is for no more than 3% of the world's population to live on less than $1.25
per day measured in 2005 dollars. See, e.g., World Bank, "Poverty Overview (Strategy)," ing of Transparency Internationa (TI) and the publication of its Corruption
http://www.worldbank.org/en/topic/poverty/overview#2 (accessed September 3, 2015). Perceptions Index (CPI) 12 raise 1ri\ernational concern about corrup-
6
"Free Exchange: Poverty's Long Farewell," The Economist, February 28, 2015. http://
tion and caused alarm (and, often, anger) in some poorly rated countries
www.economist.com/ news/finance-and-economics/21645220-goal-end ing-poverty-
2030-worthy-increasingly-out-,each-povertys (accessed September 3, 2015). Th!:! World
Bank (2015) recognizes that the goal is unrealistic and urges a focus on both overall growth • The specific goal is "Goal 16: promote just, peaceful, and inclusive societies:' The
and its distribution. A World Bank working paper, Yoshida, Uematsu, and Sobrado (2014), subgoal reads: "Substantially reduce corruption and bribery in all its forms" and the
demonstrates some of the flaws in the earlier projections. Lakner, Negre, and Prydz (2014) goal also calls on countries to fight money laundering and organized crime. United
show how a combination of policies that promote growth and provide targeted benefits to Nations, "Sustainable Development Goals," http://www.un.org/sustainabledevelopment/
the very poor can combine to produce substantial reductions in the number in absolute sustainable-development-goals/ (accessed July 22, 2015). We explain the importance of
poverty. combatting all three together in Chapter 9.
7 10 See, e.g., FATF, "High-risk and Non-cooperative Jurisdictions: FATF Public Statement -
"Institutions are the humanly devised constraints that structure political, economic and
social interaction" (North 1991: 97). These inch~de constitutions, laws, rules, customs, June 26, 2015," http://www.fatf-gafi.org/publications/high-riskandnon-cooperativejuris
and taboos. We also include entities that are commonly referred to as institutions such as dictions/documents/public-statement-june-2015.htrnl (accessed September 27, 2015)
bureaucracies, legislatures, courts, schools and other educational institutions, banks and for money laundering and financing terrorists; U.S. Department of State, Directorate of
other financial institutions, etc. Defense Trade Controls, "Country Policies and Embargoes," http://www.pmddtc.state
8
Kilby (1995) found that World Bank projects were more likely to be given an unsatisfac- .gov/embargoed_countries/index.htrnl (accessed September 27, 2015) for arms trade;
tory rating by the Bank's Operations Evaluation Department if borrower countries ranked The White House, "Presidential Determination - Major Drug Transit and Drug
poorly on cross-country measures of political instability and corruption. Knack and Producing Countries for FY 2014;• http://www.whitehouse.gov/the-press-office/2013/09/13/
Keefer (1995) examine the impact of government institutions on investment and growth. presidential-determination-major-drug-transit-and-drug-producing-countri (accessed
Their measure of government quality combines indices of corruption, expropriation risk, September 3, 2015).
11 The law is the Foreign Corrupt Practices Act of 1977, Pub. L. No. 95-213, 91 Stat. 1494.
rule oflaw, risk of contract repudiation by the government, and the quality of the bureau-
12 TI was founded in 1993 as a NGO committed to exposing and combating corruption
cracy. The study examined rates of economic growth for 97 countries over the period from
1974 to 1989. The authors show that measures of the quality of government institutions worldwide. Its Corruption Perceptions Index, a central part of that effort, is described in
do at least as well as measures of political freedoms, civil liberties, and the frequency of greater detail on their website and later in this chapter. The international role of TI is also
political violence in explaining investment and growth. discussed in Chapter 14. See www.transparency.org for further information.
6 Introduction What Is Corruption and Why Does It Matter? 7
(Johnston 2005). Finally, the intellectual underpinnings of developmen I!, onomists call it utility maximization. Whatever t4e label, societies differ
policy began to recognize the key role of public institutions (e.g., Olson i11 the way they define and channel self-interest. Endemic corruption sug-
1996). The macropolicy prescriptions of. the "Washington Consensus" g1·sts a pervasive failure to tap self-interest for legitimate and productive
proved to be insufficient to stimulate growth ' and to alleviate poverty.13 purposes.
Development economists began to reach out to the fields of political science We can go a long way towttrd understanding development failures
and sociology and to incorporate work on the functioning of institutions hy understanding how a country's institutions manage or misman-
into their conceptual framework; this led them to confront corruption as a .,gc self-interest, and how self-interest interacts with generous and
particularly obvious pathology. public-spirited motivations. The best case for the social value of self-interest
The tensions between the capacities of developing countries and the i, the archetypal competitive market where self-interest is transmuted into
requirements of international aid and lending organizations arise, in part, productive activities that lead to efficient resource use. The worst case
from the diverse histories and cultures of the countries involved. To crit- is war - a destructive struggle over wealth that ends up destroying the
ics, the international organizations do not appreciate local customs and resource base that motivated the fight in the first place. In between are situ-
institutions and fail to adapt their programs to fit individual countries' .it ions in which people use resources both for productive purposes and to
special circumstances. Although this is undoubtedly true in many cases, gain an advantage in dividing up the benefits of economic activity - called
that claim is not the end of the story. Some countries' institutions are "rent seeking" by economists (e.g., Bhagwati 1974; Krueger 1974; Tullock
poorly adapted even to their own stated development goals, and oth- L993; Khan and Jomo 2000; Ngo and Wu 2009).
ers manifestly ·neglect the interests of ordinary people 'or of important We explore the interaction between productive economic activity and
subgroups. unproductive rent seeking by focusing· on the universal phenomenon of
Other critics question the goals of the international community, arguing corruption in the public sector. 14 Corruption, of course, also takes place in
that economic growth is a narrow and incomplete measure of well-being the private sector with no government officials involved, and it often has
and that international institutions tend not to take into account local con- very damaging consequences. 15 Such activities, although not the focus of
ditions and traditions (e.g., Stiglitz 2003; Easterly 2013). But even if one our book, remain an important subject for research and policy reform that
accepts that criticism, wide differences remain across and within countries should complement our emphasis on the public sector. To us, 2_ublic-sector
in health, education, economic opportunity, and environmental quality. corruption deserves special emphasis because it undermines developmen-
Whatever one's standards of value, they vary widely around the world and tal and distributional goals and conflicts with democratic and republican
are rising and fahling at different rates. We do not argue here for a standard values.
~f universal value - be it per capita income, "human flourishing" in A. K.
Sen's terminology (Sen 1999), ethical universalism, or impartiality. Rather ....
I. What Is Corruption?
we aim to show that whatever the goals of an institution or polity, corrup-
tion can undermine those goals. Corruption has many connotations and interpretations, varying by time
We begin with a basic fact of human motivation. Differences in culture and place, as well as discipline. Box 1.1 provides some examples of corrupt
and basic values exist across the world, but there is one human trait that is acts; it is an illustrative rather than a comprehensive list. 16 To encompass the
both un\versal and central to explaining the divergent experiences of dif-
ferent countries. That motivating trait is self-interest. Critics call it greed.
14 Ironically, although self-interest is a basic assumption in economics, macroeconomic
models typically assume a disinterested "benevolent social planner." Constructivists look
13
The Washington Consensus, articulated by Williamson (1990), includes standard macro- more carefully at how policy decisions are made on both personal and political levels.
15
economic prescriptions (reducing barriers to trade, establishing an independent central See, e.g., Tillman (2009) and Argandona (2003).
16
bank with a goal of controlling inflation, investing in human capital and infrastructure, For a more complete list of terms with definitions and examples, see Transparency
etc.) plus privatization and deregulation. "Washington" here stands for the World Bank International, 2009, "The Anti-Corruption Plain Language Guide;' available at http://files
and the IMF, not the U.S. government. See Rodrik (2006, 2008) for a critique, a richer the- .transparency.org/content/download/84/335/file/2009_TIPlainLanguageGuide_EN.pdf
oretical framework, and the incorporation of a broader range of policy options. (accessed June 28, 2014).
8 Introduction What Is Corruption and Why Does It Matter? 9
Box 1.1. Types of Corruption
embezzlement Theft from the employer (firm, government, or NGO)
bribery The explicit exchange of money, gifts in kind, or by the employee.
favors f~r rule breaking or as paym~_n t for benefits
kleptocracy An autocrati<?state that is managed to maximize the
that should legally be costless or be allocated on
personal wealth of the top leaders.
terms other than' willingness to pay. Includes both
bribery of public officials and commercial bribery of influence Using one's power of decision in government to
1 peddling extract bribes or favors from interested parties.
private firm agents. '
I fl
extortion Demand qf a bribe or favor by an official as a sine q~a conflicts of Having a personal stake in the effects of the policies
noh-for doing his or her duty or for breaking a rule. interest · one decides.
We treat extortion as a form of brtbery where the
bribe taker plays an active role. (Sometimes the rule is
created by the extortionist in order to exact the bribe.)
wide range of meanings, we start with Tl's definition of corruption as: "the
exchange of The exchange of one broken rule for another.
