World Development Vol. 34, No. 7, pp.

1254–1270, 2006
Ó 2006 Elsevier Ltd. All rights reserved
0305-750X/$ - see front matter
www.elsevier.com/locate/worlddev
doi:10.1016/j.worlddev.2005.12.002

The Impact of Collective Action

on Economic Development: Empirical Evidence
from Kerala, India
PATRICIA JUSTINO *
University of Sussex, Falmer, Brighton, UK
Summary. — Public participation is generally assumed to be a necessary if not fundamental con-
dition for development. Most studies do not, however, question the kind and level of collective
action that would be desirable. This paper provides a novel empirical analysis of the impact of
collective action on economic development using data for South India. The paper models collective
action as endogenous to economic development and distinguishes between its static and dynamic
properties. The results show that while excessive activism may harm state income, collective action
in the context of responsible bargaining systems may contribute toward increased economic devel-
opment.
Ó 2006 Elsevier Ltd. All rights reserved.

Key words — collective action, economic development, time series, panel data, Kerala, India

1. INTRODUCTION level of participation is desirable, and very little

Collective action is typically understood as
a concerted effort by individuals or groups
to attain certain objectives (see Tarrow, 2003).
This is the view adopted in public economics
(see Esping-Andersen, 1990; Olson, 1971),
work on the action of labor unions (see Free-
man & Medoff, 1984), interest group theories
(Buchanan & Tullock, 1962), research on com-
mon resource management (see Meinzen-Dick
& Di Gregorio, 2004), work on collective vio-
lence (e.g., McAdam, 1982; Tilly, 1993, 2003),
theories of democracy formation (e.g., Prze-
worski, 1985) and recent research on public
action if there is development (Drèze & Sen,
1991). In particular, the participation of the
poor in the process of development is viewed
by many as an essential element of successful
development strategies (Ahmad, Drèze, Hills,
& Sen, 1991; Drèze, Sen, & Hussain, 1995;
World Bank, 2001). Participation in social
and political decisions provides individuals
with a sense of value and identity and is an
important means to voice the needs of vulnera-
ble population groups. However, the develop-
ment literature rarely asks what type or what
1254
work exists that establishes empirically the
impact of public participation and collective ac-
tion on economic development. 1 In particular,
there is an absence of empirical evidence on
whether that impact is direct or whether it de-
pends on specific transmission mechanisms,
* I wish to thank Rathin Roy, Chris Cramer, Barbara
Harriss-White, Ron Herring, Julie Litchfield, Subir
Sinha, Manny Teitelbaum, Alan Winters, and three
anonymous referees for their very useful comments and
discussions, M.D. Asthana, Hiranya Mukhopadhyaya,
B. Narayanan, and G. Omkarnath for inestimable assi-
stance during fieldwork in India and S.R. Meena, at the
National Crime Records Bureau, and V. Parameswa-
ram, at the Ministry of Labor, for their assistance and
permission to access various data records. I am also
grateful to Professor Chandan Mukherjee at the Centre
for Development Studies at Thiruvananthapuram,
Kerala, for his hospitality during the initial stages of
this project. This paper has received financial support
from the British Academy (PDF/2002/288). Final revi-
sion accepted: December 8, 2005.
 THE IMPACT OF COLLECTIVE ACTION ON ECONOMIC DEVELOPMENT
what the static and dynamic properties of those
mechanisms are and whether collective actions
act as an external shock to the economy or
are endogenous to the process of economic
development. This paper addresses these
gaps in the literature using empirical evidence
from South India. The paper compares and
contrasts two concrete forms of collective
action (industrial strikes and politically moti-
vated riots) used as forms of public activism
to exert political pressure. The paper revisits
the ‘‘development model’’ of the south Indian
state of Kerala and asks whether and how col-
lective actions affect the development of poor
economies.
The choice of Kerala as a case study was
motivated by the unique characteristics of its
development strategy, where thanks to the
actions of organized public demands, high lev-
els of social development coexist with very low
incomes. This paper builds on the large litera-
ture on Kerala’s ‘‘development model’’ (Franke
& Chasin, 1992; Gulati, 1990; Heller, 2000;
Kannan, 1995; Oommen, 1993; Pillai, 1994;
Ramachandran, 1997). In addition, the paper
contributes to the literature on the impact of
collective action and participation on develop-
ment (e.g., Ahmad et al., 1991; Meinzen-Dick
& Di Gregorio, 2004; World Bank, 2001), by
providing a rigorous empirical analysis of
important endogenous, dynamic characteristics
of collective action.
Public participation and collective action are
likely to be influenced by the process of eco-
nomic development of each given society. At
the same time, forms of collective action will af-
fect the way governments respond to popula-
tion demands which, in turn, will impact on
the type of development policies implemented
(see Esping-Andersen, 1990; Olson, 1971;
Przeworski, 1985; Tarrow, 2003; Tilly, 2003).
We have empirically tested this endogeneity
assumption for Kerala and India with positive
results. We also distinguish between static
and dynamic impacts of collective action.
Depending on their intensity, some forms of
collective action may have immediate negative
or positive effects on the economy. Immediate
negative effects include, for instance, the
destruction of private or public property,
increases in economic uncertainty and in the
risk of private investment (Alesina & Perotti,
1996). Immediate positive effects can take place
when governments respond to population de-
mands by conceding social subsidies to the
1255
poor as a form of pursuing electoral advantages
(see Esping-Andersen, 1990). These effects may
change with time. Initial negative effects may,
in the long term, lead to increases in aggre-
gate national income if collective actions result
in considerable economic gains for a signifi-
cant number of disadvantaged individuals
(Esping-Andersen, 1990; Gurr, 1970; Tilly,
1993). Similarly, immediate positive effects can
be offset in the long term if forms of collective
action cause excessive turmoil and persistent
high levels of socio-economic instability
(see Freeman & Medoff, 1984). We found
evidence for divergence between static and dy-
namic effects of collective action in Kerala
and India.
The paper is organized as follows. Section 2
briefly reviews the main characteristics of Ker-
ala’s development process over the last three
decades. Section 3 discusses the role and evol-
ution of organized collective movements in
Kerala and their impact on state income since
the early 1970s. In Section 4, we analyze
empirically the relationship between collective
action and Kerala’s economic development,
using time-series analysis techniques. This
empirical analysis is based on state-level data
on various types of organized collective actions
discussed in Section 2. Collective action is
modeled as endogenous to the process of eco-
nomic growth in Kerala and the endogeneity
assumption is explicitly tested. In Section 5,
the results for Kerala are compared with those
for a panel of 14 major Indian states. The re-
sults of this comparative analysis provide
important insights into the relationship be-
tween collective action and economic develop-
ment in poor economies. They suggest that
in the presence of well functioning and
responsible bargaining systems, some forms
of collective action may have an important
endogenous role to play in the reduction of
poverty and the promotion of economic
growth. Section 6 summarizes the evidence
and concludes the paper.
2. KERALA’S ‘‘MODEL OF
DEVELOPMENT’’
The state of Kerala, in South India, has occu-
pied an eminent position in the development
debate since the early 1970s due to its singular
development strategy. Unlike most developing
economies, Kerala’s policy makers followed a
1256 WORLD DEVELOPMENT

