VOTES, PARTIES,

AND SEATS
A Quantitative Analysis
of Indian Parliamentary
Elections, 1962–2014

Vani Kant Borooah

Votes, Parties, and Seats

vk.borooah@ulster.ac.uk
vk.borooah@ulster.ac.uk
 Vani Kant Borooah

Votes, Parties, and

Seats
A Quantitative Analysis of Indian Parliamentary
Elections, 1962–2014

vk.borooah@ulster.ac.uk
Vani Kant Borooah
School of Economics
Ulster University
Newtownabbey, County Antrim, United Kingdom

ISBN 978-3-319-30486-1 ISBN 978-3-319-30487-8 (eBook)

DOI 10.1007/978-3-319-30487-8

Library of Congress Control Number: 2016943277

© The Editor(s) (if applicable) and The Author(s) 2016

This work is subject to copyright. All rights are solely and exclusively licensed by the
Publisher, whether the whole or part of the material is concerned, specifically the rights of
translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on
microfilms or in any other physical way, and transmission or information storage and retrieval,
electronic adaptation, computer software, or by similar or dissimilar methodology now
known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this
publication does not imply, even in the absence of a specific statement, that such names are
exempt from the relevant protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information
in this book are believed to be true and accurate at the date of publication. Neither the pub-
lisher nor the authors or the editors give a warranty, express or implied, with respect to the
material contained herein or for any errors or omissions that may have been made.

Printed on acid-free paper

This Palgrave Macmillan imprint is published by Springer Nature

The registered company is Springer International Publishing AG Switzerland

vk.borooah@ulster.ac.uk
In memory of my father, Dev Kant Borooah (1914–1996): scholar, poet,
politician.

vk.borooah@ulster.ac.uk
vk.borooah@ulster.ac.uk
 PREFACE

My life changed irrevocably at the age of two when my father was elected
as the Member for Nowgong to India’s Lower House of Parliament (the
Lok Sabha) in the Indian General Election of 1951, and we moved from,
what was then, the sleepy backwater of the state of Assam to the hurly-
burly of metropolitan life in India’s capital, New Delhi. In those years
of living in New Delhi, discussion of electoral politics and parliamentary
affairs was very much the staple of conversation within the home, and I
grew up with an easy familiarity with terms like ‘whipping members into
lobbies’, ‘lame-duck sessions’, ‘waving order papers’, and—indignity of
indignities—‘naming by the Speaker’. Parliament and elections were, so
to speak, ‘in my blood’. Years later, after I moved to England and became
an academic economist, when politics ceased to be part of life’s quotidian
rhythm, my interest in parliamentary elections did not wane. This book is
the product of that undimmed interest.
The foundations of this book lie in a set of data which records the
details of the election result for each candidate, for all the constituencies,
in every Lok Sabha General Election from 1962 to 2014. The edifice built
upon this foundation, and discussed in this book, is the result of inter-
rogating these data. The central purpose of this interrogation was to give
shape to the notion of ‘electoral efficiency’ by which is meant the capacity
of a party to convert votes into parliamentary seats. Parliamentary elec-
tions in India—and also elections to its state assemblies—are conducted
under the First-Past-The-Post (FPTP) system: a single representative for
each of 543 constituencies is elected—on the basis of obtaining the largest

vii

vk.borooah@ulster.ac.uk
viii PREFACE

number of votes of all the candidates contesting that constituency—as a

member of the Lok Sabha for that constituency.
The disjoint, under the FPTP electoral system, between the votes
obtained by a party and the seats won by it frequently causes surprise,
sometimes bordering on consternation. Most recently, in the UK General
Election of May 2015, the Scottish National Party won 56 seats in the
House of Commons on the back of just under 1.5 million votes, while in
the same election, the UK Independence Party received nearly 4 million
votes and were rewarded with just one seat. Unlike a proportional elec-
toral system, in which a party’s share of the total vote is a good predictor
of its share of parliamentary seats, the relation between a party’s votes and
its seats in an FPTP system works in mysterious ways. The primary pur-
pose of this book, as captured in its title, is to throw light on this relation-
ship for Indian parliamentary elections.
Given that India’s two main political parties—the Indian National
Congress (INC) and the Bharatiya Janata Party (BJP)—receive, between
them, over half the national parliamentary vote, the analysis in this book
is restricted to a comparison of the relative electoral efficiencies of these
two parties. This leads to a further constraint. The BJP made its electoral
debut in the 1984 Lok Sabha elections, winning just two seats, but really
got into its stride in the 1989 Lok Sabha elections when it won 85 seats.
Consequently, a great deal, but not all, of the analysis in this book is a
comparison of the INC and the BJP and, consequently, restricted to the
eight Lok Sabha elections of 1989, 1991, 1996, 1998, 1999, 2004, 2009,
and 2014.
The tenor of this book, consistent with my métier as an academic econ-
omist, is analytical, based upon a rigorous examination of the data. In
the process, I have drawn heavily upon the methodology of economics
and statistics to shed light on electoral outcomes in India. Chapter 3 uses
systems estimation techniques to predict the probabilities of the INC and
BJP winning elections—and Chap. 5 uses systems estimation techniques
to predict the vote shares of the INC and the BJP—in constituencies con-
tested by both parties; Chap. 4 uses Bayesian methods to analyse the issue
of anti-incumbency; Chap. 6 refines the concept of the ‘Cubic Law of
Elections’ to develop the concept of the ‘amplification coefficient’ which
amplifies votes into seats; and Chap. 7 measures vote concentration and
vote inequality with particular reference to the decomposition of inequal-
ity and carries out two significant simulations with regard to the inter-
constituency distribution of the INC and BJP vote.

vk.borooah@ulster.ac.uk
 PREFACE ix

In writing this book, my primary debt of gratitude is to Dr. Surjit Bhalla

and to Dr. Kai Gehring for providing me with the data upon which the
analysis in this book is based. Thanks are also due to Ulka Athale-Smit
for research assistance and to Vidya Borooah for commenting on parts
of the manuscript. Needless to say, I am solely responsible for this book’s
shortcomings.
I am grateful to my publisher Rachel Sangster for encouraging me
to write this book, to Gemma Leigh for her editorial work, to Jayashree
Ramamoorthy for overseeing the book’s production and to an anonymous
referee whose comments greatly improved the book’s presentation and
contents.

Vani K. Borooah
Belfast
January 2016

vk.borooah@ulster.ac.uk
vk.borooah@ulster.ac.uk
 CONTENTS

1 Introduction 1

2 The Anatomy of Indian Parliamentary Elections 17

3 Estimating the Likelihood of Winning Parliamentary

Constituencies 45

4 Incumbents, Challengers, and Electoral Risk 77

5 Analysis of Vote Shares 97

6 The Relationship Between Votes and Seats 115

7 The Concentration and Distribution of Votes 129

8 Conclusions 151

Index 155

xi

vk.borooah@ulster.ac.uk
vk.borooah@ulster.ac.uk
 LIST OF FIGURES

Fig. 2.1 Independent members of the Lok Sabha, 1962–2014 31

Fig. 2.2 Share of votes received by independent candidates in the
total vote: 1962–2014 32
Fig. 2.3 The criminal charge status of candidates in the 2004 and
2009 Lok Sabha elections 33
Fig. 3.1 INC and BJP predicted probabilities of winning marginal
constituencies with no independent candidates or just
two parties contesting 59
Fig. 3.2 INC and BJP predicted probabilities of winning marginal
constituencies at different rates of voter turnout 60
Fig. 3.3 INC and BJP predicted probabilities of winning marginal
constituencies at different vote shares 61
Fig. 3.4 Total number of seats contested by the INC and the BJP:
1984–2014 63
Fig. 3.5 Seats contested by the INC and the BJP in Hindi-speaking
states 69
Fig. 3.6 Seats contested by both the INC and the BJP in
Hindi-speaking states 70
Fig. 4.1 Net migration of seats (losses less gains) from the INC and
the BJP positive values represent a net outflow; negative
values a net inflow 84
Fig. 4.2 RR for the INC and the BJP compared 86
Fig. 4.3 OR for the INC and the BJP compared 87
Fig. 4.4 BF values for the INC and the BJP compared 88
Fig. 4.5 IRR for the INC and the BJP compared 92

xiii

vk.borooah@ulster.ac.uk
xiv LIST OF FIGURES

Fig. 4.6 IOR for the INC and the BJP compared 92
Fig. 4.7 IBF values for the INC and the BJP compared 93
Fig. 7.1 INC vote shares and vote share inequality: 1989–2014 140
Fig. 7.2 BJP vote shares and vote share inequality: 1989–2014 140
Fig. 7.3 Between income inequality as a proportion of total
inequality, decomposition by major states 143
Fig. 7.4 INC and BJP under- and over-performance with respect
to seats when each received an equal number of votes*
(*Negative and positive values represent, respectively,
under-and over-performance) 148

vk.borooah@ulster.ac.uk
 LIST OF TABLES

Table 2.1 Composition of the 16th Lok Sabha, by political party 23

Table 2.2 Average constituency size and turnout and inequality
in the distribution of Inter-constituency size and
turnout: 1962–2014 24
Table 2.3 Average turnout by major Indian states in the 2014
Lok Sabha elections 26
Table 2.4 Average number of candidates in a constituency: 1962–2014 28
Table 2.5 Average number of candidates in a constituency by
major Indian states in the 2014 Lok Sabha elections 30
Table 2.6 The distribution of candidates with reported criminal
charges by state, 2004 and 2009 Lok Sabha elections 35
Table 2.7 The distribution of Lok Sabha members with reported
criminal charges, by state: 2004 and 2009 Lok Sabha elections 36
Table 2.8 Party affiliation of members to the 14th Lok Sabha with
reported criminal charges 37
Table 2.9 Party affiliation of members to the 15th Lok Sabha with
reported criminal charges 38
Table 2.10 Party affiliation of members to the 16th Lok Sabha with
reported criminal charges 39
Table 3.1 Number of marginal constituencies in Lok Sabha elections:
1962–2014 47
Table 3.2 Marginal constituencies by major Indian states in the 2014
Lok Sabha elections 48
Table 3.3 Logit estimates for the probability of the INC winning
marginal constituencies 51
Table 3.4 Logit estimates for the probability of the BJP winning
marginal constituencies 53

xv

vk.borooah@ulster.ac.uk
xvi LIST OF TABLES

Table 3.5 INC and BJP predicted probabilities of winning marginal

constituencies 56
Table 3.6 Election-on-election changes in the likelihood of winning
marginal constituencies 57
Table 3.7 INC and BJP predicted probabilities of winning marginal
constituencies as the incumbent and non-incumbent parties 58
Table 3.8 Constituencies contested by the INC and the BJP 62
Table 3.9 Bivariate probit estimates for the likelihood of the INC and the
BJP winning constituency elections 65
Table 3.10 INC and BJP predicted likelihood of winning by year 67
Table 3.11 INC and BJP likelihood of winning at different constituency
vote shares 68
Table 3.12 INC and BJP likelihood of winning by year, non-Hindi and
Hindi-speaking major states 72
Table 3.13 INC and BJP predicted likelihood of winning at different
constituency vote shares: Hindi- and non-Hindi-speaking
major states 73
Table 4.1 Winning and incumbency outcomes for the INC and the
BJP: 1962–2014 81
Table 4.2 Incumbency/non-incumbency performance of the INC
in Lok Sabha elections 82
Table 4.3 Incumbency/non-incumbency performance of the BJP
in Lok Sabha elections 83
Table 4.4 Risk ratio and Bayes Factor calculations for the INC in
Lok Sabha elections 85
Table 4.5 Risk ratio and Bayes Factor calculations for the BJP in
Lok Sabha elections 86
Table 4.6 Inverse risk ratio and inverse Bayes Factor calculations for the
INC in Lok Sabha elections 90
Table 4.7 Inverse risk ratio and inverse Bayes Factor calculations for the
BJP in Lok Sabha elections 91
Table 5.1 Seemingly unrelated regression estimates for vote shares of
the INC and the BJP 99
Table 5.2 Equation statistics from the SURE equation 101
Table 5.3 The predictive performance of the SURE Model, 1989–2014 102
Table 5.4 Vote shares of incumbents and challengers: BJP and INC 104
Table 5.5 Differences in vote shares between BJP and INC incumbents
and challengers 105
Table 5.6 INC and BJP vote shares in Lok Sabha constituencies in
Hindi-speaking and non-Hindi-speaking states 108
Table 5.7 Vote shares of incumbents and challengers: BJP and INC
in Hindi-speaking states 109

