Economic growth, law, and corruption: evidence from India.
Bhattacharyya, Sambit ; Jha, Raghbendra
INTRODUCTION
Is corruption influenced by economic growth? Are legal institutions
effective in curbing corruption? As corruption and economic growth are
arguably simultaneously determined, one key question is the issue of
causation. Mauro (1995) in his seminal contribution argues that
corruption acts as a disincentive for investments and as a result harms
growth over the long run. He uses the Business International indices on
corruption, red tape, and the efficiency of the judicial system to
measure corruption for the period 1980-1983 in 70 countries. In
contrast, here we compute corruption using a two-step procedure and
Transparency International data. (1) First, we compute an average of the
percentage of respondents answering yes to the questions on direct
experience of bribing, using a middleman, perception that a department
is corrupt, and perception that corruption increased over time for eight
different sectors (banking, land administration, police, education,
water, Public Distribution System (PDS), electricity, and hospitals}.
Second, we average these averages over all the eight sectors to generate
one observation per state and per time period. A higher value of the
corruption measure implies higher corruption.
The advantage of the Transparency International data over any other
dataset is threefold. First, the Transparency International data allow
us to examine the impact of economic growth and law on corruption in
each of the above-mentioned sectors separately. Second, it allows us to
make a distinction between corruption perception and corruption
experience. Third, it also allows us to separate out the effects of
growth on bribing and the use of a middleman. No other datasets would
allow us to undertake this empirical exercise. However, the limitation
is that it only offers a small sample size relative to cross-country
datasets on corruption.
By plotting the data in Figure 1 we indeed observe that economic
growth and corruption are negatively related across 20 Indian states and
over the period 2005 and 2008. However, the causality could run in both
directions. High levels of corruption and weak institutions could reduce
growth (Mauro, 1995). In contrast, one can also argue that economic
growth creates additional resources that allow a country or a state to
fight corruption effectively. Indeed, there is a large literature
documenting causality in both directions. Mauro (1995} argues that
corruption acts as a disincentive for investments and as a consequence
limits growth. Alternatively, rapid modernization of the economy
improves institutional quality and makes it easier for the state to
detect, monitor, and punish corruption (Lipset, 1960). (2) Furthermore,
rapidly growing per capita income also increases the personal
opportunity costs of corruption (Treisman, 2000). To address endogeneity
we use rainfall as an instrument for economic growth.
[FIGURE 1 OMITTED]
Theory suggests that the causal effects of GDP growth, per capita
GDP growth, and GDP levels on corruption could be different. Growth
could impact corruption by modernizing institutions and reshaping
opportunity costs and personal incentives (Lipset, 1960; Treisman,
2000). The effect could also be non-dynamic with levels of GDP rather
than growth accounting for a change in corruption (Hall and Jones,
1999). In this paper we focus on the impact of GDP growth on corruption.
The second key question is how effective legal institutions are in
curbing corruption. Our panel dataset on corruption covering 20 Indian
states and the periods 2005 and 2008 offers an opportunity to
empirically test this effect. The Right to Information Act (RTI) in
India came into effect on October 12, 2005, which is after the
conclusion of our 2005 corruption survey in January. The act ensures
citizens' secure access to information under the control of public
authorities. In addition, the accompanying Citizens' Charter makes
it legally binding for all government agencies to publish a declaration
incorporating their mission and commitment towards the people of India.
An obvious question is how RTI in India is linked to corruption. It
is quite common in India that citizens visiting some government offices
for certain legitimate services would either be not listened to or would
be given a vague response. On many occasions the officials would raise
irrelevant objections to simple applications for water connections. The
government officials would resort to such tactics because they are
either seeking a bribe or hinting that the citizen should pay a corrupt
middleman to get the job done. Without these payments, the application
would be delayed under flimsy verbal objections from the official. (3)
RTI empowers citizens to write a letter to the Public Information
Officer (PIO) of the relevant government department seeking answers to
questions such as why the application for a water connection is delayed.
Under the RTI Act the citizen is also entitled to ask for a daily
progress report on the water connection application, the names and
designations of officials with whom the application is lying during the
time under scrutiny, proof of receipt and dispatch of the application
from the office of each of these officials, what is the maximum time
limit according to the departmental rules for a water connection
application to be processed, if these rules are violated then which
official is responsible, an official assessment report on the possible
violation of the published model code of conduct of the department, and,
if a violation has occurred, then what action would be taken against the
guilty. In the event of a citizen writing such letter to the PIO, it
becomes extremely difficult for the department to provide answers to
such probing questions within the RTI time limit without taking action
against the responsible officials. The department would also try to
avoid situations where an inadequate or delayed written response leads
to the violation of the RTI Act, which is punishable by law. Therefore,
the general experience is that the job is done without any further delay
the moment a citizen files such a letter with the PIO. The officials are
also aware of the power of RTI, and, therefore, they are much more
cautious and less inclined to seek bribes.
[FIGURE 2 OMITTED]
By design, our dataset offers us the opportunity to test the effect
of the law on corruption using two time series data points in our
dataset, one before and the other after the law came into effect.
Indeed, in Figure 2 we do notice that corruption declined significantly
in 2008. However, this may also be due to some uncontrolled factors. The
only way to find out is by controlling for additional factors that may
be influencing corruption.
In this paper, using a panel dataset covering 20 Indian states and
the periods 2005 and 2008 we estimate the causal effects of economic
growth (4) and law on corruption. Since different states have
experienced different growth patterns and different levels of
corruption, India represents an ideal testing ground to examine the link
between economic growth and corruption. To tackle endogeneity concerns
we use rainfall as an instrument for economic growth. We notice that
rainfall is a positive predictor of growth. This is in line with the
view that rainfall contributes positively to economic growth. Rainfall
perhaps also satisfies the exclusion restriction of an instrumental
variable (IV) as it shows very low correlation with factors such as
inequality and poverty through which, potentially, it could also affect
corruption. (5) To capture the effect of law on corruption, we use a
time dummy and control for other nationwide changes that may be
affecting corruption. This is a valid strategy as the RTI came into
effect after the completion of Transparency International's 2005
corruption survey. Our results indicate that economic growth reduces
overall corruption experience as well as corruption in banking, land
administration, education, electricity, and hospitals. It also reduces
overall bribes and bribes in the above-mentioned sectors. However,
growth has little impact on corruption perception. This is supportive of
the view that corruption perceptions in developing economies are often
biased upwards. In contrast, the RTI negatively impacts both corruption
experience and corruption perception. Our basic result holds after
controlling for state fixed effects and various additional covariates
(eg literacy, Gini coefficient, poverty head count ratio, mining share
of state GDP, primary sector share of state GDP, state government
expenditure as a share of state GDP, newspaper circulation, and total
number of telephone exchanges). It is also robust to the use of flood
affected area, flood affected population, flood affected crop area, and
total number of flood affected households as alternative instruments and
outlier sensitivity tests.
We make the following four original contributions in this paper.
First, by using a panel dataset on corruption across Indian states and a
Limited Information Maximum Likelihood (LIML) IV estimation method we
are able to estimate the causal effect of economic growth on corruption.
Controlling for state fixed effects and additional covariates also
allows us to tackle potential omitted variable bias. To the best of our
knowledge, ours is the first panel data study of economic growth and
corruption covering Indian states. Second, using a time dummy and
exploiting the construction of our dataset we are able to estimate the
corruption curbing effect of the RTI law in India. This is an important
finding that has policy implications not just for India but also for
other comparable developing economies suffering from endemic corruption.
To the best of our knowledge, no other empirical study on corruption in
India provides evidence of this nature. Third, using sector-wise
disaggregated data we are able to estimate the causal effects of
economic growth and law on corruption in banking, land administration,
police, education, water supply, PDS, electricity, and hospitals. This
in our view is an entirely new finding. Fourth, we are able to
separately estimate the effects of economic growth and law on corruption
experience and corruption perception and we do find that they are
different. We notice that economic growth has very little influence on
corruption perception. Our finding adds to a small but growing body of
evidence on the difference between corruption perception and corruption
experience (see Olken, 2009).