Table 1. Quality of life indicators, poverty and state income in Kerala

Kerala India Low-income countries United States
Birth rate (per 1,000) (1995) 18.0 28.3 39.8 22.8a
Death rate (per 1,000) (1995) 6.0 9.0 9.0 5.9a
Infant mortality rate (per 1,000) (1996) 13.0 72.0 69.0b 6.0b
Life expectancy (1989–93) 72.0 59.4 63.0b 77.0c
Literacy rate (%) (1991)
Male 93.6 64.1 46.0 >95.0
Female 86.2 39.3 35.0 >95.0
Poverty rate (%) (1973–74)
Rural 59.2 55.7 – –
Urban 62.7 48.0 – –
Poverty rate (%) (1999–2000)d
Rural 9.4 26.8 – –
Urban 20.3 24.1 – –
State domestic producte
1970 1,325 1,387f – –
1980 1,508 1,579f – –
1990 1,815 2,112f – –
1999 2,591 2,876f – –
Source: Government of India (1998), Government of Kerala (1996), World Bank (1997, 2001), United Nations
(1997); Government of India, National Accounts Statistics. Poverty rates (headcount indices) from NSSO (various
years).
a
Refers to 1994.
b
Refers to 1997.
c
Refers to 1999.
d
The 1999–2000 poverty estimates refer to the 30-day recall period.
e
Refers to per capita state domestic product at factor cost at 1980–81 constant prices in rupees.
f
Refers to average for 14 major Indian states: Andhra Pradesh, Assam, Bihar Gujarat, Karnataka, Kerala, Madhya
Pradesh, Orissa, Punjab, Rajasthan, Tamil Nadu, Uttar Pradesh, and West Bengal.

successful ‘‘basic-needs-first’’ strategy that pri-
oritized the improvement of socio-economic
standards of its population (Table 1) and
implemented extensive land reforms (Herring,
1983). This has resulted in remarkable improve-
ments in living standards and sharp decreases
in poverty in Kerala (Table 1), both in the orga-
nized and unorganized sectors (Gulati, 1990;
Heller, 2000). This development strategy has
been largely accredited to the actions of effec-
tive collective movements, both in the orga-
nized and unorganized sectors, which have
resulted in a very successful interaction between
public demand and a pro-active state (Heller,
2000; Prakash, 1994).
Despite these successes, state income in Ker-
ala is low even in relation to the Indian average
(Table 1). Low economic growth has been
accompanied by weak performances in all the
economic sectors (especially, the agriculture
and manufacturing sectors), and has been
aggravated by a fiscal crisis and increases in
unemployment (particularly, amongst the edu-
cated population) in the state since the mid-
1970s. 2 Economic difficulties have, in turn,
started to have significant negative effects on
the implementation of further social redistribu-
tive policies (George, 1993).
Kerala’s disappointing economic perfor-
mance has been largely attributed to the
strength of militant movements (George,
1998; Kannan, 1990; Oommen, 1993; Prakash,
1994; Tharamangalam, 1998). At the same time
that organized forms of collective action have
resulted in extraordinary social developments,
they have also led to demands for unsustainable
increases in social protection expenditure and
higher wages, 3 and to increased risk of disrup-
tion, which have discouraged private invest-
ment and have led companies to establish
themselves in other Indian states with more sta-
ble labor markets (Nair, 1994; Sankaranaraya-
nan & Bhai, 1994). These accounts highlight
interesting questions concerning the relation-
ship between collective action and economic
development in Kerala.
 THE IMPACT OF COLLECTIVE ACTION ON ECONOMIC DEVELOPMENT
3. COLLECTIVE ACTION AND
DEVELOPMENT IN KERALA
Forms of collective action often take place in
Kerala via the state’s democratic electoral pro-
cess (see Heller, 1995). Since the formation of
the state in 1956, the political scene in Kerala
has been dominated by two main parties: the
Indian National Congress (INC) and the Com-
munist Party of India (CPI). In 1964, the latter
party split into two parties: CPI and CPI-
Marxist. Left-wing governments dominated
Kerala’s politics in the late 1960s and late
1970s, while the Congress party dominated
the period during 1970–77. After 1977, no sin-
gle party controlled the state’s political arena.
Since then Kerala’s population has voted alter-
natively for left- and right-wing parties every
5 years (Electoral Commission of India). This
process has maintained alive both left- and
right-wing governments’ commitment to the
state’s model of development.
This historical electoral process has simulta-
neously led to and resulted from structured col-
lective movements, both in the organized and
unorganized sectors, in the form of militant
peasant associations and labor unions (Heller,
1995, 1996, 2000; Herring, 1983; Nair, 1994).
The role of labor unions in Kerala illustrates
familiar questions about the economic effi-
ciency of organized labor movements (Freeman
& Medoff, 1984). Labor unions have consti-
tuted, in most economies, an important pres-
sure group in the establishment of labor
standards. They have also been used as an
important means for workers to voice their
demands and needs. However, it has not
been clearly established whether labor unions
encourage increases in productivity, promote
democracy in the work place and protect the
interests of the more vulnerable population
groups, or whether unions act as rent-seeking
institutions that protect their own interests.
Kerala’s experience with labor movements
throughout the last three decades illustrates
both views.
Until the mid-1960s industrial relations were
relatively better in Kerala than in most other
states, thanks to the democratic relations and
industrial culture trade unionism (especially in
the unorganized sector) which helped to insti-
tute. At that time, the role of peasant associa-
tions and labor unions was central to the
design and implementation of land reforms,
minimum wage legislation, institutionalized
bargaining procedures and general welfare
1257
measures that empowered the lower classes,
including the promotion of significant social
protection measures in the unorganized sector
(Gulati, 1990; Heller, 1996). However, after
the split of the Communist Party in 1964, trade
unionism tactics started to rely on intimidating
actions, go-slow practices, short-duration wild-
cat strikes and unofficial strikes, rather than
concern with longer term issues such as unem-
ployment, declining industries or increase in
wage inequalities (Nair, 1994). This was fuelled
by a rapid process of unionization and an in-
crease in the number of trade unions, as well
as the emergence of caste and category-craft
based unions and sectorial pressure groups,
which led to unhealthy forms of competition
between unions, inter-union rivalry and the
consequent division of the trade union move-
ment in Kerala (Heller, 1995; Nair, 1994). As
a result, the number of trade unions in Kerala
is rather large in comparison to the rest of India
(Table 2).
Labor market instability in Kerala decreased
in the late 1970s and throughout the 1980s,
thanks to ‘‘politics of class compromise’’
(Heller, 1996). 4 This progression is illustrated
in Figure 1. This figure provides a comparison
of the number of strikes and lockouts per
1,000 factory workers for every year during
1973–99 in Kerala and India. Strikes and lock-
outs are defined as work stoppages that involve
10 or more workers (in the formal sector),
whether directly and/or indirectly, and exclude
closures not related with industrial disputes (In-
dian Labour Statistics, 1994, p. 199). The data
include the number of both strikes nd lock-
outs. 5 Figure 1 shows a pronounced decrease
in the number of strikes and lockouts per
1,000 factory workers in Kerala since the early
1990s. Nonetheless, in 1999, Kerala still regis-
tered some of the highest numbers of strikes
and lockouts per 1,000 factory workers in India
(see also Table 2). 6
Although data on strikes provide a good
illustration of the degree of industrial action
in Kerala, the data are likely to underestimate
the true extent of collective action in other sec-
tors of society. Another way of assessing the ex-
tent of collective action in Kerala is to examine
the extent of politically motivated riots in the
state. The Indian Penal Code defines riots as
collective acts of violence by an assembly of five
or more people directed against government
institutions, law and order, other people, and
property. 7 This is the definition underlying
the data used in this paper. 8
1258 WORLD DEVELOPMENT

