The effect of environmental decentralization on

polluting industries in India

Stefania Lovo
January 2014
Centre for Climate Change Economics and Policy
Working Paper No. 160
Grantham Research Institute on Climate Change and
the Environment
Working Paper No. 143
The Centre for Climate Change Economics and Policy (CCCEP) was established
by the University of Leeds and the London School of Economics and Political
Science in 2008 to advance public and private action on climate change through
innovative, rigorous research. The Centre is funded by the UK Economic and Social
Research Council and has five inter-linked research programmes:
1. Developing climate science and economics
2. Climate change governance for a new global deal
3. Adaptation to climate change and human development
4. Governments, markets and climate change mitigation
5. The Munich Re Programme - Evaluating the economics of climate risks and
opportunities in the insurance sector

More information about the Centre for Climate Change Economics and Policy can be
found at: http://www.cccep.ac.uk.

The Grantham Research Institute on Climate Change and the Environment was
established by the London School of Economics and Political Science in 2008 to
bring together international expertise on economics, finance, geography, the
environment, international development and political economy to create a world-
leading centre for policy-relevant research and training in climate change and the
environment. The Institute is funded by the Grantham Foundation for the Protection
of the Environment and the Global Green Growth Institute, and has five research
programmes:
1. Global response strategies
2. Green growth
3. Practical aspects of climate policy
4. Adaptation and development
5. Resource security

More information about the Grantham Research Institute on Climate Change and the
Environment can be found at: http://www.lse.ac.uk/grantham.

This working paper is intended to stimulate discussion within the research community
and among users of research, and its content may have been submitted for
publication in academic journals. It has been reviewed by at least one internal referee
before publication. The views expressed in this paper represent those of the
author(s) and do not necessarily represent those of the host institutions or funders.
 The effect of environmental decentralization on
polluting industries in India.
Stefania Lovo
Grantham Research Institute for Climate Change and the Environment, London
School of Economics

January 7, 2014

Abstract
This paper examines the unintended effects of the 2006 reform of
the Environmental Impact Assessment (EIA) process in India using
firm-level data for the period 1998-2012. The reform favored a de-
centralization process by delegating the responsibility over environ-
mental clearance of certain activities to state-level authorities. The
results show that variations in the strength of environmental enforce-
ment across states has resulted in an increase of births for polluting
industries affected by the reform in states with lower level of enforce-
ment.

1 Introduction
This paper examines the impact of the 2006 Environmental Impact Assess-
ment (EIA) reform on firm decisions concerning births and locations. The
reform delegates the responsibility over environmental clearance of certain
activities, previously under the control of the central government, to newly
This research is part of the green growth programme at the Grantham Research In-
stitute on Climate Change and the Environment, which is funded by the Global Green
Growth Institute, as well as the Grantham Foundation for the Protection of the Envi-
ronment, and the Economic and Social Research Council through the Centre for Climate
Change Economics and Policy. I would like to thank the participants of the GGGI seminar
and GRI Green Growth Workshop at LSE for their useful comments and suggestions.

3
established state-level authorities. While environmental standards are de-
cided by the central government, most of the environmental compliance
monitoring and enforcement in India was already a responsibility of state-
level Pollution Control Boards (SPCB). The 2006 EIA reform has expanded
this decentralized model by assigning additional tasks to state-level author-
ities. Though this could potentially lead to a more transparent and efficient
environmental clearance process, it could also have unintended effects on
firms’ decisions.
Environmental enforcement varies notably across Indian states. Differ-
ences arise possibly due to variations in socio-economic and political con-
ditions across states but also due to state-specific technical and financial
constraints. This paper examines whether differences in environmental en-
forcement influence firm births across states through the EIA process strin-
gency. The key findings are derived from the fact that the reform of the EIA
has affected only certain polluting activities, while for others the process has
remained almost unchanged, and that pre-reform differences in environmen-
tal enforcement across states are likely to be indicative of differences in the
stringency of the EIA process across states. The empirical approach is based
on a firm-level latent-startup model where the number of births is a function
of time-varying state and sector level characteristics.
The estimations compare pre- and post-reform births, conditional on the
strength of environmental enforcement in each state. Environmental enforce-
ment is measured using a composite index obtained by aggregating various
state-level indicators of institutional quality, civic participation and institu-
tional capacity. The identification strategy relies on the presumption that,
while firm births in sectors affected by the reform should respond to vari-
ations in environmental enforcement after the implementation of the 2006
EIA reform, no effects are expected on non-polluting firms and on polluting
industries not affected by the decentralization process. The estimations are
based on the population of registered firms born during the period 1998-
2012. Although the formal sector contributes to only a small fraction of
total Indian output, large firms are the only ones subject to environmental
clearance, since smaller informal firms tend to operate outside the control
of pollution control authorities.
The results show a significant increase in births of polluting firms in
states with lower enforcement levels relative to states with higher enforce-
ment levels. This relocation effect has ambiguous consequences for the In-
dian economy. The relocation of firms across the territory may negatively
impact the Indian economy if it implies that firms choose inferior locations
that increase operational costs and decrease efficiency (Becker and Hender-

4
son, 2000). On the other hand, firm relocation may be socially beneficial
since firms might locate in less populated areas where either pollution dam-
ages are lower or economic growth benefits are higher and offset the envi-
ronmental costs. It is, however, not in the scope of this paper to quantify
all the welfare effects resulting from polluting firms locating in states with
lower enforcement.
The remainder of the paper is organized as follows. Section 2 describes
the policy background and the reform of the EIA process introduced in 2006.
Section 3 provides a brief overview of the relevant literature. The empirical
model is described in section 4 while the data are presented in section 5.
The empirical results are discussed in section 6 and section 7 concludes.

2 Environmental policies in India

Environmental protection rights and duties are incorporated into the Indian
Constitution which mandates that the State and every citizen shall protect
and improve the environment. India has an elaborate set of laws relat-
ing to environmental protection that dates back to the Water Act in 1974.
The central government, through the Ministry of Environment and Forests
(MOEF) and the Central Pollution Control Board (CPCB), is in charge of
planning and formulating national policies and standards. Their implemen-
tation and enforcement are decentralized and are the responsibility of the
State Pollution Control Boards.
Besides in addition to the role undertaken by central and state institu-
tions, Indian citizens benefit from a unique approach to the enforcement of
environmental laws by exercising their constitutional right to a healthy en-
vironment in the form of Public Interest Litigations (PILs) before the Court
of Justice. PILs have resulted in some environmental improvements on one
side, (Kathuria, 2007) but have also contributed to increase the amount of
work for state authorities because of court-ordered directives (OECD, 2006).
A compulsory Environmental Impact Assessment (EIA) was first intro-
duced in India with the Environmental Protection Act of 1986, but became
effective only in 1994 when the MOEF passed a major legislative measure
(Panigrahi and Amirapu, 2012). The main purpose of the EIA is to inform
decision makers and the public about the environmental implications of a
particular project (Panigrahi and Amirapu, 2012). The EIA process has
been notably modified with the introduction of the 2006 EIA notification.
These changes brought by the new notification are the focus of this study.
The 2006 EIA notification introduces, among others, one major change:

