class notes institutions

advertisement
Social capital, Institutions and development
Table of Contents
Introduction .......................................................................................................................................... 2
2. Corruption and economic growth .................................................................................................... 2
2.1. Causes of corruption ................................................................................................................. 2
2.2 Consequences of corruption ....................................................................................................... 5
3 Institutions, social capital and development ..................................................................................... 7
3.1 Social capital .............................................................................................................................. 7
3.1.1 What is social capital ? ....................................................................................................... 7
3.1.2 Social capital and Growth ................................................................................................. 11
3.2 Institutions................................................................................................................................ 14
4. Measures and consequences of corruption: a micro approach....................................................... 22
References .......................................................................................................................................... 26
1
Introduction
The chapter on endogenous policy has shown that politicians react to incentives in a relatively
efficient manner, like other agents, given their objective functions and the constraints they
face. This naturally generates a follow-up question, namely what type of institutions are likely to
provide adequate incentives for politicians? Under what conditions do poor governance and
corruption develop?
A central question is the role of colonialism in the national institutional-development
trajectories. On one hand, colonialism destroyed the local elite, leaving behind infrastructure but no
elite able to govern, due to lack of transfer of skills or social capital. On the other hand, colonialism
has allowed, in some cases, to transplant growth-friendly institutions developed in Europe. The net
effect is ambiguous.
If there are many institutional-quality indices on the market, establishing a causal relationship
between institutional quality and the performance of economic systems (as proxied by GDP per
capita) is very difficult due to severe endogeneity problems. The literature on institutions is
characterized by an extraordinary creative search for instumental variables. For instance, Mauro
(1995) used ethno-linguistic fragmentation to instrument corruption; Acemoglu, Johnson and
Robinson (2001, 2002) used the mortality of European settlers between the 17th and 19th centuries
to instrument their ability to transplant lasting institutions; Iyer (2005) used the death of monarchs
without male heirs to instrument British interference in local affairs in India; Feyrer and Sacerdote
(2009) used the trade winds’ prevalence and stability to instrument European colonization of small
islands.
2. Corruption and economic growth
2.1. Causes of corruption
There are many possible causes, but in some cases (e.g. India) the causal chain looks like:
o Increase in regulatory administration power
o Increasingly binding budget constraint
o Degradation of civil servants’ purchasing power
o Spread of corruption.
This is illustrated in Figure 1.
Figure 1: Corruption, regulation and civil-service wages
2
Average publicservice wages
Low bribe demand
Low corruption
High corruption
High bribe
demand
Extent of State
regulatory powers
Low bribe supply
High bribe supply
Multiple equilibria in corruption are also possible if the likelihood of sanctions decreases with the
proportion of corrupt individuals in public the administration. For instance, suppose that the
probability that an individual accepts a bribe depends on the proportion of corrupt colleagues in the
administration. Suppose that the probability to accept a bribe at the individual level depends on the
proportion of corrupt colleagues/superiors through an S-shaped function, as in Figure 1.
Equilibrium has to be along the 45o line, where the proportion of corrupt civil servants is equal to
the probability that anyone of them is corrupt. With the S-shaped behavioral function cutting the
45o line three times, there are three candidate equilibria. However, one of them (the middle one) is
unstable, as an increase in the proportion of corrupt individuals raises the probability of being
corrupt by even more, which itself raises the proportion of corrupt individuals, and so on. It is easily
verified that the two other equilibria (low- and high-corruption equilibria), by contrast, are stable.
Figure 1 : Multiple equilibria in corruption
Pr(i  Z)
f(C)
Stable equilibrium with high corruption
Unstable equilibrium
Stable equilibrium with low corruption
C
3
There is also an issue with the multiplicity of “predators”. In Asia (and in Russia under
communism, according to Vishny and Shleifer 1993), anecdotal evidence is that one corrupt official
would ask for a bribe, after which the firm would be largely left in peace. By contrast, in Africa,
multiple predators will come, one after the other, apparently without coordination, and ask for
bribes, the combined effect of which is to choke the business. To see how this can work, consider a
corrupt official who comes to a factory and assesses the owner’s willingness to pay from the
factory’s size, i.e. its employment. The owner anticipates this and sets employment optimally taking
into account the bribe demand. Let us model this as a Stackelberg game in which the corrupt
official first sets a bribe-demand function, which is then known to the factory owner who sets
employment optimally.