vk.borooah@ulster.ac.uk
 LIST OF TABLES xvii

Table 5.8 Vote shares of incumbents and challengers: BJP and INC in
non-Hindi-speaking states 110
Table 5.9 Differences in vote shares between BJP and INC incumbents
and challengers, Hindi-speaking states 111
Table 5.10 Differences in vote shares between BJP and INC incumbents
and challengers, non-Hindi-speaking states 112
Table 6.1 BJP and INC election results for eight Lok Sabha elections: 1989–
2014 117
Table 6.2 Constituencies contested by the INC and the BJP in Lok
Sabha elections: 1989–2014 118
Table 6.3 Vote and seat ratios and values of the amplification coefficient
in Lok Sabha elections: 1989–2014 121
Table 6.4 Vote and seat ratios and values of the amplification coefficient
in Lok Sabha elections for Hindi-speaking statesa: 1989–2014 124
Table 6.5 Vote and seat ratios and values of the amplification coefficient
in Lok Sabha elections for non-Hindi-speaking statesa:
1989–2014 125
Table 7.1 The contributions of the major states to the total Lok Sabha
votea 132
Table 7.2 Values of the Herfindahl, entropy, and dissimilarity indices
for the vote contribution of major statesa 134
Table 7.3 Inequality in the distribution of constituency vote shares 139
Table 7.4 BJP and INC seats for eight Lok Sabha elections, 1989–2014,
under an equally distributed scenario 144
Table 7.5 BJP and INC seats for eight Lok Sabha elections, 1989–2014,
under an equal voters’ scenario 147

vk.borooah@ulster.ac.uk
 CHAPTER 1

Introduction

Abstract Borooah reviews the major developments in Indian politics

from independence in 1947 till the latest Parliamentary General Election
of 2014. He discusses the three pillars of India’s identity—democracy,
unity, and secularism—and shows the influence of democracy on India’s
unity and on the concept of secularism that it has adopted.

In his celebrated speech, delivered to India’s Constituent Assembly on

the eve of 15 August 1947, to herald India’s independence from British
rule, Jawaharlal Nehru, India’s first Prime Minister, famously asked if the
newly independent nation was ‘brave enough and wise enough to grasp
this opportunity and accept the challenge of the future’. If one conceives
of India, as many Indians would, in terms of a trinity of attributes—demo-
cratic in government, secular in outlook, and united by geography and a
sense of nationhood—then, in terms of the first of these, it would appear
to have succeeded handsomely.
Since the Parliamentary General Election of 1951, which elected the first
cohort of members to its lower house of Parliament (the Lok Sabha), India
has proceeded to elect, in unbroken sequence, another 15 such cohorts
so that the most recent Lok Sabha elections of 2014 gave to the country
a government drawn from members to form the 16th Lok Sabha. Given
the fractured and fraught experiences with democracy of India’s immedi-
ate neighbours (Pakistan, Bangladesh, Sri Lanka, Nepal, and Myanmar)
and of a substantial number of countries which gained independence from

© The Editor(s) (if applicable) and The Author(s) 2016 1

V.K. Borooah, Votes, Parties, and Seats,
DOI 10.1007/978-3-319-30487-8_1

vk.borooah@ulster.ac.uk
2 V.K. BOROOAH

colonial rule, it is indeed remarkable that independent India has known no

other form of governmental authority save through elections.
Elections (which represent ‘formal democracy’) are a necessary, but
not a sufficient, condition for ‘substantive democracy’. In a ‘substantive
democracy’, citizens not only vote, but also, having elected their represen-
tatives, continue to have a sense of involvement in public affairs engen-
dered by a sense that their views are heeded by those they have chosen
to represent them (Huber et al. 1997). By corollary, substantive democ-
racy requires one to listen to opposing points of view and to respond to
these in measured tones. The shrillness of public discourse in India, both
within Parliament and outside it, may give the impression that substan-
tive democracy, as opposed to its formal counterpart, is still a distant goal
for the country. And yet, as Corbridge et al. (2013) note, India has pro-
gressed a long way along the road to becoming a genuine democracy: peo-
ple participate in politics, are more conscious of their rights, and are aided
in this by a plethora of rights-based bodies and movements (Cornwall and
Nyamu‐Musembi 2004).
India’s real achievement in promoting substantive democracy has been
at the level of the village. The Constitution (73rd Amendment) Act of
1993 made it mandatory for all villages to have a village council (Gram
Sabha) consisting of all registered voters on the electoral roll of that vil-
lage. The Gram Sabha was to be entrusted with the power of supervis-
ing the functioning of the elected village panchayat and to approve the
panchayat’s development plan for the village and the associated budget.
Consequently, in addition to voting, electors in villages had another form
of political participation: they could attend Gram Sabha meetings and
participate in its discussions. The 73rd Amendment Act stipulated that
one-third of seats in the village panchayat should be reserved for women
and disadvantaged groups like the Scheduled Castes and Scheduled
Tribes.1
In delivering India’s ‘democratic achievement’, Nehru’s role in estab-
lishing the primacy of Parliament cannot be exaggerated. Even as stern a
critic of Nehru as Anderson (2012) admitted that:

Nehru’s greatness, it is generally felt, was to rule as a democrat in a non-

Western world teeming with dictators. Preceptor to his nation, he set an
example from which those who came after him could not long depart.
Tutored by him, Indian democracy found its feet, and has lasted ever since.
That by conviction Nehru was a liberal democrat is clear. Nor was this a

vk.borooah@ulster.ac.uk
 INTRODUCTION 3

merely theoretical attachment to principles of parliamentary government.

As Prime Minister, he took his duties in the Lok Sabha with a conscientious
punctilio that put many Western rulers to shame, regularly speaking and
debating in the chamber, and never resorted to rigging national elections
or suppressing a wide range of opinion. So much is incontestable. (p. 26)

It is, of course, true that Nehru burnished his credentials as a demo-

crat at a period when the party he headed—the Indian National Congress
(INC)—dominated India’s political landscape. In the Lok Sabha elections
of 1951, the INC received 45 percent of the vote and won 75 percent of
the seats to the 1st Lok Sabha (364 seats out of a total of 489). Its clos-
est rivals in that General Election were Independent candidates and the
Socialist Party who took, respectively, 37 and 12 seats on the back of,
respectively, 16 and 12 percent of the total vote. The Communist Party of
India received only 3.3 percent of the vote, but managed to win 16 seats.
The dominance of the INC was unchanged in subsequent General
Elections. In the General Election of 1957, the INC received 45 percent
of the national vote to take 75 percent of the available Lok Sabha seats
(371 seats out of the total of 494); in the 1962 General Election, the INC
received 48 percent of the vote and won 361, out of an available 494, seats
(75 percent), while in the 1967 General Election, which was the first after
Nehru’s death in 1964, the INC received 41 percent of the vote to win
283 out of a total of 520 seats (54 percent). The dominance of the INC in
Indian parliamentary elections survived the 1971 General Election—when
on the back of nearly 44 percent of the national vote, it won 352 seats in
a House of 546—before coming to a juddering halt in 1977.
The Lok Sabha elections of 1977 were held after the longest period
between two successive elections in India, the last elections being held
six years earlier in 1971. It offered voters an opportunity to express
their views on (Jawaharlal Nehru’s daughter) Indira Gandhi’s imposi-
tion of a state of Emergency in India in 1975.2 This was triggered by the
Allahabad High Court setting aside Mrs Gandhi’s election as the member
for Rae Bareli, in the state of Uttar Pradesh, in the Lok Sabha election
of 1971 because it found that as Prime Minister, she had illegally used
the machinery of government for electoral purposes. Instead of stepping
aside in favour of a caretaker Prime Minister, pending the outcome of
her appeal to the Supreme Court, Mrs Gandhi imposed an ‘Emergency’.
It lasted 21 months, from June 1975 till March 1977, and during this

vk.borooah@ulster.ac.uk
4 V.K. BOROOAH

period ‘elections were suspended, political and civil organisations were

disbanded, and the media was gagged’ (Corbridge et al. 2013).
The electorate’s verdict in the Lok Sabha elections of 1977 could not
have been clearer: the vote share of the INC fell from 44 percent in the
1971 General Election to 35 percent in 1977 with a corresponding fall
from 352 to 154 in the number of Lok Sabha seats held by the party. Mrs
Gandhi lost her Rae Bareilly seat, and at the age of 81, Morarji Desai
became India’s fourth Prime Minister.3
Prior to the Lok Sabha election of 1977, the main opposition to the
INC came from Independent candidates, the communists, and a party,
espousing economic liberalism (the Swatantra Party), which had come
to prominence in the Lok Sabha election of 1962 winning 18 seats with
nearly 8 percent of the national vote. After the 1977 Lok Sabha election,
however, a new form of opposition emerged in the shape of a coalition of
parties, of various ideologies, which came together solely for the purpose
of winning elections by fielding common candidates. This amalgam was
called the Janata Party, and it formed the post-1977 government with
Desai as Prime Minister, but with Charan Singh, leader of the Bharatiya
Lok Dal (BLD)—one of the Janata Party’s most powerful constituents—
and Home Minister and also Deputy Prime Minister to Desai, waiting in
the wings to take over.
If proof was ever needed of the futility of relying for stable government
on a coalition of partners, united by nothing except electoral convenience,
and led by persons of overweening political ambition, then the Janata
Party government of 1977 provided it in abundance. Within two years,
the BLD, by threatening to withdraw its support from the government,
made Desai’s position untenable; his resignation in July 1979 was quickly
followed by Charan Singh taking over as Prime Minister. Singh’s sup-
port, however, quickly haemorrhaged away, and he lasted just three weeks.
Fresh elections were scheduled for 1980. The INC, under Mrs Gandhi,
returned in triumph winning that election and (under the leadership of
her son Rajiv Gandhi, who succeeded her as Prime Minister, after her
assassination in October 1984) winning the next election, in December
1984, as well.
The Lok Sabha elections of 1989 were significant for three reasons.

1. They marked the emergence of the Bharatiya Janata Party (BJP) as a

serious political force when it won 85 Lok Sabha seats in that election;

vk.borooah@ulster.ac.uk
 INTRODUCTION 5

it, thereby, redeemed itself after the indignity of winning just two seats
in its electoral debut in the previous Lok Sabha elections of 1984.
2. The 1989 election and the 1998 Lok Sabha elections bookended a
period of parliamentary instability during which in a span of 10 years,
India voted in five Lok Sabha elections: 1989, 1991, 1996, 1998, and
1999.4
3. In the decade after the 1989 Lok Sabha elections, India experienced—
what Jaffrelot (2003) termed—a ‘silent revolution’ as lower-status
groups increasingly captured political office and used political power to
alter the balance of power between the upper and the lower castes.
Each of these aspects is discussed, in turn, below.