Table 2. Collective action in selected Indian states, 1973 and 1999a

Workers’ unionsb Strikes and Riotsd
lockoutsc
1973 1999 1973 1999 1973 1999
Andhra Pradesh 5.17 27.7 0.46 0.13 0.07 0.04
Assam 3.44 6.94 0.21 0.15 0.23 0.14
Bihar 3.65 3.75 1.16 0.04 0.20 0.11
Gujarat 1.49 2.54 0.34 0.13 0.02 0.04
Karnataka 2.79 3.79 0.34 0.04 0.09 0.15
Kerala 15.0 22.2 1.00 0.14 0.26 0.21
Madhya Pradesh 3.33 0.38 0.61 0.02 0.06 0.04
Maharashtra 3.03 4.48 0.76 0.01 0.04 0.06
Orissa 4.56 2.07 0.56 0.11 0.08 0.04
Punjab 4.95 1.56 0.43 0.04 0.00 0.00
Rajasthan 8.39 7.81 0.51 0.07 0.20 0.31
Tamil Nadu 4.79 4.79 0.71 0.15 0.13 0.09
Uttar Pradesh 4.43 3.49 0.55 0.05 0.15 0.04
West Bengal 7.58 9.90 0.48 0.21 0.21 0.06
Source: Data on strikes and lockouts and workers’ unions from Government of India, Ministry of Labour, Annual
Report (various years), Government of India, Ministry of Labour, Indian Labour Yearbook (various years),
Government of India, Ministry of Labour, Indian Labour Statistics (various years). Data on riots from the Gov-
ernment of India, National Crime Records Bureau, Crime in India (various issues).
a
This table includes descriptive statistics for 1973 and 1999 for illustrative purpose only. The analysis in the paper is
based on a time-series data for every year during 1973–99.
b
Workers unions = no. of workers’ unions per 1,000 factory workers.
c
Strikes and lockouts = volume of strikes and lockouts per 1,000 factory workers.
d
Riots = volume of riots per 1,000 people.

1.2
Kerala
India
1
Strikes per 1000 workers

0.8

0.6

0.4

0.2

0
73

75

77

79

81

83

85

87

89

91

93

95

97

99
19

19

19

19

19

19

19

19

19

19

19

19

19

19

Figure 1. Number of strikes and lockouts per 1,000 factory workers in India and Kerala, 1973–99. Source: Government
of India, Ministry of Labour, Annual report (various years), Government of India, Ministry of Labour, Indian labour
yearbook (various years), and Government of India, Ministry of Labour, Indian labour statistics (various years).
 THE IMPACT OF COLLECTIVE ACTION ON ECONOMIC DEVELOPMENT 1259

Not all riots need necessarily be politically
motivated. However, riots in India (and Kera-
la) have often been characterized by more or
less violent public manifestations driven by eco-
nomic and social discontent, as well as specific
political agendas by competing political parties
(Brass, 1997, 2003; Das, 2003; Heller, 1995,
2000; Joseph, 1999; Prabhash, 2000; Wilkinson,
2004). 9 In some of the most influential studies
of rioting in India, Paul Brass (1997, 2003) has
shown how the dynamic process of riot produc-
tion in India resembles well-established theories
of collective action by Olson (1971), Tarrow
(2003), Tilly (1993, 2003), and McAdam
(1982), in that ‘‘riots have become a common
and even anticipated form of collective action’’
(Brass, 2003, p. 12), and the means to pursue
political advantage (see alsoEngineer, 1984,
2004). This is very much the case in Kerala
(e.g., Heller, 1995; Prabhash, 2000). In recent
years, the media has reported increased levels of
religion- and caste-based rioting. 10 Although
the data provided by the NCRB do not allow
the disaggregation of riots according to differ-
ent motivations, these are likely to constitute
a small percentage of total rioting in Kerala,
where levels are lower than in other Indian
states (see Varshney, 2002, Chapter 4). In the
specific context of Kerala (and India), riots as
reported in the NCRB data are thus likely to
be a good representative variable of levels of
collective action.
Kerala has some of the highest number of
riots per 1,000 people in India (Table 2). This
is partly due to the fact that riots are likely to
be more accurately registered in Kerala than
in other states, given the state’s higher levels
of socio-political development, media interven-
tion and government accountability. However,
it is also noticeable that the number of riots
per 1,000 people in Kerala has been more or
less constant since 1973 around a mean value
of 0.21 (with a small standard deviation of
0.03). Values after 1988 have been consistently
higher than the values during the 1970s and
the earlier part of the 1980s (Figure 2). This
suggests that although workers’ unions and
the government have reached important com-
promises, forms of organized collective action
are still pronounced and are likely to both re-
flect and impact on Kerala’s social and eco-
nomic performance.
4. ECONOMETRIC ANALYSIS
Kerala is typically portrayed as an example
of how poor economies can address impor-
tant individual basic needs without requiring
large budgets. However, as discussed above,

0.3

0.25
Riots per 1000 people

0.2

0.15

0.1

0.05
Kerala
India
0
73

75

77

79

81

83

85

87

89

91

93

95

97

99
19

19

19

19

19

19

19

19

19

19

19

19

19

19

Figure 2. Number of riots per 1,000 people in India and Kerala, 1973–99. Source: Government of India, National Crime
Records Bureau, Crime in India (various issues).
 1260
Table 3. Marginal effects for Keralaa
Dep var: NSDPb Dep var: Dep var: Dep var: Dep var: Dep var: Dep var: Dep var: Hud
[3SLS] (1) NSDPb (2) NSDPb NSDPb (4) NSDPb NSDPb (6) Hrc [3SLS] (7) [3SLS] (8)
[3SLS] (3) [3SLS] (5)
State income 0.644 0.062
Lagged state income 0.019 0.028 0.027 0.025 0.006 0.046** 0.435 0.571
Volume of riots 0.311*** 0.018 0.279*** 0.015 0.165*** 0.016 2.144** 1.082
Lagged volume riots 0.009 0.016 0.014 0.025** 0.946 0.241
Volume of strikes 0.026 0.001 0.030 0.002 0.010 0.006 0.563* 0.219
Lagged volume of strikes 0.009 0.003 0.012** 0.006 0.313** 0.206
Expenditure on 0.067*** 0.022* 0.074*** 0.023* 0.060*** 0.027*** 0.906** 0.262

WORLD DEVELOPMENT
social services
Number of 0.001*** 0.001*** 0.008 0.003
workers’ unions
Change of government 0.001 0.002** 0.064 0.001
*** *** *** ***
Trade openness 0.035 0.026 0.035 0.027 0.042*** 0.032*** 0.142 0.391
Education enrolments 0.021 0.0002 0.020 0.004 0.033* 0.031*** 0.310 0.52
Rural poverty
Urban poverty
Trend variable (year) 0.0002 0.0002 0.0004 0.0002 0.005 0.0003 0.042 0.020*

