... exploiting the increase in the legal punishment fo

1
Lecture 2 Corruption Institutions and
Development - P. Vicente University of
Oxford http//www.pedrovicente.org/Teaching/Oxford
/instdev.htm
2
1 Corruption and Rent-Seeking

• To introduce both concepts we use an example on
  the financing of a public project (Besley, 2006)
• Suppose that a community of N individuals has to
  make a single social decision whether to build a
  public project
• This decision is denoted by with
  denoting the case in which the project
  is constructed
• If the project is built, the government uses a
  head tax
• We assume there are two groups of citizens
• Those who value the project and receive utility
  b from it and those who do not value the project
  (0 utility)
• The citizens who value the project constitute
  proportion of the population
• We assume that all citizens have equal income, y
• The project costs c to implement (assume
  , i.e. each citizen is able to bear the
  per capita cost of the project).

3

• In order to motivate the presence of the
  government in this problem
• Suppose the project is to be funded through
  private non-mandatory contributions by each
  citizen (where )
• If , the project is implemented
  otherwise it is not
• Any surplus from contributions is divided
  equally across all citizens
• Analyzing Nash Equilibria of the contribution
  game
• No citizens apart from those valuing the project
  will contribute for the project
• No aggregate contribution amount surpassing c
  will arise
• If then there is always a NE where
  for all citizens - no citizen would like
  to deviate since his gain from the project is not
  enough for him to bear all the cost
• If , there is a NE with equal
  contributions summing c again, think of a
  deviation by one citizen
• No provision of the project (market failure)
  comes in this model from a coordination failure

4

• If indeed it is the government who decides on
  the provision of this project
• We will see no change in aggregate welfare if it
  decides not to implement it
• We will see change if the project
  is realized
• (from having change for all
  individuals valuing the project and change
  for the remaining players)
• This means that the project is worthwhile if
• If this condition holds, then any political
  mechanism in which the project fails to go ahead
  can be considered a government failure

5

• We define corruption as a situation where a
  monetary payment (bribe) is paid to the policy
  maker to influence the policy outcome
• Suppose the policy maker can earn a private
  monetary rent of for setting
  regardless of whether the project is worthwhile
  or not
• We assume this payment (fixed for
  simplicity) is a transfer made by a subset of
  organized citizens (think of the lobbying model
  of last lecture for more detail)
• Suppose that . If this happens,
  the policy maker will always want to implement
  the project, as the rent he receives is higher
  than the per capita contribution to the project
• The utility of the citizen depends on whether
  they finance the transfer
• Suppose that a fraction , of citizens
  who derive positive utility from the project,
  finances this transfer on an equal per capita
  basis

6

• The payoffs are then
• for the policy maker
  (assuming he derives 0 utility from the project)
• for those who
  favor the project and pay bribes
• for those who favor the project
  but do not pay bribes
• for those who do not favor the
  project
• Assuming that
  , we have individually rational transfers, so
  that both organized citizens and policy maker are
  better off with corruption

7

• If corruption increases social
  surplus relative to any situation yielding
• If we would have corruption
  decreasing aggregate payoff (efficiency)
• If corruption would only imply a
  change in the distribution of payoffs (not only
  across policy maker and organized citizens, given
  the payment r, but also across those in favor of
  the project and those against, since the project
  would be implemented)
• The main point here is that corruption does not
  necessarily imply a change in the aggregate
  payoff (efficiency), though there are always
  distributional effects
• However, if the government is benevolent (e.g.
  in the situation above, where the
  policy maker maximizes his payoff and happens to
  be a citizen who does not favor the project),
  i.e. does the right thing in the absence of
  corruption, then we have efficiency problems
  stemming from corruption

8

• We now present a version of the classic model of
  rent seeking due to Tullock (1980)
• Suppose that citizen i can pay to influence
  the policy maker regarding the implementation of
  the project (in favor or against it)
• Crucially we assume that is a real resource
  cost (e.g. labor time), which cannot be
  appropriated by the policy maker (we assume this
  agent does not get any positive or negative
  payoff from this influence)
• We assume that each citizen commits resources
  in favor of the project and
  against the project
• Define and
• Then assume that the probability that the
  project goes ahead is captured by the function