' abuse of an entrusted power for private gain:' This definition captures the
favors
principal-agent_problem at the root of all types of economic and political
nepotism Hiring a family member or one with close social ties, corruption - bribery, embezzlement, nepotism, influence peddling, con-
rather than a more qualified but unrelated applicant. flicts of interests, accounting fraud, electoral fraud, and so forth. The key
cronyism Preferring me~bers ofone's group - ra~ial/ethnic, term is "entrusted power;' which refers to the tasks one is expected to per-
religious, polrt'ical, or social - over members of other form - reviewing permit applications, pass~ng laws, or hearing legal cases,
groups in job-related decisions. for example - according to certain rules, written or otherwise. This power
A decision based on any of the preceding' types of may be entrusted by an employer to an employee, or by the populace to a
judicial fraud
corruption, or threats to the judge, rather than the government leader. If one abuses entrusted power, the rules are broken, and
the principal's stated goals are subverted. The harm takes two forms: first, in
merits of the case.
many cases the corrupt official acts inconsistently with his or her mandate,
accounting Intentional deception regarlling sales or profits and second, even if he or she only takes acceptable actions in response to a
fraud (usually in order to boost stock prices). payoff, the official has sold a benefit that was not supposed to be provided
electoral fraud Manipulation of election results, through vote buying on the basis of willingness to pay.17 Thus, corruption includes both accept-
or threats to the electorate, or by falsification or ing a bribe in return for certifying an unsafe building and demanding a
destruction of votes. bribe as a condition for approving a fully compliant structure. It includes
Any activity that undermines the legal requirements embezzling contract funds so a promised infrastructure project is delayed -
public service
of public service delivery even if no bribes are paid. and over budget, as well as the simple theft of public funds in a way that
fraud
For example, teachers might provide students with inflates public budgets but with little noticeable effect on the level of public
the correct answers or change students' responses services.
on standardized tests (usually in order to ensure
funding). Health care providers might prescribe 17
Banerjee, Hanna, and Mullainathan (2013) and Hodgson and Jiang (2007) make rule
unnecessary tests or invent patients to increase breaking the central feature of their respective definitions. We wish to be clear, however,
reimbursements. Civil servants might neglect their that the benefit provided in return for a bribe may not break any formal rules. Rule break-
jobs for private-sector work, steal supplies for resale, ing might only consist of the payment of the bribe and the corresponding distortions in
_ _! _ _ _ 1__ - _ ....... 1- .... .., .,, ~ .. ..-. , ____ ... ,..~.-Lr the distribution of the benefits and costs of public policies.
 10 Introduction What Is Corruption and Why Does It Matter? 11
We recognize, however, that some polities mn be so riddled with l'Ommitment, we analyze the normative consequences of corruption under
self-dealing that the populace cannot really be said to have "entrusted" different background conditions.
power to politicians and officials. This can occur either because too much
power is in the hands of self-interested, wealth-maximizing rulers - for
II. Incentives for Corruption
example, pre-Arab. Spring governments in the Middle East 18 - or because
the institutional framework is so weak and chaotic that there is no power We focus on corruption in the fpublic sector, ranging from grand to petty
"entrusted" to anyone, as in the case of Somalia from 1991 to 2012. Some l'Orruption and covering many different types of public/private interactions. CUA{.\
governments and institutions establish goals that most of us would abhor, (; rand corruption involves a small number of powerful players and large sums ~p
but efforts to undermine them can still be corrupt in bur sense, even if we of money. The corrupt seek government contracts, privatized firms, and con-
would applaud those who try to subvert these goals. 19 A weak or autocratic l'Cssions; they pay legislators to pass favorable laws and cabinet ministers and
state fuels corruption, and the level of corruption, in turn, makes reform .1gency heads to enact beneficial regulations. Heads of state may engage in out-
difficult and undermines public trust in government institutions, produc- right embezzlement of public funds without the direct involvement of dishon-
ing a vicious cycle. ' l'St private firms.
Some work on corruption starts with a strong coitJ.mitment to a pafticular Petty corruption is easier for ordinary citizens to observe and experience.
view of government legitimacy - most prominently the work of Rothstein 'Ihus, bribes might be paid to avoid speeding tickets, evade taxe{ or gain access ~
and his colleagues (e.g., Rothstein and Teorell 2008) and ofMungiu-Pippidi to government services. Government job offers and routine procurement coh-
r
(2013, 2014). Rothstein focuses on\ impartiality as a central normative goal l racts may favonel;itives, cronies, and friends with few qualifications. Grand
for the state. Mungiu-Pippidi stresses "ethical universalism;' but Jhe con- .ind petty corruption may be linked together in hierarchical bureaucracies;
cepts are similar, and they are analogous to North, Wallis, and Weingast's wrruption at one level can support and encourage corruption elsewhere in
(2009) "open access orders" and Acemoglu and Robinson's (2012) "inclusive the organization.
institutions:' Government actions and institutions that violate these norms We concentrate on bribes and kickbacks, but we recognize that large gray .-QA
are then labeled corrupt. 20 We, instead, study a range of institutional struc- areas exist, and we discuss some of the most troublesome in later chapters. We
tures that can produce incentives for payoffs and self-dealing'. Analysis of do not claim to have necessarily located the most harmful abuses of power and
the incentives for bribes, kickbacks, and other forms of self-dealing are then invite more research on the impact of borderline behavior, such as campaign
an input into both specific anticorruption policies and rbroad-based efforts spending, cronyism, and conflicts of interest. 21
at state reform. An implication of both Rothstein's and Mungiu-Pippidi's
work is that if bribery undermines a ruler's effort to favor a tight elite and
11
leads to a more impartial or universalism distribution of public benefits, See Yao (2002), who intentionally expands the definition of corruption and argues that
these other forms of corruption are at least as harmful to society. Explicit corruption refers
then it is not corrupt. Of course, they argue that such cases are unlikely to bribes, in which the quid pro quo is well-defined, while implicit corruption refers to
to occur, but we do not want to rule out that possibility by definition. nepotism and cronyism, in which the employee hired by virtue of connections receives
Rather than associating clean government with a particular normative wages in excess of his or her productivity. Notice, however, that Yao's analysis focuses on
what is essentially another form of personal benefit. The main distinction is the long-term
and vaguely defined nature of the transaction.
18
See, e.g., Slackman (2011) oA Egypt under Mubarak. An excellent example of the difficulty of distinguishing implicit corruption from
19
Corruption that undermines detestable laws is referred to as "noble cause corruption" acceptable business practices is the controversy over Western banks' hiring practices
(Miller 2005). One example of noble cause corruption is bribery to save Jews in Nazi in China. Several of these banks had special hiring tracks for the sons and daughters
Germany (Rose-Ackerman 1978: 9; Hodgson and Jiang 2007: 1049). If "noble cause cor- of top Chinese officials. 1n condemning this practice, Chinese law enforcement offi-
ruption" is widely seen as acceptable, or when corrupt acts are interpreted as "noble;' this cials point to the possibility that these hires were either quid pro quos for the approval
indicates a need to change the underlying institutions, but, of course, in such cases, gov- of particular deals or else improved the banks' future prospects. J. P. Morgan's practices
ernments are very unlikely to want such change. They may focus on high-profile prosecu- hit the newspapers in the summer of 2013, and in May 2014 Hong Kong's Independent
l tions instead. Commission against Corruption arrested Morgan's former head of investment banking.
20
'i Easterly (2013) includes many examples of corrupt acts in the process of exposing world- One internal e-mail mentioned the "existing and potential business opportunities" that
wide oppression, but he stops short oflabeling the norm violations themselves as "corrupt:' could arise from hiring the son of a key official. See Neil Gough and Michael Forsythe,
12 Introductio11
Figure 1.1 provides a schematic diagram of the loci of typical corrupt
acts. Each arrow shows the flow of illicit gains in monetary or equiva-
lent terms; the label on each arrow indicates what is gained in exchange
(except in the cases of embezzlement or fraud, when only the embezzler or
defrauder gains). "Government Treasury" represents all government funds
from any source.
Many heads of state (presidents, prime ministers, etc.) have stolen gov-
ernment funds throughout history. As explained in Chapter 8, kleptocracy
is the extreme case in which the state is organized purelr, to maximize the c
head of state's gains. In somewhat more institutionalized settings, the head
of state may derive illicit gains by playing a direct role in public procure-
ment or in the approval of foreign direct investment (FDI) projects by, for
example, charging a "consulting fee" for every contract approved. Members
of the legislature may embezzle directly from the resources they control,
accept "gifts" from firms or lobbyists in exchange for supporting or .oppos- Q) •
ing particular laws, or distribute resources to the electorate in order to .,;
influence votes·.
C:
0
:,
a.

~,
.Q
Q.
1 ·..:1 ., E ~
u w 0
Government officials are the heads of national or decentralized offices, ~ ~' .s
.2l ~
such as customs administration, public healthcare programs, public edyca- .s u ~
0. e! .;
tion, or regulatory agencies. They have the power to design public tenders
or select firms for projects, and may take (or demand) kickbacks in this E
0
~
(I)

~
~
(J)
('(l

/. i5 (I)
00

~
;:s
0
V)
Q)
process. They also oversee the bureaucrats charged with applying taxes or
u g_
~ ~
I ,i \ A
"'""
..... c c
regulations. If corruption is top-down, the higher-up official takes the bribe ..... (I) Q)
Q)
... E
a, E
or kickback and gives instructions to the bureaucrats, possibly sharing the C
.§ 1:l I <ii
bribe with them. For example, in customs administration, the port admin- .2l >
0
"" E
UJ 1 CJ
istrator may take a bribe from an importer, and instruct the customs agent
at a particular gate to allow a specific shipment through without inspection
(or to inspect all shipments by a competing importer). Conversely, corrup- ·
tion may flow from the bottom up: the customs agents take bribes and share &
"Former Chair of JP Morgan China Unit Is Arrested;' New York Times, May 21, 2014.
http://dealbook.nytimes.com/2014/05/21 /former-top-china-jpmorgan-banker-said-to-
be-arrested-in-hong-kong/ (accessed September 27, 2015). The time line with links to
other articles in the New York Times is at "Inquiries of JP Morgan's Hiring in China,"
New York Times, March 23, 2014, http://www.nytimes.com/interactive/20l3/11/l4/
business/dealbook/14chase-asia.html (accessed September 27, 2015). Deutsche Bank
has also been investigated for similar practices and other international banks have been
implicated as well. See Arno S,:huetze, "Regulators Investigate Deutsche Bank in China
'Princeling' Probe:' Reuters, June 5, 2014; AFP. "US agencies probe big bank} on China nep-
otism," The West Australian, June 4, 2015, https://au.news.yahoo.com/thewest/business/
world/a/28331871/us-agencies-probe-big-banks-on-china-nepotism/ (accessed June 9,
2015).