Our economic growth and corruption result is related to a large
literature on corruption and development that follows from the seminal
contribution by Mauro (1995). (6) However, note that our focus here is
to estimate the causal effect of economic growth on corruption and not
the other way around. Our law and corruption result is also related to a
growing literature on democratization and corruption as it emphasizes
the role of accountability. For example, Treisman (2000) shows that a
long exposure to democracy reduces corruption. Bhattacharyya and Hodler
(2010), using a game theoretic model and cross-national panel data,
estimate a reduced-form econometric model and show that resource rent is
bad for corruption although the effect is moderated by strong democratic
institutions. In contrast, Fan et al. (2009) show that decentralized
government may not increase accountability and reduce corruption if the
government structures are complex. In a similar vein, Olken (2007) also
shows that top-down government audit works better than grassroots
monitoring in Indonesia's village roads project. Therefore, our
results contribute to a policy debate that is not only important for
India but also for other comparable developing economies. The estimates
are not directly comparable as there are significant differences in
scale (microeconomic or macroeconomic), scope (national or
international), and nature (theoretical, empirical, or experimental) of
these studies.
Finally, our results are also related to a large literature on
institutions and economic development (see Knack and Keefer, 1995; Hall
and Jones, 1999; Acemoglu et at., 2001; Rodrik et al., 2004;
Bhattacharyya, 2009). The major finding of this literature is that
economic institutions (such as, property rights, contracts, regulation,
and corruption) are one of the major drivers of long-run economic
development. Besley and Burgess (2000, 2004) provide evidence that land
property rights and labour market institutions have significant effects
on economic performance across states in India. In this paper we
estimate the magnitude of the relationship when causality runs in the
opposite direction from economic growth to institutions.
The remainder of the paper is structured as follows: the next
section discusses empirical strategy and the data. The section after
that presents the empirical evidence and various robustness tests. The
last section concludes.
EMPIRICAL STRATEGY AND DATA
We use a panel dataset covering 20 Indian states and the periods
2005 and 2008. Our basic specification uses corruption data for the
years 2005 and 2008. Economic growth for the periods 2005 and 2008 are
growth in GDP (7) over the periods 2004-2005 and 2007-2008,
respectively. To estimate the causal effects of economic growth and law
on corruption we use the following model:
[c.sub.it] = [[alpha].sub.i] + [delta][[beta].sub.t] +
[[gamma].sub.1][[??].sub.it] + [X'.sub.it][GAMMA] +
[[epsilon].sub.it] (1)
where [c.sub.it] is a measure of corruption in state i at year t,
[[alpha].sub.i], is a state dummy variable covering 20 Indian states to
control for state fixed effects, [[beta].sub.t] is a dummy variable that
takes the value 1 for the year 2008 to estimate the impact of the
introduction of the RTI Act (on October 12, 2005), [[??].sub.it] is
economic growth in state i over the period t-1 to t, and [X.sub.it] is a
vector of other control variables. A high value of [c.sub.it] implies a
high level of corruption. The motivation behind including state fixed
effects is to control for time invariant state-specific fixed factors
such as language, culture, and ethnic fractionalization.
The main variables of interest are [[??].sub.it] and the time dummy
variable [[beta].sub.t]. Therefore [[gamma].sub.1] and [delta] are our
focus parameters. In theory, we would expect [[gamma].sub.1] to be
significantly negative as faster growing states are able to use
additional resources to curb corruption. The coefficient estimate
[delta] is expected to capture the effect of the RTI Act. This is
equivalent to a before and after estimation strategy in panel data
econometrics. Ideally one would like to compare the effect of RTI on
corruption before and afterwards in the areas affected by the law, and
then compare this to the effects before and afterwards in the areas not
affected by the law. Unfortunately this is not feasible here as the RTI
law came into effect nationally. In other words, there is no comparison
group here since the law was introduced at the same time in all
locations. Nevertheless, the strategy implemented here is credible at
the macro level.
To illustrate the before and after strategy, let [c.sub.1it] be the
corruption outcome in state i at time t when the RTI Act is in effect.
Similarly, let [c.sub.2it-1] be the corruption outcome in state i at
time t-1 when the RTI Act is not in effect. Note that these are
potential outcomes, and in practice we only get to observe one or the
other. One can express the above as:
E[[c.sub.1it]|i, t = 1, [[??].sub.it] = [bar.y], [X'.sub.it] =
[bar.X]] = [[alpha].sub.i] + [delta] and E[[c.sub.2it-1]|i, t - 1 = 0,
[[??].sub.it] = [bar.y], [X'.sub.it] = [bar.X]] = [[alpha].sub.i]
(2)
Given that E([[epsilon].sub.it]|i, t) = 0. The population before
and after estimates yields the causal effect of the RTI Act [delta] as
follows:
E[[c.sub.1it]|i, t = 1, [[??].sub.it] = [bar.y], [X'.sub.it] =
[bar.X]] - E[[c.sub.2it-1]|i, t - 1 = 0, [[??].sub.it] = [bar.y],
[X'.sub.it] - [bar.X]] = [delta] (3)
This can be estimated by using the sample analog of the population
means. If the RTI law is effective in curbing corruption then we would
expect [delta] to be negative.
Data on corruption are from the Transparency International's
India Corruption Study 2005 and 2008. The study was jointly conducted by
Transparency International India and the Centre for Media Studies both
located in New Delhi. The survey for the 2005 report was conducted
between December 2004 and January 2005 and the survey for the 2008
report was conducted between November 2007 and January 2008. The survey
asks respondents whether (i) they have direct experience of bribing,
(ii) they have used a middleman, (iii) they perceive a department to be
corrupt, and (iv) they perceive corruption has increased over time. (8)
These questions are asked to on average 750 respondents from each of the
20 states. Respondents are selected using a random sampling technique
covering both rural and urban areas. In aggregate the 2005 survey
interviews 14,405 respondents spread over 151 cities and 306 villages of
the 20 states. In contrast, the 2008 survey covers 22,728 randomly
selected Below Poverty Line (BPL) respondents across the country. One
could argue that this brings in issues of measurement error that will
bias our estimates downwards. This is formally known as the attenuation
bias, which is driven by measurement error. So what we estimate in the
presence of measurement error is in fact less in magnitude than the true
effect. Furthermore, if the measurement error follows all classical
assumptions (in other words, random) then our estimates will remain
unaffected. Nevertheless, measurement error problem can be mitigated
using the IV strategy, and we remedy this problem using rainfall as an
instrument. Rainfall is geography-based and therefore exogenous or
uncorrelated to the measurement error. Hence rainfall can serve as a
valid instrument to remedy measurement-error-driven attenuation bias.
The Appendix section 'Measurement error and instrumental variable
estimation' shows algebraically how the IV strategy could
potentially remedy the measurement error problem.
Our aggregate measure of corruption [c.sub.it] is computed in two
steps. First, an average is computed of the percentage of respondents
answering yes to the questions that they have direct experience of
bribing, using a middleman, perception that a department is corrupt, and
perception that corruption increased over time for eight different
sectors: banking, land administration, police, education, water, PDS,
electricity, and hospitals. (9) Second, these averages are also averaged
over all the eight sectors to generate one observation per state and per
time period. Ideally, one should weigh the sectors with their respective
usages. But in the absence of reliable usage statistics at the state
level, we compute averages with equal weights. This may not be a cause
for concern as services from all of these sectors are widely used by
citizens. Note that the sector-level disaggregated data are utilized in
Table 4, and Table 5 treats corruption perception and corruption
experience separately. Corruption experience measure is the average of
the questions on 'direct experience of bribing' and
'using a middleman'. Corruption perception measure is the
average of the questions on 'perception that a department is
corrupt' and 'perception that corruption increased over
time'.
The state of Bihar turns out to be the most corrupt in our sample
with 59% of respondents reporting corruption in 2005. In contrast
Himachal Pradesh is the least corrupt with only 17% of the respondents
reporting corruption in 2008. It appears that police, land
administration, and PDS are among the most corrupt sectors in our
dataset. Kerala and Himachal Pradesh come out to be the least corrupt
states in most of the cases. In contrast Bihar, Jammu and Kashmir,
Madhya Pradesh, and Rajasthan register high levels of corruption.