9

• In the symmetric Nash Equilibrium of this rent
  seeking game, nobody who favors the project
  commits any resources to influencing against, and
  vice-versa.
• The payoff of a citizen i who favors the project
  and contributes
• The payoff of citizen j who opposes the project
  and contributes
• We now solve for the NE where every citizen puts
  in the same effort level, i.e. we take FOCs of
  the above and equate them

10

• This means the key magnitude in defining the
  probability that the project is implemented is
  defined by the ratio of the per capita cost of
  construction of the project and the (per
  capita by definition) benefit for those in favor
  of the project b
• Total expenditure on rent seeking at the
  equilibrium is then (using last expression and
  one of the two FOCs)
• This implies the following ex-ante surplus at
  the NE

11

• If , we see that
• If the project was to be implemented without
  rent seeking, we are now in a worse position
  given uncertainty of implementation and real
  resource costs of influencing
• If the project was not to be implemented without
  rent seeking, we may have a better aggregate
  ex-post payoff if the project goes ahead and
• If , and
• If the project was to be implemented without
  rent seeking, we may be in a better position if
  the project does not go ahead and
• If the project was not to be implemented without
  rent seeking, we have a lower aggregate payoff
  given the possibility of implementation and the
  costs of influencing.

12

• If the government was going to do the right
  thing in the first place (benevolent
  government), we can only be in a worse situation
  when we have rent seeking. This is given
• the possibilities of doing the wrong thing,
  i.e. the same kind of situation we had for
  corruption
• the additional real costs of influencing that
  help defining rent seeking activities
• Indeed as defined by Buchanan, Tollison and
  Tullock, 1980, rent-seeking corresponds to
  resource wasting activities of individuals in
  seeking transfers of wealth through the aegis of
  the state
• Contrasting this definition with the usual
  definition of corruption, abuse of public office
  for private gain, we see that although
  potentially non-intersecting (above examples),
  rent seeking can be seen as an inefficiency
  qualification of certain activities that are
  considered corruption
• Corruption less well-defined in economics, but
  easier for empirics

13
2 Corruption as an Agency and Competition Problem

• We start by looking at corruption as an agency
  problem. The analysis here is due to an example
  by Becker and Stigler (JLS, 1974), the first
  economics paper addressing corruption
• Let p be the probability of detecting
  malfeasance (corruption) at any period
• b is the monetary value of the gain to
  enforcers (public officials) from malfeasance
• We take a discount rate r as given
• Let be outside-opportunity earnings for
  enforcers at time i of their career
• The problem is to find the minimum enforcer
  salary that would discourage malfeasance
  (at each moment in time)

14

• We start from the final period (n) of the career
  of a given enforcer
• At that time, he can receive with
  certainty or, if he engages in malfeasance,
  assuming he is risk neutral, he considers
  expected income with probability
  and with probability p (dismissal). We
  can therefore compute
• 
  (1)
• as the minimum wage that discourages malfeasance
• From the point of view of period , we
  have
• Substituting (1) in the above expression, we
  find , which can be generalized
  (continuing the backward process) to the other
  periods as
• 
  for

15

• The income as an enforcer in the first
  periods is higher than in the outside opportunity
  by an amount that is
• negatively related to the probability of
  detection
• positively related to the gain from malfeasance
• The excess of income in the last period can be
  thought as a pension, stemming from the
  increasing attractiveness of malfeasance as the
  enforcer approaches n
• When the enforcer approaches n, he may afford to
  be riskier in the sense that if dismissed he
  loses a lower future additional income from being
  an enforcer. This pension is then
• (note that the premium at all periods until
  is equal to foregone interest on this pension,
  )

16

• Finally, we can write the present value of the
  lifetime income as an enforcer as
• where would be implemented as
  an entrance fee for the enforcer profession,
  making it equally attractive as the outside
  opportunity, though still free of malfeasance if
  the period-by-period wages characterized above
  were to be implemented
• This would be the optimal contract for an
  enforcer so that malfeasance or corruption would
  not arise
• This assumes imperfect observation from the
  wage-giver, the principal (given probability of
  detection)
• Therefore, although not explicitly writing about
  it, Becker and Stigler were already tackling what
  was later labeled as an agency problem