5
polluting projects/firms are classified into two categories based on the po-
tential impacts on human health and natural resources. Projects classified
as category A have to undertake the EIA at the national level, as per the
1994 notification, while category B projects are referred to the State Envi-
ronmental Impact Assessment Authority (SEIAA) of the state in which the
project is located. The classification of projects into these two categories
is based on three main criteria. The first criterion defines those projects
that are of exclusive competence of either state or national authorities. Any
project within the petroleum refining industry, for instance, should under-
take the EIA at the central level, while projects in the integrated paint indus-
try are the exclusive responsibility of state authorities. A second criterion
distinguishes projects in terms of capacity. Large coke oven plants (above
150,000 tonnes per year), for example, are under the authority of the cen-
tral government, small ones are referred to the SEIAA in which the project
is located. A third criterion categorizes projects on the basis of whether
they are located within or outside a notified industrial area. Projects in the
leather/skin/hide processing industry, for example, are subject to state level
EIA only if located within an industrial estate/area. Table 1 summarizes the
five different groups of activities defined in accordance to the three criteria
mentioned above. The detailed list of projects and activities, as reported in
the official 2006 EIA notification, is provided in table 8 of the Appendix.

Table 1: Classification of sectors according to EIA status

# Group Category Description Sectors (NIC)

All sectors not re-
1 No EIA No EIA Not subject to EIA
ported below

Exclusively subjected to 111, 112, 232, 233,

2 Central A
central-level EIA 269, 2412, 2421

101, 102, 120, 131,

132, 141, 142, 155,
State-level EIA if of
3 Small/Large A/B 231, 271, 272, 273,
small capacity
401, 451, 452, 453,
454, 2694
State-level EIA if located
182, 1911, 2411,
4 EI A/B within industrial estates
2413, 2423, 2424
(IE)

Exclusively subjected to 1542, 2101, 2422,

5 State B
state-level EIA 2430, 2439

6
 SSEIAAs were constituted in each state at different points in time and
all projects were treated as category A in absence of a notified state au-
thority. Table 9 of the Appendix reports the date of establishment of each
SSEIAA. The 2006 notification initiated a process of decentralization of the
EIA procedure which could reduce the burden on the central government
and accelerate the approval process. However, the decentralized powers
could either be misused if state governments intend to actively pursue in-
dustrialisation for their respective state, or be ineffective if state authorities
lack technical and financial powers.
The EIA process is subdivided into four stages. The first stage (Screen-
ing) affects only category B projects and is aimed at determining whether
a project requires an environmental impact assessment report. Projects re-
quiring EIA are categorized as B1 while the others are termed B2 and submit
a much shorter application form. Although guidelines for the screening pro-
cess are provided by the MOEF, there is still lack of clarity on this stage
of the process. The second stage (Scoping) involves either the central or
state authority in determining the terms of references covering all relevant
environmental concerns for the preparation of the EIA. The third stage re-
quires a public consultation through both a public hearing in the proximity
of the site and invitations of written responses from the concerned stakehold-
ers. The final stage (Appraisal) involves the scrutiny of the EIA application
which can result in either approval or rejection of the project. Each EIA
contains the environmental management plan, which becomes part of the
business permit and is binding upon the company that is carrying out the
activities. Therefore, it sets the point of departure for future supervision
and enforcement as it determines what can be enforced.

3 Decentralization, environmental regulation and

firm location
The decentralization of environmental regulation is often justified by the
intention to form a better understanding of local environmental problems,
to promote a more transparent and efficient use of natural resources and
to increase local participation due to higher homogeneity of common needs
(Cistulli, 2002). There are, however, well recognized constrains on the suc-
cessfulness of any decentralization process such as weak administrative or
technical capacity, lack of financial resources, poor coordination between
national and local policies and the risk of local elite capture.
While the theoretical literature has studied extensively the trade-offs in-

7
volved in the decentralization of the decision making process (Besley and
Coate (2003); Oates (2002)) fewer studies have empirically investigated its
consequences. Brunnermeier and Levinson (2004) review the empirical lit-
erature and find that while earlier cross section studies tend to find no sig-
nificant effect of environmental regulation on firms’ decisions, more recent
studies that use panel data estimations find evidence of firms responding to
variations in regulation within a country. Many empirical studies focus on
variations in regulation across counties and states of the United States (List
et al., 2003). Becker and Henderson (2000), for example, study variations in
air quality regulation across counties and find that there has been a signifi-
cant relocation of polluting firms from more to less polluted areas. Sigman
(2005) shows that the decentralization of environmental authorities in the
United States has led to a 4% increase in water degradation downstream of
states that had the authority to issue and enforce permits for point source
polluters.
This literature is related to that on the pollution-heaven hypothesis. Sev-
eral studies have analysed the impact of cross-country differences in environ-
mental regulation on the location of foreign direct investment. The results
are mixed. Javorcik and Shang-Jin (2003), for example, analyze investment
flows to 25 economies in Eastern Europe and the former Soviet Union and
find no evidence that FDI from polluting industries is more likely to go to
countries with weaker environmental regulations. On the other hand, Dean
et al. (2009) find evidence of pollution heaven behavior in China for invest-
ment in highly polluting industries from ethnically Chinese countries, while
no effect was found for investment from other sources.
Fewer studies have focused on firm’s location decisions in developing
countries. Duvivier and Xiong (2013), for example, studies trans-boundary
pollution in China where environmental policy is decentralized. Similarly to
India, while the central government sets the standards, local governments are
in charge of monitoring and sanctioning. The authors analyze the location
choice of polluting firms in one of the most polluted province in China and
find that polluting firms tend to locate in counties that share a border with
another province. Similarly, Lipscomb and Mobarak (2007) analyze rivers
water quality across jurisdictions in Brazil and find a significant strategic
polluting behavior around borders.
The first study to analyse firms’ location decisions in response to envi-
ronmental regulation in India was conducted by Mani et al. (1997). The
study finds that the number of new plants is not affected by the stringency
of environmental regulation at the state level. A positive correlation be-
tween a measure of enforcement and the number of new plants, however,

8
suggests that the variable might be capturing the quality of state govern-
ment rather than environmental enforcement. Moreover, the data date back
to 1994 when there was very little enforcement of environmental regula-
tion across all states since prosecution could only occur through the judicial
system (Lipscomb, 2008). There is, however, evidence of Indian firms ad-
justing their behavior in response to changes in environmental regulation
over time. Lipscomb (2008), for example, analyses the response of multi-
product firms to changes in enforcement at state level1 . The author finds
that firms react to increase stringency by increasing the share of product
portfolio allocated to clean products. High productivity firms invest in new
and cleaner products and gain from an increase in enforcement. Kathuria
(2007) finds that an increase in informal regulation, measured by local news
coverage of pollution-related events and the number of public interest litiga-
tions filed, has reduced industrial pollution in the state of Gujarat. On the
other hand, however, formal regulation, measured by the number of staff
allocated to a region, was found not to affect polluting behavior.
Finally, this paper relates to the broader literature on firms’ location
choices. This is a vast literature that covers various aspects such as ag-
glomeration economies (Brülhart et al., 2012), variations in corporate taxes
(Barrios et al., 2012) and infrastructure (Holl, 2004). In India, Lall and
Chakravorty (2005) estimate a cost function at the plant level where eco-
nomic and geographic factors are included to explain firms agglomeration.
The authors find that only local economy diversity matters as a cost-reducing
factor for medium and large industries. The empirical model presented be-
low builds upon this well established literature and takes advantage of the
institutional setting of the 2006 EIA reform to analyze the impact of differ-
ences in environmental enforcement on firms’ location decisions.