14 Introduction What Is Corruption and Why Does It Matter? 15
a portion with the port administrator to avoid punishment. Similar rela- !, (.cH1111ry level measures can take one only so far. Once citizens and 01 e
tionships may exist between doctors and hospital administrators or teach- n111 ion.ti actors are alerted to the overall problem, reform requires more .a,..._
ers and school principals or superintendants. 111·r.tl sense that corruption exists. A high level of corruption indi- M)-\.~
Some individuals (and firms) are willing to bribe their way out of legal thol i,omething is wrong with the state's underlying institutions and ~ .....
punishment. If attempting to bribe the arresting officer fails, the court clerk ii iVC!>; il signals a need for structural reform - not just more vigorous
or judge might be more amenable to such a deal. If all else fatls, the prison n!• 11, l•11wnt. Measurement needs to discover how corruption operates
guards may accept bribes to allow contraband to enter the prisons, to per- li c11l.1r sectors and to estimate how it undermines public programs.
mit extra conjugal visits, or even to facilitate escapes. h n1c;1s111cs, if properly designed, can help one to recommend reforms
Nepotism and bribery connected to hiring and appointments may occur I to I 1i1, k progress over time.
in both the public and private sectors. This provides a gain to the person In I l11~1"11.iplcrs to follow we will focus on empirical studies of particular
hired: a higher salary than is available in the market, job security, or excel- !n1t1 in p.11Licular countries. We believe that such research is the key to
lent benefits, including access to bribes. Conflicts of interest may occur live 11'11lrm at the country level. However, before considering corrup-
at all levels of government: legislators may hold stock in companies that t th1· microlevel, this chapter provides an overview and assessment of
benefit from their legislation; regulators (or their relatives) may own firms 1 ountry data. We describe the methodologies, present some data,
they regulate; and police officers may frequent businesses owned by known pl.1111 their limitations as measures of corruption. Our goal is to pro-
criminals. These conflicts may facilitate outright corruption, but even if 111 I lio 1,·,1dcr with a rudimentary understanding of each measure, in order
they do not, they can distort public choices. lunh" the inferences drawn in academic studies and the popular press.
I·
Firms engage in many kinds of corruption. They may give kickbacks to thlH t,1~1 l1<lll, we describe Tl's widely cited CPI and the similar Control of
the head of state to gain preference for important projects; pay off th_e leg- llPI l,111 1n<licator ( CCI) of the World Bank Institute. The appendix to
islature to influence laws in their favor; bribe officials and bureaucrats to h,,pln provides more details and covers other cross-country measures
get an inside edge in public tenders or to evade taxes and regulations; and 111pl Ion.
pay the judiciary and law enforcement to avoid punishment. Other types of
corruption occur exclusively in the private sector, for example, when a sales
A. The Corruption Perceptions Index and the
agent bribes a purchasing agent to favor his firm's product.
Control of Corruption Indicator
!Hn~l popular measure of corruption is Tl's CPI, which it has pub-
Ill. Cross-Country Corruption Measures:
I ,1111n1.11ly since 1995 and which is available on the TI website
Perceptions and Surveys
w;11•1111s1i.1rcncy.org). The CPI is a compilation of data from other
Combating corruption is possible only if one has some way to document t h,,t ilrC merged to generate a single number for each country. 23 The
the status quo and to measure change. 22 Corruption includes a wide range 1111w measured on a scale of Oto 100, with a higher score signifying
of different activities, and because most corrupt actors seek to conceal their 111 uptionY Certain countries - the Nordic countries, New Zealand,
actions, objective measures are difficult to find, but even perceptions can
be valuable. If observers believe that corruption is endemic, that belief Ii A11111,,· tndcx is normalized to have the same mean and standard deviation; then a
may influence economic decisions and indicate fundamental problems in p!t• 1w,:1,1111· I\ taken for each country and the CPI is rescaled to fit the 0-100 range. The
1li111l11l1111y w,1s somewhat different before 2012.
the legitimacy of the state's institutions and practices. It places a burden of
oru ~II U. the CPI was reported on a scale from O to 10, where O meant "highly cor-
proof on the state to demonstrate the contrary. lllpi " 111111 10 meant "very clean:· TI is an international organization that advocates for
I
thl' l1111t1ol of rnrruption worldwide. TI collects data from a number of different surveys
1
lh,1! 11111,1 ly ll'port business and expert perceptions of corruption in various countries.
22 11111 111' lhl' underlying data sources also cover the overall business environment -
Lord Kelvin is attributed with saying, "If you cannot measure it, you cannot improve it."
("Lord Kelvin/On Measurement;' Quotations, http://zapatopi.net/kelvin/quotes/#meas, klllll 1tlm11t red tape, the quality of the courts, etc. Respondents rank the coun-
accessed September 27, 2015). li h'. A 1111 11 ,l ,tic from excellent to poor. See Transparency International, "Corruption
 16 Introduction What Is Corruption and Why Does It Matter? 17
and Singapore, in particular - have consistently scored near the top, while lll'i' I 1. Corruption indices for 2013: The Corruption Perceptions Index vs. the
others are ranked less well year after year. CPI scores tend to persist over 111i11l ol Corruption Indicator.
time, with only a few countries showing marked improveµient or deteri-
',... oration. This persistence is due partly to the periodicity of the underlying
data - some sources are not available on a yearly basis, so the same year is IJSA 17/175
used to calculate various editions of the CPI - and partly to the circular
nature of the surveys. Although some surveys instruct respondents not to
I
_
1
consider the CPI when responding, it is likely that the previous CPI scores
,: • 1· for a country influence the perceptions of corruption of the respondents.
l
;-..
f': " .... Furthermore, corruption tends to persist because participants expect it to
1
do so. Expectations are often based on previous experience, so if a particu-
lar public service has required bribery in the past, those seeking the service
~ - will anticipate that this practice will continue. As we explain in,Chapter 7,
1
culture plays a role in the persistence of such expectations.
The other major cross-country index is the World Bank's CCL The CCI
is also a compilation, including most of the same sources and countries as
the CPI (Kaufmann, Kraay, and Mastruzzi 2010). 25 The methodology for lmlln 85/175
constructing the CCI is somewhat different, but the two indices are highly
c_, 111ru1100/175
correlated,26 and scores generally fall within the margin of error of each M11~lco 102/175
other. The CCI is reported as a normalized distribution, with a zero mean
and a standard deviation equal to one. This form has the advantage of not
imposing arbitrary cutoff points at the top and the bottom of the scale, but
it is centered at zero each year. Hence, it cannot measure global trends, 27 but
can only show how countries fare relative to each other. fh CO!llll 136/175
Figure 1.2 shows the results of the CPI and the CCI side by side for 2013. 28
The least corrupt countries according to the CPI were Denmark (92), New
Perceptions Index 2012: Technjcal Methodology Note," http://www.transparency.org/files/
content/ pressrelease/2012_ CPITech n ica!Methodology N ote_EMBARGO _EN. pd f
(accessed September 27, 2015). For an assessment of the new methodology and com- 111lnn 173/175
parison to the old methodology, see Saisana and Saltelli (2012), available at http://
0 20 40 60 &t 100
files. transparency. org/ content/ download/ 534/2217 /file!JRC_Statistical_Assessment_
CPI 2014
CPI2012_FINAL.pdf (accessed June 28, 2014).
25
The CCI and related information are available at the World Bank's Worldwide Governance
Indicators site: http://info.worldbank.org/governance/wgi/index.aspx#doc-sources (aecessed
September 27, 2015).
26
For the data collected in 2013, the correlation between the two was 0.987. This is identical
11111d (91 ), Finland (89), and Sweden, Switzerland and Norway (tied at
to the correlation between the CPI and the CCI the previous year. 1111· most corrupt were Somalia and North Korea (tied at 8), Sudan
27
Ostensibly, the new CPI methodology allows comparisons over time, but the pre-2012 11 )1 Mghanistan (12), and South Sudan (15). On the CCI, the least cor-
CPI data do not. -+
28
iipt 11·1·n· Denmark (2.41), New Zealand (2.35), Sweden and Norway (tied
Note that TI uses the year the data are published (2014) while the World Bank Institute
uses the year the data were collected (2013) in assigning a year to the data. Our graph I ),,211), and Finland (2.19); the most corrupt were Equatorial Guinea
refers to 2013, but the data from TI are reported as the 2014 index. 1,,-; I), Somalia (-1.58), Libya (-1.52), Sudan (-1.49), and Afghanistan
18 Introduction What Is Corruption and Why Does It Matter? "- 19
~
'
Figure 1.2 (continued) 11ih'd States is ranked best of t~e six countriesi nd Sudan last. Russia,
[\.h·xico, and China are in the lower half of each graph, but whereas
r,111ks better than China and Mexico on the CPI, India is below these
t11111t ries on the CCI.