17

• We now relate corruption with the level of
  competition in the provision of the government
  good. This is due to Shleifer and Vishny (QJE,
  1993)
• We consider a simple model of one
  government-produced good like a passport
• We assume this good is homogeneous and that
  there is a demand for this good from
  the private agents
• Let the official government price for this good
  be p
• We assume this good is sold by an official, who
  represents the government and can refuse to
  provide the good
• We assume the government official does not face
  any risk of detection/punishment
• for simplicity, he does not bear any cost from
  producing the good

18

• Two cases
• Corruption without theft, where the official
  turns over the official price to the government
  (i.e. p is the marginal cost for the official in
  this case)
• Corruption with theft, where the official hides
  the sale to the government and keeps all he
  receives for it (i.e. zero marginal cost to the
  official)

19

• We now consider an extension of this setting to
  allow for private agents in need of several
  complementary government goods
• Formally consider first a joint monopolist
  agency (collusion) that sets the prices (gross of
  bribes) and of two government goods
• Let and be the quantities of these goods
  sold, and the official prices be denoted by
  and
• (per unit bribes can be written as
  and )
• The joint monopolist agency sets (analysis
  for is symmetric) from
• where and are marginal revenues
  from the sale of goods 1 and 2
• When the two goods are complementary we have
  , which means
• (i.e. a lower bribe on 1 aimed at increasing
  profits on 2).

20

• Suppose alternatively that permits 1 and 2 are
  allocated by independent agencies. Since each
  agent takes the others actions as given, at the
  independent agencys optimum
  ,
• This means that under independent sale, the per
  unit bribe is higher and the output lower than at
  the joint monopolist optimum
• Under a third alternative, each one of the
  several complementary government goods can be
  supplied by at least two government bodies
• This means that if an official asks for a bribe,
  the possibility of having a lower bribe exists
  with another official
• This drives bribes down to zero at the
  equilibrium (just like in Bertrand competition)
• We can then conclude that the level of the bribe
  will be lowest with competition, intermediate
  with collusion among officials, and highest with
  independent monopolistic sales
• Competition in the provision of government goods
  is therefore key to contain corruption

21
3 Macro Empirical Research

• Measuring corruption at the macro-level
• Compilation of country-level indices of
  corruption from middle nineties has brought an
  enormous degree of attention towards
  cross-country studies on the causes and
  consequences of corruption in the 90s/early 2000s
• Today there are mainly two institutions
  publishing annual country indicators of
  corruption Transparency International (from
  1995) and the World Bank Institute (from 1996)
• These indices are aggregations of indices from
  multiple sources, involving field in-country
  surveys and international organizations internal
  (expert) polls

22

• Consequences of corruption
• Main reference Mauro (QJE, 1995)
• First paper providing evidence of a negative
  impact of corruption on economic growth, through
  a negative effect on investment
• Since institutions and economic growth may be
  thought to evolve jointly
• Use of an index of ethnolinguistic
  fractionalization ELF (measuring the probability
  that two persons drawn randomly from a countrys
  population will not belong to the same
  ethnolinguistic group) as an instrumental
  variable
• ELF is highly correlated with corruption and
  other institutional variables
• Think of Shleifer and Vishnys comparison of
  collusive public agencies and independent sale
• ELF can be assumed exogenous to economic
  variables (that not through institutional
  variables)

23
4 Micro Empirical Research
24
Validity of Macro Indices

• Fisman and Miguel (JPE, 2006)
• These authors develop an empirical approach for
  evaluating the role of both social norms and
  legal enforcement in corruption by studying
  parking violations among United Nations diplomats
  living in New York
• Natural Experiment mission personnel and their
  families benefit from diplomatic immunity, a
  privilege that allowed them to avoid paying
  parking fines prior to Nov. 2002
• All else constant diplomats/same demographic
  profile same location (close to the UN building)