4 Empirical model
This section adapts the so called latent-startup model (Becker and Hender-
son, 2000) to represent the possible responses of Indian polluting firms to
the policy changes introduced with the 2006 EIA notification. While envi-
ronmental clearance is required also for the expansion or modernization of
existing polluting firms, this paper considers only the effects on the birth
of new polluting firms. The model assumes that potential entrepreneurs
are spatially immobile and decide whether to set up a firm in a particular
1
Enforcement is measured by the percentage of polluting firms which as been closed
by state and year.

9
sector and location. The alternative model, the footloose-startup model, in-
stead, considers the decision about which location to select once investors
have already decided to set-up a company. Empirically, the two models are
equivalent (Brülhart et al., 2012). At each point in time, an entrepreneur
acts as maximizing its net expected present value and compares the sunk
cost of entry with the expected profits in a particular sector and location.
The expected profits, πf ijt , of firm f in sector i state j at time t depends
on the characteristics of the sector and location of the firm at the time of
establishment, xijt , and on the expected relative compliance costs, cj , which
refers to expected future monitoring, reporting and punishment costs which
varies according to the location. Sunk costs are, for simplicity, only repre-
sented by the cost of complying with the EIA application, sijt . Assuming
a linear approximation of profits, the expected net present value can be
written as follows:
npvf ijt = πf ijt − Sjit + f ijt = α00 xijt + α1 cj + β1 sjt + f ijt , (1)
where f ijt is a random disturbance. Expected compliance costs depend on
the relative level of enforcement at state level, cj = f (Ej ), and are assumed
to be constant, in relative terms, over time, i.e. enforcement is allowed to
increase or decrease over time but the ranking across countries in terms
of environmental stringency is expected to be unaffected. Set-up costs are
instead expected to change over time, but only for category B firms. The
introduction of the 2006 EIA notification has decentralized the process of
environmental clearance for category B firms introducing a new source of
variation across states. The model assumes that before 2006, set-up costs
were identical across states because the EIA was conducted at the central
level for all firms. After 2006, environmental clearance costs depend on Ej
only for firms in category B.
Set-up costs after the implementation of the reform can, therefore, be
re-written as a function of the enforcement capacity in each state:
npvf ijt = α0 xijt + β0 f (Ej ) + β1 (DT × Ej )) + f ijt , (2)
where DT is a dummy variable indicating the years following the implemen-
tation of the EIA notification at time T . After T , set-up costs can be higher
or lower than pre-reform costs. States with low levels of enforcement are
expected to impose lower environmental clearance costs. For some firms,
start-up costs could drop to zero if the screening process conducted by the
SEIAA indicates that the project is exempted from the EIA process.
The expected effect of this decentralization process for category B firms
is twofold. Some states might impose more stringent conditions than those

10
imposed previously by the central government resulting in a reduction of
births in high enforcing states (deterrence). On the other hand, some states
might conduct a laxer EIA in order to promote industrialization, or due to
technical and financial constraints, facilitating the birth of new polluting
firms (attraction). Both forces lead to relative higher birth rates in lower-
enforcing states compared to higher-enforcing states.
Following Becker and Henderson (2000), the model can be represented
as a reduced form equation where the total number of firms born in each
sector, state and year, nijt , is a function of the above mentioned variables
and a set of state (gj ), sector (di ), and time (wt ) fixed effects:

nijt = exp(α0 xijt + β1 ((DT × Ej )

(3)
+γ 0 di + δ 0 gj + ρ0 wt ).

The above equation can be estimated separately for each group of sectors
reported in table 1. We expect the coefficient β1 to be negative only for firms
affected by the decentralization process (Category B: group 5) while being
insignificant for non-polluting firms (i.e. not included in the reform, group
1) and those undertaking the EIA at central level (Category A: group 2).
For groups 3 and 4 the effect is ambiguous as they contain a mixture of
category A and B firms.
In practice, the model is estimated by pooling the five groups of sectors
and interacting all variables by group dummies. In doing so the results
are equivalent to those obtained by estimating the model separately but
have the advantage of allowing for a statistical comparison of the coefficients
across groups. Moreover, the pooled model can be related to a heterogeneous
treatment effect model where the treated sectors are those included in group
3, 4 and 5 and the control group includes non-polluting sectors and sectors
subject to central-level EIA (group 1 and 2). The treatment effect is allowed
to be heterogeneous depending on the level of enforcement in each state prior
to the implementation of the 2006 reform.
Because the model includes state, year and sector fixed effects it is not
possible to identify the effects of pure location, time and sector-specific vari-
ables. All specifications will control for the average share of firms in each
year and location to control for state-level growth patterns. Additional
controls will be discussed in the next sections. Although the EIA notifica-
tion was introduced in September 2006, the decentralization process could
not actually take place unless a SEIAA was created. Because most of the
SEIAAs were established between 2007 and 2008 (table 9 in the appendix),
the variable DT will take values one for 2008 and subsequent years. This

11
approach, de facto, compares the average number of new firms born before
and after the implementation of the EIA notification, conditioned on the
level of enforcement in each state.
The model is estimated using a Poisson pseudo-maximum-likelihood es-
timator with robust standard errors (Wooldridge, 1991), which allows for the
discreteness of the dependent variable and the large number of zeros. The
estimator produces consistent estimates under relatively weaker assump-
tions than a standard Poisson model, i.e. only the conditional mean need
to be correctly specified2 . The model is also estimated using a standard lin-
ear model (OLS) after log-transforming the dependent variable. While the
log-transformation does not alter the multiplicative relationship between
the explanatory variables and allows for double clustering, it has the dis-
advantage of dropping all cells with zero births. In this specification, the
dependent variable is the ratio of new firms in each sector and location over
the total number of new firms in the sector as it better deals with differences
in growth rates across sectors. The number of births is, however, also used
in an alternative specification.

4.1 Identification issues

At the aggregate level, enforcement capacity is both the cause and the con-
sequence of firms’ location choices. A larger amount of polluting firms may
increase awareness about pollution and lead to increasing pressure to control
pollution. On the other hand, however, more polluting firms may put pres-
sure on the capacity of state-level authorities to deal with non-compliance
and reduce the ability of the authorities to monitor and punish polluters.
By considering only the number of new firms created each year in each state
and sector this problem is substantially reduced. Moreover, the measures
of enforcement considered in the regressions are time-invariant and there-
fore do not lead to a spurious statistically correlation between changes in
enforcement over time and changes in the number of new firms. Although
all specifications control for the presence of state, location and year-level
2
Another possible solution is to assume that the dependent variable follows a negative
binomial distribution, a test is available to assess whether this model is preferable to a
standard Poisson model (Duvivier and Xiong, 2013). Over-dispersion may also arise from
an excess of zeros (zero inflation) which means that the dependent variable reports more
zeros that what is assumed by the Poisson model. An excess of zeros might be explained
by the fact that some states are not suitable locations for particular types of firms. Table
10 in the appendix reports the estimates using a zero-inflated negative binomial model
where the stock of firms in the previous period is used to differentiate between suitable
and unsuitable locations.