Economic growth [[??].sub.it] is defined as the growth in real GDP
of the states over the periods 2004-2005 and 2007-2008, respectively. We
use real GDP as our preferred measure instead of real GDP per capita to
compute growth rates because aggregate growth of the economy
(modernization effect) is more likely to have an impact on corruption at
the macro level than per capita growth. Nevertheless, we also use per
capita GDP growth to estimate the model and our results are robust. Real
GDP growth data are from the Planning Commission. Our growth variable
varies between -4.2 % in Bihar in 2005 and almost 17% in Chhattisgarh in
2005.
As economic growth here is arguably endogenous, one key question is
the issue of reverse causation. Corruption, as argued by many including
Mauro (1995), may dampen growth through the investment channel. In that
case, a simple OLS estimate of our model would be biased. In order to
estimate the causal effect of economic growth on corruption we need to
implement the IV estimation strategy. In particular, we need to identify
an exogenous variable that is correlated with economic growth but
uncorrelated with the error term [[epsilon].sub.it] in the model, that
is, this exogenous variable would affect corruption exclusively through
the economic growth channel. This is commonly known as the exclusion
restriction. Indeed, finding such a variable is a challenge in itself.
But we are fortunate to have log rainfall (ln [RAIN.sub.it_1]) from the
Compendium of Environmental Statistics published by the Central
Statistical Organization. We notice that In [RAIN.sub.it-1] is
positively related to economic growth and the relationship is
statistically significant. This is in line with the view that rainfall
positively contributes to economic growth. Furthermore, In
[RAIN.sub.it-1] is geography-based and therefore is exogenous. However,
rainfall may affect corruption through channels other than economic
growth. Poverty and inequality are such examples. Rainfall may lead to
reduction in poverty, which may in turn lead to a reduction in
corruption. Better rainfall and better agriculture growth may also
increase inequality leading to an increase in corruption. In such a
situation the rainfall instrument may not satisfy the exclusion
restriction. To eliminate such possibility, we check the correlation
between the rainfall instrument and poverty and inequality. It turns out
to be 0.17 and 0.38, respectively, which suggests it is unlikely that
rainfall would affect corruption through the poverty and inequality
channels. Therefore, it is safe to conclude that In [RAIN.sub.it-1] can
serve as a valid instrument. However, if the relationship between In
[RAIN.sub.it-1] and [[??].sub.it] is not strong enough then it may lead
to the weak instruments problem. Staiger and Stock (1997) and Stock and
Yogo (2005) show that if the instruments in a regression are only weakly
correlated with the suspected endogenous variables then the estimates
are likely to be biased. Instruments are considered to be weak if the
first stage F-statistic is less than Stock-Yogo critical value. The LIML
Fuller version of the IV method is robust to weak instruments. We
implement the LIML method to estimate our model. Moreover, we operate
with a relatively small sample of 40 observations and the LIML estimates
are robust to small samples. Therefore, the risk of a significantly
large bias due to weak instruments is minor.
Finally, another potential concern is about the power of the
diagnostic tests with limited degrees of freedom. LIML estimates adopted
here are best suited for this purpose as they have robust and powerful
small sample properties. Nevertheless, we also perform the following two
tests to be certain about the validity of our conclusions. First, we
adopt Hendry et al.'s (2004) least square dummy variables approach
and our results are robust. This method can be implemented using the
following two steps. The first step is to estimate the model using LIML
and identify all the statistically insignificant state dummy variables.
Then the second step is to re-estimate the model using LIML but without
the statistically insignificant state dummies. The advantage is that
this significantly improves the power of the tests. Second, we estimate
the model without any state dummies and our results are robust. These
results are reported in columns 9 and 10 of Table 7.
The time dummy is used to capture the effect of the RTI Act. One
can certainly dispute whether our time dummy is solely picking up the
effect of RTI and Citizens' Charter. It is possible that other
nationwide changes introduced around this time are also affecting
corruption. In that case the estimate on the time dummy is also picking
up the effects of factors other than the RTI. Even though plausible, it
is hard to identify significant national policy changes during this time
other than the RTI that may affect corruption. Nevertheless, to tackle
this issue we also control for literacy, Gini coefficient, poverty head
count ratio, mining share of GDP, primary sector share of GDP, state
government expenditure, newspaper circulation, and total number of
telephone exchanges as additional control variables. Therefore it is
perhaps safe to say that [delta] is indeed capturing the effects of RTI.
Detailed definitions and sources of all variables are available in
Appendix section 'Data description'. Table 1 reports
descriptive statistics of the major variables used in the study. The
Appendix section 'Sample and state codes' provides a list of
20 states covered in the study and presents a map of Indian states.
EMPIRICAL EVIDENCE
Table 2 reports Kolmogorov-Smirnov test results for the equality of
distributions of corruption over the time periods 2005 and 2008. The
test shows that the distribution of corruption across states has changed
over the two time periods. This may be driven by the variation in
economic growth across states. In Table 3 we try to ascertain this by
estimating equation 1 using OLS and LIML Fuller IV methods. Column 1
reports the OLS estimates and column 3 presents estimates of the model
using In [RAIN.sub.it-1] as an instrument for economic growth. Our
suspicion that economic growth can be endogenous is supported by the
endogeneity test reported at the bottom of column 3. We notice that
economic growth has a negative impact on corruption. Ceteris paribus,
one sample standard deviation (4.1% points) increase in economic growth
in an average state would reduce corruption by 1.8% points. In other
words, our model predicts that an increase in the growth rate of Bihar
from -4.2% in 2005 to 16% in 2008 would reduce corruption from 59% in
2005 to 50.3% in 2008. According to our dataset, Bihar's actual
corruption in 2008 is 29 %. Therefore, the estimated coefficient on
economic growth explains 29 % of the actual decline in corruption in
Bihar over the period 2005-2008.
The coefficient on the year 2008 dummy captures the effect of RTI.
Our estimates suggest that RTI has a negative impact on corruption and
the effect is statistically significant. In particular, ceteris paribus,
the RTI Act reduces corruption in an average state by 18.5 % points. To
put this into perspective, the RTI Act explains approximately 62 % of
the actual decline in corruption in Bihar over the period 2005-2008.
(10) This is indeed a large effect.
Note that IV coefficient estimates are typically larger than the
OLS estimates. This is not surprising given that IV estimates are
correcting for the measurement-error-induced attenuation bias in OLS.
In column 4 we use per capita GDP growth instead of aggregate GDP
growth and our result remains unaffected. The evidence here supports the
hypothesis that rapidly growing per capita income also increases the
personal opportunity costs of corruption. Note that we also estimate the
model using 5-year average growth rates instead of economic growth over
the periods 2004-2005 and 2007-2008. Our results are robust to this
test. Results are not reported here but are available upon request.
Column 2 reports the OLS estimate of this model.
How good is our In [RAIN.sub.it-1] instrument? Panel B in Table 3
shows that it is positively correlated with economic growth. Therefore
it can serve as an instrument provided it satisfies the exclusion
restriction. In other words, rainfall affects corruption exclusively
through the economic growth channel. However, rainfall may affect
corruption through channels other than economic growth. Poverty and
inequality are such candidates. Rainfall may lead to reduction in
poverty, which may in turn lead to a reduction in corruption. Better
rainfall and better agriculture growth may also increase inequality
leading to an increase in corruption. In such situation, the exclusion
restriction would be violated.
In Table 4 we ask whether the effect of economic growth and law on
corruption is uniform across all sectors of the economy. In particular
we look at corruption in banking, land administration, police,
education, water supply, PDS, electricity, and hospitals. Indeed there
are more sectors in an economy that may have chronic corruption problem,
and we admit that our list is far from being comprehensive. However, it
should be noted that our study is the first attempt to look at
corruption at a disaggregated level in India using panel data and we are
constrained by data availability. The results indicate that the RTI Act
had an impact on all sectors examined in this study. However, the
magnitude of the predicted decline varies from a 20.4% points in
policing to 6.2% points in the PDS. In contrast, the effect of economic
growth is far from being uniform. Banking, land administration,
education, electricity, and hospitals register a statistically
significant negative effect of economic growth on corruption. However,
the effect is insignificant in case of policing, water supply, and PDS.