R-squared 0.929e 0.993 0.949e 0.993 0.973e 0.998 0.949e 0.962e

No. of observations 24 24 24 24 24 24 24 24
a
Refers to the percentage impact of a unit change in independent variable on state income.
b
NSDP = logarithmic function of real per capita net state domestic product at 1980–81 constant prices.
c
Hr = percentage of the population below the poverty line in rural areas.
d
Hu = percentage of the population below the poverty line in urban areas.
e
This is not the ‘‘true’’ R-squared. The R-squared reported in 3SLS models is determined using the actual values of the structural model for LNSDP and not the
instruments of the endogenous variables in the right-hand side of the equation.
*
Significant at the 10%.
**
Significant at the 5%.
***
Significant at the 1%.
 THE IMPACT OF COLLECTIVE ACTION ON ECONOMIC DEVELOPMENT
although this strategy was quite successful in
the early stages of Kerala’s development, it is
not clear whether it has proved to be a sustain-
able strategy in the long term. In this section,
we examine some of the reasons behind this
outcome from a collective action perspective.
In particular, we model empirically the rela-
tionship between forms of collective action
and economic development in Kerala using a
time-series model that covers for every year
the period during 1973–99. 11 We explicitly
model collective action as endogenous to Kera-
la’s economic development, represented by the
level of state income in each year. The model al-
lows us to also distinguish between short- and
long-term effects of collective action on state in-
come. This dynamic process is described by
Y t ¼ a þ uY t1 þ bX t1 þ cZ t þ dW t þ t þ et ;
ð1Þ
where the dependent variable, Yt, represents the
logarithmic function of per capita net state
domestic product at 1980–81 constant prices.
We have allowed current levels of state income
to depend on past levels of state income. This is
a standard assumption in the economic growth
literature (Barro, 1991). a is a constant, et is the
error term, and t is the trend variable. This var-
iable is used to correct for the presence of deter-
ministic trends in the time-series variables. 12
Xt1 is a vector of endogenous lagged regres-
sors illustrating the dynamic impact of collec-
tive action on the level of state domestic
product. Collective action is represented by
the number of riots per 1,000 people and the
number of strikes and lockouts per 1,000 fac-
tory workers. 13 We have decided to use the
volume of collective action per unit of popula-
tion rather than the absolute number of strikes
and riots in order to enable, in the next section,
comparison between Kerala and other Indian
states, with very different population sizes.
Zt is a vector of endogenous independent
variables at time t. It includes the current num-
ber of riots and strikes and lockouts per unit of
population, and a proxy for the redistributive
social development program in place in Kerala.
This is given by the level of public expenditure
on social services (logarithmic function of per
capita expenditure on social services at 1980–
81 constant prices). 14 This variable is treated
as endogenous given the discussion of Kerala’s
redistributive model in Section 2.
Wt is a vector of exogenous independent vari-
ables at time t. It includes a measure for open-
1261
ness of the Indian economy, given by the
all-India ratio of imports and exports over na-
tional domestic product (per capita at 1980–
81 constant prices), to control for the increase
in trade liberalization and national economic
changes that took place in India after the
mid-1980s. 15 We expect this variable to cap-
ture the effects of economic changes not related
to collective action or redistributive decisions.
In order to control for the impact of human
capital and productivity on economic growth
(Barro, 1991), we have added a measure for
the level of education in Kerala, given by the
per capita number of individuals enroled in
primary and secondary education.
The results for the model are presented in
Table 3, columns (1), (3), and (5). For compar-
ison reasons, the specification in column (1) ex-
cludes lagged values of collective action. These
have been added to the model in column (3).
Column (5) illustrates yet another specification
of the model. In this specification, we have
added two further collective action variables
discussed in the previous section. These include
the number of trade unions (per 1,000 factory
workers) in period t. We have also incorporated
into the model a political variable, which takes
the value 1 if in that year there has been a gov-
ernment change and 0 if not. 16 This controls
for the influence of individual political parties
on economic development and collective ac-
tion. Columns (2), (4), and (6) include estimates
for the same three model specifications where
all independent variables are treated as exo-
genous. We present these results for compara-
tive reasons. 17
We have used a three-stage least-squares mod-
el (3SLS) to estimate model (1). This method
uses instrumental variables to construct consis-
tent estimates that allow for the correlation of
the explanatory variables with the disturbance
term caused by the presence of endogeneity in
some independent variables. In stage 1, we pre-
dicted values for the instrumented variables
from the regression of each of the endogenous
variables on all exogenous variables in the sys-
tem, plus three other variables in order to guar-
antee the system of equations to be identified.
These were rural and urban poverty, 18 and real
per capita development expenditure (at constant
1980–81 prices). 19 These variables provide
good instruments for both collective action and
redistribution in Kerala as, given the discussion
in the previous sections, the three endogenous
variables are likely to be closely associated with
the level of poverty in the state, as well as to
1262 WORLD DEVELOPMENT
aggregate development expenditure. Stage 2
consisted in the estimation of the covariance
matrix of the disturbances, based on the residu-
als of the two least-squares estimation of the
structural equation (1). In stage 3, we used gen-
eralized least-squares estimation procedures
using the covariance matrix estimated in stage
2 and replacing the endogenous variables in
model (1) by the instrumental variables esti-
mated in stage 1 (Greene, 2000). The results
listed in Table 3 are marginal effects. They illus-
trate the percentage change in state income that
results from a unit change in the levels of each
independent variable.
(a) Results
The results of the 3SLS estimation (column
(1), Table 3) show that the volume of riots
has a negative and statistically significant im-
pact on the state’s level of income. This result
is consistent across all model specifications (col-
umns (3) and (5)). The marginal effect of riots
on state income decreases as additional inde-
pendent variables are included in the model.
The impact is, however, still large in column
(5): a one unit increase in the level of riots in
Kerala decreases state income by 16.5%. The
impact of lagged values of riots on Kerala’s
state income is positive in column (3) but nega-
tive in column (5). The coefficient is always sta-
tistically insignificant.
The impact of strikes and lockouts in column
(1) is positive but not statistically significant.
This is also true in column (3). The coefficient
becomes negative in column (5), suggesting that
the previous positive coefficient was determined
by omission variable biases in the other two
model specifications. The coefficient is, how-
ever, never statistically significant. Kerala’s
level of state income is, however, positively
associated with increases in the lagged levels
of strikes and lockouts. This coefficient is statis-
tically significant in column (5) and shows that
one more strike per 1,000 people in any given
period is associated with 1.2% increase in state
income in the next period.
In addition, Kerala’s state product has been
negatively affected by the number of workers’
unions in the state. The impact is quite small
in comparison with other measures of collective
action. This result suggests that the fragmen-
tation of Kerala’s labor organizations into
several competing unions has had significant
negative consequences on the state’s economic
performance (see Nair, 1994). One hypothesis
is that union fragmentation may have resulted
in loss of cohesion and instability in labor rela-
tions, which are, in turn, likely to harm labor
productivity (Nair, 1994; Sankaranarayanan
& Bhai, 1994; Tharamangalam, 1998). The
coefficient for the change of government vari-
able is positive but statistically insignificant
across all models.
Additional control variables have also af-
fected Kerala’s economic performance in the
last three decades. As expected, Kerala has
benefited from the process of economic liberali-
zation that has taken place across India. Edu-
cation also has a positive and statistically
significant impact on state income in column
(5). The coefficient is negative and statistically
insignificant in columns (1) and (3), possibly if
due to omission biases in the model. More sig-
nificant is the impact of public expenditure on
social services on Kerala’s state income. This
coefficient is positive and statistically significant
across all model specifications. Depending on
the control variables included into model (1),
expenditure on social services has contributed
to between 6% and 7.4% increases in Kerala’s
state income during 1975–99. This result dis-
proves the presence of a trade-off between
redistribution and economic growth in
Kerala. 20
(b) The endogeneity assumption
We have explicitly tested the endogeneity
assumption. Granger causality tests showed
strong signs of double causality between (i)
state income levels and real per capita expendi-
ture on social services, (ii) state income levels
and the volume of riots, and (iii) state income
levels and the volume of strikes in Kerala, at
less than the 1% level of significance. This clear
indication of endogeneity implies that the right-
hand side regressors will be correlated with the
disturbance term, causing standard OLS esti-
mates of model (1) to be inconsistent. The com-
parison between the results of the 3SLS and the
OLS estimators shows the magnitude of the
estimation error we would incur if the endoge-
neity nature of collective action is not taken