12
unobservables, unobserved heterogeneity could still be a concern. Neverthe-
less, the regressor of central interest in the estimations reported below is a
three-way interaction term, between state-level enforcement, a dummy vari-
able indicating the post-implementation period and a group dummy that
is less subject to endogeneity problems. Moreover, the results are tested
for robustness to the inclusion of additional control variables that should
capture other sources of unobserved variation over time, such as changes in
average real wages and in the number of special economic zones.
In cross-section studies it is often argued that failing to control for cor-
ruption creates a problem of omitted variable bias (Dean et al., 2009). High
corruption often implies lower environmental stringency but may also act
as a deterrent for new investments. This is not a concern in this study.
Corruption is included as a measure of environmental stringency since it is
the best available measure of the quality of state-level institutions and there
are no reasons to expect that its deterrence effect should vary before and
after the implementation of the EIA notification.

5 Background and Data

This section describes the data used to measure differences in environmental
enforcement across states and the firm-level data used to analyze the rela-
tionship between the decentralization of the EIA process and firm births.

5.1 State-level environmental enforcement measures

Although environmental standards for industrial pollution are determined
by the central government, evidence suggests that there are large differ-
ences across states in terms of enforcement and compliance (OECD (2006);
World Bank (2006)). Variations arise from socio-economic differences across
states but also from differences in commitment and technical and financial
capacity of state-level environmental authorities. We adopt five measures
of environmental enforcement aimed at capturing state-level differences in
institutional capacity, civic participation and institutional quality. These
measures are reported in table 2.
The choice of these indicators was constrained by data availability. In-
stitutional capacity is measured by the number of monitoring stations per
million people. The data, taken from the IndiaStat database, refer to the
year 2007 and reveal a significant variation across states, ranging from 0.02
per million people in Bihar to 2.12 per million people in Himachal Pradesh.

13
 Table 2: Measures of environmental enforcement by state

State NGOs Judgements Corruption Articlesa Stationsa Index

Andhra Pradesh 29 4 4 213 21 2.27
Assam 7 0 15 9 12 -1.54
Bihar 2 3 20 13 2 -1.25
Chandigarh 2 2 4 5
Chhattisgarh 3 0 6 4 9 -1.38
Delhi 22 2 11 166 11 2.27
Goa 0 0 13 3
Gujarat 7 4 3 146 20 0.81
Haryana 3 1 13 21 5 -1.31
Himachal Pradesh 4 2 2 3 11 -0.14
Jammu & Kashmir 6 0 19 3 3 -1.85
Jharkhand 2 0 14 5 6 -1.89
Karnataka 17 3 17 247 14 0.73
Kerala 7 0 1 155 16 0.05
Madhya Pradesh 12 4 18 43 26 0.03
Maharashtra 26 4 5 165 42 1.83
Meghalaya 1 0 0 2
Odisha 17 3 9 8 12 0.42
Puducherry 1 0 2 3
Punjab 1 1 7 25 15 -1.05
Rajasthan 12 0 16 6 18 -1.31
Tamil Nadu 29 2 12 443 16 1.89
Uttar Pradesh 24 4 10 111 35 1.22
Uttarakhand 4 1 2 2
West Bengal 15 2 8 120 21 0.20
a
The indicator is divided by population before constructing the index

When formal regulation is weak, informal regulation through civic partic-

ipation can play an important role. This is particularly true in India where
a democratic system allows the formation of groups and NGOs, the press
is relatively free and people are empowered with the use of public interest
litigations to demand interventions of the judiciary system. These features
are particularly relevant for this study since citizens are given an active
role in the EIA procedure through a public hearing stage. Three measures
of civic participation are adopted: the number of environmentally oriented
NGOs, the number of newspaper articles mentioning environmental-related
news and the number of judgments passed by the supreme and high courts
related to environmental disputes. NGOs play an important role in shaping
the socio-political discourse in India and there are several examples of how
these organizations have successfully promoted environmental disclosure and

14
raised awareness of governments and the general public (UNESCAP, 2000)3 .
The number of environmentally-oriented NGOs was also used in Javorcik
and Shang-Jin (2003) to measure variation in strength of environmental
enforcement across countries. Another measure of public concern over envi-
ronmental issues is represented by the number of newspaper articles covering
topics related to industrial pollution. The number of newspaper articles in
each state and year was obtained by conducting a search across all English-
language Indian newspapers contained in the database Factiva for the period
1998-2006. Each search included a set of common keywords, such as closure,
court, order, fine etc., and the name of the State Pollution Control Board,
e.g. Bihar State Pollution Control Board. The variable used to construct
the enforcement index was obtained by calculating the cumulative number
of articles referring to each State Pollution Control Board for the entire pre-
reform period. Finally, it was noted that Indian citizens can benefit from a
unique approach to enforce environmental law by exercising a constitutional
right before the Supreme Court and the High Courts in the form of Pub-
lic Interest Litigations (PIL). Unfortunately, it was not possible to obtain
the number of PILs filed in each state, but the number of judgments of the
Supreme and High courts offers a reasonable proxy. The list of judgments re-
lated to environmental issues was obtained from the Judgments information
system of the Supreme and High courts of India. Judgments were manually
assigned to each state based on the location of the firms or the pollution
control boards involved in the court case.
To measure institutional quality we used the corruption index at state-
level provided in a study by the Centre for Media Studies issued by Trans-
parency International India for the year 2005 (CMS, 2005). While this index
is our preferred measure of corruption, it is not available for the Union ter-
ritories of Chandigarh, Goa, Meghalaya, Puducherry and Uttarakhand that
are, therefore excluded from part of the analysis4 .
All enforcement measures are time-invariant and, when possible, refer
to the pre-reform period. They are aggregated into one unique measure of
state-level enforcement through principal component analysis. The use of
principal component analysis techniques is appealing because these multiple
3
The list of Indian NGOs was obtained from an online database: http://ngosindia.com/
accessed in June 2013.
4
We also tested an alternative measure of institutional quality constructed as the num-
ber of cases of persons arrested under the prevention of corruption act and related sections
that have obtained charges. The information was obtain from the India Bureau of Crime
and was available for all states. We also tested the robustness of the results to the exclusion
of corruption from the enforcement index.

15
Table 3: Environmental enforcement index: principal component analysis

Component Eigenvalue Proportion of Variance Cumulative

Comp1 1.876 0.375 0.375
Comp2 1.492 0.298 0.674
Comp3 0.883 0.177 0.850
Comp4 0.479 0.096 0.946
Comp5 0.270 0.054 1.000
Variable First component
NGOs 0.638
Judgments 0.519
Corruption Index -0.299
Total articles/Population 0.481
Stations/Population 0.060

variables are correlated and aim at capturing the same concept, i.e. the level
of environmental enforcement. Table 3 shows that, as expected, all measures
but the corruption index are positively related to the latent environmental
enforcement measure. The first principal component explains about 37% of
the total variance in the data. The eigenvalue of the first principal compo-
nent is close to two, thus we retain only the first component which will be
referred to as enforcement index. The index ranges between -1.9 and 2.2 and
takes higher values in states where environmental enforcement is stronger.