In Table 5 we check whether there is a difference between actual
corruption experience and corruption perception. Indeed, we find that
the effect of economic growth on corruption is not uniform across actual
experience and perception. Panel A reports estimates with actual
corruption experience. Note that corruption experience here is the
average of answers to the questions on 'direct experience of
bribing' and 'using influence of a middleman'. In
addition to affecting overall corruption experience, economic growth
appears to reduce corruption experiences in banking, land
administration, education, electricity, and hospitals. The effects on
police, water supply, and the PDS is statistically insignificant. The
observed pattern is very similar to Table 4. This suggests that our
corruption results reported in Tables 3 and 4 are driven by actual
corruption experiences. Panel B reports estimates with corruption
perception. Note that corruption perception here is the average of
answers to the questions on 'perception that a department is
corrupt' and 'perception that corruption has increased'.
We notice that economic growth has little effect on corruption
perception and, in case of policing, it appears to have increased
corruption perception, (11) This is in line with the view that perpetual
pessimism with regards to government services tends to shape corruption
perception in developing economies, and any impact that economic growth
may have on actual corruption is often overlooked. Our result is broadly
in line with the findings of Olken (2009} who also reports differences
in corruption perception and corruption experience in Indonesia, another
developing economy.
The effect of RTI on corruption experience and corruption
perception is somewhat uniform. However, the magnitude of the effect
varies across sectors. We notice that the effect of RTI on corruption
experience is greater than its effect on corruption perception in case
of overall corruption, land administration, and PDS. In contrast, the
reverse is observed in case of banking, police, education, water supply,
electricity, and hospitals.
In Table 6 we dissect corruption experience even further and
examine the effect of growth on bribes and the usage of middlemen
separately. The results are similar to Table 5, Panel A. In addition to
affecting overall bribes, economic growth appears to reduce bribes in
banking, land administration, education, electricity, and hospitals.
However, the effects on police, water supply, and PDS are statistically
insignificant. The time dummy remains significant throughout,
highlighting the importance of RTI. The results are similar for
middlemen usage.
In Table 7 we add additional covariates into our specification to
address the issue of omitted variables. In column 1 we add literacy as
an additional control variable. The rationale is that literate citizens
are relatively more empowered to fight corruption. Our result survives.
Poverty and inequality may also increase corruption. To check whether
this has any effect we add Gini coefficient and poverty head count ratio
as additional controls in columns 2 and 3. Our result remains
unaffected. Natural resources in general and resource rent in particular
may also increase corruption (see Ades and Di Tella, 1999; Treisman,
2000; Isham et al., 2005; Bhattacharyya and Hodler, 2010). To check we
add mining share of GDP and primary sector share of GDP in columns 4 and
5 and our results are robust. High levels of government expenditure may
increase corruption as corrupt officials now have access to more
resources to usurp. It can also work in the opposite direction with the
government now able to engage more resources into auditing. Indeed we do
notice evidence in support of the latter in column 6 with state
government expenditure having a significant negative impact on
corruption. This is in line with Olken (2007) who shows that government
audit reduces corruption in Indonesia. Nevertheless, more importantly
our economic growth and law results remain unaffected. In column 7 we
test whether controlling for the effect of media would alter our result.
Media and an active civil society may reduce corruption. We try to
capture this effect using newspaper circulation. Our main result
survives. Column 8 tackles the view that telecommunication revolution in
India may have triggered this decline in corruption by eliminating the
middleman and reducing discretionary power of corrupt officials. To
capture this effect we use number of telephone exchanges as a control
variable, and our results survive.
Note that we perform two further tests. First, we test whether our
results are driven by influential observations. We identify influential
observations using Cook's distance, DFITS, and Welsch distance
formula and eliminate them from the sample. Our result remains
unaffected. Second, we estimate the model using flood affected area,
flood affected population, flood affected crop area, and total number of
flood affected households as alternative instruments. Our results
survive this test. Furthermore, note that the results remain unaffected
if these instruments are used in conjunction with rainfall and the
Sargan tests are satisfied. These results are not reported but are
available upon request.
Overall these empirical findings support our prediction that both
economic growth and RTI have negative impacts on corruption. However,
the effect of the RTI Act is more uniform than the effect of economic
growth.
CONCLUDING REMARKS
We study the causal impact of economic growth and law on
corruption. Using a panel dataset covering 20 Indian states and the
years 2005 and 2008 we are able to estimate the causal effects of
economic growth and law on corruption. To tackle endogeneity concerns we
use rainfall as an instrument for economic growth. Rainfall is a
positive predictor of growth which is in line with the view that
rainfall contributes positively to economic growth. It also affects
corruption through the economic growth channel reasonably exclusively.
To capture the effect of law on corruption we use a time dummy and
control for other nationwide changes, which may be affecting corruption.
Our results indicate that economic growth reduces overall corruption as
well as corruption in banking, land administration, education,
electricity, and hospitals. However, growth has little impact on
corruption perception. In contrast the RTI negatively impacts both
corruption experience and corruption perception. Our basic result holds
after controlling for state fixed effects and various additional
covariates, literacy, Gini coefficient, poverty head count ratio, mining
share of state GDP, primary sector share of state GDP, state government
expenditure as a share of state GDP, newspaper circulation, and number
of telephone exchanges. It is also robust to the use of alternative
instruments and outlier sensitivity tests.
Our results have important policy implications not just for India
but also for other comparable developing economies. Our findings imply
that economic forces have an important role in reducing corruption.
Therefore, macro policies to promote economic growth not only improve
overall living standards, but they also enhance the quality of public
goods by reducing corruption. This perhaps works through the following
channels. First, it provides the government with additional resources to
fight corruption. Second, it also reduces the incentives for corruption
at the micro level by raising the opportunity cost. More micro level
research is certainly called for to find out whether the data supports
these conjectures.
Legislation such as the RTI Act in India is also important in
curbing corruption. On the one hand it empowers citizens' and
breaks the information monopoly of public officials. Therefore, it
prevents corrupt public officials from misusing information to advance
their own interest. On the other hand, it provides the government with
more power and public support for conducting top down audit of corrupt
departments. There is evidence that the latter works effectively in a
developing economy environment (Olken, 2007).
Finally, more caution is required in the measurement of corruption.
Our results indicate that there is a fair bit of difference between
actual corruption experience and corruption perception in developing
economies. Therefore over-reliance on one or the other may be
counterproductive. We do not stand alone on this, as other studies also
indicate that perception and actual corruption tends to vary
significantly (Olken, 2009). Measuring corruption appropriately in our
view is crucial in furthering our understanding of corruption.
Acknowledgements
We gratefully acknowledge comments by and discussions with Paul
Burke, Ranjan Ray, Takashi Kurosaki, Peter Warr, and Conference and
seminar participants at the Australian National University, University
of Oxford, and ISI Delhi. We thank the editor, Josef Brada, and an
anonymous referee for their comments. We also thank Rodrigo Taborda for
excellent research assistance. All remaining errors are our own.
APPENDIX
Data description
Corruption [[c.sub.it]]: Corruption is computed using a two-step
procedure. First, an average is computed of the percentage of
respondents answering yes to the questions that they have direct
experience of bribing, using a middleman, perception that a department
is corrupt, and perception that corruption increased over time for eight
different sectors: banking, land administration, police, education,
water, Public Distribution System (PDS), electricity, and hospitals.
Second, these averages are also averaged over all the eight sectors to
generate one observation per state and per time period. Higher value of
the corruption measure implies higher corruption.
Source: India Corruption Study 2005 and 2008, Transparency
International.
Corruption in Banks [[c.sup.BANKS.sub.it]]: Corruption computed in
the same fashion as [c.sub.it] but only for the banking sector.
Source: India Corruption Study 2005 and 2008, Transparency
International.
Corruption in Land Administration [[c.sup.LAND.sub.it]]: Corruption
computed in the same fashion as [c.sub.it] but only for the land
administration sector.
Source: India Corruption Study 2005 and 2008, Transparency
International.
Corruption in Police [[c.sup.POLICE.sub.it]]: Corruption computed
in the same fashion as [c.sub.it] but only for police.