Let L be employment and let the production function be f  L    L   L2 ; let w be the wage rate
and b the bribe. The factory owner solves
max L   L, b   p  L   L2    w  b  L
(1)
With first-order condition
 p  2 L  w  b
or
L b 
pwb
2
Knowing this, the official solves
b  p  w  b 
2
max b bL  b  
(2)
with FOC
b
1
 p  w 
2
(3)
Now suppose that there is a bunch of corrupt officials, each asking for a bribe bi . Let B   i bi
Optimal employment for the factory owner is now
L  B 
pwB
2
(4)
and a representative corrupt official solves




bi  p  w  B  bi  p  w   j i b j  bi 
max bi bi L  B  

2
2

 bi p  wbi   j i bi b j  bi 2
2
with FOC  p  w   j i b j  2bi , or
4
(5)
bi   p  w  B
(6)
Suppose there are two such officials. Then B  2bi so
bi 
1
2
 p  w  and B   p  w  ;
3
3
(7)
Suppose now that there are three. Then B  3bi and
bi 
1
3
 p  w  and B   p  w  .
4
4
(8)
That is, as the number of predators rises, each individual bribe ( bi ) goes down but the total (B) goes
up, until the point where it chokes completely the firm. This is a story applying to societies
characterized by lack of communication, lack of trust, and lack of central authority, as many of
Africa’s fragile states are (e.g. the Democratic Republic of Congo, the Central African Republic,
Benin, etc.).
2.2 Consequences of corruption
Mauro (1995) attempts to estimate the effect of corruption on growth adding an index of corruption
at the right of a Barro growth equation. Estimation:
gi ,196085  0  1Yi ,1960  5 I i   2 Si ,1960  3ni  4CORRUPTi  ui
where gi is the growth rate of GDP, Yi represents the effect of convergence if the coeff. is
negative, Si is the secondary enrolment rate and ni is the population growth. CORRUPT is an
average index of corruption for the period between 1960-85. Estimation problems:
o Reverse causality (growth can alter the preference for the present which itself affects corruption)
o Omitted variable (more general governance aspects, institutions etc.)
How to instrument? Mauro uses an index of ethno-linguistic fragmentation calculated by Soviet
sociologists in the 1960s. The index’s formula is given by
I
ELF  1   si 2
(9)
i 1
where si is the ethno-linguistic part of the group i within the population of the country. It is time
invariant (has been calculated only once and does not vary much over time anyway) and gives the
probability that two individuals chosen randomly from the population belong to two different
ethno-linguistic groups). Why use ethnic fragmentation as an instrument for corruption? Banerjee
and Pande (2010) show that corruption correlates with the ethnic polarization of political parties.
The intuition is that when parties (and voters) define themselves by their ethnic group, tribe, clan,
etc., these characteristics tend to supersede the intrinsic qualities of politicians (like honesty). In
other words, people are willing to vote for corrupt representatives as long as they belong to the
same ethnic group. Indeed, Figure 2 shows the correlation between country corruption rankings in
Transparency International’s classification (on the vertical axis) and an index of the ethnic
5
polarization of political parties calculated by the Freedom House, a U.S. think tank (on the
horizontal axis). The correlation is striking.
Figure 2: Relationship between ethnic fragmentation of political parties and corruption
Source : Banerjee and Pande (2010)
Regression results, using either OLS or two-stage least squares (2SLS), are shown in Table 1.
Table 1: Regression results, growth and corruption
6
3 Institutions, social capital and development
3.1 Social capital
3.1.1 What is social capital ?
There has been huge debate in the economics literature on whether “social capital” is a useful
(meaning measureable) concept or just a buzzword. Essentially the idea is that representing human
behavior through a prisoner’s dilemma where everyone cheats whenever possible is a reduction that
does not correctly represent it. Experiments show that people are altruistic, i.e. that they “do good”
even independently of reputational considerations, and that they trust each other, albeit to varying
degrees. These two behavioral regularities are incompatible with standard assumptions of
microeconomics.