1.1 THE RISE OF THE BJP

The rise of the BJP was significant in two respects: (i) it offered voters,
in the form of Hindu nationalism, an alternative to the ‘secular’ model
propagated by Nehru and which was the bedrock of the INC’s ideology5
and (ii) for the first time, there was the prospect of two-party democracy
in India. After two short-lived attempts to be a party of government (fol-
lowing the Lok Sabha elections of 1996, a BJP-led government lasted 13
days, and following the 1998 Lok Sabha elections, a BJP-led government
lasted just over a year), the BJP, as the senior partner in a coalition of other
parties, was at last able to offer the country stable government in 1999
when Prime Minister Vajpayee’s government saw out the full five years of
the 13th Lok Sabha and the Lok Sabha elections in 2014 ushered in a BJP
majority government with Narendra Modi as Prime Minister.

1.2 POLITICAL INSTABILITY DURING THE 1990S

In terms of government, the outcome of the 1989 election was that the
INC, even though it was the largest single party, went into opposition and a
minority ‘National Front’ government, with V.P. Singh as Prime Minister,
was formed with support from the Leftist Parties and the BJP. Subsequent
infighting within the parties comprising the ‘National Front’, in conjunc-
tion with the BJP withdrawing its support over the Ayodhya temple issue,
resulted in the government resigning after losing a vote of confidence
in November 1990. The new government was another minority govern-
ment, with Chandra Shekhar as Prime Minister, this time supported by
the INC. However, within the next few months, the INC withdrew its

vk.borooah@ulster.ac.uk
6 V.K. BOROOAH

support—on the charge that the government was ‘spying’ on the INC’s
leader, Rajiv Gandhi—paving the way for the dissolution of the 9th Lok
Sabha and the start of the General Election campaign of 1991.
After the 1991 elections, the INC, with 244 seats, formed the govern-
ment (with Narasimha Rao as Prime Minister after the INC’s dynastic
heir, Rajiv Gandhi, was assassinated in May 1991) with the BJP, on 120
seats, as the main opposition. This government lasted its parliamentary
term, and the 1996 General Election followed five years later.
The INC made a poor fist of it in the 1996 elections: even though it
won a larger share of the vote than the BJP (28.8 percent compared to
20.3 percent), it ended up winning fewer seats (140 compared to the
BJP’s 161). With the BJP, as the largest single party, unable to form a gov-
ernment—Atal Bihari Vajpayee lasted just 13 days as Prime Minister—and
the INC, as the next largest party, refusing even to try, the outcome was a
minority government. This was formed as a coalition of several smaller par-
ties and labelled the ‘United Front’ with Deve Gowda as Prime Minister.
The United Front excluded both the BJP and the INC but was sup-
ported by the latter. In April 1997, the INC withdrew its support to
the United Front, which was increasingly beset by internal wrangling
between its constituent parties, but agreed to support another United
Front coalition (which included Lalu Prasad Yadav’s Rashtriya Janata Dal
and Mulayam Singh Yadav’s Samajwadi Party as constituents) with Inder
Gujral as Prime Minister. Eleven months later, after Gujral had refused to
accede to the INC’s demand to drop three ministers from the Dravida
Munnetra Kazhagam (DMK) party from his government, the INC with-
drew its support, and this government also collapsed: the 11th Lok Sabha
was dissolved, and in February 1998, a fresh set of parliamentary elections
were held to elect the 12th Lok Sabha.6
The 1998 elections continued the low fortunes of the INC: it obtained
the same share of the national vote as the BJP (26 percent) but ended up
with fewer seats (141 compared to the BJP’s 182). The outcome of the
election was a coalition government, headed by the BJP with Atal Bihari
Vajpayee as Prime Minister. By the end of the year, this coalition also
collapsed as one of its partners—the All India Anna Dravida Munnetra
Kazhagam (AIADMK) with 18 seats—withdrew its support. This led to
the General Election of September 1999 to elect the 13th Lok Sabha.7
Since the Lok Sabha elections of 1999, India has enjoyed stable govern-
ment with each government completing its five-year term. The 1999 elec-
tion resulted in a coalition government, labelled the National Democratic

vk.borooah@ulster.ac.uk
 INTRODUCTION 7

Alliance (NDA), with the BJP’s Atal Bihari Vajpayee as Prime Minister.
The Lok Sabha elections of 2004 and 2009 resulted in an INC-led
coalition of centre-left parties, labelled the United Progressive Alliance
(UPA): respectively, UPA-I and UPA-II. After the Lok Sabha elections of
2014, Narendra Modi became Prime Minister as head of a BJP majority
government.

1.3 THE RISE OF THE LOWER CASTES

Foremost amongst these lower caste groups who achieved political
prominence in the 1990s were the ‘Other Backward Classes’ (Jaffrelot
2003). These were castes that were not ‘forward castes’—in the sense of
belonging to the Brahmin, Kshatriya, or Vaishya varnas—but who were
not, unlike the Scheduled Castes, considered ‘untouchable’. Originally,
they were mobilised by the upper caste INC, but they now mobilised
themselves against the INC. In the context of Indian politics, the ‘Other
Backward Classes’ is a useful electoral category encapsulating the lower
castes above the pollution line who try, by voting along caste lines, to carve
political space for themselves.8
The catalyst for this ‘silent revolution’ was the Mandal Commission’s
Report of 1980 which ‘recommended that, in addition to the 23 percent
of government jobs reserved for the Scheduled Castes (SC) and Scheduled
Tribes (ST), a further 27 percent should be reserved for the (OBC). In
1990, V.P. Singh’s government announced plans to implement this rec-
ommendation triggering a wave of ‘anti-Mandal’ rioting in India. In 1992,
India’s Supreme Court, in Sawhney Vs. The Union of India, upheld jobs
reservation for the Other Backward Classes but ruled that: (i) reservation
was not to extend to more than 50 percent of the population and (ii) that
groups within the (OBC) category who were manifestly not disadvan-
taged (the ‘creamy layer’) were to be excluded from reservation’.
The implementation of the Mandal Commission’s Report cemented
social identity into the basic structure of Indian politics by establishing, for
nearly half of India’s population, a clear association between social status
(based on caste) and economic status (based on education and employ-
ment). To belong to a ‘reserved category’—the Scheduled Castes, the
Scheduled Tribes, and the Other Backward Classes—meant access to edu-
cation and jobs on terms which were more favourable than those avail-
able to persons who did not belong to these categories. It was, therefore,
important to preserve, and be aware of, one’s caste identity because thanks

vk.borooah@ulster.ac.uk
8 V.K. BOROOAH

to ‘reservation’—the scope of which was greatly extended by the Mandal

Commission—it was the proverbial goose that laid golden eggs. Any polit-
ical party that even remotely suggested that the policy on ‘reservation’
should be diluted or phased out—for example, on the grounds that it was
unfair to persons in the non-reserved categories or that it might impact
adversely on quality—would, in effect, be committing electoral suicide.9
Nonetheless, a conversation that Indians need to have amongst them-
selves—but, given the emotive nature of the topic, probably will not
have—is about the costs and benefits of reservation in a post-Mandal
society in which the reserved category is defined so widely. There can be
little doubt that but for reservation, many students, currently in institu-
tions of higher education for which entry is highly competitive, would
not be there (Vishnu 2015).10 It is perfectly reasonable to ask whether the
presence of the ‘less qualified’ dilutes the quality of education offered by
such institutions with the result that they turn out, for example, doctors
and engineers who have been subjected to less than rigorous intellectual
scrutiny. To ask such a question is not to suggest that they should not be
admitted. There may well be compelling social arguments, including the
need to blunt discrimination against persons from the backward castes,
why reservation should be continued. These reasons should, however, be
spelled out and set alongside the costs, if any, of reservation. The ‘golden
eggs’ argument on its own is a poor reason for continuing with reservation
in its present form.

1.4 DEMOCRACY AND SECULARISM

Another aspect that impacts on electoral outcomes in India is its attitude
towards religions, enshrined in its commitment to being a secular nation.
Secularism in India, however, means something very different from that
in France which, too, prides itself on being secular. The French attitude to
secularism, enshrined in its principle of laïcité, actively prevents religious
interference in state affairs. This dates back to the French Revolution of
1789 and is traditionally understood as a way of controlling the Catholic
Church (Hussey 2014). In India, however, secularism, in operational
terms, requires avoiding doing or saying anything which might ‘hurt a
community’s religious sentiments’.11 So, in France, for example, pupils
and staff are forbidden, since February 2004, to conspicuously wear any
religious symbols at school, in particular, headscarves by Muslims. From
April 2007, these rules were also applied to public employees, and from

vk.borooah@ulster.ac.uk
 INTRODUCTION 9

April 2011, the niqab was banned in public places. All these laws were
based on the principle of secularism (laïcité).
In India, on the other hand, the wearing of religious symbols is not
only permitted, but also anyone who objected to, or took action to pre-
vent, such practices, by ‘hurting the (relevant) community’s religious sen-
timents’, would be viewed as non-secular or, to use a term popular in
India, ‘communal’. In France, the right of the magazine Charlie Hebdo
to publish anti-Islamic cartoons is regarded as a triumph of secularism; in
India, the ban on Salman Rushdie’s Satanic Verses, because it would, or
might, offend Muslim sentiments, is also a victory for secularism. On the
French view of secularism, the sacred is trumped by the profane; from the
Indian perspective, the sacred invariably trumps the profane.
There can be little doubt that some actions taken in France, on the prin-
ciple of laïcité, are insensitive to cultural differences and may even appear
provocative. The decision of some schools not to offer their Muslim and
Jewish pupils any dietary alternative when pork is the main item on the
schools’ menu is a good example of cultural insensitivity.12 It is safe to say
that on an Indian view of secularism, a similar situation would not arise.
In India, however, problems arise when what practitioners of a religion
find hurtful is taken to unreasonable, indeed unacceptable, extremes. The
reluctance to employ persons from the lower castes to cook school meals,
in order not to offend upper-caste sensibilities that food touched by lower
caste hands is rendered unclean, is an example of such pathology. On the
French principle of laïcité upper caste children would have to eat food
cooked by lower caste persons or else go hungry. The Indian attitude is
to tiptoe around the problem and, with much handwringing, attempt to
square the circle by expressing sympathy for both points of view.
A consequence of secularism in India is that each religion has an incen-
tive to preserve its identity in undiluted form—immune to any proposals
for reform—because such proposals, by ‘hurting its sentiments’, would fall
foul of the secular principle. The upshot is that the same heightened sense
of identity that reservation policies provide the backward caste groups is
provided by secularism to religious groups.
Arguably, Muslims in France and India have, in different ways, been
most affected by each country’s particular concept of secularism. In France,
Muslims, more than other religious groups, have been subject to the full
rigour of laïcité in terms of how they lead their public and personal lives.
In India, the policy towards Muslims has been one of non-interference,
most particularly with respect to Muslim personal law.