Source: India Corruption Study 2005 and 2008, Transparency
International.
Corruption in Education [c.sup.EDUC.sub.it]: Corruption computed in
the same fashion as [c.sub.it] but only for education sector.
Source: India Corruption Study 2005 and 2008, Transparency
International.
Corruption in Water [[c.sup.WATER.sub.it]: Corruption computed in
the same fashion as [c.sub.it] but only for the water supply sector.
Source: India Corruption Study 2005 and 2008, Transparency
International.
Corruption in PDS [[c.sup.PDS.sub.it]]: Corruption computed in the
same fashion as [c.sub.it] but only for the public distribution system.
Source: India Corruption Study 2005 and 2008, Transparency
International.
Corruption in Electricity [[c.sup.ELEC.sub.it]]: Corruption
computed in the same fashion as [c.sub.it] but only for the electricity
sector.
Source: India Corruption Study 2005 and 2008, Transparency
International.
Corruption in Hospitals [[c.sup.HOSP.sub.it]]: Corruption computed
in the same fashion as [c.sub.it] but only for hospitals.
Source: India Corruption Study 2005 and 2008, Transparency
International.
Corruption Experience Measures: Corruption experience measures are
the average of answers to the questions on 'direct experience of
bribing' and 'using influence of a middleman'.
Source: India Corruption Study 2005 and 2008, Transparency
International.
Corruption Perception Measures: Corruption perception measures are
the average of answers to the questions on 'perception that a
department is corrupt' and 'perception that corruption has
increased'.
Source: India Corruption Study 2005 and 2008, Transparency
International.
Economic Growth [[[??].sub.it]]: Real growth rate in state GDP
measured in 2009 constant prices.
Source: Planning Commission, Government of India.
Log Rainfall [In [RAIN.sub.it-1]]: Log of rainfall across states
measured in millimeters.
Source: Compendium of Environmental Statistics, Central Statistical
Organisation, Ministry of Statistics and Programme Implementation.
Flood Area: Total area affected by flood in 1994 and 1996 measured
in millions of hectares.
Source: Central Water Commission, Government of India.
Flood Population: Total population affected by flood in 1994 and
1996 measured in millions.
Source: Central Water Commission, Government of India.
Flood Crop Area: Total crop area affected by flood in 1994 and 1996
measured in millions of hectares.
Source: Central Water Commission, Government of India.
Flood Household: Total number of households affected by flood in
1994 and 1996 measured in millions of hectares.
Source: Central Water Commission, Government of India.
Literacy: Literacy rate for 2002 and 2005.
Source: Selected Socioeconomic Statistics India 2006, Central
Statistical Organization, Table 3.3.
Gini Coefficient: Gini coefficient urban for the periods 1999-2000
and 2004-2005.
Source: Planning Commission.
Poverty Head Count Ratio: Percentage of population below poverty
line (rural and urban combined).
Source: Planning Commission.
Mining Share of GDP: Mining sector share of state GDP.
Source: Handbook of Statistics on the Indian Economy, Reserve Bank
of India.
Primary Sector Share of GDP: Primary sector share of state GDP.
Source: Handbook of Statistics on the Indian Economy, Reserve Bank
of India.
State Government Expenditure: State government expenditure as a
proportion of state GDP.
Source: Indian Public Finance Statistics, Ministry of Finance.
Newspaper Circulation: Number of registered newspapers in
circulation.
Source: Registrar of Newspapers, Government of India.
Telephone Exchange: Number of telephone exchanges.
Source: Ministry of Information and Broadcasting, Government of
India.
Sample and state codes
Andhra Pradesh (AP), Assam (AS), Bihar (BH), Chhattisgarh (CG),
Delhi (DL), Gujarat (GJ), Haryana (HR), Himachal Pradesh (HP), Jammu and
Kashmir (JK), Jharkhand (JH), Karnataka (KT), Kerala (KL), Madhya
Pradesh (MP), Maharashtra (MH), Orissa COS), Punjab (PJ), Rajasthan
(RJ), Tamil Nadu (TN), Uttar Pradesh (UP), West Bengal (WB).
Measurement error and instrumental variable estimation
Assume that the true relationship between corruption and growth is
[c.sub.it] = [[alpha].sub.i] + [alpha][[beta].sub.t] +
[[gamma].sub.1][[??].sub.it] + [[epsilon].sub.it]. However the
corruption variable has measurement error so that [[??].sub.it] =
[c.sub.it] + [[theta].sub.it] (where [[theta].sub.it], is the time
varying measurement error). Because of measurement error we only observe
[[??].sub.it] and not true corruption [c.sub.it]. So we estimate the
model [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] While
estimating this model using fixed effects we would difference the data
and get [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] The
parameter estimate of interest would be, [MATHEMATICAL EXPRESSION NOT
REPRODUCIBLE IN ASCII] and p lim [[??].sub.1] = [[gamma].sub.1] +
(bias). Therefore, estimating the model using OLS would yield biased
estimates. If we use rainfall instrument [Z.sub.it] that is correlated
with [DELTA][[??].sub.it] but orthogonal (or uncorrelated) to
[DELTA]][[theta].sub.it], [DELTA][[beta].sub.t] and
[delta][[epsilon].sub.it] then cov([Z.sub.it], [DELTA][[theta].sub.it])
= cov([Z.sub.it], [DELTA][[epsilon].sub.it]) = 0. Then we would get,
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] and [MATHEMATICAL
EXPRESSION NOT REPRODUCIBLE IN ASCII] since cov([Z.sub.it],
[DELTA][[beta].sub.t] = cov([Z.sub.it], [DELTA][[epsilon].sub.it]) = 0.
Therefore, the IV strategy could potentially remedy the measurement
error problem with 2008 corruption data.
REFERENCES
Acemoglu, D, Johnson, S and Robinson, J. 2001: The colonial origins
of comparative development: An empirical investigation. American
Economic Review 91 (5): 1369-1401.
Ades, A and Di Tella, R. 1999: Rents, competition, and corruption.
American Economic Review 89(4): 982-993.
Bardhan, P. 1997: Corruption and development: A review of issues.
Journal of Economic Literature 35(3): 1320-1346.
Besley, T and Burgess, R. 2000: Land reform, poverty reduction and
growth: Evidence from India. Quarterly Journal of Economics 115(2):
389-430.
Besley, T and Burgess, R. 2004: Can labor regulation hinder
economic performance? Evidence from India. Quarterly Journal of
Economics 119 (1): 91-134.
Bhattacharyya, S. 2009: Unbundled institutions, human capital and
growth. Journal of Comparative Economics 37: 106-120.
Bhattacharyya, S and Hodler, R. 2010: Natural resources, democracy
and corruption. European Economic Review 54: 608-621.
Dabla-Norris, E. 2000: A game--theoretic analysis of corruption in
bureaucracies, IMF Working Paper No. WP/00/106.
Fan, S, Lin, C and Treisman, D. 2009: Political decentralization
and corruption: Evidence from around the world. Journal of Public
Economics 93: 14-34.
Hall, R and Jones, C. 1999: Why do some countries produce so much
more output per worker than others? Quarterly Journal of Economics
114(1): 83-116.
Hausman, J, Stock, J and Yogo, M. 2005: Asymptotic properties of
the Hahn-Hausman test for weak instruments. Economics Letters 89(3):
333-342.
Hendry, D, Johansen, S and Santos, C. 2004): Selecting a regression
saturated by indicators. University of Oxford, Unpublished manuscript.
Huntington, SP. 1968: Political order in changing societies. Yale
University Press: New Haven.
Isham, J, Pritchett, L, Woolcock, M and Busby, G. 2008: The
varieties of resource experience: Natural resource export structures and
the political economy of economic growth. World Bank Economic Review
19:141-174.
Knack, S and Reefer, P. 1995: Institutions and economic
performance: Cross-country tests using alternative institutional
measures. Economics and Politics 7: 207-227.
Leite, C and Weidmann, J. 1999: Does Mother Nature corrupt? Natural
resources, corruption and economic growth, IMF Working Paper No.
WP/99/85.