5.2 Firm-level data

The firm-level data used to compute the number of births in each sector
and year are collected in the Orbis database by Bureau Van Dijk and are
originally provided by the Centre for Monitoring Indian Economy (CMIE).
The database covers the universe of registered companies, i.e. all companies,
public or private, that are registered under the Companies Act, 1956 at
the Registrar of Companies (RoCs). It records about 140,000 companies
created between 1998 and the end of 2012. Although registered companies
account for only 20% of all firms in India, which tend to be very small
and operate under informality, these companies are the most likely to be
subjected to pollution controls as only large and medium-sized facilities have
the required environmental clearance permits. Most small-scale industries
operate without any consent (OECD, 2006).
The analysis considers only companies belonging to the manufacturing
and energy sectors. While the database provides very little information on
companies characteristics, such as assets, employment etc., it was possible

16
to obtain important information using the corporate identification number
(CIN) that the Ministry of Corporation assigns to each registered company
and that combines information on the year of establishment, state, 5 digit
industry code (National Industrial Classification, NIC), ownership type and
a registrar code. In 1998 the Indian Statistical office adopted a substan-
tially different sector classification which also affected the sector definition
contained in the CIN code. To avoid problems of misclassification of some
firms, the analysis only considers firms established after 1998. This does
not constitute a major drawback since prior to 1997 there was very little
enforcement of environmental regulation across all states (Lipscomb, 2008).
A birth is defined as the registration of a new company in the Registrar
of Companies of the Ministry of Corporation. Companies are assigned to
the five groups reported in table 1 based on the sector they operate in. Un-
fortunately, it is not always possible to assign a particular activity or project
listed in the 2006 EIA notification to a specific sector. Projects/activities
descriptions are sometimes too broad or too narrow to perfectly match a
sector as defined in standard industrial classifications. It was, however, pos-
sible to recover some useful information from a previous draft of the EIA
notification, which was circulated before the official approval of the reform
that provides a concordance table between sectors classification and activ-
ities using the National Industrial Classification (NIC). NIC is the main
classification used by the Indian Statistical Office and form part of the CIN
number. The concordance table was later removed from the official EIA
notification. The list provided in the draft notification was supplemented
by manually matching activities that did not report a corresponding sector
code. The detailed sector-activity concordance used in the analysis is re-
ported in table 8 in the Appendix while a summary is provided in table 1.
Sectors are considered at different digits based on the highest level of aggre-
gation that allowed a one-to-one matching between activities and sectors.
Some sectors, however, were dropped because of ambiguous matching with
listed activities and are reported in table 11 of the appendix. The exclusion
of these sectors, however, does not significantly affect the results.
The information contained in the Orbis database does not allow for the
identification of production capacity or of whether a company is located
within an industrial estate. Therefore, it was not possible to distinguish
whether a company belonging to group 3 and 4 (defined as Capacity and
IE) had undertaken the EIA at the central or state level. One attempt,
however, will be made in the next sections to distinguish large from small
companies. The number of births in each group and year is reported in table
4. The same information is reported by state and year in table 12 in the

17
 Table 4: Number of new firms by category and year of incorporation

Year 1 2 3 4 5
Year Without EIA Central-level Capacity IE State-level Total
1998 2346 160 1045 399 116 4066
1999 2438 208 1273 483 141 4543
2000 1780 154 1038 422 99 3493
2001 1497 115 1128 343 106 3189
2002 1747 154 1400 458 101 3860
2003 2284 207 2200 590 128 5409
2004 3133 225 3660 706 193 7917
2005 4097 335 5799 952 231 11414
2006 3966 351 6556 877 255 12005
2007 4944 337 7084 1020 250 13635
2008 4802 397 7984 833 230 14246
2009 5058 350 4653 699 194 10954
2010 6792 437 7297 980 311 15817
2011 7405 477 7045 1226 328 16481
2012 6534 404 5232 1007 209 13386
Total 58823 4311 63394 10995 2892 140415

Appendix. We excluded the states and union territories of Andaman and

Nicoba, Lakshadweep, Manipur, Mizoram, Nagaland, Tripura, Daman and
Diu, Dadra and Nagar Haveli and Arunachal Pradesh because of insufficient
firm level data and the lack of information on most enforcement measures.
These states, however, represent only about 1% of the Indian population.

6 Firm births and environmental enforcement

This section begins presenting the results of estimating equation 4 using a
linear model. The results are reported in table 5. Columns 2 to 6 show
the effects of each of the five individual enforcement measures used in the
construction of the environmental enforcement index used in column 1 and
7. All specifications include state, year and sector fixed effects capturing
state, time and sector level shocks and trends5 .
The results show that the decentralization of the EIA process has affected
firm births in those sectors subject to state-level environmental clearance.
This conclusion is reinforced when considering the possibility that states
5
Standard errors are clustered at sector level. Sectorial clustering appears to be the
most relevant source of correlation since the dependent variable is the share of new firms
in the sector. The last column, however, reports standard errors clustered at both state
and sector level.

18
applying a more stringent environmental clearance process may attempt
to mitigate its negative effects by offering fiscal incentives to new plants.
Considering the results reported in column 7, while no effect is found for
non-polluting firms and for new firms subject to central-level EIA (No EIA
and Central), the strength of environmental enforcement is shown to have
a larger negative effect after the implementation of the 2006 EIA reform
for firms in sectors of exclusive competence of the SEEIA authority (State)
and for those firms subject to state-level EIA if located within an industrial
estate (EI). The effects are not statistically different among these latter two
categories.
The results are consistent across all different individual measures of en-
forcement, although, in few instances, some coefficients are not significant.
The coefficient related to the third group of sectors (Capacity), those sub-
ject to state-level EIA only if of small capacity, is also negative but much
smaller and not statistically significant. It is likely that many of the compa-
nies included in this category are of large capacity, this issue, however, will
be further explored below. In order to interpret the magnitude of the effects
we can consider that the average gap between states in terms of the envi-
ronmental index score is 0.25. Therefore, an increase in enforcement that
would, on average, allow a state to catch up with the next higher ranked
state would lead to a decrease in the share of new polluting firms between
4% and 6% every year. This is a relevant effect considering that the overall
average annual change in births is about 9-10%.