into consideration. Notably, the OLS coeffi-
cients underestimate the extent of the impact
of current levels of rioting on state income (col-
umns (5) and (6)). The OLS coefficients also
underestimate the effect of strikes and lockouts
on state income, as well as the effect of public
expenditure on social services, trade openness
and, to a lesser degree, education enrolments.
 THE IMPACT OF COLLECTIVE ACTION ON ECONOMIC DEVELOPMENT
(c) Interpretation of the results
The above results illustrate the complexity of
the relationship between collective action and
economic development. On the one hand, col-
lective action in the form of riots has negatively
affected Kerala’s economic performance, pos-
sibly due to the uncertainty such actions will
cause on the establishment of private property
rights and the socio-political stability of the
state. The effect has been enhanced by the pro-
cess of union fragmentation in Kerala’s labor
relations. On the other hand, increases in
strikes have been associated with increases in
the level of state income.
This differentiation in the impact of different
forms of collective actions on state income has
two potential explanations. The first is the
motivation behind each type of action. The re-
sults in columns (1), (3), and (5) show that in-
creases in public expenditure on social services
have contributed positively toward Kerala’s
economic performance. This result is in line
with recent analyses of the impact of redistribu-
tion on economic growth, which have shown
that pro-poor redistributive policies may stimu-
late economic growth due to their positive
externalities on the size of internal markets, la-
bor productivity, and accumulation of human
capital (Justino, 2003; Murphy, Shleifer, &
Vishny, 1989; Perotti, 1993). If the implementa-
tion of these policies is a direct result of work-
ers’ political activism (see Esping-Andersen,
1990; Freeman & Medoff, 1984; Gurr, 1970),
then we would indeed expect the coefficient
for the relevant collective action variable (i.e.,
strikes) to be positive.
The second reason is the level of economic
and social instability caused by certain types
of collective actions. In particular, high levels
of instability may affect negatively investment
decisions and ultimately economic growth
(Alesina & Perotti, 1996). Strikes and lockouts
are likely to affect labor relations and invest-
ment climate and have, for that reason, been
pointed out as one of the reasons behind Ker-
ala’s recent poor economic performance (e.g.,
Kannan, 1990). Our results do not, however,
confirm this hypothesis. If anything, Kerala’s
poor economic record are likely to be associ-
ated with the level of rioting in the state as
riots are often responsible for the destruction
of livelihoods and markets, loss of trust and
social cohesion and the waste of significant
human and economic resources (see Brass,
2003).
1263
The results may reflect some measurement
error. As discussed previously, data on strikes
and lockouts are likely to underestimate the ex-
tent of strikes and lockouts in Kerala. This
should not, however, invalidate the above re-
sults given their strong statistical significance
and the size of the coefficients. These coeffi-
cients will, at the very least, represent a mini-
mum impact of strikes and lockouts on
Kerala’s state income. More accurate data on
strikes and lockouts are likely to increase their
positive impact. On the other hand, estimates
of riots in Kerala (and elsewhere in India) are
registered by the police and do not depend on
accurate self-reporting. This allows reasonable
confidence in the accuracy of the data. Given
the nature of rioting in Kerala (and India),
these data may, however, be influenced by
political pressure. We have tested carefully the
data against this hypothesis by examining the
correlation between riot incidence and the like-
lihood of one of the two main political parties
winning state elections. This coefficient had a
very small estimate of 0.0174, which was statis-
tically insignificant at 1%, 5%, and 10% levels
of significance. We have estimated similar tests
for every single one of the major 14 Indian
states, with similar results. Coefficients were
higher in other states than in Kerala but none
were statistically significant at levels well above
10%. These tests allow us to conclude with rea-
sonable levels of confidence that Kerala’s eco-
nomic development has indeed been hampered
by public actions in the form of riots. Labor
market actions have in contrast contributed
positively toward the promotion of economic
development in the state.
In order to further test the robustness of the
results discussed above, we have examined the
impact of the forms of collective action on pov-
erty, as the poor would generally be those who
benefit the most from larger redistributive pol-
icies, better wages and similar demands that
motivate workers to strike and populations to
riot (Barron et al., 2004; Brass, 2003). We have
examined more closely the direct impact of
riots and strikes on rural and urban poverty
in Kerala (columns (7) and (8), respectively).
The results show that increases in the current
level of riots have had positive and statistically
significant effects on rural poverty in Kerala.
The lagged coefficient is also positive but not
statistically significant. On the other hand,
increases in strikes are associated with long-
term decreases in rural poverty. The current
and lagged coefficients are both negative and
1264 WORLD DEVELOPMENT
statistically significant. These results strongly
suggest that, even if some riots are motivated
by the demand for better living conditions, they
may at the same time destroy important sources
of income, depending on the level of violence of
each period of unrest. In contrast, increases in
strikes are correlated with decreases in rural
poverty. This suggests that strikes have contrib-
uted positively toward the increase in the living
standards of some population groups, which
would in turn explain the positive coefficient
of this collective action variable in column (5).
The fact that only lagged values of strikes affect
state income is intuitively plausible as it would
take some time for pro-poor policies to impact
on the rest of the economy. Our results do not
allow us to infer about the urban sector as the
coefficients in column (8) are statistically insig-
nificant. This could be a result of the size of the
rural sector in Kerala. 21
These results have important implications for
the explanation of Kerala’s recent poor eco-
nomic performance. Our results do not support
the widely proposed hypothesis that excessive
activism in the labor market has been responsi-
ble for Kerala’s poor economic performance.
Labor disputes have been positively associated
with levels of state income, most likely because
they have resulted in improvements in the living
standards and incomes of the poor. Our results
indicate therefore that collective action in the
labor market can be an important vehicle for
poverty reduction and improvements in human
development via important redistributive ef-
fects. This empirical evidence supports previous
analytical discussions in, for instance, Heller
(1996, 2000) and Herring (1983).
What then explains Kerala’s poor economic
performance? From a collective action perspec-
tive, two of our results suggest some answers to
this question. We found that excessive levels of
unionism have affected the state’s economic
performance. Large numbers of unions may
not necessarily result in increases in strikes.
Nonetheless, they may affect labor productivity
if they imply breakdowns in labor relations or
increases in caste- and class-based grievances
(see Nair, 1994). We also found that the volume
of politically motivated riots may be partially
responsible for Kerala’s low levels of economic
growth. 22 Kerala has some of the highest num-
bers of riots per 1,000 people in India (Table 2).
Remarkably, the large literature on poverty
and development, in India in general and Ker-
ala in particular, does not entail any significant
discussions of this potentially important con-
straint to socio-economic development. Our
results point strongly toward further research
in this area.
5. PANEL ANALYSIS
How unique is the relationship between orga-
nized forms of collective action and state in-
come that takes place in Kerala? This section
analyzes that relationship in the context of the
whole of India for the period during 1973–99,
using a panel of 14 major Indian states (Andhra
Pradesh, Assam, Bihar Gujarat, Karnataka,
Kerala, Madhya Pradesh, Orissa, Punjab,
Rajasthan, Tamil Nadu, Uttar Pradesh, and
West Bengal). 23 The main objective of this sec-
tion is to examine the differences between Ker-
ala’s development strategy and that of other
Indian states, and investigate whether the role
of collective action is specific to Kerala’s model
of development or whether broader lessons can
be derived. The model used in the previous sec-
tion can be modified easily to encompass the
panel dimension of the Indian data. The result-
ing model is
Y it ¼ ai þ uY i;t1 þ bX i;t1 þ cZ it
þ dW it þ N t þ eit ; ð2Þ
where the vectors X, Z and W are interpreted as
before. i represents individual states. ai are the
state effects, with i = 1, . . . , 14. The new vector
Nt is a vector of national-level independent
variables, invariant across the various states.
eit is, as before, the error term, which varies
unsystematically across time and individuals. 24
All variables are the same as in model (1) for
Kerala, except for the election variable. In the
above model, the election variable is shared
by all states that take the value 1 if the Indian
National Congress Party had the majority of
seats in the Lok Sabha. 25
Model (2) can be estimated by standard
fixed-effects panel estimation techniques. 26 As
before, we have modeled collective action and
redistributive policies as endogenous vari-
ables. 27 We have used two techniques to model
endogeneity. The first is the fixed-effects two-
stage least-squares models adapted to panel
data developed by Baltagi (1995, Chapter 7).
The second is the generalized method of mo-
ments (GMM) (Arellano & Bond, 1991), as
model (2) contains lagged endogenous vari-
ables. Both 2SLS and GMM models are pre-
sented in Table 4. This table includes two
 THE IMPACT OF COLLECTIVE ACTION ON ECONOMIC DEVELOPMENT 1265