The way altruism is typically measured in experiments through variants of the “dictator game”
where participants are given a lump-sum amount of money which they may or may not share with
anonymous other participants, with no possibility of getting the favor returned and one-shot
interaction. That is, the game is purged of any personal feeling and of any reputational or
reciprocity consideration Typically, people do give.
Trust is typically measured in experiments through variants of the “trust game” where participants
in room A are given a lump-sum amount of money, say ten francs, and can give any of it to
anonymous participants in room B, which then receive three times the amount (thirty francs).
7
Room-B participants can then return any fraction of that to their room-A donor. The behaviour of
room-A participants measures trust, while that of room-B participants measures trustworthiness.
Cross-country studies attempting to measure differences in trust have typically relied on surveys
(there are many) rather than experimental evidence. One would hope that survey evidence acurately
reflects how people behave. Unfortunately, evidence on the consistency between individual survey
responses and individual behaviour in experiments is frankly all over the place (see Algan and
Cahuc 2013), except that trust as reported in surveys seems to correlate more with trustworthiness
(second-mover behavior) than to trust itself (first-mover behavior) in the trust game.
Table 2: Individual correlates of trust
Source: Algan Cahuc 2013, Table 2.
8
Keeping in mind that what people say (in surveys) may differ from what they do (in experiments), a
simple regression of self-reported trust across individuals gives results shown in Table 2. Very few
individual characteristics correlate with trust, namely education, income and age (all positively).
The regression’s explanatory power is very low, with an R2 at 2.7%. Country fixed effects make
the R2 go up significantly; indeed, much of the variation in trust comes from cross-country
differences rather than individual ones. This motivates a quest for the determinants and
consequences of variations in trust between countries.
One of the classic studies is that of Knack and Keefer (1997 QJE), who attempted through the usual
strategy to find a link between social capital and growth based onanswers to the following questions
from the World Values Survey (1981 and 1990-91):
“Generally speaking, would you say that most people can be trusted, or that you can't be too
careful in dealing with people? ".
Proportion of Yes in the world: 36%, deviation of 14%, and remarkable fall in the United States,
55-60% in the fifties to 38-39% in 1990. Confidence Index (TRUST) of Knack and Keefer from
average national responses to this question.
Figure 3 : Trust and civic cooperation
The Civic Index (CIVIC) is based on responses to the following question:
"Would you say that the following behaviors can always be justified, never be justified or
something in between:
o
o
o
o
Claiming government benefits which your are not entitled to
Avoiding fare on public transport
Cheating on taxes if you have a chance
Keeping money that you have found
9
o
Failing to report damage you’ve done accidentally to a parked vehicle.”
coded responses of 1 (never justifiable) to 10 (always justified) in the WVS but recoded upside
down in K & K; therefore, higher index means more citizenship.
In a classic study, Fisman and Miguel (2007) were able to relate survey-reported measures of social
capital with the behavior of people through a natural experiment. They looked at the number of
parking violations by UN diplomats in New York City over 1997-2005. UN diplomats enjoy
immunity, which means that parking violations, although recorded in NYPD files, are not enforced.
The 1’700 UN diplomats together accumulated 150’000 unpaid parking tickets resulting in fines
totaling over $18 million. Let ni be the number of unpaid parking violations by the mission of
country i in a year, si the mission’s size in number of diplomats, ci the corruption index of country
i as measured by Transparency International, and R  i  the region to which country i belongs
(Africa, etc.). Fisman and Miguel estimated the following regression:
ln 1  ni     1 ln  si    2ci  3GDPpci   FERi   ui
(10)
Results are shown in
Table 3: The number of parking violations correlates with home-country corruption
Source: Fisman and Miguel (2007).