vk.borooah@ulster.ac.uk
10 V.K. BOROOAH

As regards the latter, a Muslim man in India can divorce his wife by
simply saying talaq (divorce) thrice and the All India Muslim Personal
Law Board declared in September 2015 that there was no scope of change
in the triple talaq system.13 Notwithstanding court judgements to the
contrary, Muslim husbands who divorce their wives are not required to
pay them alimony. This is due to the (INC inspired) Muslim Women
(Protection of Rights on Divorce) Act, 1986 which gave Muslim women
the right to maintenance for only three months after divorce after which
the onus of their maintenance was on their relatives.14 Yet, no attempt
is made to establish basic rights for Muslim women, in the form of pro-
tection from arbitrary divorce or maintenance payments in the event of
divorce, because it would be tantamount to attacking ‘Muslim identity’
and, therefore, fall foul of the secularism principle. This is notwithstanding
the fact that Article 44 of the Indian Constitution specifically requires the
state to secure for its citizens a uniform civil code throughout the territory
of India.
The failure to bring Muslims into mainstream life in India has, in fact,
failed Muslims. The Sachar Committee (2006), in its report to the gov-
ernment of India, quantified and highlighted the backwardness of Indian
Muslims. This report drew attention to a number of areas of disadvantage:
inter alia, the existence of Muslim ghettos stemming from their concern
with physical security; low levels of education engendered by the poor
quality of education provided by schools in Muslim areas; pessimism that
education would lead to employment, difficulty in getting credit from
banks; and the poor quality of public services in Muslim areas. In conse-
quence, as the Sachar Committee reported: one in four of 6- to 14-year
old Muslims had never attended school; less than 4 percent of India’s
graduates were Muslim, notwithstanding that Muslims comprised 13 per-
cent of India’s population; and only 13 percent of Muslims were engaged
in regular jobs, with Muslims holding less than 3 percent of jobs in India’s
bureaucracy.15
One of the reasons for protecting Muslim identity in India is because
it is acknowledged that Muslims—who, according to the 2011 Census,
comprise 14 percent of the population, with about 170 million adher-
ents—play a crucial role in determining electoral outcomes in India. On
one estimate, Muslim voters play a decisive role in determining the out-
comes in about 100 (of the total of 543) constituency elections.16 At the
same time, there are a large number of parties, national and regional, com-
peting for the Muslim vote: inter alia, the INC, the Rashtriya Janata

vk.borooah@ulster.ac.uk
 INTRODUCTION 11

Dal, the Samajwadi Party, the Aam Aadmi Party, the All India Majlis-e-
Ittehadul Muslimeen, and the All India United Democratic Front, with the
latter two being explicitly Muslim parties. Paralleling the earlier discussion
on reservations, any political party in India that suggested measures that
might, even remotely, be construed as an attack on Muslim identity would
have to suffer the consequences of losing the Muslim vote.

1.5 DEMOCRACY AND NATIONAL UNITY

The last point of note is that there is the feeling that is generated in most
Indians that by voting in elections which are regular and frequent, and in
large part acknowledged to be free and fair, they are collectively involved
in a national project of some importance. The turnout in Indian elections
is high: 67 percent of voters exercised their franchise in the 2014 Lok
Sabha elections and the average turnout, over the 14 Lok Sabha elections
between 1962 and 2014, was 59 percent. The next chapter discusses the
myriad reasons that bring voters to the polling booth—party loyalty, the
desire to reward or punish candidates, and the prospect of material gain—
but, in general, as Banerjee (2014) argues, ‘for many Indian voters, voting
is not just a means to elect a government…rather the very act of voting is
seen by them as meaningful, as an end in itself, that expresses the virtues
of citizenship, accountability, and civility that they wish to see in ordinary
life, but rarely can’ (p. 3).
Needless to say, an appreciation of possessing the right to elect one’s
government is not spread evenly across the country. If one views India,
as Rudolph and Rudolph (2002) do, as a multinational federation, rather
than as a nation state, with more in common with the European Union
than with the USA, then it is not surprising that at various times, some
parts of the country have been excluded from the democratic process:
elections in Assam and Punjab could not be held during the 1984 Lok
Sabha elections (but were held in 1985); elections in Assam could not
be held during the 1989 Lok Sabha elections because electoral rolls were
incomplete; elections in Punjab could not be held during the 1991 Lok
Sabha elections (but were held in 1992); and the 1991 Lok Sabha elections
in Jammu and Kashmir were boycotted in all its constituencies.17
Some commentators, most recently Anderson (2012), have seized
upon these aberrations—and on the civil unrest that engendered them
and upon the raft of legislation enacted to suppress such unrest—to draw
attention to the deficiencies of Indian democracy. It is undoubtedly true

vk.borooah@ulster.ac.uk
12 V.K. BOROOAH

that many of these legislative measures, listed in Anderson (2012),18 sub-

orn the democracy that they purport to defend. Of these, none is more
odious than the Armed Forces (Special Powers) Act (1958)—which con-
tinues to be in effect in Jammu and Kashmir and in India’s north-eastern
states—under which members of the armed forces can act as though they
were in a situation of war: able to stop, search, arrest, and kill without
judicial accountability.
That said, the other side of the ledger is that since independence in
1947, India has elected 16 successive Parliaments under the aegis of nearly
10,000 constituency elections. It has never known military rule nor have
Indian political leaders, even in the darkest days of the Emergency of
1975–77, ever harboured presidential ambitions. In terms of geography,
it has known serious unrest in only four regions: Jammu and Kashmir,
Punjab, the northeast, and the swathe of districts running from Jharkhand
to Andhra Pradesh which have come under the influence of Maoist guer-
rillas. Some of these conflicts were in the past (Punjab), and in others (in
the northeast and in Jammu and Kashmir), efforts at a negotiated settle-
ment are under way. Balancing the books, it would take an extraordinary
degree of pessimism to envisage a dystopian future for Indian democracy.

NOTES
1. In response to the burden of social stigma and economic backwardness
borne by persons belonging to India’s ‘untouchable castes’, the
Constitution of India allows for special provisions for their members.
These are mainly in the form of reserved seats in the national Parliament,
state legislatures, municipality boards, and village councils (panchayats);
job reservations in the public sector; and reserved places in public higher
educational institutions. Articles 341 and 342 include a list of castes enti-
tled to such benefits, and all those groups included in this list—and subse-
quent modifications to this list—are referred to as the ‘Scheduled Castes’.
Similarly, Articles 341 and 342 also include a list of tribes entitled to similar
benefits, and all those groups included in this list—and subsequent modi-
fications to this list—are referred to as the ‘Scheduled Tribes’.
2. After Lal Bahadur Shastri’s untimely death, she was India’s third Prime
Minister. This count excludes Gulzarilal Nanda, who was interim Prime
Minister twice: first, from 27 May to 9 June 1964, after Nehru’s death and
Shastri’s appointment, and then, from 11 to 24 January 1966, after
Shastri’s death and Mrs Gandhi’s appointment.
3. See the previous note on Gulzarilal Nanda.

vk.borooah@ulster.ac.uk
 INTRODUCTION 13

4. If both the 9th and the 10th Lok Sabha had lasted their parliamentary
terms of five years, there would have been three fewer elections: in 1989,
1994, and 1999.
5. The 42nd Amendment (1976) to the Indian Constitution, passed under
Mrs Gandhi’s government, declared India to be a secular country.
6. The DMK was allegedly criticised by the Jain Commission’s inquiry into
Rajiv Gandhi’s assassination in 1991 though since the Jain Commission’s
report was never made public, the allegation could not be substantiated.
7. Its demands were that: a former naval chief, who had been sacked, should
be reinstated; the Defence Minister George Fernandes should be relieved
of his portfolio; and a Joint Parliamentary Committee probe should be
ordered (Frontline, 24 April—7 May 1999, http://www.frontline.in/
static/html/fl1609/16090160.htm, accessed 26 November 2015).
8. It is a political cliché in India to view a person’s caste as an important deter-
minant of the party he/she will vote for. In her eponymous book, Chandra
(2004) asks why ethnic parties succeed. The Bahujan Samaj Party (BSP)
is the main Indian political party which espouses the cause of the Scheduled
Castes—who comprise 17 percent of India’s population—against that of
the upper castes. It employs the same methods of caste mobilisation in
every state—all of which have the same electoral system—but meets with
different degrees of success in different states. In one, Uttar Pradesh, it has
formed governments; in a second group of states, it obtains moderate lev-
els of support, but not enough to form a government; in a third group of
states, it draws a blank. Chandra’s answer is that the elites amongst the
Scheduled Castes weigh the advantages, in terms of access to the state
patronage system, of voting for their ‘own’ party, the BSP, against voting
for another party. If the BSP falls short in this calculation, then it fails to
attract votes from even its own ethnic group, the Scheduled Castes. The
conclusion of her analysis is that the caste basis for voting cannot be taken
for granted—it depends upon the circumstances.
9. As happened with the BJP when, on the eve of the 2015 Bihar Assembly
elections, one of its senior leaders asked for a rethink on the policy on res-
ervation: he suggested that a ‘non-political committee’ be set up to exam-
ine who needs the benefit of reservation and for how long (NDTV, 22
September 2015, http://www.ndtv.com/india-news/rss-chief-mohan-
bhagwats-statement-on-reservation-sparks-debate-1220171, accessed 26
November 2015)
10. For example, Vishnu (2015) reports that in the academic year 2014–15
the elite Indian Institutes of Technology admitted 2,029 students from the
Scheduled Castes and 856 students from the Scheduled Tribes of whom
only 432 and 80, respectively, would have secured admission in an open
competition based on examination performance.

vk.borooah@ulster.ac.uk
14 V.K. BOROOAH

11. There are convoluted attempts to define the Indian concept of secularism.
Bhargava (2010), for example, defines a secular state as ‘not anti-religious
but existing and surviving only when religion is no longer hegemonic…it
allows freedom of religion but is itself free from religion’. It is difficult to
see how such a platitudinous definition distinguishes the Indian version of
secularism from the French type.
12. The Guardian, 13 October 2015, ‘Pork or Nothing: How School Dinners
are Dividing France’, http://www.theguardian.com/world/2015/
oct/13/pork-school-dinners-france-secularism- children-religious-
intolerance, accessed 30 November 2015.
13. First Post, http://www.firstpost.com/india/no-scope-of-change-in-triple-
talaq-system-says-all-india-muslim-personal-law-board-2419482.html,
accessed 27 November 2015.
14. The Hindu, http://www.thehindu.com/thehindu/2003/08/10/sto-
ries/2003081000221500.htm, accessed 27 November 2015.
15. In 1865, Napoleon III gave Algerian Muslims the right to be governed, in
non-criminal cases, by Islamic law rather than the French Civil Code with
the result that Muslims ‘had no control or stake in the country in which
they lived’ (Hussey 2014).
16. Deutsche Welle, 10 April 2014, http://www.dw.com/en/muslims-to-
play-key-role-in-indian-elections/a-17558549, accessed 28 November
2015.
17. Srinagar, Ladakh, Baramulla, Anantnag, Jammu, and Udampur
18. The Preventive Detention Act (1950), Armed Forces (Special Powers) Act
(1958), Unlawful Activities (Prevention) Act (1967), Prevention of Insults
to National Honour Act (1971), Maintenance of Internal Security Act
(1971), National Security Act (1980), Terrorism and Disruptive Activities
(Prevention) Act (1985), Prevention of Terrorism Act (2002), and the
Unlawful Activities (Prevention) Amendment Act (2004).