Table 4. Marginal effects for Indian panel

Dep var: LNSDP (2SLS) Dep var: LNSDP (2SLS)
(with Kerala) (without Kerala)
2SLS (1) GMM (2) 2SLS (3) GMM (4)
Lagged state income 0.095*** 0.018 0.091*** 0.029
Volume of riots 0.101** 0.107** 0.100** 0.056*
Lagged volume riots 0.037 0.063** 0.035 0.028
Volume of strikes 0.018** 0.014*** 0.019** 0.012**
Lagged volume of strikes 0.003 0.002 0.002 0.005
Exp on social services 0.0002 0.0003 0.003 0.0003
Number of workers’ unions 0.0002 0.0002 0.0003 0.0001***
Election results 0.0002 0.001 0.006 0.004
Trade openness 0.020*** 0.010** 0.023*** 0.016***
Education enrolments 0.053 0.048 0.026 0.004

State effects Yes Yes Yes Yes

Within R-squared 0.914 – 0.912 –
Between R-squared 0.769 – 0.813 –
Overall R-squared 0.771 – 0.787 –
No. of observations 70 70 65 65
Note: Test for first order autocorrelation in column (2): Pr > z = 0.3703. Test for second order autocorrelation in
column (2): Pr > z = 0.7379. Test for first order autocorrelation in column (4): Pr > z = 0.4728. Test for second
order autocorrelation in column (4): Pr > z = 0.6047.
*
Significant at the 10%.
**
Significant at the 5%.
***
Significant at the 1%.

model specifications for each technique. Col- underlying development motivations—as