25

• Fisman and Miguels results
• parking violation corruption measure is strongly
  positively correlated with other (survey-based)
  country corruption measures and that this
  relationship is robust to conditioning on region
  fixed effects, country income, and a wide range
  of other controls, including government employee
  salary measures
• This suggests that home country corruption norms
  are an important predictor of propensity to
  behave corruptly among diplomats
• by exploiting the increase in the legal
  punishment for parking violations after Nov 2002,
  enforcement matters too it led to immediate and
  massive declines of approximately 98 percent in
  parking violations
• evidence in favor of convergence to
  zero-enforcement with tenure in New York

26
(No Transcript)
27
(No Transcript)
28
Measurement

• Reinikka and Svensson (QJE, 2004)
• describe and analyze the results of a survey
  tool to track the flow of public resources to
  intended destinations (Public Expenditure Track
  Survey)
• focus on a large public educational program in
  Uganda - a capitation grant to cover schools
  nonwage expenditures - financed and run by the
  central government, using district offices as
  distribution channels
• comparison of disbursed flows from the central
  government (intended resources) with the
  resources actually received by schools
• 250 primary schools were surveyed, and data on
  receipts were collected for 19911995
• Unique panel data set to study the level and
  determinants of local capture

29

• Reinikka and Svenssons results
• on average, schools received only 13 percent of
  central government spending on the program most
  schools received nothing the bulk of the grants
  was captured by local government officials (and
  politicians)
• large variation in grants received across
  schools actual spending, unlike budget
  allocations, is regressive, as schools in
  better-off communities experience a lower degree
  of capture
• suggestive of a bargaining model between schools
  and local officials
• the poor seem to be harmed by corruption

30
(No Transcript)
31
(No Transcript)
32
Incentives

• Olken (JPE, 2006) addresses
• the relative effectiveness of different types of
  incentives on diminishing corruption
• top-down based on employer-created monitoring
  
• bottom-up based on accountability towards the
  public
• while using direct measures of corruption (not
  relying on subjective/perception-based data)
• More specifically, Olken studies a number of
  Indonesian villages where a World Bank-financed
  road project was implemented
• Corruption is measured by having samples of
  roads constructed under the project evaluated
  independently by engineers and compared with
  reported expenses
• Research Questions
• How does corruption respond to an exogenous
  change in the likelihood of government-induced
  audits?
• How does it respond to a change in the level of
  demand for accountability by villagers?

33
(No Transcript)
34

• Olkens results
• Increase probability of an audit (from 4 to
  100) resulted into substantial reductions in
  corruption (8)
• Both in over-invoicing of labor and un-accounted
  for materials
• But jobs given to family members increased
• Both public accountability interventions implied
  small decreases in corruption
• However, they did imply differential movements
  in the mechanisms of corruption invitations
  alone decreased labor-hidden corruption

35
(No Transcript)
36
Efficiency

• Bertrand et al (QJE, 2006) address the issue of
  whether corruption is efficient or not (back to
  our first model in the lecture)
• Back to the analysis of the consequences of
  corruption, at the micro level of Mauro (1995).
• Bertrand et al follow a sample of applicants
  through the process of obtaining a drivers
  license in New Delhi (India)
• In the end surprise independent driving tests
  were submitted to all participants this was key
  for the derivation of welfare consequences of
  corruption
• Main Research Questions
• Can corruption be used to speed up the process
  of getting a license (grease the wheels)?
• Do bad drivers use bribes to get a license
  (efficiency question)?

37
(No Transcript)
38

• Bertrand et als results
• Apart from confirming a high level of corruption
  (participants paid on average more than twice the
  official fees to obtain their driving license),
  it was found that individuals who wanted to get a
  license faster did so (grease the wheels
  confirmed).
• Corruption also allowed the circumvention of
  socially useful rules (inefficiency)
• 69 of the bonus group failed the independent
  driving test (61 in the control group)
• Being a good driver (lesson group) barely
  increased the likelihood of obtaining a license