Thus, the behavior of diplomats in a zero-enforcement environment seems indeed to correlate
closely with levels of corruption at home. The differences are huge, ranging from 246 violations per
10
diplomat per year (almost one a day) for Koweit, 139 for Egypt, 124 for Chad, to 16 for Chile or
Tunisia, 4 for Argentina or Singapore, 1 for Germany, and zero for Switzerland, Greece, Israel or
Norway. Interestingly, enforcement seems to have a role to play, as the 2002 Clinton-Schumer
amendment which made it possible to tow diplomatic cars and cut U.S. aid to the country by the
amount of unpaid fines, led to a substantial reduction in parking violations across the board.
Who Can You Trust?
« If you lose your wallet in Norway or Denmark, chances are good that some kindly person will return it to
you. In Italy, Switzerland or eastern Germany, don't bet on it.
These are conclusions drawn from a frankly unscientific experiment by Reader's Digest, which
"accidentally" dropped some 200 wallets, each containing about $50 worth of cash, plus family snapshots
and contact numbers of the putative owners. In a score of cities, small towns and suburbs across Europe, the
wallets were left in such places as zoos, gas stations, supermarkets, telephone booths and cafes.
In all, 58 percent of the finders returned the wallets, either directly to their owners or through an official
intermediary—a policeman, hotel receptionist or the like. In America, where the experiment was conducted
in twelve large and small cities, 67 percent of the wallets came back.
Nordics appear to be more honest than Mediterraneans; few wallets dropped in Italy came back. No other
geographical or demographic pattern emerged. The rate of return was middling in Britain, France and the
Netherlands. Across Europe, men were as prone as women to return the wallets, the young as likely as the
old, immigrants and the poor as likely as the rich.
Dens of thieves? Tidy Lausanne, Switzerland, and Weimar, home of Goethe and Schiller, turn out to be the
places where you least want to lose your wallet. In Lausanne, one of only two wallets to be returned was
handed in by an Albanian. In the Hague, one of five wallets that were never seen again had been dropped in
front of the International Court of Justice. »
Source : The Economist [check exact reference]
3.1.2 Social capital, income, and growth
Across countries, Figure 4 shows that social capital and income levels are closely correlated. Thus,
a regression of income on trust returns a positive and significant coefficient. However, the
identification of a causal relationship from trust to income based on such an exercise is very very
weak, because there is a huge problem of confounding influences and endogeneity.
Figure 4: Social capital correlates across countries and regions with per capita income
(a) Countries
(b) Regions
11
Source: Algan Cahuc (2013).
Similarly, one can test the correlation between growth and trust. Like in the trade-and-growth
literature, consider a Barro growth equation where Si ,1980 and Pi ,1980 are respectively secondary- and
primary-school enrolment rates in 1980
git  0  1Yi ,1980  2 Si ,1980  3Pi ,1980  4TRUSTi  5CIVICi
6TRUSTi * Yi ,1980  ui
Note that trust and civic behavior tend to correlate with each other across countries :
Figure 5 : Social capital (Trust) and growth
Note: This is a “partial correllogram”, i.e. the influence of control variables is filtered out.
Knack and Keefer (1997) estimated this regression on a cross-section of countries, in column 5 by
instrumenting trust with ethnolinguistic fragmentation:
12
Table 1 : Social capital and Growth : Regression results
Note: standard errors in parentheses
Source: Knack and Keefer (1997)
Again, the problem is that in a cross-section, the identification of a causal mechanism going from
trust to growth (or income, it is the same problem) is very weak. In order to filter out unobserved
heterogeneity between countries, one would need country fixed effects; to filter out reverse
causation and omitted variables affecting both income and trust (both are what we call endogeneity
problems), one would need an instrumental variable.
Tabellini (2010) proposed to address both problems by going from the country to the region level,
which makes it possible to add country fixed effects, and by instrumenting region-level trust by past
levels of institutions and education. Past education was proxied by literacy rates around 1880, and
past institutions by constraints on executive power (checks and balances) between 1600 and 1850
(Figure 6).