umns (1) and (2) include all 14 states, whereas
columns (3) and (4) exclude Kerala in order
to test the possibility of the panel results being
driven by specific characteristics of this state.
The results vary marginally when Kerala is
removed from the panel.
Table 4 shows quite interesting results. Con-
trary to the Kerala analysis, all forms of collec-
tive action result in positive economic benefits
across India. With the exception of column
(1), both short- and long-term effects of riots
and strikes on state income are positive. This
may well be due to differences in the nature of
rioting in Kerala and other Indian states. There
is good reason to believe that riots in Kerala
have a large political foundation motivated by
party competition (Heller, 1995; Prabhash,
2000). Forms of rioting in other Indian states
are likely to result from similar basis (see Brass,
2003), but also (in larger numbers than in
Kerala) from separatist movements and clashes
between different ethnic and religious groups
and different castes (see Hardgrave, 1993;
Varshney, 2002). Although these forms of riots
may be violent and destructive, those with
for instance recent adivasis movements or
public actions leading to the implementation
of reservations for backward castes in gov-
ernment jobs—may result in overall improve-
ments in living standards. These must,
however, remain as conjectures until better
disaggregated data on riots in India are made
available. 28
These results can of course also be due to
measurement error in the collective action vari-
ables. The analysis in this section requires a cer-
tain level of comparability in the data on strikes
and riots across the 14 Indian states. There are
good reasons to question the comparability of
the data. As discussed previously, data on
strikes and lockouts are self-reported. On the
other hand, data on riots are collected by the
local police and may be subject to political
pressure. This is more so in states with lower
incidence of media influence, lower degrees of
literacy and numeracy rates, more fragmented
or absent labor unions and lower levels of law
and order and state government accountability.
In order to address these potential concerns, we
have carefully tested the above results to the
1266 WORLD DEVELOPMENT
removal and inclusion of certain states. We
report the results of our regressions with and
without Kerala in Table 4, but we have under-
taken similar tests for each of the 14 states
without finding any statistically significant dif-
ferences between these and the results reported
in this section. 29 The results in this section
emphasize, therefore, the importance of forms
of collective action for economic development.
They show, furthermore, that the relationship
between collective action and economic devel-
opment may have different static and dynamic
effects in different contexts, depending on the
volume and type of activism.
6. CONCLUSIONS
Most development texts, including recent
World Bank and United Nations publications,
argue that public participation in the process
of decision making is necessary, if not funda-
mental, for the implementation of successful
development strategies. These studies do not,
however, consider that collective action can
take different forms and be based on different
motivations. Moreover, the development litera-
ture rarely provides rigorous, empirically based
analysis, of the extent and importance of pub-
lic participation on economic development.
This paper examined empirically the impact of
collective action and public participation on
economic development in South India. Our
analysis uncovered important aspects of that
relationship, notably the endogenous nature
of collective action and its different static and
dynamic effects. The paper entails, therefore,
substantial implications for the understanding
of the relationship between collective action
and economic development.
Overall we concluded that collective action
may have important endogenous, positive ef-
fects on economic development. In India, both
labor disputes and politically motivated riots
have raised state income levels. In Kerala, in
general, labor market activism has resulted
in significant benefits for the poor, as well as
in improvements in economic growth. A partic-
ular beneficial outcome in Kerala was the
implementation of social policies that pro-
moted human development and widespread so-
cio-economic security. This confirms the widely
accepted view that collective action and public
participation are important conditions for suc-
cessful economic development. We showed,
however, that this relationship may not hold
for all economies at all times. The analysis of
the Kerala dataset and its comparison with
the Indian panel showed that collective action
and public participation are not always positive
conditions for economic development. Our
empirical analysis demonstrated that the rela-
tionship between collective action and eco-
nomic development will depend on the type
and the level of collective action, as well as spe-
cific underlying motivations behind different
types of collective action. Large numbers of
uncoordinated workers’ unions may break
cohesion in labor markets and offset the poten-
tial gains of more concerted actions, whilst
certain types of politically motivated forms of
civil unrest may entail potentially destructive
effects.
It is undeniable that the participation of vul-
nerable groups in the national decision-making
process may help to address economic and so-
cial difficulties faced by those typically excluded
from fundamental, social, economic, and polit-
ical institutions, as well as increasing aggregate
economic performance. There is, nonetheless,
a fine line between public participation and
unsustainable socio-political instability, as
demonstrated by the example of Kerala. This
may have important policy implications. In
particular, the empirical results presented and
discussed in this paper signal the need to com-
plement usual development policies with the
establishment of strong institutions that medi-
ate the interests and needs of all population
groups. Avoiding potentially negative out-
comes from collective action will require the
establishment of well-functioning bargaining
processes between the state, industry, and civil
organizations, as well as the promotion of ac-
tions that encourage civil responsibility from
all actors.
We are still far from understanding the full
extent and implications of the relationship be-
tween collective action, public participation,
and economic development. We expect that
the empirical evidence provided in this paper
may generate enough interest and motivate fur-
ther research on both the nature of collective
action and public participation, and the ana-
lysis of these relationships in the context of
other case studies. In particular, further empir-
ical evidence will be fundamental for better in-
formed policy actions by civil organizations,
local governments, and the international com-
munity.
 THE IMPACT OF COLLECTIVE ACTION ON ECONOMIC DEVELOPMENT
NOTES
1. An exception is empirical work on the impact of
labor union actions (see Freeman & Medoff, 1984).
2. See George (1993) and Lizy (1990) for analyses of
Kerala’s fiscal crisis. Government of Kerala (1996),
Isaac and Tharakan (1995), Prakash (1994), Ramachan-
dran (1997) and Sankaranarayanan and Bhai (1994)
provide extensive discussions of the difficulties faced by
Kerala’s agricultural and industrial sectors. NSSO
(1997) and Tharamangalam (1998) discuss in further
detail the possible causes and consequences of Kerala’s
high unemployment rates.
3. In 1998, Kerala’s wages ranged between Rs. 19.5 and
Rs. 114.16 per day, well above the Indian average. These
wage rates have been steadily increasing since 1975
(Kerala’s Ministry of Labour, Annual Report, various
years).
4. The decrease in strikes and lockouts in the late 1990s
is also likely to be the result of the ban on bundhs (short
duration wildcat strikes) by the High Court in 1997.
5. It is not possible to separate the number of strikes
from the number of lockouts for the whole dataset at the
state level and thus assess the relative strength of
workers’ unions versus employers. We do not think,
however, that this shortcoming will significantly affect
our conclusions as the number of lockouts at the all-
India level is very small. These data are self-reported and
are, therefore, likely to underreport the true extent of
strikes and lockouts in Kerala (and India) (see Indian
Labour Statistics, 1994 and discussion in Justino, 2001).
This issue is not however a serious problem in this paper
given the strong statistical significance of the results. We
will discuss this issue further in Section 4.
6. In terms of absolute numbers of strikes and
lockouts, Kerala is roughly in the middle of the
distribution amongst Indian’s 14 largest states included
in Table 2. Indian states vary, however, widely in terms
of the size of their total and working population.
Comparisons should therefore only be made for num-
bers of strikes and lockouts per number of factory
workers as illustrated in Table 2. In this case, Kerala’s
position in the distribution goes up quite significantly.
7. See Ranchhoddas, Thakore, and Vakil (1971), cited
in Brass (2003). See also Gurr (1970).
8. The data were kindly provided by the National
Crime Records Bureau (NCRB), part of the Indian
Ministry of Home Affairs, and refers to the number of
1267
riots per 1,000 people. These data were collected by the
author from a collection of published records held by the
NCRB since 1969, and updated yearly. The data are net
of juvenile delinquency as those types of rioting are
typically not politically driven (private communication
with S.R. Meena at NCBR). Juvenile delinquency
comprises less than 1% of total rioting in Kerala (and
India). We have tested our results to the exclusion of
these data and found no statistically significant changes.
9. For similar evidence in Indonesia, see Barron,
Kaiser, and Pradhan (2004).
10. See for instance www.rediff.com/news. See also
Human Rights Watch (2001) and Justino (2004).
11. The choice of the period of analysis was driven by
data availability.
12. We tested for the presence of stochastic trends.
Augmented Dickey–Fuller tests did not show the pres-
ence of unit roots at the 5% level of significance,
suggesting that we have deterministic rather than
stochastic trends (Greene, 2000).
13. Strikes usually take place within the industrial
sector, whereas riots refer to the whole population.
14. These data are published in the Reserve Bank of
India’s annual bulletins. The variable includes the
amount of real per capita public expenditure on educa-
tion, health, family welfare, welfare of scheduled castes,
schedule tribes and other backward classes, labor
welfare, social security, and welfare and nutrition.
15. Economic liberalization has been the most signif-
icant economic change that has taken place in India in
the recent years. For discussion of the liberalization
process in India, see Srinivasan (2001).
16. This variable is provided by Kerala’s Assembly
Election Database (http://assembly.expert-eyes.org).
17. The OLS regressions in columns (2), (4), and (6)
exhibited serial correlation. This has been corrected by
using a two-step Cochrane–Orcutt transformed regres-
sion estimator (see Greene, 2000).
18. Poverty is measured by the headcount index
derived from household consumption expenditure data
in the National Sample Surveys (NSS) (see Deaton,
2001). This variable has been linearly interpolated in
1268 WORLD DEVELOPMENT
order to avoid gaps in the NSS data. As the data have
not been collected on some years, interpolation does not
lead to loss of efficiency in the estimators (Greene, 2000).
19. This variable includes both capital and noncapital
public expenditure and is provided by the Reserve Bank
of India.
20. See Ravallion and Datt (1999) for a recent discus-
sion of the ‘‘redistribution versus growth’’ debate.
21. Note that in Kerala labor disputes take place both
in urban and rural areas, where the role of peasant
associations has been significant (Gulati, 1990; Oom-
men, 1993).
22. There will be of course other reasons behind
Kerala’s low state income. Other possible areas of