I ho d,1ta come from surveys or questionnaires, applied to residents,
USA 32/210 1.28 IIH!II~ ll•aders (CEOs), or "country experts;' sometimes complemented
h111d statistics" drawn from official sources. The C CI \ ses a broader
lli..l' pl of corruption, covering victimization and anticorruption institu- ,
lll h as electoral integrity and freedom of the press, whereas the CPI
ll\1il C tightly focused on concepts of corruption associated with bribery,
1hea il'ment, and political influence. The addition of these factors likely
pl11i11, India's fall in the rankings as one moves from the CPI to the CCI.
tlll' Appendix to this chapter for more details on the sources.)
llw C PI performed the important function of helping to put corruption
1111111 011 the agenda of international organizations and domestic reform-
In t h1· 1990s. Through dissemination by the popular press, it also raised
--0.35 China 109/210
w,m·ncss of citizens around the world, who in many countries have
1111111d,·d greater accountability and transparency. The indices have been
-0.48 Mexico128/210 1,Jx h'nsively by researchers to identify the causes of corruption, or con-
--0.56 India 134/210
nw h'. to determine the effect of corruption on variables of interest, such
' ii >I' 111 growth rates.
lhll1111· discussing some of these results, however, it is important to keep
nlnd the limitations of composite indices (Andersson and Heywood
HIH ). I·1rst of all, it is not clear exactly what is being measured. Many differ-
--0.99 Russia 175/21 O 11 ll,11.1 sources are included in each index, but not every country has data
1lli1hl1· from each source. 30 Hence, "corruption"..may b~ re indicative
,11d .:orruption in some countries and of petty corruption in others; it
!!Id
~11mf ta greater risk of political instability resulting from corruption
-1.49..., ~ Sudan 207/210 I • 111111· .,nd a higher probability that businesses will have to pay bribes in
-2.0 -1 .50 -1 .00 --0.50 0.00 0.50 1.00 1.50 2.00 2.50
tlit11 1i A, macroindices, they tell one nothing about the details of how cor-
CCI 2013 1p1in11 operates. Just as gross domestic product (GDP) per capita masks
tlui 11111,nH.' distribution by region, economic sector, and social class, the
Note: The CPI is listed by TI as for 2014, but the underlying data are actually from 2013.
TI data used with permission. l'l 1111d <'. Cl make no distinction between corruption in the police force
Sources: Based on data from Transparency,International, Corruption Perceptions Index 1 u1 Ht11111s, and political corruption; nor do they differentiate between cor-
2014 and World Bank, World Governance Indicators 2013. 1pt ln11 th.it only represents a transfer of funds and corruption that also
11,to, l!o tlw allocation of resources. Similarly, these indices do not directly
(-1.43) .29 For illustrative purposes, we highlight six countries: the United 1111' the volume of bribes, the incidence of corruption, or its impact.
States, Russia, Mexico, China, India, and Sudan. Note that i'n each graph,
29
Note that four of the five worst-ranked countries on either index are postconflict coun- "" In lo be included in the CPI, three sources must be available; for inclusion in the
tries; see Chapter 10.
 20 Introduction What ls Corruption and Why Does It Matter? 21
.i.,lly distortionary impact on ~onomic and 'political life? How impor-
Second, most of the sources behind the corruption measures are subjec-
tive and can be influenced by visible scandals that do not reflect underlying
-
111· outright payoffs compared to more subtle types of influence such
conditions. 31 As a result, a worse score may reflect freedom of the press 1•11yis111 and lobbying? Cross-~ountry indices tell us something about
rather than necessarily higher "real" levels of corruption. Furthermore, the 111111 I ions in state/society relations, but little about the details. Empirical
underlying methodology used to produce some of the component parts is lhlk, lh,,t reveal the mechanisms at work usually focus on single coun-
proprietary and not transparent. (See the discussion of "expert surveys" in h11i 111 'l'clors. They cannot be easily generalized. The level of bribes is
the appendix.) the l ritical variable in any case. One wants to know not just how much
Third, a country's score is not expressed in cardinal units, such as dollars. p,1id, but also what was purchased with the payoff. For that, one needs
Rather it is a unit-less constructed number that tries to capture a coun- h!h\lkd u>untry-by-country and sector-by-sector analyses. This book is an
try's position on a continuum from high to low levels of corruption. Yet, in lhllilpl lo set the agenda for such efforts and to draw lessons from the work
empirical work the CPI is often used as if it were a cardinal number, so that hill uh,•,1dy exists. Only if we look at the fine structure of political and
a one-point difference is taken to be the same, no matter where on the dis- 1i111111t systems, can we go from a showing that "corruption" is harmful
tribution that difference falls. Thus, the sizes of coefficients should be given Hi understanding of how it operates in different contexts. 33 Given that
little weight; the direction and significance of correlations are what matter. l,·dgl', reform programs can attack corruption where it has the worst
Finally, some criticize the index for being culturally biased and not rec- lk, h ,111d where marginal gains are high relative to marginal costs. ·We
ognizing that some transactions viewed as corrupt in wealthy, market econ- ,1w 1111 l'Xisting work in individual countries to illustrate our arguments
omies are acceptable in other countries, and vice versa. That may indeed be 11u•111111g the causes and consequences of corruption and to recommend
true, as we discuss in Chapter 7, but some actions are universally under- 1111111s I lowever, 'the existing collection of cases is not sufficient. We need
stood to be corrupt. Even countries that tolerate grease payments of vari- ptcmatic knowledge of how corruption and self-dealing affect the
\.
ous kinds do not legally condone huge kickbacks paid to political leaders 1•rr,111,,11 of government programs and private markets.
in connection with major contracts and concessions. The only exceptions l!I Rhmt, the index scores are not policy tools in and of themselves. The
would be states that are the personal fiefdoms of the ruler and his family. Iii'! ween reform policies and the index numbers are complex and
Of course, the indices are not an overall measure of the impact of private \I No government ought to have as its goal an improvement of X
wealth on public power. An index that tried to capture those interrelation- 111111 in its CPI score. Reform requires more focused measurements
ships, many of which are perfectly legal in developed countries, would pro- 11118 to targeted policies. In response to this need, many country- and
duce a different ranking (cf. Sandoval-Ballesteros 2013). lo1'·1opl•dfic instruments have been developed since the late 1990s. These
Given the range of behaviors covered by the concept of corruption, it is luil(' rnkrosurveys of firms and individuals (which permit the identi-
not clear what it means for a country to rank poorly on a corruption index. 32 li1111 of characteristics associated with corrupt behavior), experiments
Does it mean that bribes are a large share of the value of contracts and 1 hrlt,l\'loral laboratories and in the field, and audits. These data have
government services? Does it mean that the proportion of deals influenced llmv,·d grl'ater insight into the causes and consequences of corruption and
by bribery is high? Does it mean that self-dealing in ail its forms has an lornwd ,tnticorruption policy in ways that country-level indices cannot.
11l11·kss, cross-country work, if interpreted with a degree of caution,
31 hdp sl'l the stage for the more focused sectoral work that we discuss in
Olk.en and Pande (2012: 482) cite the example oflndonesia where the CPI fell (indicating
increased corruption) after the fall of Suharto. They speculate that the fall may have been 1111·111 chapters, and that is the key to setting reform priorities.
the result of a freed press that was better able to report scandals. Of course, another expla
nation is that the populace became more aware of corruption as its nature changed from
centralized to competitive bribery (Chapter 8).
32
See Mendez and Sepulveda (2009) for a model that demonstrates the analytic differences 11 n,1mplc of the kind of detailed understanding needed for concrete proposals in
among contrasting definitions. The three they consider are (1) the numbh of corrupt deals, 1ii, 111.11 ~.,scs consider Tendler's (1979) report to the World Bank on graft in rural works
(2) the ratio of the number of corrupted to total deals, and (3) the total volume of bribes rn111\ ,n Bangladesh. The paper is an admirable analysis of the impact of graft on dif-
collected by corrupt officials. They show how one's evaluation of the extent of corruption 1,1 u~pn ts of a development project and a discussion of the conditions under which
can vary depending upon which metric is used in the context of their formal model. :ii p1·11pll· ,an be used as monitors of others' honesty.
22 Introduction What Is Corruption and Why Does It Matter? 23
i"111ti 1.1. Sector-specific results from the 2013 GloQal Corruption Barometer.
B. The Global Corruption Barometer
NGOs
Popular polls are one response to the criticism of composite··indices and I teliu1ous bodies
elite surveys. Survey firms conduct polls through face-to-face, telephone, or Military
online interviews or questionnaires. The questions may be concerned with Media
perceptions of corruption or actual experience. For example, the Global I .1lueatlon system
Corruption Barometer (GCB ), published by Tl, asks respondents - ordinary Mo1l,cal and health
people - both types of questions. "To what extent do you think that cor- Private sector
ruption is a problem in the public sector of this country?" is a perception Judiciary
question. The responses are coded from 1 ("no problem at all") to 5 ("very 111t:iv1I servants
serious problem''). This same question is,also asked about specific aspects of monVLeglslature
public-sector service delivery, the media, NGOs, and business. 34 Strikingly, Police .,
the vast majority of countries have a score between 3.5 and 5. Indeed, the I 'ol1tical parties
least corrupt country by this measure is Rwanda, with a score of 2, fol- 0 10 20 30 40 50 60 70
lowed by Denmark (2.2), Sudan (2.6), Switzerland (2.7), and Finland (2.9); Percentage of respondents who believe that each
institution is "corrupt" or "very corrupt"
the most corrupt are Mongolia and Liberia (tied at 4.8), Zimbabwe, Serbia,
I 1,1nsparency International, Global Corruption Barometer 2013. Results are
Russia, Paraguay, Nigeria, Mexico, and Indonesia (tied at 4.7). Only 107
ll 1111 I 1·1,000 respondents in 107 countries in 2013. TI data used with permission.
countries are represented, however, excluding many small countries and
many of the countries considered most corrupt on other indices.
