IMPACT OF POLITICAL RESERVATIONS IN WEST BENGAL LOCAL
GOVERNMENTS ON ANTI-POVERTY TARGETING
1
Pranab Bardhan
2
, Dilip Mookherjee
3
and Monica L. Parra Torrado
4
December 26, 2009
ABSTRACT
Political reservation for disadvantaged groups is believed to be a way of improving targeting of
publicly provided goods to those groups. This paper examines the impact of political reservations
for women and scheduled castes and tribe (SC/ST) candidates in local governments in West Bengal,
India between 1998-2004 on targeting to landless, low caste and female-headed households. It
differs from existing literature by differences in geographic coverage, time span, and use of self-
reported household benefits across a broad range of programs. Reservation of chief executive
(pradhan) positions in local government for women was associated with a significant worsening of
within-village targeting to SC/ST households, and no improvement on any other dimension of
targeting. Reservation of pradhan posts for SC/ST members was associated with a significant
increase in benefits received by the village as a whole, improvement in intra-village targeting to
female-headed households, and to the group (SC or ST) of the pradhan. The effects of women’s
reservations are not consistent with simple citizen-candidate or elite capture models of electoral
politics. They are consistent with a more complex hypothesis of capture-cum-clientelism which is
weakened by election of politically inexperienced women to reserved pradhan posts.
1
We thank the MacArthur Foundation Inequality Network, and National Science Foundation Grant No.SES-0418434 for
research funding, and Dr. Sandip Mitra for assistance with the survey design and implementation. This paper is based on
Chapter 2 of Monica Parra Torrado’s PhD dissertation submitted to the Department of Economics at Boston University in
2008. We also thank Shahe Emran and an anonymous referee for constructive suggestions on an earlier version.
2
Department of Economics, University of California, Berkeley
3
Department of Economics, Boston University
4
Fedesarrollo, Bogota, Colombia
2
1. Introduction
Improving governance is an essential aim of development policy. Many countries have embarked
on programs of decentralization in which local governments are given greater authority over delivery
of development programs in an effort to promote government accountability. The design of local
governance includes rules ensuring representation of minorities and women. An important goal of
gender or minority reservation of political elected positions is to improve targeting of developmental
and welfare programs to women and vulnerable groups.
However, the extent to which targeting is actually improved depends on the extent to which such
mandated reservations succeed in transferring effective power to members occupying the reserved
positions, and on the integrity and competence of such officials. It also depends on the extent to
which personal preferences of elected leaders affect actual policies and programs (stressed by citizen
candidate models of electoral politics), rather than voter preferences and needs (stressed by
Downsian models).
5
It is appropriate, therefore, to empirically evaluate the effect of reservations
implemented so far on targeting of public service delivery.
This paper studies the effect of political reservations in local governments in favor of women,
scheduled castes and tribes (SC/ST) in the Indian state of West Bengal on provision of government
services and local public goods. Political reservations at the village level were mandated by the 73rd
amendment to the Indian Constitution in 1992.
6
This amendment requires a fraction of seats and
Pradhan (chief executive) positions be reserved for SC/ST candidates, in accordance with their
demographic share in each Gram Panchayat (GP, or village level council). In addition, one-third of
GP seats and one-third of Pradhan positions are reserved for women. In West Bengal, reservations of
5
See Bardhan and Mookherjee (2008) for a discussion of these different models of electoral politics and an empirical test
of their relative validity in the context of land reform implementation in the same sample of West Bengal villages.
6
The 74
th
amendments to the Indian Constitution mandated political reservations at the (urban) municipality level in 1992
.
3
council seats were implemented since 1993 and Pradhan positions since 1998. GPs with reserved
positions are selected randomly according to a rotation schedule for successive elections.
The Indian Parliament is currently considering a bill to amend the Constitution to expand the
scope and extent of these reservations to state legislative assemblies and the national Parliament to
mirror the reservations at the local government level. The state of Kerala has recently expanded the
proportion of seats in local governments and all civic bodies reserved for women to 50%. Rajaraman
and Gupta (2008) quote the Economist which reported g
ender-based quotas in elected posts or in
political party candidate fields in force in 110 countries in 2008. Whether reservations improve
governance significantly is thus a question of considerable policy significance.
The fact that allocation of reserved seats were randomly assigned helps avoid problems of
statistical identification of cause and effect. However, since most available statistical data pertains to
outcomes of local government actions, it is a challenging task to infer from these how processes of
local governance function and the way they are modified by reservations. The problem is
compounded by the significant social, economic and political heterogeneity across different regions.