39
(No Transcript)
40
Value of Political Connections

• Fisman (AER, 2001)
• This author looks at the case of Suhartos
  Indonesia
• Natural experiment
• An event study is conducted on a string of
  rumors about former Indonesian President
  Suhartos health during his final years in office
• Comparison of the returns of firms with
  differing degrees of political exposure

41

• Fismans results
• In every case the returns of shares of
  politically dependent firms were considerably
  lower than the returns of less-dependent firms
• Furthermore, the magnitude of this differential
  effect is highly correlated with the net return
  on the Jakarta Stock Exchange Composite Index
  (JCI) over the corresponding episode
• In a pooled regression using all the events,
  allowing for an interaction between political
  dependency and event severity, the coefficient
  on this interaction term is positive and
  statistically significant, implying that
  well-connected firms will suffer more, relative
  to less-connected firms, in reaction to a more
  serious rumor
• Suggest that a large percentage of a
  well-connected firms value may be derived from
  political connections

42
(No Transcript)
43
IO

• Barron and Olken (2007)
• Examine the degree to which standard pricing
  theories from industrial organization are
  consistent with actual patterns of bribes and
  extortion payments
• Context bribes paid by truck drivers on their
  trips to and from the Indonesian province of Aceh
• Truck drivers in Aceh make a variety of illegal
  payments, including payments to police and
  military officers to avoid harassment at
  checkpoints along the roads, payments at weigh
  stations to avoid fines for driving overweight,
  and protection payments to criminal organizations
  and the police.
• Measurement enumerators accompanied truck
  drivers along their regular routes to and from
  Aceh
• From Nov 2005 to Jul 2006, 304 trips to and from
  Aceh, with more than 6,000 illegal payments
  directly observed along the routes

44

• Barron and Olkens results
• On average, drivers spent about US 40 per trip,
  or about 13 percent of the total cost of a trip,
  on bribes, extortion, and protection payments
• During the period of the survey, a major
  withdrawal of government military happened in
  Aceh
• the average bribe paid in North Sumatra
  increased significantly in response to the
  reduction in the number of checkpoints in Aceh
  (market structure matters)
• Downstream checkpoints i.e., those that are
  closest to the final destination receive higher
  bribes than upstream checkpoints i.e., those
  that are closer to the origin of the trip
  (hold-up)
• Officials at checkpoints, for example, appear to
  practice third-degree price discrimination,
  charging higher prices to those drivers with
  observable characteristics that indicate a higher
  willingness to pay (e.g. newer trucks)

45
(No Transcript)
46
Causes A Natural Experiment for the Resource
Curse

• Vicente (2007 JDE, 2009) looks at an oil
  discovery-induced natural experiment (where the
  source of exogenous variation occurs naturally)
• This enables analyzing, for a country (Sao Tome
  and Principe STP, West Africa) that departs from
  being resource-free and from having weak
  institutions, how the resource curse may begin
  and evolve
• Cape Verde, control group
• The theoretical mechanism proposed is the elite
  wants to increase corruption primarily in
  allocations that enable increasing the likelihood
  of being in power when the oil revenues arrive
• Research Questions
• Do natural resources cause corruption?
• Which are the sectors/allocations by which the
  discovery-induced flow of increased corruption
  begins?

47
(No Transcript)
48
(No Transcript)
49

• Specifications
• Difference-in-Difference
• Triple-difference

50

• Vicentes results
• Generalized increase in perceived corruption
  (ranging from 31 to 40 of the subjective scale
  used)
• Highest and most significant effects on vote
  buying, scholarships for higher education (with
  clear implications in future holding of state
  power) and customs (consumption)
• Perceptions capture well the sign of the changes
  but exaggerate magnitudes

51
(No Transcript)
52
(No Transcript)
53
5 Summary

• Corruption vs. Rent-seeking the first not
  assuming inefficiency and closer to empirics the
  second better founded theoretically
• Incentives crucial to contain corruption
• Corruption as a competition problem
• Cross-country indices are not too bad but
  measurement is fundamental (missing value is
  probably the best way to go)
• Top-down corruption vs. bottom-up some answers
  but some questions remain
• Corruption greases the wheels but is socially
  inefficient
• Natural experiments on corruption politically
  connectedness and the effect of oil