1
l,i i \ t h.1t the institutions perceived to be most corrupt - political par-
Global results (based on 114,000 responses in 107 countries in 2013) by 1111• police, public officials and civil servants, parliament or the legisla-
subsector are presented in Figure 1.3. 35 By institution, on a .global level,
llilll, 1111d the judiciary- are the very institutions charged with creating and
political parties were perceived as the most corrupt institution, while NGOs
1phnlding the law.
were perceived as the least corrupt. What is particularly worrisome in these ·1111• ,urvey also tabulates experiences by asking about respondents' use
(
I \'lit 1ous services in the past year, and in the cases in which the service
34 11,l•d, if they paid a bribe. The number of bribers divided by the number
Before engaging in cross-national comparisons using the GCB, it is important to note the
limitations of the data. First, the questions only capture low-level petty corruption experi- f '*'' (multiplied by 100) yields a bribery incidence index. 36 The global
ences, not grand corruption by high-level officials. Second, differences in reported bribery
11lh from this question are presented in Figure 1.4. Country-level results
rates might be driven in part by cultural differences in respondents' willingness to report
illicit behavior. Corruption is more openly discussed in some societies than others. There pi l"Sl'nted in the Appendix.
may also be cultural differences in what constitutes a corrupt transaction. A bribe in one
country may be considered a gift in another. We discuss those complexities in Chapters 7
and 8. Third, government institutions may vary significantly across countries, and "regis- C. Perceptions vs. Incidence
try and permit services" could represent something quite different in Turkey and Ireland,
or in Venezuela and Malaysia. Any cross-national comparison assumes that sector defini- <ilv1·11 the results from the popular polls, we can ask whether elite surveys
tions hold relatively constant worldwide. Hll "out of touch:' Figure 1.5 plots the GCB's question regarding how much
35
Global results are based on the entire sample: one response is one vote. For most coun-
tries, the sample size is approximately 1,000. Countries with significantly fewer respon-
dents are Cyprus (570), Luxembourg {502), Solomon Islands (509), and Vanuatu (505); Ilu: rnr, uption incidence index reported by the GCB measures the user-based incidence:
those with significantly more respondents are Afghanistan (2040), Australia (1200), 1I11: pnn•ntage of users who paid a bribe, independent of the number of times they used the
Bangladesh (1822), Bosnia and Herzegovina (2000), Brazil (2002), Ghana (2207), Japan r, 1·1n•. Some surveys - e.g., the Encuesta Nacional de Corrupci6n y Buen Gobierno, pro-
(1200), Korea (1500) , Moldova (1211), Pakistan (2451), Peru (1211), Romania (1143), and d11, nl hy Transparencia Mexicana - report a use-based incidence, based on the number
Ukraine (1200). China is not represented. See http://issuu.com/transparencyinternational/ 111 tlntl's the service required a bribe, divided by the number of times the service was used.
docs/2013_globalcorruptionbarometer_en?e=2496456/3903358#search (accessed June 11, 1111 1t'.sulling figure is the percentage of uses of a service that were corrupt. The distinction
2014). litlWl'l'll the two is more than semantic, and there are advantages and disadvantages to each.
 24 Introduction What Is Corruption and Why Does It Matter? 25
Figure 1.4. Global Corruption Barometer: Incidence of bribery in subsectors. Figure 1.5. Public opinion~(GCB) vs. expert opinion (CPI).
Utiltties
100
iO I •
Tax
• • •
•
mI • ••
Education • • •• • •
oI ............... • • •
Medical •• •
o ·I
• ~ . •• • ••
Land 21 0 •
Registry 21 40 .
•
.,o, y = -13.35x + 98.0S •
Judiciary 24 R2 = 0.15 •• •
Police 31
o, • • • • ••
• • •
~ ,o • •
0 5 10 )5 20 25 30 '$
Percent who paid a bribe in the last 12 months, having had contact wnh each service 0
1
1 2 3 4 5 6
Source: Based on data from Transparency International, Global Corruption Barometer
2013 Report, page 11. Results are based on 114,000 respondents in 107 countries. TI data Global Corruption Barometer: How problematic is corruption?
used with permission. t1/,· I he t-stat for the coefficient is -4.16; p-value 0.000. TI data used with permission.
11111, 1•: Based on data from Transparency International, Corruption Perceptions Index
,i I I ,1n<l Transparency International, Global Corruption Barometer 2013.
of a problem corruption is (5 = very serious problem), against the CPI
(100 = very clean). 37 The negative 'correlation we would expect is present,
but very weak. Most countries score between 4 and 5 on the GCB scale, Figure 1.6. Incidence (GCB) vs. expert opinion (CPI) .
while there is a much greater variance in the CPI data. In other words, most 100
residents believe that corruption is a serious or very serious problem, while 0
those with some cross-country experience see a good deal of variability 80
around the world. 70
.... . .
Figure 1.6 plots corruption incidence reported in popular surveys (GCB) .. . y = --0.63x + 60.39
60 .
(i.e., the percentage of users of a public service who report paying a bribe R2 = 0.44
50
for that service) against expert opinion (CPI). Here the negative relation
40 .
is much stronger. The difference between these two graphs highlights the
30 .
subjectivity of the perception question, which is made clearer in Figure 1. 7 ..• .
comparing residents' perceptions with their experiences, both from the 20
• •
GCB. The relationship between people's direct experience with corruption 10 . .
and their perceptions of corruption is quite weak. Even in countries with 0
0 10 20 30 40 50 60 70 80 90
low corruption incidence (x-axis), residents tend to perceive that corruption
Global Corruption Barometer 2013: % who paid a bribe
~(. Note: The t-stat for the coefficient is -8.41; p-value 0.0000. TI data used with permission.
Source: Based on data from Transparency International, Corruption Perceptions Index
37
" ~ We use the 2013 CPI so that both sets of data reflect the same year. 013 and Transparency International, Global Corruption Barometer 2013.
 26 Introduction What Is Corruption and Why Does It Matter? 27
Figure 1.7. Incidence (paid a bribe) vs. public perception (how much of a problem is Ollll' sources are collected less than annually, so previous years
corruption?) . I wlll'n calculating these indices. As a result, it is not surprising
C-· I I ho 11n·1.1ge changes only slowly, because some of the source data
C 6
0
" :;:, held ( n11 stant over two or more years.
~ C.
.c!l::,
~ ~ 5
lHit1dl111ts interpret the perception question to mean "How much
0 tJ • ••• • • pr.-,hle111 is corruption when it occurs?"
~ VJ ... •••
C'CI ·- .
• •• • •
co E 4 .••..•....•* ••••• I
... :: ... . .... O *: ••• •
C (I)
o-
·- .c . f .. * • . • • .• II 011 licsl' reasons, it is important to consider what each measure
-o
C. ~
• 11pll1111 represents. The best index to use depends on the questions
E~ 3 • • •
• •• • It, 1111,wl·r. Perceptions and incidence are distinct measures, and
80 • y = 0.01x + 3.91
can • Wlll"l' opinions or (lxperiences are taken into account. It would
.c ::, 2 • R2 = 0.07
.Q E pri111t·. for example, to use the GCB's bribery incidence in a dis-
C, :i:
0 1111' wa11d rnrruption. Likewise, we cannot infer from the CPI how
I 1
0 10 20 30 40 50 60 ' 70 80 1tll11(11 \' l ll1zcns or firms pay in bribes each year.
Global Corruption Barometer 2013: % who paid a bribe
Note: The t-stat for the coefficient is 2.58; p-value 0.0114. TI data used with permission. I Ill'< :osts and Causes of Corruption: An Overview of
Source: Basea on data from Transparency International, Global Corruption Barometer Cross-Country Empirical Results
2013, http://www.transparency.org/gcb20l3/in_detail
It iq1lt'd l he cross-country data, we now step back and ask if these
is a serious problem (y-axis). Why might this be so? There are at least four ,_l,in , lll'\'l'rlheless, teach us something. The indices appear to cap-
possibilities (Mocan 2008; Morris 2008) . 11h'1 l)1 l11g .1spccts of the relationship between the state, on the one
!Ill dt l11·11s and private businesses, on the other. 38 In spite of some
1. Perceptions reflect the difference between grand and petty corrup- lill! N i11dividu,ll cases, the general patterns show that some countries
tion: although people may not have to pay bribes for public services, Mt• 11t high achievers in terms of good governance and economic
they may be aware of high-level corruption, which leads them to li1I p1 ugn•ss and that others are persistent laggards. In the middle
report that corruption is a serious problem. IMC1H111llwr of ambiguous cases in which the correlation is less pow-
2. Perceptions take into account more information. Most incidence sur- hut th,J p,11 hologies of cor,ruption are felt in particular sectors and
,'. veys ask whether the respondent or someone in the respondent's fam- I• 111' H• ,vcrn mcnt performance. The indices do not explicitly indicate
'- ily has paid a bribe in the last twelve months. Even if this is not so,
'-,
p1•lll lt'~ 111 ight be effective, but they do highlight problems - both \~
'
the respondent may know someone who has paid a bribe, and that lhl' l11dd1•1Kc of bribes is high (e.g., the police) and where percep -
,I
knowledge leads to a higher perception of corruption. Perceptions are i t' hlHh , wh,llcver the reality (e.g., political parties, many very poor
,., also swayed by scandals in the media. This leads to the "paradox of
', ~.__ )
'' • distance:' People perceive government in general to be corrupt, but n1phk.1l studies undertaken in the past twenty years try to deter-
-~
they have a more positive opinion of those government programs that
'- Iii the ( ,,uses and consequences of corruption. At first, most stud-
"-- affect them directly and of the bureaucrats with whom they interact I i. H• \ t. 1 ountry analysis, but an increasing n~mber of microanalyses
(Frederickson and Frederickson 1995). I liH111• ,pccific corruption data. Figure 1.8 provides an overview of
3. Perceptions change more slowly than incidence. If an anticorruption I 1ol11i'l f111<lings from these studies, many of which we cite in sub-
campaign is undertaken, the impact should be reflected rather soon II 1. h11ptc1 ,. ·I here are a host of causes that generally interact with each
in the incidence of bribery, but for psychological reasons, people still 111 thlii di.1gram, we have divided the causes into "incentives" and
hold on to their previously formed perceptions. Perhaps they have 1ll1111 R:' hut personal ethics, of course, also plays a role. Corruption
\
not used the reformed services since the anticorruption campaign
 I
What Is Corruption and Why Does It Matter? 29
111t•rscction of situation-specific incentives, society-wide insti-
11d pt!rsonal ethics. It should be clear that the consequences of cor-
.c (/)
~.::,:. ml (1hlly for many individuals and businesses, as well as affecting
ra u,
_.,.. <l> ·c
.c ;:, ]j lltlllhtl ,tability and the etfectiveness of government spending. In
~ C
a1 ,S! Cl) Cl)(/)
u, ra t hl' .irrow may, in fact, go both ways. For example, poor rule of
ra u, iii
CDC~
EE
~ -g ~ ih11tt·s to corruption, but corruption also undermines the rule of
C) ·=o> r:.ra
.~ ra C 11p1ion enables trafficking in drugs, arms, contraband, or humans,
Cl)
4
, , ,llso actively try to corrupt the authorities. For simplicity, we
ll 011~· way arrows, but the reader should be aware that the relation-
~ li\1H1>,1 these variables are much more complex.