Figure 6: First-stage determinants of culture in Tabellini (2010)
(a) Literacy around 1880
(b) institutions in the XVIIth century
13
Source: Tabellini (2010)
Instrumental-variable estimation goes in two stages: In the first one, the explanatory variable likely
to be endogenous (trust) is regressed on its instruments (past literacy and institutions); in the second
stage, the main regression, the variable of interest (income or growth, depending on the case) is
regressed on all explanatory variables and the predicted value of the endogenous one (trust) based
on the first-stage regression. Tabellini’s first-stage regression results suggested that the predictive
power of these two historical variables on trust was substantial (so they were good instruments for
trust), while second-stage result suggested that, the “predetermined” fraction of the variation in trust
had an effect on income.
3.2 Institutions
Acemoglu, Johnson and Robinson (2001) looked at the effect of institutions on levels of income
with a similar instrumentation strategy. Consider the following equation:
ln yi  0  1Ri  Xi α  ui
(11)
where Ri is the variable measuring the quality of institutions (let’s say the legal protection of
property rights) and X i is a vector of control variables. The problem in estimating (11) is the usual
one, namely that the quality of institutions today is endogenous to the level of development.
Their hyper-clever instrumentation strategy is based on the idea that in places where European
settlers could survive, they built European-style institutions that would prove pro-growth, whereas
in places where their mortality was very high—due to diseases, mosquitos, etc.—they set up
institutions that were geared toward “rapid plundering” of the natural resources, framing the
countries for a future “natural-resource curse”.
Formally, their approach goes like this. Le i index countries, and consider estimation on a crosssection of countries, with the main equation relating the log of GDP per capita yi to the quality of
institutions Ri , and an auxiliary equation explain institutional quality on instruments Z i :
Ri  0  1Zi  X iβ  vi
ln y     Rˆ  X α  u
i
0
1
i
i
(12)
i
14
where R̂ is the predicted value of Ri in the first stage regression, and the two equations are
estimated simultaneously. The data come from records British army records, Vatican records on the
mortality of the bishops in Latin America, and other sources.
Figure 7 : Quality of institutions today and settler mortality
Indice de protection de
la propriété privée
Source : Acemoglu, Johnson et Robinson (2001)
Results are shown in Table 4.
Table 4: Per capita income and quality of institutions : OLS
15
Error! Reference source not found. (continued): 2SLS (Instrumental variable)
Source : Acemoglu, Johnson et Robinson (2001)
Colonizer and legal system
An approach (e.g. La Porta et al. 1999) which dates back to the work of Hayek (1960) argued that
legal systems have differential effects on growth. Hayek argued that the French civil law system
found its origins in the Napoleonic code whose aim was to limit the possibilities of interference by
judges in State affairs, being less conducive than the English Common Law system to the
development of democratic checks and balances. This has been tested by an empirical literature
putting to the right of the growth equations dummy variables marking the identity of the colonizer
(GB or France) or the legal system. In general, the dummy variable "colonizer = F ' or 'french legal
system' came out negative.
Table 2 : Per capita GDP, institutional quality (instrumented by the mortality of the settlers) and
legal system
16
Source : Acemoglu, Johnson et Robinson (2001)
Unfortunately, the AJR data were re-examined by Albouy (2008), with the following comment:
“[…] out of 64 countries in their sample, only 28 countries have mortality rates that originate from
within their own borders. The other 36 countries in the sample are assigned rates based on AJR’s
conjectures as to which countries have similar disease environments. These assignments are based
on weak and sometimes inaccurate foundations. Six assignments are based upon AJR’s
misunderstanding of former names of countries in Africa. Another sixteen assignments are based on
a questionable use of bishop mortality data in Latin America from Gutierrez (1986), which are based
on 19 deaths. […] If, in the hope of reducing measurement error, AJR’s 36 conjectured mortality
rates are dropped from the sample, the empirical relationship between expropriation risk and
mortality rates weakens substantially, particularly in the presence of additional covariates. Second,
AJR’s mortality rates never come from actual European settlers, although some settler rates are
available in their sources. Instead, AJR’s rates come primarily from European and American soldiers
in the nineteenth century. In some countries, AJR use rates from soldiers at peace in barracks, while
in others, they use rates from soldiers on campaign. Soldiers on campaign typically have higher
mortality from disease, and AJR use campaign rates more often in countries with greater
expropriation risk and lower GDP. Thus, AJR’s measures of mortality artificially favor their
hypothesis.” (p. 2)
This is a bit of a let-down. More recently, Feyrer and Sacerdote (2009) adopted an original
identification strategy in order to assess the effect of the colonial past on income today, using the
17
prevalence of trade winds to instrument the European colonialism. The idea is the following. We
would like to assess whether the length of the period of colonial rule has a positive or negative
effect on the income today, as the underlying idea is the same that the one of AJR, this means
colonialism was able to facilitate the transplantation of development institutions that had already
been proven successful in Europe.