Lipset, SM. 1960: Political man: The social bases of politics.
Anchor Books: Garden City.
Mauro, P. 1995: Corruption and growth. Quarterly Journal of
Economics 110: 681-712.
Olken, B. 2007: Monitoring corruption: Evidence from a field
experiment in Indonesia. Journal of Political Economy 115 (2): 200-249.
Olken, B. 2009: Corruption perceptions vs. corruption reality.
Journal of Public Economics 93: 950-964.
Rodrik, D, Subramanian, A and Trebbi, F. 2004: Institutions rule:
The primacy of institutions over geography and integration in economic
development. Journal of Economic Growth 9: 131-165.
Rose-Ackerman, S. 1999: Corruption and government: Causes,
consequences and reform. Cambridge University Press: Cambridge.
Staiger, D and Stock, J. 1997: Instrumental variables regression
with weak instruments. Econometrica 65: 557-586.
Stock, J and Yogo, M. 2005: Testing for weak instruments in linear
IV regression. In: Andrews, D and Stock, J (eds) Identification and
Inference for Econometric Models: Essays in Honor of Thomas Rothenberg.
Cambridge University Press: Cambridge, pp. 80-108.
Treisman, D. 2000: The causes of corruption: A cross-national
study. Journal of Public Economics 76: 399-457.
(1) Note that Treisman {2000) in Table 1 (p. 411) reports a
correlation of 0.87 between Business International and Transparency
International macro cross-country data.
(2) Note that Huntington (1968} disagrees with this view. He argues
that rapid modernization induces normative confusion at a time when the
emerging economic elites are striving for power and influence. As a
result growth increases corruption.
(3) Note that the official would never present these objections in
writing.
(4) Note that the Kolmogorov-Smirnov tests reported in Table 2
indicates that the distribution of corruption across states have changed
over the two time periods. Forces such as economic growth may be driving
these changes.
(5) More on this in the section 'Empirical strategy and
data'.
(6) Ades and Di Tella (1999), Rose-Ackerman (1999), Leite and
Weidmann (1999), Dabla-Norris (2000) are other important contributions
in this literature. Bardhan (1997) provides an excellent survey of the
early contributions.
(7) Note that we also use GDP per capita growth rate in Table 3 and
our results are robust.
(8) Note that the survey asks some additional questions. However,
they are not common over the two time periods in our study. Therefore we
are not including them here.
(9) Note that the India Corruption Study only reports these macro
percentages and the underlying micro data is not reported.
(10) The model predicts that corruption in Bihar should have
declined by 18.5% points due to the RTI Act. However, the actual decline
is 30% points. Therefore, the predicted decline is 62% of the actual.
(11) According to our estimates, economic growth reduced corruption
perception only in education.
SAMBIT BHATTACHARYYA [1] & RAGHBENDRA JHA [2]
[1] Department of Economics, Jubilee Building, 262, University of
Sussex, Brighton BN1 9RF, United Kingdom. E-mail:
s.bhattacharyya@sussex.ac.uk
[2] Australia South Asia Research Centre, Arndt-Corden Division of
Economics, Australian National University, Canberra 0200, Australia.
E-mail: r.jha@anu.edu.au
Table is Summary statistics
Variable Number Mean
of obs.
Corruption [[c.sub.it]] 40 32.3
Corruption in banks [[c.sup.BANKS.sub.it]] 40 22.2
Corruption in land admin. [[c.sup.LAND.sub.it]] 40 48.8
Corruption in police [c.sup.POLICE.sub.it] 40 53.4
Corruption in education [[c.sup.EDUC.sub.it]] 40 18.9
Corruption in water [[c.sup.WATER.sub.it]] 40 29.3
Corruption in PDS [[c.sup.PDS.sub.it]] 40 32.4
Corruption in electricity [[c.sup.ELEC.sub.it]] 40 30.95
Corruption in hospitals [[c.sup.HOSP.sub.it]] 40 30.8
Economic growth [[[??].sub.it]] 40 7.9
Log rainfall [ln [RAIN.sub.it-1]] 40 6.8
Variable Standard Minimum
deviation
Corruption [[c.sub.it]] 11.6 16.8
Corruption in banks [[c.sup.BANKS.sub.it]] 12.5 2.3
Corruption in land admin. [[c.sup.LAND.sub.it]] 13.9 19.2
Corruption in police [c.sup.POLICE.sub.it] 14.0 14.0
Corruption in education [[c.sup.EDUC.sub.it]] 9.9 3.2
Corruption in water [[c.sup.WATER.sub.it]] 11.95 4.1
Corruption in PDS [[c.sup.PDS.sub.it]] 10.9 10.6
Corruption in electricity [[c.sup.ELEC.sub.it]] 11.7 4.6
Corruption in hospitals [[c.sup.HOSP.sub.it]] 10.9 9.6
Economic growth [[[??].sub.it]] 4.1 -4.2
Log rainfall [ln [RAIN.sub.it-1]] 0.8 5.4
Variable Maximum
Corruption [[c.sub.it]] 59.1
Corruption in banks [[c.sup.BANKS.sub.it]] 55.0
Corruption in land admin. [[c.sup.LAND.sub.it]] 77.3
Corruption in police [c.sup.POLICE.sub.it] 80.8
Corruption in education [[c.sup.EDUC.sub.it]] 49.3
Corruption in water [[c.sup.WATER.sub.it]] 54.0
Corruption in PDS [[c.sup.PDS.sub.it]] 60.3
Corruption in electricity [[c.sup.ELEC.sub.it]] 57.0
Corruption in hospitals [[c.sup.HOSP.sub.it]] 57.8
Economic growth [[[??].sub.it]] 16.9
Log rainfall [ln [RAIN.sub.it-1]] 8.0
Table 2: Kolmogorov-Smirnov equality of distribution test over time
periods 2005 and 2008
Variable Kolmogorov- p-values
Smirnov
test
statistic
Corruption [[c.sub.it]] 0.90 0.00
Corruption in banks [[c.sup.BANKS.sub.it]] 0.45 0.02
Corruption in land admin. [[c.sup.LAND.sub.it]] 0.80 0.00
Corruption in police [c.sup.POLICE.sub.it] 0.95 0.00
Corruption in education [[c.sup.land.sub.it]] 0.60 0.00
Corruption in water [[c.sup.WATER.sub.it]] 0.45 0.02
Corruption in PDS [[c.sup.PDS.sub.it]] 0.35 0.11
Corruption in electricity [[c.sup.ELEC.sub.it]] 0.60 0.00
Corruption in hospitals [[c.sup.HOSP.sub.it]] 0.70 0.00
Notes: The Kolmogorov-Smirnov non-parametric test is to test the
hypothesis that distribution of corruption across states over the two
time periods (2005 and 2008) are identical. In other words, the null
hypothesis is [H.sub.0]:[F.sub.2005](c) = [G.sub.2008](c), where
[F.sub.2005](c) and [G.sub.2008](c) are empirical distribution
functions of corruption across states in 2005 and 2008, respectively.
The test statistic is defined as [MATHEMATICAL EXPRESSION NOT
REPRODUCIBLE IN ASCII] and can be compared with Table 55 of Biometrika
tables, Vol. 2. If the difference is large then it Leads to rejection
of the null hypothesis. Note that PDS stands for Public Distribution
System.