·C<::' rnuntry data intlicate underlying connections between the
~ 0
Cl)~
~ 2- ,t government institutions and other variables of interest. In spite
,(::-1:,j
~
1111111.,tions of these data, they provide a useful place to begin. 39
d ;:,~
I •-::,
0 ~ "ffi ;:a m• l ,11 illustrates the simple relationship between the UN's Human
~ 0:::, (/)
'&:,
en0) .au - CT~ lop111r11/ Index - an index that takes account of education and health
t:: ~2 :s ?°'
0 cii II ,IN gross national income (GNI) per capita40 - and perceived levels
.....0u C ' This correlation is one of
</) ~""0~ n 11pl ron in 2012 as measured by Tl's CPI.
...u 8
a.:::, nio~t rnbust relationships to have emerged out of corruption research
<ii
u_ ~
...c:: ::, u al ll!i~t1111 JOOS; Aki;:ay 2006; Reiter and Steensma 2010; Askari, Rehman,
=:::,
& 0 ~
...
</) a. cii <n
....
c:: 0
I Art.,., JO 12). Countries with higher levels of corruption have lower lev-
0 3: cii
u 0:::, .s of lr111llan development. Similarly, as a rule, richer countries and those
"d <ii (/) - .p
§ .~
~ tlr hlHh growth rates have less i:eported corruption and better function-
~
f!? :6 ;:, ~
IC
~ Cl) Cl)
a. .c ·<ii- ::! lilK l\•H'1·111ments (Kaufmann 2003).
g f s- <55 11ft. 11 Is to explore the mechanisms at work suggest that corruption reduces
u §' Cl)
.S:
CX? o:g I hli cfh Iiveness of industrial policies, making running a business more
TI C C
- ~ ra :::, 0 pcmiw, and thus encourages business to operate in the informal sec-
-8 ~ 'iii
So 01' i11 violation of tax and regulatory laws. 41 As more individuals and firms
~ ~ -~ Cl)
C
~ ..g
C \\'e du not attempt to review all the cross-country studies produced since Mauro's (1995)
0 i~
""~ §<ii 1111ly l·or early surveys see Bardhan (1997) and Jain (2001). See also Rose-Ackerman and
~ E a'i J'i lll'X (2013).
>, 'O
'6 o~ C I 111 more information on the Human Development Index, see the United Nations
a. Cl) 0
g ~ Undopment Programme's website at http://hdr.undp.org/en/content/human-development
<516) ~
0 a. -C a:: 1111ll'X hdi (accessed September 27, 2015).
~, C
~
.... E Cl) ·-
u ~ l.1uro ( 1995, 1997) demonstrates that high levels of corruption are associated with lower
())v. (/)
;;: E
0 cii li·wl\ of investment as a share of GDP. The corruption indices are highly correlated with
~ - Cl)
3: .~
o ra ot lll'r measures of bureaucratic efficiency, such as the level ofred tape and the quality of
- <ii .~
(/) u t lw Judiciary. As a consequence, Mauro was unable to measure the marginal effect of any
·- .c
3: one of these measures. Putting the separate indices together in a measure of bureaucratic
0
§ ~0
C
- 0 ~ 1·ll"icncy, "ifBangladesh (with a score of 4.7] were to improve the integrity and efficiency
u 0)
Cl) of lls bureaucracy to the level of that of Uruguay [score 6.8] .. . its investment rate would
rlsr by almost five percentage points and its yearly GDP growth rate would rise by over
h,ilf a percentage point (Mauro 1995: 705)." Mauro also demonstrates that highly corrupt
30 Introduction What Is Corruption and Why Does It Matter? 31
,... Figure 1.9. Corruption and development. 1~111 (I labib and Zurawicki 2002). Corrupt countries tend to suffer
(') 1.2 111,ut• bureaucratic red tape, which may be intentionally created by
0
(\J
k111g bureaucrats. 43 Asian economies are not an exception - those
ci 1.0
high u>rruption levels would have attracted more FDI if corrup-
~ lhlll lwcn lower, and their industrial policies would have been more
X •
Q) i
"O
llvc. '1 Consequently, corruption depresses economic growth (Mauro
.!: 0.8
•• • • •.if••·.:
... . ......• •• id1 )009).
EQ) • • • .•
E ••• .....•••••• • • lltH,,tn of the magnitude of these effects vary and in any case are dif-
a. 0.6 •
0 • • 11 h1 i11tnpret. Considering only GDP per capita, which is a narrower
ai
>
.• i
Q) • • lll'l' 111 development than the HDI and does not include measures of
y = 0.01x + 0.43
0
C:
0.4
. ••••• •• , R2 = 0.51 11011 ,111tl health, Dreher and Herzfeld (2005) find that an increase of
ca •• •
E ljlli1111 hy one index point (on a scale from O to 10) dampens GDP
::J
I 0.2 lh h)· I \ basis points (i.e., 0.13 percentage points) and lowers per capita
10 20 30 40 50 60 JO 80 90
hy !lround $425. Gyimah-Brempong (2002) estimates the effect to be
Corruption Perceptions Index (CPI) 2014
11 7'1 ,ind 90 basis points or just under one percentage point. Mo
Note: the t-statistic for the coefficient is 13.26; p-value 0.0000. TI data used with permi~ion. tlm.1tes the elasticity of real GDP wit~ respect to corruption to be
Sources: Based on data from United Nations Development Programme, Human
Development Index and Its Components and Transparency International, Corruption
At u>rding to Haque and Kneller (2009), the correlation is nonlin-
Perceptions Index 20i4. 1 htl11111cs more negative below the 4th percentile of GDP per capita
hO\'t' .,pproximately the 75th percentile. 45 Estimates of the relationship
Ii i,mruption and GDP or GDP growth may be sensitive not only
evade taxes, the government finds it necessary either to raise tax rates or to
engage in seigniorage, leading to inflation42 (Al-Marhubi 2000; Blackburn
c,11, ,ind control variables used, but also to the countries included.
I
linwt•vcr, that the measure of corruption is an index with no natural
and Powell 2011), a depressed national currency (Bahmani-Oskooee and
I h,rn. it is unclear how one should interpret coefficients on that vari-
Nasir 2002), and, if fixed exchange rates are in place, a higher black market
premium (Bahmani-Oskooee and Goswami 2005). It negatively affects the tlll t ht· dasticity measures that result. Furthermore, these studies do
nl\r tIll' issue of causation and the possibility of vicious and virtuous
business and investment climate so that FDI is discouraged by high corrup
tion levels (Wei 2000; Habib and Zurawicki 2002; Egger and Winner 2006)
and by differences in the corruption levels of the host country and country I\Htlwr LOmplicate matters, some countries do manage to have high
if hu111,1n development and growth despite high levels of corruption,
countries tend to underinvest in human capital by spending less on education (Mauro Iii( tl1.1t the relationship is far from deterministic. High levels of cor-
1997). He argues that this occurs because education provides less lucrative corruption 11111 t' more destructive under some conditions than others (Wedeman
opportunities than other types of more capital-intensive public spending.
Ades and di Tella (1997a) argue that an aggressive industrial policy may be partly motl
vated by the corrupt gains the policy makes available. In such cases, the direct positive h1111i s, 11"1011 regarding "greasing the wheels" versus "sanding the wheels" in the sec-
effect of the policy can be undermined by its role in increasing corruption and hen~c. l'11tr 1111,il lknefits of Corruption" in Chapter 2.
discouraging investment. Their empirical results demonstrate that in the presence of cnr 1111 \\'1·1(2000) shows that corruption acts like a tax on FD!. An increase in the cor-
ruption, the positive impact of industrial policy is halved. East Asian economies are not 11 !t'\'1·1 ii om relatively clean Singapore to relatively corrupt Mexico is the equivalent
immune from this effect. Johnson, Kaufmann, and Zoido-Lobat6n (1998: 389-91) find 111ur~,r in the tax rate of more than 20 percentage points. The statistical result holds
that higher levels of corruption are associated with a larger unofficial economy. I i\ 11.111 wuntries as well as for the others in his sample. By contrast, Egger and
42
Braun and diTella (2004) trace the causality in reverse: higher inflation causes corrup I (lOll11) find that corruption has a smaller effect on inward FD! for large (GDP),
tion, suppressing growth both directly and indirectly. Cuk.ierman, Edwards, and Tabellinl llilhnl. ,llld differently endowed countries, arguing that China's size and low wages
(1992) find that higher levels of political instability lead to higher rates of inflation. Jnsofa 111!.l tli1· lll'gative effects of corruption in attracting FDI from OECD countries.
as corruption causes political instability - which, in turn, fuels inflation - this is a tran~ thor, nllmate the effects of lagged GDP per capita on corruption, rather than
1
mission mechanism for corruption to cause inflation. I
32 Introduction What fs Corruption and Why Does It Matter? 33 "
1997: 459). For example, it can be especially detrimental if the rule 1of law 1111k environmental quality (Mauro 1998; Esty and Porter 2002;
is weak (Meon and Sekkat 2005) or in nondemocratic countries (Drury, '" y International 2011a). In a corrupt regime, economic actors
Krieckhaus, and Lusztig 2006). Although countries with strong institu- 111ples, such as those engaged in illegal businesses, have a com-
tions are usually resistant to corruption, if it does get a foothold, it can be 1dv,111tage and may dominate the business and political sectors.