REFERENCES
Anderson, P. (2012). After Nehru. The London Review of Books, 34, 21–36.
Banerjee, M. (2014). Why India votes? New Delhi: Routledge.
Bhargava, R. (2010). India’s secular constitution. In A. Vanaik & R. Bhargava
(Eds.), Understanding contemporary India: A critical perspective (pp. 19–48).
New Delhi: Orient Blackswan.
Chandra, K. (2004). Why ethnic parties succeed: Patronage and ethnic head counts
in India. Cambridge: Cambridge University Press.
Corbridge, S., Harriss, J., & Jeffrey, C. (2013). India Today: Economics, politics,
and society. Cambridge: Polity Press.

vk.borooah@ulster.ac.uk
 INTRODUCTION 15

Cornwall, A., & Nyamu‐Musembi, C. (2004). Putting the ‘rights‐based approach’

to development into perspective. Third World Quarterly, 25, 1415–1437.
Huber, E., Rueschemeyer, D., & Stephens, J. D. (1997). The paradoxes of con-
temporary democracy: Formal, participatory, and social dimensions.
Comparative Politics, 29, 323–342.
Hussey, A. (2014). The French Intifada: The long war between France and its Arabs.
London: Granta Books.
Jaffrelot, C. (2003). India’s silent revolution: The rise of the lower castes in Northern
India. New York: Columbia University Press.
Rudolph, S. H., & Rudolph, L. I. (2002). New dimensions in Indian democracy.
Journal of Democracy, 13, 52–66.
Sachar Committee Report. (2006). The social and economic status of the Muslim
community in India. New Delhi: Government of India (Cabinet Secretariat).
Vishnu, U. (2015, August 6). They get leg-up at JEE but hard landing on campus.
Indian Express.

vk.borooah@ulster.ac.uk
 Chapter 2

The Anatomy of Indian Parliamentary

Elections

Abstract Borooah discusses the twin pillars of India’s electoral system:

the Election Commission of India (ECI) which oversees and regulates the
electoral activities of political parties; and the plethora of political parties
which, through their candidates, seek the mandate of voters and, by doing
so, subject themselves to the regulatory supervision of the ECI. He then
examines features of the system of elections to India’s lower house of par-
liament with respect to the size of the electorate, the percentage of voters
that turned out to cast their vote, and the candidates that offer themselves
to the voters’ judgement. Lastly, he examines the electoral performance of
candidates who had criminal charges against them.

2.1   Introduction
In elections to India’s lower house of parliament (the Lok Sabha), a single
representative for each of 543 constituencies is elected—on the basis of
obtaining the largest number of votes of all the candidates contesting that
constituency—as the Member for that constituency. This system of elec-
tion is known as the First-Past-The-Post (FPTP) system. In this chapter,
we examine features of the system of elections to India’s lower house of
parliament (hereafter, the Lok Sabha) with respect to the size of the elec-
torate, the percentage of voters that turned out to cast their vote, and the
candidates that offer themselves to the voters’ judgement. Using recently
available data, we examine the consequences of voters being able, under

© The Editor(s) (if applicable) and The Author(s) 2016 17

V.K. Borooah, Votes, Parties, and Seats,
DOI 10.1007/978-3-319-30487-8_2

vk.borooah@ulster.ac.uk
18 V.K. Borooah

a 2013 ruling by India’s Supreme Court, to reject all available choices

by availing of the option of voting for a fictional candidate, ‘None of the
Above’ (NOTA). We also examine the electoral performance of candidates
who (following a 2003 Supreme Court ruling, requiring all candidates to
reveal, six months before they filed their candidacy papers, whether there
were outstanding criminal charges against them) had criminal charges
against them.
Before doing so, we discuss in the next two sections the twin pillars of
India’s electoral system: the Election Commission of India (ECI) which
oversees and regulates the electoral activities of political parties with the
power to proscribe any activity (or activities) it feels inappropriate to the
electoral process; and the plethora of political parties which, through their
candidates, seek the mandate of voters and, by doing so, subject them-
selves to the regulatory supervision of the ECI.

2.2   The Election Commission of India

The ECI is a body mandated under Article 324(2) of the Indian
Constitution and currently comprises a Chief Election Commissioner and
two Election Commissioners.1 Its constitutional role is the ‘superinten-
dence, direction, and control of elections’. Under the Representation of
the People Acts of 1950 and 1951, the ECI appoints the Chief Electoral
Officer in each state or Union Territory (UT), the District Election
Officer for each district, and the Returning Officer for each Lok Sabha or
Assembly constituency where the latter is responsible for the conduct of
elections in that constituency. The ECI in consultation with the state or
UT government appoints an Electoral Registration Officer who is respon-
sible for the preparation of the electoral rolls for each constituency in that
state or UT. The District Election Officer then appoints the Presiding
Officer for a particular polling station who, with the assistance of Polling
Officers, is responsible for voting at that station. In addition, the ECI may
appoint ‘observers’ to a particular constituency—either with respect to the
general conduct of that election or, more specifically, with respect to elec-
tion expenditures—who then report directly to the ECI.2
The ECI has progressively tightened its views on permissible campaign-
ing practices through its Model Code of Conduct (MCC).3 At the start of an
election period, this Code with the MCC sets out an elaborate set of param-
eters within which elections should be conducted. In general, the MCC
places strictures on the conduct of the election campaign by requiring that:

vk.borooah@ulster.ac.uk
 The Anatomy of Indian Parliamentary Elections  19

1. ‘No party or candidate shall include in any activity which may aggra-
vate existing differences or create mutual hatred or cause tension
between different castes and communities, religious or linguistic’.
2. ‘Criticism of other political parties, when made, shall be confined to
their policies and programme, past record and work. Parties and
Candidates shall refrain from criticism of all aspects of private life, not
connected with the public activities of the leaders or workers of other
parties. Criticism of other parties or their workers based on unverified
allegations or distortion shall be avoided’.
3. ‘There shall be no appeal to caste or communal feelings for securing
votes. Mosques, Churches, Temples or other places of worship shall
not be used as forum for election propaganda’.
4. ‘All parties and candidates shall avoid scrupulously all activities which
are “corrupt practices” and offences under the election law such as
bribing of voters, intimidation of voters, impersonation of voters, can-
vassing within 100 meters of polling stations, holding public meetings
during the period of 48 hours ending with the hour fixed for the close
of the poll, and the transport and conveyance of voters to and from
polling station’.
5. ‘The right of every individual for peaceful and undisturbed home life
shall be respected, however much the political parties or candidates
may resent his political opinions or activities. Organizing demonstra-
tions or picketing before the houses of individuals by way of protesting
against their opinions or activities shall not be resorted to under any
circumstances’.
6. ‘No political party or candidate shall permit its or his followers to make
use of any individual’s land, building, compound wall etc. without his
permission for erecting flag-staffs, suspending banners, pasting notices,
writing slogans etc.’.
7. ‘Political parties and candidates shall ensure that their supporters do
not create obstructions in or break up meetings and processions orga-
nized by other parties. Workers or sympathisers of one political party
shall not create disturbances at public meetings organized by another
political party by putting questions orally or in writing or by distribut-
ing leaflets of their own party. Processions shall not be taken out by
one party along places at which meetings are held by another party.
Posters issued by one party shall not be removed by workers of another
party’.

vk.borooah@ulster.ac.uk
20 V.K. Borooah

As Singh (2012) points out, the MCC was first developed in the 1960s
in the state of Kerala following a broad consensus amongst politicians
about the need for ethical ballast to the electoral vessel. Despite the fact
that it has no statutory basis, the MCC has progressed from a voluntary
agreement between political parties to a set of prescriptive rules, codified
and implemented by the ECI with the acquiescence (however, grudgingly
given) of all the political parties involved.
In 2013, the Supreme Court directed the ECI to frame guidelines with
regard to the contents of election manifestos in consultation with all the
recognised political parties. Broadly, the ECI expects that manifestos will
not seek to beguile voters by containing promises which cannot be met
and, indeed, which the party concerned has no intention of meeting. In
particular, the ECI expects that “manifestos also reflect the rationale for
the promises and broadly indicate the ways and means to meet the finan-
cial requirements for it”.
The MCC also constrains the ruling party, in particular, its govern-
ment’s ministers, from using public resources—cars, planes, helicopters,
and government personnel—for campaign purposes or to seek to influence
voters by announcing new grants (e.g., increases in pensions) and new
projects (like roads, hospitals, and schools), or to make strategic appoint-
ments (like university vice chancellors or chairpersons of public bodies).
Such constraints that the ECI places on the pre-election behaviour of the
ruling party—and, in respect of bribing and intimidating voters, also on
other parties—blunts the use of ‘vote banks’ for electoral purposes.
In the Indian context, Srinivas (1955) coined the term ‘vote banks’ to
mean the exchange of benefits and favours to groups of citizens in return
for their political support. Vote banks had three essential features: political
parties which, at the time Srinivas was writing, was essentially the INC; a
village ‘middleman’, usually a high caste landowner who was a party mem-
ber and who had an agency over groups of voters; and voter groups. There
was then a patron-client relationship between party and ‘middleman’, and
the middleman and voters, based on a system of reciprocal favours.
Favours to voters took essentially two forms: the provision of local pub-
lic goods targeted at particular groups, say a paved road or a school in a
locality in which people from a group were concentrated; the provision of
private benefits to targeted groups of (usually poor) voters, often in the
form of cash payments or gifts in kind like cycles, sewing machines, and
so on; and illegally supplying below poverty line (BPL) cards to voters

vk.borooah@ulster.ac.uk
 The Anatomy of Indian Parliamentary Elections  21

who do not qualify for these (Breeding, 2011). This raises the interesting
question, addressed by Schedler and Schaffer (2007), of how one should
distinguish between favours granted through the public purse (‘local’
public goods) and payments in cash and in kind. Indeed, even when direct
payments are made, they should not necessarily be viewed as purely com-
mercial transactions; instead, they may reflect a sociocultural relationship
between the patron and client, embodying ‘obligation and reciprocity’
and an egalitarian transfer of resources from rich to poor (Srinivas, 1955).
However, the efficacy of vote banks as an electoral instrument has been
severely blunted by the MCC in respect of its strictures on bribing and
intimidating voters. An important consequence of the MCC has, there-
fore, been that the reliance of parties in India on vote banks to deliver
electoral approval is based more on hope than on expectation since fall-
ing foul of the ECI’s strictures risks severe penalties including disquali-
fication.4 Today in India, not least because of the efforts of the ECI, as
Breeding (2011) observes, ‘vote banks are social displays of wealth on the
part of political parties to attract primarily low-income citizens; they are
gestures, historical remnants of a system in which the rules governing the
game have changed’ (p. 77).5

2.3   India’s Political Parties

Any political party wishing to contest an election in India for a seat in a
state Legislative Assembly or to the Lok Sabha must first register with the
ECI with the advantage of registering being that the (registered) party
gets preference in the matter of allotment of free symbols vis-à-vis purely
independent candidates. The ECI then classifies registered parties as ‘rec-
ognised’ or ‘unrecognised’ parties with recognition being awarded as a
‘national’ or as a ‘state’ party.
In order to be recognised as a ‘national’ party, a party must fulfil any of
the following conditions:6

1. It wins 11 Lok Sabha seats from at least three different states.

2. At a Lok Sabha General Election, it polls 6 percent of votes in four
states and also wins four Lok Sabha seats.
3. It is recognised as a ‘state party’ in at least four states.

In order to be recognised as a ‘state’ party, a party must fulfil any of the

following conditions:

vk.borooah@ulster.ac.uk
22 V.K. Borooah

1. It should win at least 3 percent of the total number of seats or a mini-

mum of three seats in the Legislative Assembly.
2. It should win at least one seat in the Lok Sabha for every 25 seats (or
fraction thereof) from that state.
3. It should obtain at least 6 percent of the total valid votes polled during
the General Election to a Lok Sabha or the state Legislative Assembly
and should, in addition, win at least one Lok Sabha, and two Legislative
Assembly seats in that election.
4. Even if it fails to win a seat to the Lok Sabha or to the state Legislative
Assembly, the party will still be recognised as a state party if it secures
8 percent or more of the votes in that state.

As of 12 February 2014, there were, on the above criteria, six rec-

ognised national parties in India: the Bharatiya Janata Party (BJP), the
Indian National Congress (INC), the Communist Party of India (CPI),
the Communist Party of India (Marxist) (CPM), the Bahujan Samaj Party
(BSP), and the Nationalist Congress Party (NCP). In addition, there were
47 recognised state parties and 1563 ‘unrecognised’ parties. 7
Table 2.1 shows the composition of the 16th Lok Sabha (i.e., formed
after the May 2014 General Election). This shows that the status of a
party—as a recognised national or state party—had little bearing on the
number of seats it held in the 16th Lok Sabha—after the BJP and INC,
the next six parties with the largest number of seats were all state parties
with a national party (the CPM) only appearing in seventh place.
In every Lok Sabha election since 1989, the majority of votes cast
accrued to the collective of the INC and the BJP. In the Lok Sabha elec-
tions of 2014, the two parties collectively received 51 percent of the vote
with the BJP winning 282 seats with 31.3 percent of the national vote and
the INC winning 44 seats with 19.5 percent of the national vote. The All
India Anna Dravida Munnetra Kazhagam came third in terms of seats,
winning 37 seats with just 3.3 percent of the national vote.