research outside the scope of this paper include the role
of industrial policies, the channeling of remittances
toward productive activities, the inadequacy of Kerala’s
educational system, amongst others. See Justino (2001).
23. Although it would have been possible to estimate
time-series results for all-India, panel analysis is a more
appropriate technique given the large heterogeneity
between all Indian states in social, cultural, religious,
economic, and political terms. These states represented
93.3% of the total Indian population in 1999–2000. The
analysis in this section uses six data points: 1973–74,
1977–78, 1983, 1987–88, 1993–94 and 1999–2000. These
dates correspond to the dates of the large sample NSSO
surveys. Our analysis focuses on these 6 years in order to
ensure consistency across all variables.
24. t exhibits a different periodicity. The usual estima-
tion procedures will still yield unbiased estimators for
the true coefficients, since variables included in the
regression model are observed for the same exact periods
(Greene, 2000).
25. The Indian National Congress Party, founded in
1885, was in power almost without opposition since
REFERENCES
Ahmad, E., Drèze, J., Hills, J., & Sen, A. K. (1991).
Social security in developing countries. Oxford: Clar-
endon Press.
Alesina, A., & Perotti, R. (1996). Income distribution,
political instability and investment. European Eco-
nomic Review, 15, 1203–1228.
Arellano, M., & Bond, S. (1991). Tests of specification
for panel data: Monte Carlo evidence and an
independence until 1977. At that time it was beaten in
the national elections by the right-wing Bharatiya Janta
Party, but recovered its position quickly in 1980. The
Bharatiya Janta Party returned to power in the 1990s
and had been the ruling party in India until the latest
elections in 2004 (Election Commission of India, http://
www.eci.gov.in).
26. Breusch–Pagan Lagrange multiplier tests for ran-
dom effects and Hausman specification tests suggest the
presence of strong fixed effects. We have not found any
statistically significant evidence for random effects.
27. The instruments are all the exogenous variables in
the equation plus, as before, rural and urban poverty
headcounts and aggregate development expenditure,
which guarantee an identified system. We have tested
the endogeneity assumption for the panel using Haus-
man specification tests (Wooldridge, 2002), and could
not reject the hypothesis that endogeneity was present in
the model. We also found signs of double causality
between the three pairs of variables in all states at less
than 1% level of significance.
28. Brass (2003) argues extensively for the need for
more appropriate definitions of riots in India, including
better definitions and operationalization of notions of
communal riots. To the best of my knowledge, no
dataset exists that allows a separation of motives for
rioting in India. Varshney (2002) and Wilkinson (2004)
have collected a groundbreaking dataset on ethnic riots
in India. These cannot, however, be directly compared
with the NCBR sets as methodologies differ. The
Varshney–Wilkinson survey is based on information
contained in the Times of India over four decades, while
the NCBR data are based on police records. Both data
sets entail significant biases as well as strengths, though
none differentiates between different types of riots.
29. These results have not been included in this paper
for reasons of space constraints but can be obtained
from the author upon request.
application to employment equations. Review of
Economic Studies, 58, 277–297.
Baltagi, B. H. (1995). Econometric analysis of panel data.
New York: John Wiley & Sons.
Barro, R. J. (1991). Economic growth in a cross section
of countries. Quarterly Journal of Economics, 106,
407–443.
 THE IMPACT OF COLLECTIVE ACTION ON ECONOMIC DEVELOPMENT
Barron, P., Kaiser, K., & Pradhan, M. (2004). Local
conflict in Indonesia: Measuring incidence and iden-
tifying patterns. World Bank, mimeo.
Brass, P. (1997). Theft of an idol: Text and context in the
representation of collective violence. Princeton:
Princeton University Press.
Brass, P. (2003). The production of Hindu–Muslim
violence in contemporary India. Seattle: University
of Washington Press.
Buchanan, J., & Tullock, G. (1962). The calculus of
consent. Ann Arbor.
Das, M. K. (2003). Kerala: Politics of pressure groups.
Economic and Political Weekly, August 2.
Deaton, A. (2001). Adjusted Indian poverty estimates for
1999–2000. Working paper. Research Program in
Development Studies, Princeton University.
Drèze, J., & Sen, A. K. (1991). Public action for
social security: Foundations and strategy. In E.
Ahmad, J. Drèze, J. Hills, & A. K. Sen (Eds.), Social
security in developing countries. Oxford: Clarendon
Press.
Drèze, J., Sen, A. K., & Hussain, A. (1995). The political
economy of hunger: Selected essays. Oxford: Claren-
don Press.
Engineer, A. A. (1984). Communal riots in post-indepen-
dence India. Hyderabad: Sangam Books.
Engineer, A. A. (2004). Communal riots after indepen-
dence: A comprehensive account. New Delhi: Shipra.
Esping-Andersen, G. (1990). The three worlds of welfare
capitalism. Cambridge: Polity Press.
Franke, R. W., & Chasin, B. H. (1992). Kerala:
Development through radical reform. New Delhi:
Institute for Food and Development Policy, Promilla
& Co., Publishers.
Freeman, R. B., & Medoff, J. L. (1984). What do unions
do? New York: Basic Books.
George, K. K. (1993). Limits to Kerala model of
development: An analysis of fiscal crisis and its
implications. Thiruvananthapuram, Kerala: Centre
for Development Studies.
George, K. K. (1998). Historical roots of the Kerala
model and its present crisis. Bulletin of Concerned
Asian Scholars, 30, 35–40.
Government of India (1998). Selected socio-economic
statistics India 1996–97. New Delhi: Central Statis-
tical Organisation, Ministry of Planning and Pro-
gramme Implementation.
Government of India (various years). Annual report.
New Delhi: Ministry of Labour.
Government of India (various years). Indian labour year
book. New Delhi: Ministry of Labour.
Government of India (various years). Indian labour
statistics. New Delhi: Ministry of Labour.
Government of India (various years). Crime in India.
New Delhi National Crime Records Bureau, Minis-
try of Home Affairs.
Government of India (various years). National accounts
statistics. New Delhi: Central Statistical Organisa-
tion, Department of Statistics, Ministry of Planning
and Programme Implementation.
Government of Kerala (1996). Kerala economic survey
1995–96. Thiruvananthapuram: State Planning
Board.
1269
Greene, W. H. (2000). Econometric analysis (4th ed.).
New Jersey: Prentice-Hall.
Gulati, L. (1990). Agricultural workers pension in
Kerala—An experiment in social assistance. Eco-
nomic and Political Weekly, 25, 339–343.
Gurr, T. R. (1970). Why men rebel. Princeton University
Press.
Hardgrave, R. (1993). India: The dilemmas of diversity.
Journal of Democracy, IV, 54–68.
Heller, P. (1995). From class struggle to class compro-
mise: Redistribution and growth in a south Indian
state. Journal of Development Studies, 31, 645–672.
Heller, P. (1996). Social capital as a product of class
mobilization and state intervention: Industrial work-
ers in Kerala, India. World Development, 24,
1055–1071.
Heller, P. (2000). The labour of development: Workers
and the transformation of capitalism in Kerala, India.
Ithaca: Cornell University Press.
Herring, R. J. (1983). Land to the tiller: The political
economy of agrarian reform in South Asia. London:
Yale University Press.
Human Rights Watch (2001). World report: India. New
York: Human Rights Watch.
Isaac, T. M., & Tharakan (1995). Kerala—The emerging
perspectives: Overview of the international congress
on Kerala studies. Social Scientist, 23, 3–36.
Joseph, A. V. (1999). Kerala: Resurgence of politics of
violence. Economic and Political Weekly, December
18.
Justino, P. (2001). Social security and political conflict in
developing countries, with special reference to the
south Indian state of Kerala. Unpublished Ph.D.
thesis, Department of Economics, School of Oriental
and African Studies, University of London.
Justino, P. (2003). Social security in developing countries:
Myth or necessity? Evidence from India. PRUS
Working paper no. 20. Poverty Research Unit at
Sussex, University of Sussex.
Justino, P. (2004). Redistribution, inequality and political
conflict. PRUS Working paper no. 18. Poverty
Research Unit at Sussex, University of Sussex.
Kannan, K. P. (1990). Kerala economy at the cross-
roads? Economic and Political Weekly, September 1–
8; 1951–1956.
Kannan, K. P. (1995). Public intervention and poverty
alleviation: A study of the declining incidence of
rural poverty in Kerala, India. Development and
Change, 26, 701–727.
Lizy, M. A. (1990). Centre-state financial relations in
India. New Delhi: Ashish Publishing House.
McAdam, D. (1982). Political process and the develop-
ment of black insurgency 1930–1970. Chicago: Uni-
versity of Chicago Press.
Meinzen-Dick, R. S., & Di Gregorio, M. (2004).
Collective action and property rights for sustainable
development. New York: IFPRI and CAPRi.
Murphy, K. M., Shleifer, A., & Vishny, R. (1989).
Income distribution, market size, and industrializa-
tion. Quarterly Journal of Economics, CIV, 537–564.
Nair, M. K. S. (1994). Trade unionism in Kerala. In
B. A. Prakash (Ed.), Kerala’s economy: Performance,
problems and prospects. India: Sage Publications.
1270 WORLD DEVELOPMENT
NSSO (various years). Sarvekshana. New Delhi: Na-
tional Sample Survey Organization.
Olson, M. (1971). The logic of collective action: Public
goods and the theory of groups. Cambridge, MA:
Harvard University Press.
Oommen, M. A. (1993). Essays on Kerala economy.
New Delhi: Oxford and IBH Publishing Co.
Perotti, R. (1993). Political equilibrium, income distri-
bution and growth. Review of Economic Studies, 60,
755–776.
Pillai, P. P. (1994). Kerala economy: Four decades of
development. India: Institute of Planning and Ap-
plied Economic Research, John Matthai Founda-
tion.
Prabhash, J. (2000). CPI(M)’s Muslim league dilemma.
Economic and Political Weekly, August 19: 3009–
3011.
Prakash, B. A. (1994). Kerala’s economy: Performance,
problems and prospects. India: Sage Publications.
Przeworski, A. (1985). Capitalism and social democracy.
Cambridge University Press.
Ramachandran, V. K. (1997). Kerala’s development
achievements. In J. Drèze, & A. K. Sen (Eds.), Indian
development: Selected regional perspectives. New
Delhi: Oxford University Press.
Ranchhoddas, R., Thakore, D. K., & Vakil, M. R.
(1971). The law of crimes (2nd ed.). Bombay:
Bombay Law Reporter.
Ravallion, M., & Datt, G. (1999). When is growth pro-
poor? Evidence from the diverse experiences of India’s
states. Washington, DC: The World Bank.
Reserve Bank of India (various years). Bulletin. New
Delhi.
Sankaranarayanan, K., & Bhai, M. (1994). Industrial
development in Kerala: Problems and prospects. In
B. Prakash (Ed.), Kerala’s economy: Performance,
problems and prospects. India: Sage.
Srinivasan, T. N. (2001). Integrating India with the world
economy: Progress, problems and prospects. Depart-
ment of Economics, Yale University.
Tarrow, C. (2003). Power in movement: Social move-
ments, collective action and politics. Cambridge Uni-
versity Press.
Tharamangalam, J. (1998). The perils of social develop-
ment without economic growth: The development
debacle of Kerala, India. Bulletin of Concerned Asian
Scholars, 30, 23–34.
Tilly, C. (1993). Contentions repertoires in Great
Britain 1758–1834. Social Science History, 17,
253–279.
Tilly, C. (2003). The politics of collective violence.
Cambridge: Cambridge University Press.
United Nations (1997). Human development report 1997.
New York: United Nations.
Varshney, A. (2002). Ethnic conflict and civic life:
Hindus and Muslims in India. Yale: Yale University
Press.
Wilkinson, S. (2004). Votes and violence: Electoral
competition and ethnic riots in India. New York:
Cambridge University Press.
Wooldridge, J. M. (2002). Econometric analysis of cross
section and panel data. Cambridge, MA: MIT Press.
World Bank (1997). World development report 1997.
Washington, DC: The World Bank.
World Bank (2001). World development report. Wash-
ington, DC: The World Bank.