19
 Table 5: Base results: log-linear model

Dependent variable: log of new firms over total new firms in the sector
(1) (2) (3) (4) (5) (6) (7)
Index NGOs Judgments Corruption News Monitoring Indexa
Baseline: No EIA 0.051** 0.009*** 0.058*** 0.002 0.008 0.007*** 0.051
(0.023) (0.003) (0.016) (0.005) (0.010) (0.002) (0.043)
After x Aj x Central 0.030 0.001 0.027 -0.014*** -0.004 0.004 0.030
(0.044) (0.008) (0.039) (0.005) (0.028) (0.005) (0.055)
After x Aj x Capacity -0.019 -0.002 -0.020 -0.008 0.003 0.001 -0.019
(0.067) (0.008) (0.041) (0.011) (0.015) (0.007) (0.062)
After x Aj x IE -0.159*** -0.015*** -0.104** 0.030* -0.035** -0.011*** -0.159***
(0.054) (0.005) (0.041) (0.015) (0.016) (0.004) (0.054)
After x Aj x State -0.193*** -0.029*** -0.166*** 0.010 -0.013 -0.020*** -0.193***
(0.063) (0.006) (0.028) (0.007) (0.022) (0.003) (0.055)

20
State Yes Yes Yes Yes Yes Yes Yes
Year Yes Yes Yes Yes Yes Yes Yes
Sector Yes Yes Yes Yes Yes Yes Yes
Observations 11561 12590 12590 11561 12590 12590 11561
Standard errors clustered at sector level in parentheses. * p < 0.1, ** p < 0.05, *** p < 0.01
a
Double clustering at sector and state level. All regressions control for the average share of new firms
in the state (not reported).
 In table 6 we test the robustness of the above results considering alter-
native specifications. The first column, considers the log of new firms as
dependent variable and produces similar findings: the strength of environ-
mental enforcement has a larger negative effect after the reform for those
sectors affected by the decentralization process. The remaining columns
include additional control variables. The results reported in column 2 are
obtained after controlling for the average wage in each state and year. The
data are provided by the Ministry of Statistics and Programme Implementa-
tion (Government of India, 2012) and are calculated from the Annual Survey
of Industries that collects information on medium and large firms in India.
Wages are deflated using state-level price indexes (source: IndiaStat). The
inclusion of this variable does not significantly affect the results. On the
other hand, changes in average wages do not appear to significantly affect
the birth of new firms (coefficients not reported). While wages are usually
found to be an important determinant of firms’ location decisions, their poor
performance in these specifications is likely to be due to the limited variation
in real wages over time, as suggested also in Becker and Henderson (2000).

Table 6: Log-linear model: additional control variables

(1) (2) (3) (4) (5)

Log of new Log of share Log of share Log of share Log of share
Baseline: No EIA 0.065 0.055 0.055 0.038 -0.014
(0.040) (0.045) (0.045) (0.047) (0.043)
After x Aj x Central 0.009 0.017 0.016 -0.001 -0.005
(0.046) (0.056) (0.056) (0.072) (0.070)
After x Aj x Capacity -0.011 -0.021 -0.030 -0.019 -0.035
(0.063) (0.067) (0.068) (0.073) (0.073)
After x Aj x IE -0.099∗ -0.150∗∗ -0.150∗∗ -0.150∗∗∗ -0.142∗∗∗
(0.056) (0.069) (0.066) (0.051) (0.052)
After x Aj x State -0.148∗ -0.200∗∗∗ -0.201∗∗∗ -0.179∗∗∗ -0.164∗∗∗
(0.084) (0.055) (0.054) (0.059) (0.061)
Average wage Yes Yes Yes No No
Minimum wage No Yes Yes No No
SEZ No No Yes No No
State Yes Yes Yes Yes No
Year Yes Yes Yes Yes No
Sector Yes Yes Yes Yes No
State time-trend No No No No Yes
Observations 11561 11561 11561 11561 11561
Standard errors clustered at sector and state level in parentheses.
* p < 0.1, ** p < 0.05, *** p < 0.01 All regressions control for the average share
of new firms in the state (not reported).

21
 Column 3 controls for changes in minimum wages in polluting sectors
(source: IndiaStat), while column 4 includes an additional variable indicat-
ing the presence of sector-specific special economic zones (source: Ministry
of Commerce and Industry, Department of Commerce). It is important to
control for the creation of new special economic zones (SEZ) as they pro-
vide notable advantages, such as tax exemptions, to new firms. India is
one the first country to recognize the importance of SEZs and the first zone
was created in 1965. The Indian government passed the SEZ act in 2005 in
order to increase investors’ confidence. The bill was implemented in 2006
and brought about a simplification of the bureaucratic procedures. While
most of the SEZs involve only the information technology sectors (55%) it is
still important to consider those that affected polluting sectors (about 20%
including also general multi-product SEZs) as they could induce possible
confounding effects if omitted. The inclusion or exclusion of both variables
has no significant effect on our main results.
The results obtained so far have helped to deal with the concern that
variations in environmental enforcement across states could proxy for differ-
ences in other state level unobservable characteristics. It is, however, pos-
sible that some differences in the policy environment remain unmeasured.
The results reported in column 5 include state-specific time trends and iden-
tify the effect of a change in the EIA process deviating from the pre-existing
state-specific trends. The effect of the 2006 EIA reform is still apparent and
mostly unchanged.
Table 7 reports the results obtained using the Poisson model. Also these
results support our central hypothesis: firm births in sector affected by
the decentralization process are more negatively affected by the strength of
environmental enforcement after the implementation of the reform, while no
changes are observed for non-affected sectors. The results are robust to the
inclusion of the additional control variables considered previously in table 6
and reported in column 2, 3, 4 and 5.
The coefficient related to those firms subject to state-level EIA if of
small capacity (Capacity) is now negative and statistically significant in one
specification (column 4). It is possible to further disaggregate this group of
firms by considering the classification of companies provided in the Orbis
database. This classification allows to distinguish firms into small, medium
and large and is based on annual turnover, total assets or total number of
employees. Although this classification is not intended to measure capacity
it can provide a reasonable approximation. The results reported in table 13
of the Appendix show that the effect is indeed much larger for small firms,
i.e. those more likely to be subject to state-level EIA. In column 2 of table

22
Table 7: Poisson Pseudo-maximum likelihood model: base specification and
additional control variables

(1) (2) (3) (4) (5)

Baseline: No EIA 0.002 0.003 0.003 0.001 0.004
(0.038) (0.038) (0.038) (0.038) (0.040)
After x Aj x Central -0.007 -0.027 -0.029 -0.032 -0.038
(0.056) (0.058) (0.058) (0.059) (0.057)
After x Aj x Capacity -0.102 -0.113 -0.113 -0.153∗∗ -0.078
(0.079) (0.073) (0.073) (0.073) (0.058)
After x Aj x IE -0.119∗∗∗ -0.113∗∗ -0.114∗∗ -0.126∗∗ -0.089∗
(0.041) (0.049) (0.050) (0.055) (0.050)
After x Aj x State -0.114∗ -0.103∗ -0.108∗ -0.110∗ -0.099∗
(0.059) (0.061) (0.057) (0.058) (0.060)
Average wage No Yes Yes Yes Yes
Minimum wage No No Yes Yes Yes
SEZ No No No Yes Yes
State time-trend No No No No Yes
Observations 24900 24900 24900 24900 24900
Robust (clustered by sector) standard errors in parentheses
∗
p < 0.1, ∗∗ p < 0.05, ∗∗∗ p < 0.01. All regressions control for the
average share of new firms in the state (not reported).