2.4   Electorates and Turnout

Compared to the UK, the size of the Indian electorate is enormous. In
the 2015 UK General Election, the average size of the electorate in a par-
liamentary constituency was just over 71,000. In the 2014 Lok Sabha
elections, the average electorate size was 1.53 million. Only one Indian
parliamentary constituency, the Laccadive Islands, with an electorate of

vk.borooah@ulster.ac.uk
 The Anatomy of Indian Parliamentary Elections  23

Table 2.1 Composition

Bharatiya Janata Party* BJP 282
of the 16th Lok Sabha, by
Indian National Congress* INC 44
political party All India Anna Dravida Munnetra AIADMK 37
Kazhagam**
All India Trinamool Congress** AITC 34
Biju Janata Dal** BJD 20
Shiv Sena** SHS 18
Telugu Desam Party** TDP 16
Telangana Rashtra Samithi** TRS 11
Communist Party of India CPM 9
(Marxist)*
YSR Congress Party** YSRCP 9
Lok Janshakti Party** LJP 6
Nationalist Congress Party* NCP 6
Samajwadi Party** SP 5
Aam Aadmi Party** AAP 4
Shiromani Akal Dal** SAD 4
Independent IND 3
Rashtriya Janata Dal** RJD 4
All India United Democratic AIUDF 3
Front**
Jammu and Kashmir Peoples JKPDP 3
Democratic Party**
Rashtriya Lok Samta Party** RLSP 3
Apna Dal AD 2
Indian National Lok Dal** INLD 2
Indian Union Muslim League** IUML 2
Janata Dal (Secular)** JD(S) 2
Janata Dal (United)** JD(U) 2
Jharkhand Mukti Morcha** JMM 2
All India Majlis-e-Ittehadul AIMIM 1
Muslimeen**
All India Namathu Rajiyam (NR) AINRC 1
Congress**
Communist Party of India* CPI 1
Kerala Congress (Mani)** KC(M) 1
Naga People’s Front** NPF 1
National People’s Party** NPP 1
Pattali Makkal Katchi** PMK 1
Revolutionary Socialist Party** RSP 1
Sikkim Democratic Front** SDF 1
Swabhimani Paksha SWP 1
Total 543

Source: Own calculations from Lok Sabha election data

*
National party
**
State party

vk.borooah@ulster.ac.uk
24 V.K. Borooah

just under 50,000, was smaller than the UK’s largest constituency—the
Isle of Wight with an electorate of 108,000. Malkajgiri in Andhra Pradesh
had an electorate of over 3 million, and nine constituencies had electorates
between 2 million and 3 million.8 The turnout in Indian elections is also
high: 67 percent of voters exercised their franchise in the 2014 Lok Sabha
elections—compared to 66 percent in the 2015 UK General Election—
and the average turnout, over the 14 Lok Sabha elections between 1962
and 2014, was 58.6 percent.9 In 2014, the turnout of voters was greater
than 80 percent in 69 constituencies, and it fell below 50 percent in only
11 constituencies.
Table 2.2 shows, for each Lok Sabha election between 1962 and 2014,
the average size of the electorate, the percentage of voters who voters in
these constituencies, and also inter-constituency inequality in the distri-
bution of these sizes and turnouts. Inequality is measured by the Gini
coefficient which is one of the most commonly used inequality measures.
If N represents the total number of constituencies and Ei and Ej are the
electorate sizes in constituencies i and j, the Gini coefficient is defined as:

Table 2.2 Average constituency size and turnout and inequality in the distribu-
tion of Inter-constituency size and turnout: 1962–2014
Year Constituency size Gini coefficient on Turnout Gini coefficient on
size (%) turnout

1962 437,876 0.062 53.3 0.132

1967 483,755 0.070 58.1 0.115
1971 529,322 0.074 53.3 0.124
1977 595,591 0.075 58.6 0.098
1980 676,505 0.071 55.4 0.105
1984 740,013 0.078 62.3 0.093
1989 945,145 0.078 59.9 0.109
1991 959,427 0.078 54.5 0.132
1996 1,091,293 0.104 57.1 0.128
1998 1,114,900 0.096 61.4 0.084
1999 1,139,641 0.097 59.6 0.097
2004 1,236,590 0.118 59.1 0.115
2009 1,320,415 0.096 59.4 0.134
2014 1,536,144 0.092 67.3 0.089

Source: Own calculations from Lok Sabha election data

vk.borooah@ulster.ac.uk
 The Anatomy of Indian Parliamentary Elections  25

1 N N
G= ∑∑ Ei − E j
2 N 2 µ i =1 j =1 2.1

In other words, the Gini coefficient is computed as half the mean of

the difference in sizes between pairs of constituencies, divided by the aver-
age constituency size (μ). So, in 2014, with a mean constituency size of
1,536,144 and a Gini value of 0.092, the difference in electorate sizes
between two constituencies chosen at random would have been 18.4 per-
cent of 1,536,144 or just under 283,000.
Table 2.2 shows that between 1962 and 2014, the average size of
the electorate increased by a factor of 3.5: from 437,876 in 1962 to
1,536,144 in 2014. Over the same period, inequality in electoral size
increased slightly: in 1962, the largest 10 percent of constituencies had an
average electorate (493,266) that was 1.25 times the average electorate of
the smallest 10 percent of constituencies (393,246); by 2014, this markup
had increased to 1.44 (1,808,886 versus 1,258,59).
In order to examine changes in turnout, we can split the elections into
two periods: 1962 to 1984 (when the BJP made its first electoral foray,
winning just two seats in the 8th Lok Sabha); and 1989 (when the BJP
secured 85 seats in the 9th Lok Sabha) to 2014 (when the BJP secured
282 seats in the 16th Lok Sabha). The average turnout increased from
57 percent in the earlier period to 60 percent in the later period, and
the turnout in constituencies in the highest and lowest deciles of turnout
increased from 71.2 percent and 42.3 percent, respectively, in 1962–84 to
76.3 percent and 44.5 percent, respectively, in 1989–2014.
Table 2.3 shows the average turnout for the 2014 Lok Sabha elections
by the major states of India. This table shows that the highest turnout
was in West Bengal (82.2 percent) and the lowest in Jammu and Kashmir
(50.6 percent) with several states recording a voter turnout in excess of
70 percent.
Considerations of voter turnout at elections raise the question of why
people bother to vote. Traditional theories of voting are based on an indi-
vidualistic model of voting. On this view of voting, it is not clear why a
rational individual, on a purely cost-benefit basis, would bother to vote:
the chances of an individual vote influencing the electoral outcome are
infinitesimally small while the costs of voting—taking time off work,
standing in a long queue—are real and not insubstantial (Downs, 1957).
However, given the far from negligible turnout witnessed in elections in

vk.borooah@ulster.ac.uk
26 V.K. Borooah

Table 2.3 Average turnout

State Turnout
by major Indian states in the
2014 Lok Sabha elections Andhra Pradesh 75.6
Assam 79.8
Bihar 56.5
Chhattisgarh 69.5
Gujarat 63.6
Haryana 71.5
Himachal Pradesh 64.4
Jammu and Kashmir 50.6
Jharkhand 63.9
Karnataka 67.7
Kerala 74.0
Madhya Pradesh 61.7
Maharashtra 60.5
Orissa 73.9
Punjab 70.7
Rajasthan 63.0
Tamil Nadu 74.0
Uttarakhand 60.7
Uttar Pradesh 58.6
West Bengal 82.2

Source: Own calculations from Lok Sabha election data

India and, indeed, throughout the world, it is clear that people do take
the trouble to vote.
One reason why people vote is because of ‘group identity’ voting which
has been analysed, for elections in Israel, by Hillman et al. (2014). In
the Indian context, the existence of vote banks goes some way towards
explaining why large numbers of people in India turn out to vote. Downs’
(1957) argument was based on the belief that the costs of voting—gather-
ing information about parties and candidates, registration, and time spent
to/from/at the polling station—were specific to the voter and were likely
to exceed the benefits from voting. The latter are in the form of collective
goods, and their benefit to a specific voter is likely to be zero.10Besley
et al. (2012) suggest that in the context of Indian villages, residents in the
Gram Pradhan’s village had greater access to public goods than residents
in other villages. However, in the context of ‘vote banks’, many of the
benefits of voting may be private benefits paid to groups of voters for their
electoral support and may be quite substantial.

vk.borooah@ulster.ac.uk
 The Anatomy of Indian Parliamentary Elections  27

The existence and implementation of the MCC, discussed earlier,

are likely to have diminished the importance of an exchange of favours,
between electors and candidates, that characterised traditional vote banks.
However, in addition to opportunistic electoral politics, based on recip-
rocal favours, there are several, more general, explanations for this para-
dox of (not) voting. As Geys (2006) observes, the instrumental theory of
voting holds that an action has value only if it affects the outcome. Sen
(1977) argued that if ‘outcome’ was narrowly defined as serving one’s
own interest, to the exclusion of any other’s, then a person acting in such
a manner might be ‘rational’ but he would also be a fool.
Indeed, Sen (1977) argued that people act out of a myriad motives,
many of which are unconnected with self-interest. One of these is ‘sym-
pathy’; another is ‘commitment’. Even if it is argued that ‘sympathy’ is
just an economic externality, Sen (1977) argues that commitment involves
a counter-preferential choice, destroying the crucial assumption that the
chosen alternative must be better than the others—‘it drives a wedge
between personal choice and personal welfare’ (p. 329). Consequently,
the high turnout in elections ‘may be guided not so much by maximisa-
tion of expected utility, but by something much simpler, viz., just a desire
to record one’s true preference’ (p. 333).
The concept of ‘expressive voting’ elaborates upon, and extends, the
view of people voting to record their preferences. In terms of ‘expres-
sive voting’, people vote not for instrumental reasons—that is to effect
change—but rather to express an opinion or a point of view, regardless
of whether that turns out to be the winning opinion. This view has been
articulated by, inter alia, Brennan and Lomasky (1993), Hillman (2010),
and Hamlin and Jennings (2011).
All this is not to say that expressive voting cannot be self-interested or
not result in change. As regards the first point, Hillman (2010) argues
that expressive utility, along with material utility, comprises total utility.
A person’s voting decision may be based simultaneously on maximising
material utility (a high-income person votes against higher tax and more
generous welfare payments) and on maximising expressive utility (a high-
income person affirms his identity). As regards the second point, if a suffi-
cient number of people express the same opinion, then social and political
change—sometimes dramatic—inevitably follows. The 2014 Indian elec-
tion results, which led to a landslide victory for the BJP under Narendra
Modi, can be interpreted as an expression of the electorate’s distaste for
the ineffectual, dynastic government led by the INC. As Banerjee (2014)

vk.borooah@ulster.ac.uk
28 V.K. Borooah

argues, ‘for many Indian voters…voting is not just a means to elect a gov-
ernment. Rather, the very act of voting is seen by them as meaningful, as
an end in itself, which expresses the virtues of citizenship, accountability
and civility that they wish to see in ordinary life, but rarely can’ (p. 3).