Table 3: Economic growth, law, and corruption
Dependent variable: Corruption [[c.sub.it]]
(1) (2) (3) (4)
Panel A
OLS estimates LIML Fuller IV estimates
Economic growth -0.33 *** -0.43 ***
[[[??].sub.it]] (0.12) (0.14)
Year 2008 dummy -18.24 *** -17.08 *** -18.48 *** -18.83 ***
(3.08) (2.59) (1.49) (1.92)
Per capita GDP growth -0.23 ** -0.39 **
(0.11) (0.21)
Endogeneity test 0.07 0.06
(p-value)
Controls: State dummies
Instruments Log rainfall
[ln [RAIN.sub.it-1]]
States 20 20 20 20
Observations 40 40 40 40
Adjusted [R.sup.2] 0.89 0.88
Panel B: First stage estimates
Economic growth Per capita GDP growth
[[[??].sub.it]
Log rainfall [ln 12.2 * 14.7 *
[RAIN.sub.it-1]] (6.55) (9.30)
F statistic 12.4 13.14
Stock-Yogo critical 24.09 23.81
value
Partial [R.sup.2] on 0.009 0.018
instruments
Controls: State dummies, Year 2008 dummy
States 20 20
Observations 40 40
Adjusted [R.sup.2] 0.76 0.57
Notes: ***, **, and * indicate significance level at 1%, 5%, and 10%,
respectively, against a two-sided alternative. Figures in the
parentheses are cluster standard errors at the state level and they
are robust to arbitrary heteroskedasticity and arbitrary intra-group
correlation. All regressions are carried out with an intercept. Sample
years are 2005 and 2008. Fuller's modified LIML estimator with [alpha]
= 1 (correction parameter proposed by Hausman et al., 2005) is used in
Panel A, which is robust to weak instruments. Endogeneity test for one
or more endogenous regressors p-values are reported. The null
hypothesis is that the specified endogenous variables can actually be
treated as exogenous. Under the null the test statistic follows [chi
square]-distribution with degrees of freedom equal to the number of
regressors tested. Note that Sargan overidentification test is not
reported for columns 3 and 4 in Panel A as we have an exactly
identified system. Stock-Yogo critical value are based on LIML size
and significance level of 5%. An F-statistic below the level of Stock-
Yogo critical value would indicate that the instruments are weak.
Partial [R.sup.2] on excluded instruments are also reported, which
measures instrument relevance.
Table 4: Economic growth, law, and corruption in different sectors
Corruption in Corruption in
banks land admin.
[[c.sup.BANKS.sub.it]] [[c.sup.LANDS.sub.it]]
LIML Fuller IV estimates
(1) (2)
Economic growth -0.46 ** -0.96 ***
[[[??].sub.it]] (0.19) (0.20)
Year 2008 dummy -9.43 *** -17.18 ***
(3.13) (3.14)
Endogeneity test 0.06 0.05
(p-value)
Controls: State dummies
Instruments Log Rainfall [ln [RAIN.sub.it-1]]
States 20 20
Observations 40 40
Corruption Corruption in
in police education
[[c.sup.POLICE.sub.it]] [[c.sup.EDUC.sub.it]]
LIML Fuller IV estimates
(3) (4)
Economic growth 0.33 -0.60 ***
[[[??].sub.it]] (0.28) (0.13)
Year 2008 dummy -20.38 *** -9.03 ***
(2.71) (1.83)
Endogeneity test 0.07 0.06
(p-value)
Controls: State dummies
Instruments Log Rainfall [In [RAIN.sub.it-1]]
States 20 20
Observations 39 40
Corruption Corruption
in water in PDS
[[c.sup.WATER.sub.it]] [[c.sup.PDS.sub.it]]
LIML Fuller IV estimates
(5) (6)
Economic growth -0.85 0.11
[[[??].sub.it]] (0.60) (0.44)
Year 2008 dummy -7.91 *** -6.15 *
(2.86) (3.33)
Endogeneity test 0.06 0.06
(p-value)
Controls: State dummies
Instruments Log Rainfall [In [RAIN.sub.it-1]]
States 20 20
Observations 39 40
Corruption in Corruption in
electricity hospitals
[[c.sup.ELEC.sub.it]] [[c.sup.HOSP.sub.it]]
LIML Fuller IV estimates
(7) (8)
Economic growth -0.76 ** -0.85 ***
[[[??].sub.it]] (0.31) (0.18)
Year 2008 dummy -11.55 *** -12.78 ***
(2.48) (2.44)
Endogeneity test 0.08 0.06
(p-value)
Controls: State dummies
Instruments Log Rainfall [In [RAIN.sub.it-1]]
States 20 20
Observations 40 40
Notes: ***, **, and * indicate significance level at 1%, 5%, and 10%,
respectively, against a two-sided alternative. Figures in the
parentheses are cluster standard errors and they are robust to
arbitrary heteroskedasticity and arbitrary intra-group correlation.
All regressions are carried out with an intercept. Sample years are
2005 and 2008. Fuller's modified LIML estimator with [alpha] = 1
(correction parameter proposed by Hausman et al., 2005) is used, which
is robust to weak instruments. Endogeneity test for one or more
endogenous regressors p-values are reported. The null hypothesis is
that the specified endogenous variables can actually be treated as
exogenous. Under the null the test statistic follows [chi square]-
distribution with degrees of freedom equal to the number of regressors
tested. Note that Sargan overidentification test is not reported as we
have an exactly identified system.
Table 5: Effect of economic growth and law on corruption experience
and corruption perception
Corruption Corruption Corruption
experience experience experience in
overall in banks land admin.
(1) (2) (3)
Panel A: LIML Fuller IV estimates with corruption experience
Economic growth -0.92 *** -0.77 *** -1.66 ***
[[[??].sub.it]] (0.21) (0.25) (0.59)
Year 2008 dummy -17.09 *** -11.55 *** -29.25 ***
(2.05) (2.19) (4.99)
Endogeneity test (p-value) 0.06 0.06 0.08
Controls: State dummies
Instruments Log rainfall [ln [RAIN.sub.it]]
States 20 20 20
Observations 40 40 40
Corruption Corruption Corruption
experience experience experience
in police in education in water
(4) (5) (6)
Panel A: LIML Fuller IV estimates with corruption experience
Economic growth -0.19 -0.87 *** -0.80
[[[??].sub.it]] (0.58) (0.09) (0.87)
Year 2008 dummy -12.09 *** -7.55 *** -7.65 **
(4.49) (1.47) (3.81)
Endogeneity test (p-value) 0.05 0.04 0.05
Controls: State dummies
Instruments Log rainfall [ln [RAIN.sub.it]]
States 20 20 20
Observations 39 40 39
Corruption Corruption Corruption
experience experience in experience
in PDS electricity in hospitals
(7) (8) (9)
Panel A: LIML Fuller IV estimates with corruption experience
Economic growth -0.13 -0.97 *** -1.79 ***
[[[??].sub.it]] (0.37) (0.29) (0.34)
Year 2008 dummy -10.39 *** -11.04 *** -7.22 ***
(3.31) (1.93) (2.74)
Endogeneity test (p-value) 0.06 0.06 0.06
Controls: State dummies
Instruments Log rainfall [ln [RAIN.sub.it]]
States 20 20 20
Observations 40 40 40
Corruption Corruption Corruption
perception perception perception in
overall in banks Land admin.
Panel B: LIML Fuller IV estimates with corruption perception
Economic growth -0.21 -0.11 -0.83
[[[??].sub.it]] (0.36) (0.45) (0.62)
Year 2008 dummy -15.35 *** -14.42 ** -12.17 ***
(2.86) (5.95) (4.11)
Endogeneity test (p-value) 0.06 0.06 0.07
Controls: State dummies
Instruments Log rainfall [In [RAIN.sub.it]]
States 20 20 20
Observations 40 40 40
Corruption Corruption Corruption
perception perception perception
in police in education in water
Panel B: LIML Fuller IV estimates with corruption perception
Economic growth 0.72 * -0.64 * -0.84
[[[??].sub.it]] (0.37) (0.34) (0.96)
Year 2008 dummy -14.27 *** -14.54 *** -12.47 ***
(3.69) (2.77) (4.73)
Endogeneity test (p-value) 0.05 0.03 0.06
Controls: State dummies
Instruments Log rainfall [In [RAIN.sub.it]]
States 20 20 20
Observations 39 40 39
Corruption Corruption Corruption
perception perception in perception
in PDS electricity in hospitals
Panel B: LIML Fuller IV estimates with corruption perception
Economic growth 0.05 -0.62 0.22
[[[??].sub.it]] (0.65) (0.54) (0.41)
Year 2008 dummy -6.67 -19.14 *** -18.12 ***
(4.44) (3.83) (2.59)
Endogeneity test (p-value) 0.05 0.06 0.06
Controls: State dummies
Instruments Log rainfall [In [RAIN.sub.it]]
States 20 20 20
Observations 40 40 40
Notes: ***, **, and * indicate significance level at 1%, 5%, and 10%,
respectively, against a two-sided alternative. Figures in the
parentheses are cluster standard errors and they are robust to
arbitrary heteroskedasticity and arbitrary intro-group correlation.