especially destructive as it undermines those institutions (Aidt, Dutta, and 111dil'S have examined the relationship between corruption, on
Sena 2008). u, h1111d, ,ind inequality and ~verty, on the other. High growth rates
Some analysts argue that, under specific conditions, corruption even i~t with rising inequality, 'th those at the bottom of the income
improves economic outcomes. If businesses and individuals face onerous lliu1 k111 ll'Ceivi ng few benefits a d the majority of the income growth
amounts of red tape, for example, corruption helps them to reduce their 1li1B tn the lop of the distribution. If corruption-fueled growth does
costs, both monetary and temporal, allowing for more innovation, busi- t1,111,l.1tl' into improved education, health care, and public infra-
ness creation, trade, and economic growth (Leff 1964). This is known as lltl'r, i11t•quality can persist over generations and can eventually slow
the "greasing the wheels" hypothesis. For countries with poor institutions, th 1\ ldt (2011) constructs a broad index of sustainable development
some empirical studies found that corruption is not detrimental to growth46 IHiWll th.it corruption has a detrimental effect. Corruption in Aidt's
and may even increase efficiency (Meon and Weill 2010) or entrepreneur- 11h1ti1111 might spur investment and growth in the short run, but this
ship (Dreher and Gassebner 2013). However, these results contradict tbe ltl tin,,· 1H'g,1tive effects in the long run if the projects chosen do little
findings of Meon and Sekkat (2005), in which corruption was inore cii!itly 11h111111· long term growth and poverty reduction. Other work explicitly
for growth when the quality of governance was poor. 47 Neeman, Paserman, 1111 lrll'quality. One study found a curvilinear relationship between
and Simhon (2008: 30) conclude that "corruption is negatively correlated ptlo11 .111d the Gini coefficient, a widely used summary measure of
with output in open economies, but not in closed economies:' With rela- lit)' (Li, Xu, and Zou 2000). The Gini coefficient ranges from zero
tion to bilateral trade, Dutt and Traca (2010) find that, although corruption wit It higher numbers representing higher inequality. Over most of
reduces trade when tariff rates are low, the opposite is true for high tariffs. ilHC, a~ corruption increases, inequality increases. This relationship
Thus, there is some evidence that if the state imposes very restrictive cules, !H .ill regions, but is strongest in Latin America, !followed by Africa
corruption helps firms and individuals circumvent them, but one should lm,,h l\rcmpong and Munoz de Camacho 2006). This is consistent with
always remember that this a second-best option. The best policy is an over- ! 1111d I\ h,1gram (2005), 48 who argue that economic elites make high level
all reform that reduces the incentives to pay bribes in the first place. lfo to maintain their privileged position in very unequal societies,
Even when corruption and economic growth coexist, payoffs introduce 11111111;1 in ,I vicious circle of corruption and inequality. Similarly, Gupta,
costs and distortions. Corrupt high-level officials support too much unpro- 1i1tdi, .ind Alonso-Terme (2002) find that corruption both increases
ductive public investment and undermaintain past investments. Corruption 11111lity ,111d depresses the income growth of the poorest 20%. However,
encourages excessive public infrastructure investment (Tanzi and Davoodi \1, 1111d Zou (2000) find that in very corrupt countries, corruption can '
1997, 1998, 2002) that "crowds out" private investment. At the same time, r' illl·quality. This is not, however, a defense of corruption. Rather, it
the public infrastructure may be of low quality, so that the expectation of t ~ Ih,1l corruption can be so entrenched that it not only lowers overall
higher growth and job creation are not realized. Highly corrupt countries I' hut ,1lso wipes out the rents that benefit the political and economic
tend to underinvest in human capital by spending less on education, and 111~1 uv,·1yonc is equally poor.
46 Aidt, Dutta, and Sena (2008). Note, however, that Aidt (2009) argues that any pos-
sible short-term individual gains are outweighed by Jong-term macroeconomic growth
concerns. i nr *'llln In the United States, Apergis, Dincer, and Payne (2010) and Chong and
47 Mendez and Sepulveda (2006) find that there is a quadratic relationship between corrup- 1il81t'l11 (2007) also find a vicious circle between corruption and inequality. Dincer and
tion and growth in free countries, with a nonzero maximum. In nonfree countries, there is ( 11111;ilp (2012) find that corruption increases inequality, but do not test for reverse causal-
no statistically significant relationship. Although their samples sizes are quite small, their it1•, A.. u,d1ng to Dobson and Ramlogan-Dobson (2012), informal sector employment
results complement other results that suggest interactions between the corruption levels 1h111:~ ,,nd may even reverse the effect of corruption on inequality; they argue that for
1
and other features of government. tl,i 1 11'.tM111 wrruption is less costly in Latin America than in other regions.
 34 Introduction What Is Corruption and Why Does It Matter? 35
The effects of corruption on the quality oflife can be extreme. Although l(:111. ' 1 The connection between government size and corruption
there are likely to be a multitude of causes, it remains troubling that "83% fohl by \\cemoglu and Verdier (2000ijwho develop a game-theory
of all deaths from building collapse i.n earthquakes over the past 30 years Id to ,how how the correction of market failures causes government
occurred in countries that are anomalously corrupt" 49 (Ambraseys and rnw•. 1, regulations are introduced and inspectors are hired; ultimately,
Bilham 2011: 153). Poorly constructed roads (Tanzi and Davoodi 1998; i~ .1 I rade-off on the margin between market failure or "government
Olken 2007, 2009) are made even more dangerous by drivers who obtain lure" n>rruption_ or other ty~e~f government malfunction. However,
their licenses through bribery (Bertrand et al. 2007). Corruption is cor- 1~1, of enforcing regulat10n are only a small share of country
related with deforestation (Barbier 2004; Bulte, Damania, and Lopez 2007; IKtlH dominated by the milit ry, pensions, education, health, and
Kishor and Damania 2007; Koyuncu and Yilmaz 2009) and environmental i•rth As we discuss in Chapter 2, in assessing regulatory corruption,
degradation, both of which contribute to global warming. Access to pota- ,·,1riables are the details of the legal regime and the capacity of the
ble water, education, medical services, and basic utilities may be compro- ,111c1,1ty, not simply budget totals. It is also unclear whether corrup-
mised (Transparency International 2006, 2008, 2013c). For a set of public 11 IN 111orc prevalent under centralized or decentralized bureaucracies
services in Peru, Recanatini (201 la: 53) finds that more corrupt services 11(1\Vcrnment structures. We discuss the contrasting research results in
tend to be of lower quality. Corruption plays a key role in migrant smug- pier I J as part of a general discussion of forms of accountability.
gling, drug trafficking, human trafficking, 50 arms trafficking, and general lhcr ,tudies switch the direction of the causal arrow and try to explain
violations _o f human rights (Levi, Dakolias, and Greenberg 2007; Chanctn f111111ry differences in the level of corruption on the basis of country
0
and Sharman 2009; Europol 2013; Organization of American States 2013a, t11\,h 1Nics. Sandholtz and Koetzle (2000), for example, find that cor-
2013b; UNODC 2013; U.S. Department of State 2014). Corruption has also 1'11011 ,, lower in countries with high levels of per capita GDP, high levels
played a critical role in laying the groundwork for financial crises (Tillman 111111111( freedom, openness to trade, a Protestant tradition, and, more
2009). Corruption undermines the legitimacy of government (Can ache and I>'! with democratic structures, especially long-lasting ones. The use
Allison 2005; Sandholtz and Taagepera 2005) and its credit rating (Connolly .11\'1 ~111ment is associated with less corruption, as transparency and
2007), as well as the trust that people place in one another (Rose-Ackerman nunt:,hility are increased and discretion decreased, often by eliminating
2001a; Rothstein and Stolle 2003). I wnl,1Ll with civil servants (Andersen 2009). Some research, which
There is some debate over the relationship between the size of gov- ll~l~II" in Chapter 7, finds that the participation of women in politics
'
ernment and the extent of corruption. Downsizing through program 111rruption, b~t, as we show, that result is not very robust and its
elimination and privatization may reduce corruption because some pro- I111pl ications are unclear.
grams no longer exist. However, if a program is merely cut back, payoffs · I I ross country work is not based on an adequate structural model
may increase in size and extent as applicants compete for the scarce sup- thu way rnrruption interacts with other features of the environment.1fe
1
ply (Rose-Ackerman 2000: 99). Pointing to the examples of the Nordic llllr~ highlight important empirical regularities, but the direction of ca\i-
countries, where low corruption and high govern~ent budgets coex- tlll!I I~ often unclear. For example, are low levels of income and growth a '-
ist, Friedman et al. (2000) show that, in a set of 69 countries, higher tax llll\ lll'l' or a cause of corruption, or both? Sometimes the causal link is
shares are associated with low corruption. According to them, low cor- pl)' lh\l'rtcd, not demonstr<\ted. In reality, it seems likely that the causal "
10w 11111~
·-' ruption induces more economic activity to occur in the formal economy both ways, often creating vicious or virtuous spirals (Lambsdorff
- where it is taxable, and in democracies citizens are willing to support Hl(i. Hml' Ackerman 2006b, 2008a; Treisman 2007a). A country may be
I high levps of public expenditures only if the government is honest and
\
c111p11kal works (Goel and Nelson 1998) found that government size, measured
49
The authors predict the expected level of corruption based on per capita GDP; "anomalously 11\1·111111t·nt spending, was positively correlated with higher corruption, but oth-
corrupt" refers to those countries that are more corrupt than predicted. u (« :,·, ling an<l 1l1acker 2005; Glaeser and Saks 2006) find no correlation. As Gerring
50
Transparency International, "Corruption and Human Trafficking," http://files.transparency till 11111, ~l'I (2005: 250) note: "big government is not necessarily corrupt government."