2.5   Independent Candidates and ‘None

of the Above’

Lok Sabha elections attract a large number of candidates to most constitu-

encies, but the record must surely be held by Nalgonda in Andhra Pradesh
and by Belgaum in Karnataka which, in 1996, fielded, respectively, 480
and 456 candidates. Apart from this bounty of candidates in Nalgonda
and Belgaum, 1996 was a bumper year for contesting Lok Sabha elections:
inter alia, the constituency of East Delhi was contested by 122 candi-
dates; Allahabad by 73; Nagpur by 60; Muzaffarpur (Bihar) by 67; Pune
by 44; and so on. Table 2.4 shows the average number of candidates in a
constituency for each of the Lok Sabha elections between 1962 and 2014.
The numbers in this table point to a secular increase in the number of can-
didates from around five per constituency for the four elections between

Table 2.4 Average number of candidates in a constituency: 1962–2014

Year Average number of Average number of Average number of
candidates in a independent candidates in party candidates in
constituency a constituency a constituency

1962 4.1 1.0 3.1

1967 4.6 1.7 2.9
1971 5.4 2.2 3.2
1977 4.5 2.3 2.2
1980 8.8 5.3 3.4
1984 10.2 7.2 3.0
1989 11.7 7.0 4.7
1991 16.4 10.4 6.0
1996 25.7 19.5 6.2
1998 8.7 3.5 5.2
1999 8.6 3.6 5.0
2004 10 4.4 5.6
2009 14.9 7.1 7.8
2014 16.2 6.0 10.2

Source: Own calculations from Lok Sabha election data

vk.borooah@ulster.ac.uk
 The Anatomy of Indian Parliamentary Elections  29

1962 and 1977 rising to about 14 per constituency for the three elections
of 2004, 2009, and 2014.
As Table 2.4 shows, this increase is partly due to the increase in the
number of independent candidates in a constituency (up from an average
of one per constituency in 1962 to six per constituency in 2014) but it is
partly also due to the increase in the number of political parties (up from
an average of three per constituency in 1962 to 10 per constituency in
2014). In 1962, of an average of four candidates per constituency, one was
an independent and three were party candidates; in 2014, of an average of
16 candidates per constituency, six were independents and ten were party
candidates. What is undoubtedly true is that the ratio of independent to
party candidates has shifted in favour of the former: in 1962, there were
three party candidates for every independent candidate, but in the elec-
tions between 1984 and 1996, party candidates were outnumbered by
independents, and in 2004 and 2009, there was approximately one inde-
pendent candidate for every party candidate.
One possibility for the rise in independent candidates is not that they
expect to win, but that they want to undermine the vote of a party can-
didate. In a closely fought election (discussed in the next chapter), the
presence of independent candidates can erode support sufficiently to have
an appreciable impact on the outcome.11 Another reason for the rise in the
number of independent candidates could be pique at being denied a party
nomination. Since being a Lok Sabha member is a rewarding job—offering,
inter alia, a good salary, generous pension benefits, government-provided
housing in the capital, and free travel across India—there is considerable
competition to be adopted as a major party’s candidate for a constituency
(‘getting a ticket’, as it is termed in India). Alas, many are called, but few
are chosen. Some of those not chosen seek to exact revenge by standing
against the official candidate who deprived them (unfairly, in their eyes)
of their opportunity.
Table 2.5 shows, for the 2014 Lok Sabha elections, the average number
of candidates in a constituency in the 20 major Indian states. The smallest
number of candidates in a constituency were in the three eastern states of
Orissa (10.3 candidates), West Bengal (12.2 candidates), and Assam (12.6
candidates) and in the northern state of Himachal Pradesh (10.5 candi-
dates). These states had also the smallest number of ­independent candi-
dates per constituency: 1.5 in Orissa, 2 in West Bengal, 2.8 in Himachal
Pradesh, and 4.2 in Assam. At the other extreme, the newer states of
Chhattisgarh and Jharkhand had a large number of candidates per constit-

vk.borooah@ulster.ac.uk
30 V.K. Borooah

Table 2.5 Average number of candidates in a constituency by major Indian

states in the 2014 Lok Sabha elections
State Total candidates Independent candidates Party candidates

Andhra 15.2 5.5 9.7

Assam 12.6 4.2 8.4
Bihar 16.2 4.2 12.0
Chhattisgarh 20.2 9.6 10.5
Gujarat 13.8 6.1 7.8
Haryana 24 12.1 11.9
Himachal Pradesh 10.5 2.8 7.8
Jammu and Kashmir 13.8 5.5 8.3
Jharkhand 18.1 5.5 12.6
Karnataka 16.5 7 9.5
Kerala 14.5 6.1 8.4
Madhya Pradesh 14.0 4.3 9.7
Maharashtra 19.7 9.3 10.4
Orissa 10.3 1.5 8.8
Punjab 20.5 9.1 11.4
Rajasthan 13.8 4.7 9.1
Tamil Nadu 22.7 13.3 9.4
Uttarakhand 15.8 5.4 10.4
Uttar Pradesh 17.1 4.7 12.4
West Bengal 12.2 2.0 10.3

Source: Own calculations from Lok Sabha election data

uency (20.2 in Chhattisgarh and 18.1 in Jharkhand), and they were joined
in this plethora of candidates by Tamil Nadu (22.7 candidates per con-
stituency), Punjab (20.5 candidates per constituency), and Maharashtra
(19.7 candidates per constituency).
Although there has been a rise in the number of independent candi-
dates over time, this has not been matched by the number of independent
members of the Lok Sabha. Figure 2.1 shows that the number of indepen-
dent members in the Lok Sabha fell from 35 in 1967 to just three—one
from Assam (Kokrajhar) and two from Kerala (Chalakudy and Idukki,
respectively)—in 2014.
Between them, independent candidates received a total of nearly 17
million votes in the 2014 Lok Sabha elections which represented 3 percent
of the total of nearly 554 million votes cast in that election. Figure 2.2,
which charts the share of independent candidates in the total of votes cast,
shows that notwithstanding the increase in the number of independent

vk.borooah@ulster.ac.uk
 The Anatomy of Indian Parliamentary Elections  31

Fig. 2.1 Independent members of the Lok Sabha, 1962–2014 (Source: Own cal-
culations from Lok Sabha election data)

candidates between 1962 and 2014 (noted in Table 2.4), the propor-
tion of the total votes going to independent candidates has seen a secular
decline from 13 percent in the Lok Sabha elections of 1962 to 3 percent in
the Lok Sabha elections of 2014.

2.5.1  None of the Above

Voting for independent candidates, arguably, expresses dissatisfaction with
political parties in effectively representing voters’ needs. A rejection of all
candidates in a constituency, on the other hand, is an unambiguous rejec-
tion of the entire political system, party and non-party, in that constitu-
ency. In September 2013, the Supreme Court of India upheld the right of
voters to reject all candidates contesting elections and directed the ECI to
provide voters with the option of casting their vote for a phantom candi-
date: NOTA. Following this directive, the Lok Sabha election of May 2014
was the first parliamentary election to incorporate the NOTA option.12
In this election, NOTA received just over a total of 6 million votes—
that is, 11 million less than the 17 million received by the collective of
independent candidates—and the three constituencies with the largest

vk.borooah@ulster.ac.uk
32 V.K. Borooah

Fig. 2.2 Share of votes received by independent candidates in the total vote:
1962–2014 (Source: Own calculations from Lok Sabha election data)

number of NOTA votes were: the Nilgiris (Tamil Nadu) with 46,559
votes comprising 5 percent of the total votes cast in the constituency;
Nabarangpur (Orissa) with 44,408 votes comprising 4.3 percent of the
total votes cast in the constituency; and Bastar (Chhattisgarh) with 38,772
votes comprising 5 percent of the total votes cast in the constituency.
The state with the largest number of NOTA votes was Uttar Pradesh
(592,211 votes), followed by Tamil Nadu (582,062 votes), Bihar (581,011
votes), and West Bengal (568,276 votes). These four states, collectively,
accounted for 39 percent of the total of NOTA votes.

2.6   Candidates with Criminal Histories or Who

Face Criminal Charges
In a landmark judgement in 2002, the Indian Supreme Court mandated
that prior to an election, all candidates running for public office should
file affidavits with the ECI in which they would report criminal histories
or pending criminal charges for any offense punishable with imprisonment
of two years or more; these affidavits were to be lodged six months before
the individual filed his/her candidacy papers.13 Since these rulings, there

vk.borooah@ulster.ac.uk
 The Anatomy of Indian Parliamentary Elections  33

have been three Lok Sabha elections—2004, 2009, and 2014: informa-
tion on the ‘criminal status’ of all the candidates in the 2004 and 2009
Lok Sabha elections was collected by Golden (2014) and made available
through the Inter-University Consortium for Political and Social Research
(ICPSR) at the University of Michigan; and information on the criminal
status of candidates in the 2014 Lok Sabha elections was available from the
Association for Democratic Reforms (2014).14
Figure 2.3 shows that in the 2004 election, 8.7 percent of the candi-
dates (475 out of 5435) reported a criminal charge (hereafter, ‘CC candi-
dates’); in the 2009 election, 11 percent were CC candidates (893 out of
8070); and in the 2014 election, 17 percent were CC candidates (1401
out of 8180). Consequently, there would appear to be strong evidence
that the proportion of CC candidates in the total of candidates for Lok
Sabha elections is on the rise.
The proportion of CC candidates was, however, unevenly distributed
over the states. Table 2.6 shows the proportion of CC candidates, in the
total number of candidates, for every state in India. The outlier states in
this table were Bihar and Jharkhand—remembering that Jharkhand was

Fig. 2.3 The criminal charge status of candidates in the 2004 and 2009 Lok
Sabha elections (Source: Own calculations from Golden (2014) for the 2004
and 2009 elections and Association for Democratic Reforms (2014) for the
2014 election)

vk.borooah@ulster.ac.uk
34 V.K. Borooah

carved out of erstwhile Bihar in 2004—where, respectively, 20 and 23

percent of candidates reported a criminal charge against them.15
The fact that persons with reported criminal charges were candidates
for Lok Sabha elections begs the question of how they fared in these elec-
tions. Of the 543 Lok Sabha members elected in 2004 (to the 14th Lok
Sabha) and in 2009 (to the 15th Lok Sabha), respectively, 128 and 129
members—or 24 percent of the total—had reported criminal charges
(hereafter, CC members). Since, as Fig. 2.2 showed, there were a total
475 CC candidates in the Lok Sabha elections of 2004, and 893 CC can-
didates in the Lok Sabha elections of 2009, respectively, 27 and 14 percent
of such candidates were elected to the 14th and 15th Lok Sabha. After the
May 2014 Lok Sabha election, there were 1401 CC candidates of whom
185 (13 percent) were elected to the 16th Lok Sabha. Conversely, the
proportions of non-CC candidates elected were: 8.3 percent to the 14th
Lok Sabha, 5.8 percent to the 15th Lok Sabha, and 5.3 percent to the 16th
Lok Sabha.
Table 2.7 shows the interstate distribution of CC members of the 14th
and 15th Lok Sabha (i.e., after the 2004 and 2009 elections). The five
states that stand out in this table are: Uttar Pradesh (27 out of 80 were CC
members in the 15th Lok Sabha), Bihar (18 out of 40 were CC members
in the 15th Lok Sabha), Maharashtra (12 out of 48 were CC members in
the 15th Lok Sabha), Andhra Pradesh (9 out of 42 were CC members in
the 15th Lok Sabha), and Tamil Nadu (9 out of 39 were CC members in
the 15th Lok Sabha). In sum, these five states supplied 75 of the total of
129 CC members (58 percent) in the 15th Lok Sabha.
Tables 2.8, 2.9, and 2.10 show the party affiliations of CC members
for, respectively, the 14th, 15th, and 16th Lok Sabha. Of the 128 CC
members in the 14th Lok Sabha, 52 (41 percent) were supplied by the
two main parties, the BJP (28 CC members) and the INC (24 CC mem-
bers). The 15th Lok Sabha saw these parties increase their supply of CC
members: now 71 of the total of 129 CC members (55 percent) belonged
to the BJP (36 CC members) or the INC (35 CC members). Not to be
underestimated either is the contribution of the smaller parties in supply-
ing CC members. One in three of the 36 Samajwadi Party (SP) members
in the 14th Lok Sabha reported a criminal charge while 8 of the 12 Shiv
Sena members were CC members. In the 15th Lok Sabha, 8 of the 20
Janata Dal (United) members, 6 of the 21 Bahujan Samaj Party mem-
bers, and 8 of the 23 SP members reported a criminal charge. The 16th

vk.borooah@ulster.ac.uk
 The Anatomy of Indian Parliamentary Elections  35

Table 2.6 The distribution of candidates with reported criminal charges by state,