All regressions are carried out with an intercept. Sample years are
2005 and 2008. Fuller's modified LIML estimator with [alpha] = 1
(correction parameter proposed by Hausman et al., 2005) is used, which
is robust to weak instruments. Endogeneity test for one or more
endogenous regressors p-values are reported. The null hypothesis is
that the specified endogenous variables can actually be treated as
exogenous. Under the null the test statistic follows [chi square]-
distribution with degrees of freedom equal to the number of regressors
tested. Note that Sargan over identification test is not reported as
we have an exactly identified system.
Table 6: Effect of economic growth and law on bribes and middleman
usage
Bribes Bribes Bribes in
overall in banks land admin.
(1) (2) (3)
Panel A: LIML Fuller IV estimates with bribes
Economic growth -0.93 *** -0.87 *** -1.26 ***
[[[??].sub.it]] (0.26) (0.21) (0.39)
Year 2008 dummy -17.34 *** -11.31 *** -27.36 ***
(3.65) (2.20) (4.65)
Endogeneity test (p-value) 0.06 0.06 0.09
Controls: State dummies
Instruments Log rainfall [In [RAIN.sub.it]
States 20 20 20
Observations 40 40 40
Bribes in Bribes in Bribes
police education in water
(4) (5) (6)
Panel A: LIML Fuller IV estimates with bribes
Economic growth -0.14 -0.83 *** -0.69
[[[??].sub.it]] (0.48) (0.09) (0.80)
Year 2008 dummy -13.62 *** -6.21 *** -7.64 **
(4.16) (1.92) (3.86)
Endogeneity test (p-value) 0.06 0.04 0.05
Controls: State dummies
Instruments Log rainfall [In [RAIN.sub.it]
States 20 20 20
Observations 39 40 39
Bribes Bribes in Bribes in
in PDS electricity hospitals
(7) (8) (9)
Panel A: LIML Fuller IV estimates with bribes
Economic growth -0.11 -1.17 *** -1.69 ***
[[[??].sub.it]] (0.37) (0.20) (0.31)
Year 2008 dummy -10.39 *** -11.68 *** -7.14 ***
(3.32) (1.76) (2.05)
Endogeneity test (p-value) 0.06 0.07 0.06
Controls: State dummies
Instruments Log rainfall [In [RAIN.sub.it]
States 20 20 20
Observations 40 40 40
Middleman Middleman Middleman in
overall in banks land admin.
Panel B: LIML Fuller IV estimates with middlemen usage
Economic growth -0.71 ** -0.62 ** -0.98 **
[[[??].sub.it]] (0.32) (0.24) (0.32)
Year 2008 dummy -18.48 *** -11.29 *** -21.13 ***
(2.13) (2.49) (4.32)
Endogeneity test (p-value) 0.06 0.06 0.07
Controls: State dummies
Instruments Log rainfall (In [RAIN.sub.it]]
States 20 20 20
Observations 40 40 40
Middleman Middleman Middleman
in police in education in water
Panel B: LIML Fuller IV estimates with middlemen usage
Economic growth -0.13 -0.71 *** -0.49
[[[??].sub.it]] (0.47) (0.14) (0.62)
Year 2008 dummy -12.89 *** -4.68 *** -6.62 **
(4.79) (1.91) (3.18)
Endogeneity test (p-value) 0.06 0.04 0.05
Controls: State dummies
Instruments Log rainfall (In [RAIN.sub.it]]
States 20 20 20
Observations 39 40 39
Middleman Middleman in Middleman
in PDS electricity in hospitals
Panel B: LIML Fuller IV estimates with middlemen usage
Economic growth -0.09 -0.97 *** -1.80
[[[??].sub.it]] (0.28) (0.26) (0.38)
Year 2008 dummy -10.86 *** -11.09 *** -7.48 ***
(3.76) (1.96) (2.65)
Endogeneity test (p-value) 0.06 0.06 0.06
Controls: State dummies
Instruments Log rainfall (In [RAIN.sub.it]]
States 20 20 20
Observations 40 40 40
Notes: ***, **, and * indicate significance level at 1%, 5%, and 10%,
respectively, against a two-sided alternative. Figures in the
parentheses are cluster standard errors and they are robust to
arbitrary heteroskedasticity and arbitrary intra-group correlation.
All regressions are carried out with an intercept. Sample years are
2005 and 2008. Fuller's modified LIML estimator with [alpha] = 1
(correction parameter proposed by Hausman et al., 2005) is used, which
is robust to weak instruments. Endogeneity test for one or more
endogenous regressors p-values are reported. The null hypothesis is
that the specified endogenous variables can actually be treated as
exogenous. Under the null the test statistic follows [chi square]-
distribution with degrees of freedom equal to the number of regressors
tested. Note that Sargan over identification test is not reported as
we have an exactly identified system.
Table 7: Economic growth, law, and corruption: Robustness with
additional covariates
Dependent variable: Corruption [[c.sub.it]]
(1) (2) (3)
LIML Fuller IV estimates
Economic -0.34 *** -0.39 *** -0.44 ***
growth [[[??].sub.it]]
(0.05) (0.12) (0.13)
Year 2008 -19.12 *** -19.58 *** -18.62 ***
dummy
(2.02) (1.75) (1.81)
Endogeneity 0.06 0.07 0.06
test (p-value)
Controls: State dummies
Additional Literacy Gini Poverty head
controls: coefficient count ratio
Instruments Log rainfall [In [RAIN.sub.it]]
States 18 20 20
Observations 36 40 40
Dependent variable: Corruption [[c.sub.it]]
(4) (5) (6)
LIML Fuller IV estimates
Economic -0.44 *** -0.48 ** -0.17 ***
growth [[[??].sub.it]]
(0.16) (0.22) (0.02)
Year 2008 -18.83 *** -18.06 *** -15.51 ***
dummy
(2.27) (2.24) (2.18)
Endogeneity 0.06 0.06 0.06
test (p-value)
Controls: State dummies
Additional Mining Primary State government
controls: share of sector share expenditure
GDP of GDP ***(-)
Instruments
States 20 20 19
Observations 40 40 38
Dependent variable:
Corruption [[c.sub.it]]
(7) (8)
LIML Fuller IV estimates
Economic -0.76 *** -0.64 ***
growth [[[??].sub.it]]
(0.06) (0.21)
Year 2008 -17.23 *** -19.91 ***
dummy
(1.70) (3.19)
Endogeneity 0.06 0.07
test (p-value)
Controls: State dummies
Additional Newspaper Telephone
controls: circulation exchange
Instruments
States 18 14
Observations 36 28
Dependent variable:
Corruption [[c.sub.it]]
(9) (10)
LIML Fuller IV estimates
Economic -0.47 *** -1.22 **
growth [[[??].sub.it]]
(0.16) (0.67)
Year 2008 -18.21 *** -18.41
dummy
(1.94) (1.94)
Endogeneity 0.06 0.07
test (p-value)
Controls: State dummies --
without AP, TN, WB
Additional -- --
controls:
Instruments
States 20 20
Observations 40 40
Notes: ***, **, and * indicate significance level at 1%, 5%, and 10%,
respectively, against a two-sided alternative. Figures in the
parentheses are cluster standard errors and they are robust to
arbitrary heteroskedasticity and arbitrary intra-group correlation.
All regressions are carried out with an intercept. Sample years are
2005 and 2008. Fuller's modified LIML estimator with [alpha] = 1
(correction parameter proposed by Hausman et al., 2005) is used, which
is robust to weak instruments. Endogeneity test for one or more
endogenous regressors p-values are reported. The null hypothesis is
that the specified endogenous variables can actually be treated as
exogenous. Under the null the test statistic follows [chi square]-
distribution with degrees of freedom equal to the number of regressors
tested. Note that Sargan over identification test is not reported as
we have an exactly identified system. Also note that columns 9 and 10
report the Hendry et al. (2004) procedure and the estimates without
state dummies. These procedures are described in the section
'Empirical strategy and data'.