.org/content/download/ 111/447/file/2011_ 3_ TI_ CorruptionapdHumanTrafficking_ 111ill1111 lo c;oel and Nelson (2011), the effect depends on how both corruption and
EN.pdf (accessed October 8, 2015). ,vn 111111·111 \lt.C are measured.
 36 Introduction What Is Corruption ana Why Does It Matter? 37
caught in a corruption trap where corruption breeds more corruption and 111iv1::, arise within public programs. We identify pathologies that recur
discourages legitimc1:te business investil}ent. Corruption limits growth and 1:1 Iors, drawing specific examples from a range of concrete situa-
destroys trust in government, and low growth and distrust of the state fuel ,,_ "' Wl• show that corruption can create inefficiencies and inequities
.J
and seem to justify corruption. Conversely,· low corruption aids growth, I l~ .. ,1 hest, inferior to legally 'established payment schemes. Reforms
and higµ growth creates a societal demand to lower corruption even n·d11, 1• the incentives for bribery and increase the risks of engaging
further. Vicious spirals are not, of course, inevitable, but they are a risk, I t 1111 upt 11m. The goal is not the elimination of corruption per se but an
and escaping them is usually difficult. Such spirals will not be evident in 1pn,,·,·111l'.lll _in t~e overall effi~ency'. fairness, and legit~acy of the state.
cross-country analyses although they may be behind some of the results. 1lw 11,1,d 1'11m1nat1on of corrupti~n will never be worthwhile, but steps can
A related empirical issue concerns the relative power of political ver- 1i1kt111 lo limit its reach and reduce the harms it causes.
sus economic actors in determining the divisions of corrupt gains. Ihl• ,,·11111d <limension recognizes that corruption has different meanings
Cross-country indices provide no direct evidence of how the benefits lllfr11·11l societies. One person's bribe is another person's gift. A political
are shared. Following John Joseph Wallis, one can distinguish between le,•111 public official who aids friends, family members, and supporters
"Systematic corruption ... when politics corrupts economics ... [and] Venal 111 praiseworthy in some societies and corrupt in others. As econo-
corruption ... when economics corrupts politlcs" (Wallis 2006: 25, italics l•IN, w1· l,tnnot provide an in-depth analysis of the role of culture and
added). If those with political power distribute economic power, systematic 1-l!ll y 111 the development of corruption, but we can point out when the
corruption may exist; when those with economic power influence policy 111 l he past no longer fits modern conditions. Our aim is not to set a
or law, it is venal. 52 Of course, few systems will be pure examples of either IIIVl'ii,,11 ,t.indard for where to draw the legal line between praiseworthy
type - systematic and venal corruption tend to coexist - but more research rnd illegal, unethical bribes. Rather, we isolate the factors that should
on the division of gains would be extremely worthwhile. 11111} till' choice. Culture and history are explanations, not excuses. Every
Unfortunately, even when the statistical difficulties are well-handled, 111Hry h.,s experienced high levels of corruption at some point, but many
empirical regularities based on cross-country indices are of limited use to fi_,111111 ,I way to reduce both the amount of corruption and the impact
policy makers. They can raise consciousness about the negative impact of 111 ,;ol.icty. 55 At the same time, we recognize that corruption can influ-
corruption on growth, productivity, and the distribution of wealth, but they 11llurc, especially trust and honesty. If corruption increases, it has an
are of little use in designing anticorruption strategies. Designing policies Ilic 1111pact on societal values, leading to cynicism.
around such studies, with their imperfect data sets and aggregated mea- (he rllinl <limension considers how the basic structure of the public and
sures, seems _£roblematic. 53 In what follows, we will focus, instead, on the 1•1 lvlihl t-1·1 tors produces or suppresses corruption. We examine the rela
costs and benefits of reforms in particular sectors and for spe<;:ific types of hip hl'tween corrupt incentives and democratic institutions and dis
government actions. Ilic 11•l,1tive bargaining power of public and private organizations and
llv1d11.d ,1Ctors. Reform at this level may well require changes in both
II~! It 111 ional structures and the underlying relationship between the mar
V. Plan of the Book
11111d till' state.
' -, \ W~ analyze the problem of corruption along four dimensions. The first takes '
(the basic institutions of state and society as given and asks how corrupt
52
See also Khan (1996, 2006) and Johnston (2005). 1n Johnston's typology influence markets I hmu 11111·1 l'stcd in specific sectors are referred to Campos and Pradhan (2007); 1'1's
are an example of venal corruption, while systematic corruption is more characteristic of /n/111/ C: ormption Report series (available at http://www.transparency.org/research/
elite cartels and official moguls. For oligarch and clan corruption both types of corruption i (11< (1:,,l'd September 28, 2015); Graycar and Smith (2011); S0reide and Williams
are likely to be pervasive. (WI'!), ltuw, Ackerman and S0reide (2011); Klitgaard (1988); and the sector-specific sill'\
53
. There is also some skepticism over whether the corruption and GDP growth correlation is ll11krd 1111tli-1 "Focus Areas" on Tl's homepage (http://www.transparency.org/) (accesst•d
driven by faulty measurement, specifically the use of perceptions-based corruption mea- pir111ltl'1 28, 2015).
sures. Treisman (2007b) and Aidt (2009) find no strong relationship between corq1ption (:1,,. ,1·1 .ind Goldin (2006a) for a series of essays on how the United States reduci:d
experiences and growth. i!i 111pt 11111 during the late nineteenth and early twentieth centuries.
 38 Introduction
The final ,section of the book turns to the difficult issue of achieving
reform. Even if a government is aware of corruption, it may have no incen-
CHAPTER 1 APPENDIX
tive to undertake reform, unless domestic or international actors exert pres-
sure to do so. Proposals for reform lead to the problem of domestic political
will. Good ideas are useless unless someone is willing to implement them.
Which domestic conditions are most likely to convince leaders that fight
,,.,.,'i Country Measures of Corruption
\ ing corruption is worthwhile? We draw some lessons from successful and
sustainable policies carried out in the past. Although no two countries fac
the same set of background conditions, 'modern-day reformers can learn
\
something from the historical record. We bring in the international com
0
munity, aid and lending organizations, cross-national civil society groups,
such as TI or Global Witness, and multinational economic and political
bodies. For some countries, especially those at low levels of development, tlx, wt· l'xplain some of the individual surveys that are used
the role of multinational businesses is critical. If th~se firms collaborate in thti ('l'I .,nd the CCL This is not a comprehensive list of data
. maintaining corrupt regimes, they undermine development goals. Finally, •II 11pt Ion .ind related topics: there are now scores of data sets
we assess inten;ational efforts to control money laundering, often assoc! \ l
ountry to geographically specific, and more are devel-
ated with high-level corruption. i!', 1 '!lit·
purpose of this appendix is merely to give an over-
This book does not end with a compilation of "best practices:' Instead,
it suggests a range of alternatives that reformers must tailor to the goals o I llm, in ,llphabetical order, the data sources used to calculate
reform and the conditions in individual countries and sectors. Combating l'O~ilc Indices for corruption corresponding to 2013, as well as
corruption is a means to an end. That end may be efficient production and tlw I Ypr ol data and the number of countries included in each.
development, impartial and equitable government, human development :Ill' expert opinion or executive surveys - commonly
and flourishing, or goals related to the performance of a particular sector 1!111s111veys"; only five public surveys are used in calculat-
such as health, education, or national defense. The appropriate reforms lhruo 1,t' which are regional. The sources used in calculating
need to be tailored both to the immediate incentives surrounding the cor
rupt act, and to the broader institutional context - both formal political and Iii I, ,,vailable at Tl's Anti-Corruption Research Network, "Datasets;'
market institutions and the informal institutions arising from a society' 11 n t·.11, hnl.'lwork.org/resources/datasets (accessed September 27, 2015).
culture. , 11!.ll ···• (2011) identify three stages in the historical development of
11111 c 111n11 (I) composite indices such as the CPI; (2) "comparative
We stress one fundamental lesson. Reform should not be limited to th
111,.· 111s' that allow for cross-country comparisons in space and time;
creation of "integrity systems" or "anticorruption agencies:' Instead, fun pc,lf,t:, ,l.'ctor-specific microanalyses, which aim to examine the causes
damental changes in the way government operates ought to be at the heart i 1111• l'll<'lls of policies in specific contexts. We have not followed this
of the reform agenda. The primary goal should be to reduce the underly wt ,,re more interested in explaining what the cross-country data
:u l' ~l'wrnl methodologies in the third category that are omitted here, "
ing incentives to engage in corruption ex ante, not to tighten systems of ex 111 li1111tnl to Public Expenditure Tracking Surveys (PETS), which com-
post control. Enforcement and monitoring are needed, but they will hav lt,(1111 ticd 111 onl.' level of government to those received or disbursed at the
little long-term impact if reforms do not reduce the basic conditions that IM11l~.k11 1111d Smith 2004; Reinikka and Svensson 2006; Sundet 2008); cost
h 111111p,u~· projected budgets in public works to actual costs (Engerman
encourage payoffs. If these incentives and institutions remain, the elimina 110<,, 1lyvhjcrg 2007; Flyvbjerg and Molloy 2011); mismatches between
f\on of one set of "bad apples" will soon lead to the creation of a new group ir1111111r ,111d cumulative public expenditure on il)frastructure (Golden and
of corrupt officials and private bribe payers. 1•ht•'l~I .,11,llh of roads, comparing core samples to materials reportedly used
\ !10\/). prnpnrtional convictions on corruption charges (Corporate Crime
,litf"'' ,,nd Saks 2006); and the number of newspaper articles related to
l•i! 11, I ~1111; C,rntzkow, Glaeser, and Goldin 2006).