The source of variation exploited statistically is between countries, which creates a problem of
selection (or endogeneity, here the two are the same thing): Europeans are likely to have settled
earlier and longer in countries with higher development potential (favourable climate, etc.). Under
OLS, we risk picking up the correlation between initial conditions and future development rather
than the causal effect of colonization. Therefore, what is needed is an instrumental variable
correlated with the probability and the length of European colonization but not with
contemporaneous income.
European ships with square sails performed poorly when sailing
windward and sailed mostly downwind. Therefore, they followed trade
winds, which blow from East to West around 15o of latitude North and
South. Islands located along the trade winds were therefore more
likely to be colonized, while the prevalence of the trade winds has no
influence on per capita GDP today. The prevalence of trade winds,
which is affected by many local factors (ocean currents, microclimates etc.) is therefore a good instrument: A good predictor of the
endogenous variable (colonization) but satisfying the exclusion
restriction (no direct causal effect on income today).
Feyrer and Sacerdote use a cross-sectional sample of 81 small
islands. As a whole, a longer period of colonization is associated to a
higher income level (Figure 6 ):
Figure 8 : colonization years and current income
18
Source : Feyrer et Sacerdote (2009)
Are winds a good instrument? As said, they must be well correlated with European colonization
(instrument strenght) but have no effect on the income of today (exclusion restriction). With regard
to the first question, the Figure 9 shows a clear correlation between dominance of the trade winds
and period of European colonization.
Figure 9: Predominance of easterly winds (Trade winds) and European settlement
Prevailing Westerly
winds (jet stream)
Prevailing Easterly
winds (trade winds)
Note: The horizontal scale measure the (normalized) deviation from West winds, not Easterly winds; a negative figure
means prevailing East winds. Circles represent islands in the Atlantic, triangles are islands in the Pacific and squares are
islands in the Indian Ocean.
Source: Feyrer and Sacerdote (2009)
As to the exclusion restriction, since the end of the 19th century the Navy (and therefore trade) is
clearly not depending on winds anymore, which plays in favor of the validity of the instrument. On
19
the other hand, we can see that climates under the trade winds are very nice (see map) and favorable
to tourism. This creates a potential problem, as we might be picking up a causal link between
potential tourism and income instead of a link between European colonization and income today as
we thought we were. We still bump into the usual identification problem.
Figure 10: Trade winds map
Lattitude des alizés nord
Equateur
Lattitude des alizés sud
Be that as it may, regression results are at least consistent with the story, whether with OLS or IV:
Table 5: Regression of the level of current income over the length of the period of colonization
Source : Feyrer et Sacerdote (2009)
20
In addition, late colonization (during or after the age of enlightenment) seems to be associated with
a more pronounced positive effect:
Table 3: Timing and effect of colonization
Source : Feyrer et Sacerdote (2009)
The Spanish and Portuguese colonization also seems to have had less than positive effects, which is
not very surprising given that they were the first colonizers and the institutions that they could
transplant were essentially those of the Inquisition-not necessarily favorable to growth.
21
Table 4: Colonizers and effects of colonization
Source : Feyrer et Sacerdote (2009)
4. Measures and consequences of corruption: a micro approach
In the light of the uncertainties in the identification of the causal relationships in cross sections of
countries, recent literature focused its attention on micro data. Sequeira and Djankov (2010) is a
good example, showing the cost of corruption in the port of Maputo in terms of efficiency. SD
show that corruption at the port of Maputo encourages businesses to choose rather Durban even
when they are closer to Maputo (see Figure 11).