13 the results are obtained using the Poisson estimator and show that the
negative effect previously attributed to the category ”Capacity” is entirely
capture by small firms. This confirms once again that the observed negative
effects are attributable to the decentralization process that has affected only
some polluting firms (Category B) established after 2007.
As noted earlier, no changes are observed for firms belonging to non-
polluting sectors. While this finding supports the underlying hypothesis,
this control group includes a large number of sectors and, although on av-
erage the effect is close to zero, it could conceal significant individual sector
responses. To provide further evidence in support to our hypothesis, the
effect is estimated separately for each of the 18 non-polluting sectors at two-
digit level. The extended model was estimated using both the linear and
Poisson estimators. A negative and significant coefficient was found only
for the tobacco (-0.214) and the office, accounting and computing machin-
ery (-0.371) sectors. It is reasonable to expect these two sectors to not be
indirectly affected by the reform as their appear to be quite disconnected
from category A sectors. The negative effects could, instead, be associated
to pre-existing trajectories in birth rates that will be analyzed in the next
section.

23
 Among the 18 sectors mentioned above we consider also the automobile
industry. This sector was initially included in the draft EIA reform, officially
circulated in September 2005, but later removed from the final version of
the notification. Because all projects in the automobile sector were assigned
to state-level authorities, the fact that we do not find any effect of the
intervention suggests that there were no anticipatory effects of 2006 EIA
reform.

6.1 Time-varying effects

The results reported above are obtained by comparing the average number
of births before the implementation of the reform with the average number
of births after the reform, conditional on the level of enforcement in each
state. While a significant difference would indicate that on average the
pattern of births has changed across the two periods, the change could also
be attributed to pre- or post-reform events that can significantly affect the
average number of new firms. In this section we explore the effects that
differences in the strength of environmental enforcement at state level have
in each point in time. This is done by interacting the enforcement index
with year dummies.

Figure 1: Impact of enforcement index over time

The results are summarized in figure 1 and are obtained using a linear

24
specification with a log-dependent variable applied separately to each group
of sectors. The graph reports the coefficients of the interaction terms for each
of the 5 groups of sectors. Statistical significance at a level lower than 10%
is indicated by a grey circle. Considering those sector of exclusive compe-
tence of state-level authorities (State), the results show that environmental
enforcement to have a significant negative effect from the end of 2007, while
for the period prior to the reform the effect is often close to zero or not
statistically significant. The drop after the implementation of the reform
is large and the coefficients remain negative and significant for the entire
post-reform period. Similarly, the coefficients related to those companies
subject to state-level EIA if located within an industrial estate (IE) show
a significant drop between 2006 and 2008. The coefficients remain negative
thereafter although not always significant. No significant differences are ob-
served for the group labeled ”Capacity”, probably for the reasons already
mentioned above, and for non-polluting firms.

Figure 2: Impact of enforcement index over time

Figure 2 reports the results of a similar analysis conducted for the sectors
tobacco and office, accounting and computing machinery that were found to
exhibit a significant drop in average births after the implementation of the
EIA reform. The graph reveals a negative trend in births beginning in 2003,
two years before the implementation of the reforms for the tobacco sector. In

25
the case of the machinery sector, in particular, the coefficients remain almost
stable before and after the implementation of the reform. This indicates that
the effects previously observed for these two sectors cannot be attributed to
the EIA reform and confirms our hypothesis that, while the decentralization
of the EIA process has affected firm births in polluting sectors subject to
state-level EIA, no unintended effects are observed on non-polluting sectors.

7 Conclusions
The decentralization of environmental regulation and enforcement has the
potential to lead to a more efficient and transparent system of pollution
control and management. On the other hand it carries the risk of inducing
a race-to-the bottom competition among local authorities and to suffer from
the consequences of heterogeneous financial and technical constraints across
the territory.
This paper has investigated whether the decentralization of the EIA
process in India initiated in 2006 has produced unintended effects on the lo-
cation decisions of polluting firms. The results have shown that the reform
has produced a significant change in the number of firm births in polluting
sectors leading to relatively higher birth rates in states with lower enforce-
ment levels. The results are robust to different specifications and are not
driven by other pre- or post- reform shocks.
These findings show that firms are responsive to variations in environ-
mental enforcement within India but are not intended to prove that states
actually use environmental stringency as a competitive instrument since in-
stitutional, technical and financial constraints also play an important role in
determining the effectiveness of environmental enforcement. Moreover, the
results neither suggest that a centralized approach (as the one in place before
the reform) would be optimal since no attempts were made to compare the
two enforcement models. The net social welfare effects of the 2006 reform
were not quantified and room is left for further research on the environmen-
tal and socio-economic consequences of the environmental decentralization
process in India.

26
Appendix

Table 8: Activities listed in the EIA notification, their categorization and

the corresponding NIC classification

27
Table 8 – continued from previous page

28
Table 9: Date of constitution of State EIA Authority

State Date of constitution of SEIAA

Andhra Pradesh 4th July 2007
Arunachal Pradesh 27th March 2008
Bihar 7th February 2011
Chandigarh 21st August 2009
Chhattisgarh 9th January 2008
Dadra and Nagar Haveli 11th October 2007
Daman & Diu 11th October 2007
Delhi 30th July 2008
Goa 15th April 2010
Gujarat 12th June 2007
Haryana 21st April 2008
Himachal Pradesh 11th October 2007
Jammu and Kashmir 8th January 2008
Jharkhand 20th December 2012
Karnataka 11th June 2007
Kerala 3rd November 2011
Madhya Pradesh 8th January 2008
Maharashtra 21st April 2008
Meghalaya 23rd July 2007
Nagaland 15th April 2010
Orissa 17th November 2008
Puducherry 13th December 2007
Punjab 19th November 2007
Rajasthan 30th July 2008
Sikkim 8th July 2008
Tamil Nadu 3rd March 2008
Uttar Pradesh 12th July 2007
Uttarakhand 22nd September 2008
West Bengal 13th April 2007

29
 Table 10: Results obtained employing a negative binomial model

(1) (2) (3) (4)

Baseline: No EIA 0.046∗ 0.047∗ 0.047∗ 0.029
(0.028) (0.028) (0.028) (0.029)
After x Aj x Central 0.017 0.016 0.018 0.008
(0.057) (0.058) (0.059) (0.073)
After x Aj x Large 0.002 0.000 -0.007 0.006
(0.059) (0.058) (0.060) (0.064)
After x Aj x District -0.131∗ -0.130∗ -0.131∗ -0.111
(0.075) (0.075) (0.073) (0.087)
After x Aj x State -0.095∗ -0.096∗ -0.101∗∗ -0.091
(0.054) (0.054) (0.051) (0.059)
inflate
Constant 1.387∗∗∗ 1.377∗∗∗ 1.379∗∗∗ 1.382∗∗∗
(0.137) (0.135) (0.135) (0.135)
lnalpha
Constant -0.621∗∗∗ -0.626∗∗∗ -0.627∗∗∗ -0.629∗∗∗
(0.082) (0.082) (0.082) (0.082)
Observations 21075 21075 21075 21075
Sector-clustered standard errors in parentheses
∗
p < 0.1, ∗∗ p < 0.05, ∗∗∗ p < 0.01

Table 11: Sectors excluded from the analysis because of ambiguities in their
classification