This limits the scope of the evaluation based on statistical data to effects of political reservations on
the outcomes of local governance, such as measures of performance with regard to delivery of public
services or targeting of different benefit schemes administered by local governments.
A number of papers have already examined this issue in different settings, following the seminal
work of Chattopadhyay and Duflo (2003, 2004) for selected districts in states of West Bengal and
Rajasthan. These include Besley, Pande, Rahman and Rao (2004), Besley, Pande and Rao (2005)
and Ban and Rao (2008) in the context of three South Indian states, Rajaraman and Gupta (2009) in
four central Indian states (Orissa, Madhya Pradesh, Chattisgarh and Rajasthan), and an earlier paper
4
of ours (Bardhan, Mookherjee and Parra Torrado (2005)) in the state of West Bengal.
7
Given the
substantial heterogeneity of impact that one would expect across different regions or kinds of
programs administered by local governments, it is necessary to examine whether the findings of
existing studies are specific to their respective contexts.
This paper focuses on the state of West Bengal, using the same set of villages as in our earlier
study. We revisit these villages to take advantage of substantially better data. We use a household
survey rather than data provided by local governments concerning distribution of benefits. To the
extent there may be corruption and diversion of private good benefits away from intended
beneficiaries, government records may conceal the actual pattern of targeting. With regard to local
public goods such as roads and drinking water (key services provided by local governments),
analyses based on government records are not detailed enough to record their location and proximity
to different household groups. Asking individual households to identify various local government
programs that they have significantly benefitted from provides a way of assessing intra-village
targeting of these local public goods.
Another major weakness of our previous study is that it was based on government records for
1998, the very first year of the Pradhan reservations. A newly installed chief executive is likely to
take some time to learn the job and settle into the task of administration. Even if they have a distinct
impact their effects may not be discernible in the first year or two. This paper is based on a
household survey carried out in 2003-04, five or six years after the original Pradhan reservations in
1998. By pooling the data from 1998-2004, this enables a more comprehensive assessment of the
wave of Pradhans elected to reserved seats in 1998.
7
We should also mention the work of Munshi and Rosenzweig (2008) who stress the importance of the size of caste groups on
competence and commitment of the representatives elected from such groups. They exploit the effects of the randomized
reservations for SC/ST groups to demonstrate this. Hence they are not concerned with estimating the effect of the reservations
per se. Their results indicate that the effectiveness of these reservations depend on the relative size of the SC/ST group within the
village. To incorporate this possibility, we estimate a regression which permits the effect of the reservations to vary with the
demographic share of SC/ST groups.
5
Yet another difference from existing literature concerns geographic coverage, the range of local
government programs covered, and measures of targeting. Chattopdhyay and Duflo, the only other
authors studying West Bengal, focus only on Birbhum, a single district (out of 18 districts in the
state). Our study covers all the 16 agricultural districts in the state, including Birbhum. We exclude
only Kolkata, an urban area, and Darjeeling a hill district. Our use of a household survey enables us
to study a wider range of local government programs including private benefits such as housing and
toilets constructed, employment provided in public works programs, below-poverty-line (BPL) cards,
IRDP loans and agricultural minikits distributed. It also enables us to assess the targeting of local
public goods directly based on household responses. Chattopadhyay and Duflo study impacts of the
reservations on the extent of congruence between allocation of local government expenditures across
different programs and preferences expressed by women vis-à-vis men; they do not examine
distribution of private benefits, or the effect on targeting to landless or SC/ST groups.
Our main findings are the following. With regard to effect of women reservations, we find no
improvement in any dimension of targeting, and a worsening of intra-village targeting to SC/ST
groups. In contrast, we find a significant positive effect of SC/ST Pradhan reservation on per capita
benefits in the village as a whole, and on intra-village targeting to female headed households, as well
as the group (SC or ST) for whom the position is reserved. The improvements in village-level
benefits partially redressed a systematic tendency for higher level governments to allocate lower
benefits to villages with high SC/ST populations. Joint reservations of Pradhan position for women
SC/ST candidates resulted in a mixture of the respective effects of reservations for women and
SC/ST: an improvement in village average benefits, and deterioration in a number of dimensions of
intra-village targeting.
Hence women reservations resulted in some deterioration of targeting, while SC/ST reservations
resulted in some improvements. Particularly surprising is the absence of any significant effect of the