Figure 11 : Location of firms with respect to Maputo and Durban
22
Source : Sequeira Djankov (2010)
The two ports are comparable:
Table 5 : Technical characteristics of Maputo and Durban ports
Source : Sequeira Djankov (2010)
But corruption levels are extremely different:
Figure 12 : Bribes distribution in Maputo and Durban
23
Source : Sequeira Djankov (2010)
The data source on corruption is a detailed IFC survey:
“The IFC hired well-established clearing agents to track all bribe payments to o_cials in a random
sample of 1,300 shipments, between March 2007 and July 2008. Clearing agents recorded detailed
information on the date, time of arrival and clearance of each shipment; on expected storage costs at
the port; on the size of the client _rm and on a wide range of cargo characteristics such as its size,
value and product type. They also noted the primary recipients of bribes, the bribe amounts
requested and the apparent reason for a bribe request, ranging from the need to jump a long queue
of trucks to get into the port, to evading tariffs or missing important clearance documentation. For a
random subset of shipments, the IFC hired local observers who accompanied clearing agents
throughout the clearing process to verify the accuracy of the data. These observers began
shadowing clearing agents several weeks before the tracking study took place in order to become
familiarized with all clearing procedures. To avoid any suspicion, the observers were similar in age
and appearance to any other clerk who normally assists clearing agents in their interactions with
customs.” (p. 16)
Corruption strongly influenced the choice of operators:
Figure 13
Probability of choosing Maputo depending on the ratio of costs to get there (distance)
24
Tariff faibles
Tariff élevés
Choix hypothétique sans
corruption (p = 0.5)
Choix observés
(p = 0.2)
Source : Sequeira Djankov (2010)
SD estimate that the average cost of re-routing the cargo by Durban in order to avoid corruption in
Maputo is equal to eight times the average amount of bribes.
25
References
Acemoglu, Daron, Simon Johnson, and James A. Robinson (2001), “The Colonial Origins of
Comparative Development: An Empirical Investigation”; American Economic Review 91, 1369–
1401.
Algan, Yann, and P. Cahuc (2013), “Trust, Growth and Well-being: New Evidence and Policy
Implications”; in P. Aghion and S. Durlauf, eds., Handbook of economic growth; Amsterdam:
Elsevier.
Barnerjee, Abhijit, and R. Pande (2007), “Ethnic Preferences and Politician Corruption”; Harvard
University, KSG working paper rwp07-031.
Feyrer, James, and B. Sacerdote (2009), « Colonialism and modern income : Islands as natural
experiments » ; Review of Economics and Statistics 91, 245-262.
Fisman, Raymond and E. Miguel (2007), “Corruption, Norms and Legal Enforcement: Evidence
from Diplomatic Parking Tickets”; Journal of Political Economy 115, 1020-1048.
Iyer, Lakshmi (2005), “The Long-Term Impact of Colonial Rule: Evidence from India,” Harvard
Business School working paper no. 05-041.
Knack, Stephen, and Philip Keefer (1997), “Does Social Capital Have an Economic Payoff? A
Cross-Country Investigation”; Quarterly Journal of Economics 112, 1251-1288.
La Porta, Rafael, Florencio Lopez-de-Silanes, Andrei Shleifer, and Robert Vishny (1997), “Legal
Determinants of External Finance,” Journal of Finance 52:3, 1131–1150.
Mauro, Paolo (1995), “Corruption and Growth,” Quarterly Journal of Economics 100:3, 681–712.
Sequeira, Sandra, and Simeon Djankov (2010), An Empirical Study of Corruption in Ports; MPRA
Paper No. 21791.
Shleifer, Andrei & R. Vishny (1993), “Corruption”; Quarterly Journal of Economics 108, 599-617.
Tabellini, Guido (2010), “Culture and Institutions: Economic Development in the Regions of
Europe”; Journal of the European Economic Association 8, 677-716.
26
Download