NIC code Description

103 Extraction and agglomeration of peat [incl. digging of peat]
181 Manufacture of wearing apparel, except fur apparel
192 Manufacture of footwear.
261 Manufacture of glass and glass products
281 Manufacture of structural metal products, tanks, reservoirs and steam generators
289 Manufacture of other fabricated metal products; metal working service activities
2410 Manufacture of basic chemicals
2420 Manufacture of other chemical products
2424 Manufacture of soap and detergents
2429 Manufacture of other chemical product n.e.c.
2102 Manufacture of corrugated paper
2109 Manufacture of other articles of paper

30
 Table 12: Number of new firms by category and state

State Without EIA Large Central-level EI State-level Total

Andhra Pradesh 3281 326 4909 1406 259 10181
Assam 301 52 488 27 18 886
Bihar 454 30 1733 124 17 2358
Chandigarh 326 16 325 180 22 869
Chhattisgarh 265 56 1041 49 13 1424
Delhi 11984 512 12018 1823 605 26942
Goa 207 14 258 28 12 519
Gujarat 6200 803 4110 1346 215 12674
Haryana 1349 61 743 200 59 2412
Himachal Pradesh 134 10 514 83 14 755
Jammu & Kashmir 123 12 177 24 5 341
Jharkhand 242 29 1053 39 7 1370
Karnataka 3474 169 3291 407 161 7502
Kerala 1179 96 1639 274 55 3243
Madhya Pradesh 1075 164 2065 387 73 3764
Maharashtra 12822 875 11306 2061 651 27715
Meghalaya 48 13 170 7 4 242
Odisha 531 78 2165 91 27 2892
Puducherry 89 6 86 33 9 223
Punjab 1202 78 858 197 61 2396
Rajasthan 1919 186 4711 435 75 7326
Tamil Nadu 5185 266 3898 652 269 10270
Uttar Pradesh 2121 135 1776 668 92 4792
Uttarakhand 217 19 163 50 13 462
West Bengal 4095 305 3897 404 156 8857
Total 58823 4311 63394 10995 2892 140415

31
Table 13: Small versus large companies in the ”Capacity” group

(1) (2)
OLS Poisson
Baseline: No EIA 0.038 0.001
(0.047) (0.038)
After x Aj x Central -0.001 -0.032
(0.072) (0.059)
After x Aj x Large 0.003 -0.001
(0.050) (0.049)
After x Aj x Small -0.058 -0.261∗
(0.095) (0.136)
After x Aj x IE -0.150∗∗∗ -0.126∗∗
(0.051) (0.055)
After x Aj x State -0.179∗∗∗ -0.110∗
(0.059) (0.058)
Observations 12596 30000
Average wage Yes Yes
Minimum wage Yes Yes
SEZ Yes Yes
State Yes Yes
Year Yes Yes
Sector Yes Yes
Clustered standard errors in parentheses.
∗
p < 0.1, ∗∗ p < 0.05, ∗∗∗ p < 0.01

32
References
Barrios, S., H. Huizinga, L. Laeven, and G. Nicodme (2012). International
taxation and multinational firm location decisions. Journal of Public Eco-
nomics 96 (11), 946–958.

Becker, R. and V. Henderson (2000). Effects of air quality regulations on

polluting industries. Journal of Political Economy 108 (2), 379–421.

Besley, T. and S. Coate (2003). Centralized versus decentralized provision

of local public goods: a political economy approach. Journal of Public
Economics 87 (12), 2611 – 2637.

Brülhart, M., M. Jametti, and K. Schmidheiny (2012). Do agglomeration

economies reduce the sensitivity of firm location to tax differentials? The
Economic Journal 122 (563), 1069–1093.

Brunnermeier, S. B. and A. Levinson (2004). Examining the evidence on

environmental regulations and industry location. The Journal of Envi-
ronment & Development 13 (1), 6–41.

Cistulli, V. (2002). Environment in Decentralized Development: Economic

and Institutional issues. Training Materials for Agricultural plannings no.
44. FAO, Rome.

CMS (2005). India Corruption Study 2005 to improve governance. Trans-

parency Internation India.

Dean, J. M., M. E. Lovely, and H. Wang (2009, September). Are foreign

investors attracted to weak environmental regulations? evaluating the
evidence from China. Journal of Development Economics 90 (1), 1–13.

Duvivier, C. and H. Xiong (2013, 3). Transboundary pollution in china: a

study of polluting firms’ location choices in hebei province. Environment
and Development Economics FirstView, 1–25.

Government of India (Various years (1998-2012)). Annual survey of indus-

tries. Technical report, Ministry of Statistics and Programme implemen-
tation, Central Statistics Office.

Holl, A. (2004, May). Manufacturing location and impacts of road transport

infrastructure: empirical evidence from spain. Regional Science and Urban
Economics 34 (3), 341–363.

33
Javorcik, B. S. and W. Shang-Jin (2003). Pollution havens and foreign direct
investment: Dirty secret or popular myth? The B.E. Journal of Economic
Analysis & Policy 3 (2), 1–34.

Kathuria, V. (2007). Informal regulation of pollution in a developing coun-

try: Evidence from India. Ecological Economics 63 (23), 403 – 417.

Lall, S. V. and S. Chakravorty (2005). Industrial location and spatial in-

equality: Theory and evidence from India. Review of Development Eco-
nomics 9 (1), 47–68.

Lipscomb, M. (2008). The erfect of environmental enforcement on product

choice and competition. theory and evidence from india. Working pa-
per no. 08-13, Center for Economic Analysis, University of Colorado at
Boulder.

Lipscomb, M. and A. M. Mobarak (2007). Decentralization and the politi-

cal economy of water pollution: Evidence from the re-drawing of county
borders in brazil *. Technical Report 67, IPC Working Paper Series.

List, J. A., D. L. Millimet, P. G. Fredriksson, and W. W. McHone (2003).

Effects of environmental regulations on manufacturing plant births: Ev-
idence from a propensity score matching estimator. The Review of Eco-
nomics and Statistics 85 (4), pp. 944–952.

Mani, M., S. Pargal, and M. Huq (1997). Is there an environmental ”race to

the bottom”? Evidence on the role of environmental regulation in plant
location decisions in India. Technical report, The World Bank.

Oates, W. E. (2002). A reconsideration of environmental federalism. In

L. John A. and A. de Zeeuw (Eds.), Recent advances in Environmental
Economics. Edgar Elgar.

OECD (2006). Environmental compliance and enforcement in India: Rapid

assessment. Technical report, OECD, Paris.

Panigrahi, J. K. and S. Amirapu (2012). An assessment of EIA system in

India. Environmental Impact Assessment Review 35 (0), 23 – 36.

Sigman, H. (2005). Transboundary spillovers and decentralization of envi-

ronmental policies. Journal of Environmental Economics and Manage-
ment 50 (1), 82 – 101.

34
UNESCAP (2000). State of environment in asia and the pacific 2000. Tech-
nical report, Asian Development Bank and United Nations.

Wooldridge, J. M. (1991). Specification testing and quasi-maximum-

likelihood estimation. Journal of Econometrics (48), 29–55.

World Bank (2006). India, strengthening institutions for sustainable growth:

country environmental analysis. Technical report, Report No. 38292-IN,
South Asia Environment and Social Development Unit, the World Bank,
Washington D.C.

35