2004 and 2009 Lok Sabha elections
State Total Candidates with Percentage of total
candidates a charge candidates with a
criminal charge

Andaman and Nicobar Islands 23 5 21.7

Andhra Pradesh 848 52 6.1
Arunachal Pradesh 21 1 4.8
Assam 274 13 4.7
Bihar 1134 230 20.3
Chandigarh 31 0 0
Chhattisgarh 280 9 3.2
Dadra and Nagar Haveli 15 5 33.3
Daman and Diu 10 3 33.3
Delhi 289 21 7.3
Goa 34 5 14.7
Gujarat 521 76 14.6
Haryana 370 21 5.7
Himachal Pradesh 54 0 0
Jammu and Kashmir 164 7 4.3
Jharkhand 431 101 23.4
Karnataka 599 34 5.7
Kerala 394 56 14.2
Laccadive Islands 8 1 12.5
Madhya Pradesh 723 45 6.2
Maharashtra 1231 97 7.9
Manipur 28 0 0
Meghalaya 16 1 6.3
Mizoram 7 0 0
Nagaland 8 0 0
Orissa 257 33 12.8
Pondicherry 49 7 14.3
Punjab 360 29 8.1
Rajasthan 531 36 6.8
Sikkim 11 1 9.1
Tamil Nadu 1394 87 6.2
Tripura 31 2 6.5
Uttar Pradesh 2506 312 12.5
Uttarakhand 130 7 5.4
West Bengal 723 71 9.8
Total 13,505 1368 10.1

Source: Own calculations from Golden (2014)

vk.borooah@ulster.ac.uk
36 V.K. Borooah

Table 2.7 The distribution of Lok Sabha members with reported criminal

charges, by state: 2004 and 2009 Lok Sabha elections
State Total members CC members CC members
2004 & 2009 2004 2009

Andaman and Nicobar Islands 1 0 1

Andhra Pradesh 42 6 9
Arunachal Pradesh 2 0 0
Assam 14 0 2
Bihar 40 15 18
Chandigarh 1 0 0
Chhattisgarh 11 2 2
Dadra and Nagar Haveli 1 0 1
Daman and Diu 1 1 0
Delhi 7 2 0
Goa 2 1 0
Gujarat 26 7 7
Haryana 10 1 1
Himachal Pradesh 4 0 0
Jammu and Kashmir 6 0 1
Jharkhand 14 7 7
Karnataka 28 6 6
Kerala 20 7 7
Laccadive Islands 1 0 0
Madhya Pradesh 29 6 5
Maharashtra 48 20 12
Manipur 2 0 0
Meghalaya 2 0 0
Mizoram 1 0 0
Nagaland 1 0 0
Orissa 21 3 5
Pondicherry 1 0 0
Punjab 13 5 2
Rajasthan 25 3 1
Sikkim 1 0 0
Tamil Nadu 39 8 9
Tripura 2 0 0
Uttar Pradesh 80 24 27
Uttarakhand 5 0 1
West Bengal 42 4 5
Total 543 128 129

Source: Own calculations from Golden (2014)

vk.borooah@ulster.ac.uk
 The Anatomy of Indian Parliamentary Elections  37

Table 2.8 Party affiliation of members to the 14th Lok Sabha with reported
criminal charges
Party Total number Number of
of members CC members

Indian National Congress (INC) 145 24

Bharatiya Janata Party (BJP) 138 28
Communist Party of India (Marxist) (CPM) 43 7
Samajwadi Party (SP) 36 12
Rashtriya Janata Dal (RJD) 24 11
Bahujan Samaj Party (BSP) 19 8
Dravida Munnetra Kazhagam (DMK) 16 5
Shiv Sena (SHS) 12 8
Biju Janata Dal (BJD) 11 1
Communist Party of India (CPI) 10 2
Nationalist Congress Party (NCP) 9 5
Janata Dal (United) (JD(U)) 8 2
Shiromani Akali Dal (SAD) 8 4
Jharkhand Mukti Morcha (JMM) 5 5
Lok Janshakti Party (LJNSP) 4 1
Marumalarchi Dravida Munnetra Kazhagam (MDMK) 4 1
All India Forward Bloc (AIFB) 3 1
Janata Dal (Secular) (JD(S)) 3 1
All India Majlis-e-Ittehadul Muslimeen (AIMIM) 1 1
Kerala Congress (KEC) 1 1
Totala 500 128

Source: Own calculations from Golden (2014)

Total refers to only those parties with at least one CC member

a

Lok Sabha saw the number of CC members rise to 187 which comprised
34 percent of the total strength of the House.16
The election of candidates with reported criminal charges to the Lok
Sabha raises the further question of how they performed as legislators.
This issue has been examined, with respect to the 14th Lok Sabha, by
Gehring et al. (2015). Their first conclusion was that compared to non-
­CC members of the Lok Sabha, the attendance record of CC members was
about 5 percent lower. There was, however, no difference in the amount of
‘parliamentary activity’—raising questions and participating in debates—
between CC and non-CC members of the 14th Lok Sabha.
The Indian government operates a Member of Parliament Local Area
Development (MPLAD) Scheme under which members of the Lok Sabha

vk.borooah@ulster.ac.uk
38 V.K. Borooah

Table 2.9 Party affiliation of members to the 15th Lok Sabha with reported
criminal charges
Party Total number of Number of CC
members members

Indian National Congress INC 206 35

Bharatiya Janata Party BJP 116 36
Samajwadi Party SP 23 8
Bahujan Samaj Party BSP 21 6
Janata Dal (United) JD(U) 20 8
All India Trinamool Congress AITC 19 2
Dravida Munnetra Kazhagam (DMK) 18 4
Communist Party of India (Marxist) (CPM) 16 2
Biju Janata Dal BJD 14 4
Shiv Sena SHS 11 2
All India Anna Dravida Munnetra Kazhagam 9 3
(AIADMK)
Independent IND 9 1
Nationalist Congress Party NCP 9 2
Telugu Desam Party (TDP) 6 2
Rashtriya Lok Dal (RLD) 5 2
Rashtriya Janata Dal (RJD) 4 3
Shiromani Akali Dal (SAD) 4 1
Janata Dal (Secular) (JD(S)) 3 1
All India Forward Bloc (AIFB) 2 1
Jharkhand Mukti Morcha (JMM) 2 2
Telangana Rashtra Samithi (TRS) 2 1
All India Majlis-e-Ittehadul Muslimeen (AIMIM) 1 1
Marumalarchi Dravida Munnetra Kazhagam 1 1
(MDMK)
Viduthalai Chiruthaigal Katch (VCK) 1 1
Totala 522 129

Source: Own calculations from Golden (2014)

Total refers to only those parties with at least one CC member
a

can suggest—up to an amount of ₹5 crore (£0.5 million) per year—to the

Collector of the district, in which their constituencies lie, public works
that might benefit their constituents.17 Gehring et al. (2015) analysed
the utilisation of the MPLAD Scheme by individual members of the Lok
Sabha and found that CC members had a utilisation rate of monies under
MPLAD Scheme that was 7 percent lower than that of non-CC mem-

vk.borooah@ulster.ac.uk
 The Anatomy of Indian Parliamentary Elections  39

Table 2.10 Party affiliation of members to the 16th Lok Sabha with reported
criminal charges
Party Total number of Number of CC
members members

Bharatiya Janata Party (BJP) 282 98

Indian National Congress (INC) 44 8
All India Anna Dravida Munnetra Kazhagam 37 6
(AIADMK)
All India Trinamool Congress (AITC) 34 7
Biju Janata Dal (BJD) 20 3
Shiv Sena (SHS) 18 15
Telugu Desam Party (TDP) 16 6
Telangana Rashtra Samithi (TRS) 11 5
Communist Party of India (Marxist) (CPI 9 5
(M))
YSR Congress Party (YSRCP) 9 5
Lok Janshakti Party (LJSP) 6 4
Nationalist Congress Party (NCP) 6 5
Rashtriya Janata Dal (RJD) 4 4
Jammu and Kashmir Peoples Democratic 3 1
Party (JKPDP)
Janata Dal (United) (JDU) 2 1
Independents 3 2
Others 39 12
Total 543 187

Source: Own calculations from Association for Democratic Reform (2014)

bers. The overall conclusion must be that although CC candidates have

a better chance of being elected than non-CC candidates (13 percent to
5 percent for the 16th Lok Sabha), once elected, they do not serve their
constituents, both for reasons of attendance and for reason of constitu-
ency improvement, as conscientiously as do non-CC members of the Lok
Sabha.

2.7   Concluding Remarks

This chapter set out some of the salient features of the Indian electoral
landscape beginning with the regulator, in the form of the ECI, and pro-
ceeding to the candidates, both party and non-party. The importance of
the ECI in administering, managing, and controlling elections in India
cannot be underestimated. For example, purely in terms of administra-

vk.borooah@ulster.ac.uk
40 V.K. Borooah

tion and management, elections in May 2014 to the 16th Lok Sabha were
organised in nine phases beginning on 7 April 2014 and concluding on 12
May 2014 with the results being declared on 16 May 2014. Nearly 815
million persons were eligible to vote, of whom nearly 550 million voted,
using 930,000 voting centres deploying 1.4 million Electronic Voting
Machines (EVM). All this required the ECI to engage 2 million workers
to oversee the electoral process.
Mozaffar and Schedler (2002) argue that ‘good elections are impos-
sible without effective electoral governance’, and it is precisely such gover-
nance that the ECI seeks to provide. So much so that, Rudolf and Rudolf
(2002) place the ECI alongside the Supreme Court and the Presidency as
an enforcer of rules that ‘safeguard the legitimacy of the political system’
and suggest that the cabinet and parliament have ceded pride of place to
these three regulatory institutions.
While many of the duties of the ECI are technical and administrative,
the MCC provides a moral compass for the conduct of electoral politics
in India. In so doing, the ECI has mutated from a referee enforcing rules,
agreed to by others, to a regulatory body which makes rules which others
have to obey (Singh, 2012). In assuming this role, it has been aided by
the Supreme Court ruling that under Article 324(2) of the Constitution,
the ECI has ‘a reservoir of powers where the law was silent’ (Singh, 2012).
Some find the authoritarian nature of the ECI’s mode of operation to
be troubling. For example, Chaterjee (2006) feels that by riding rough-
shod over local culture and practices, the ECI has gone too far in the
direction of sanitising and cleaning politics. Yet others feel that at critical
moments, the ECI has proved toothless. After his alleged ‘hate speech’ in
the Pilibhit constituency in March 2009, the ECI advised the BJP not to
adopt Varun Gandhi as its parliamentary candidate in that constituency for
the Lok Sabha elections of 2009; this advice was ignored and Mr Gandhi
went on to become the Lok Sabha member for Pilibhit. More generally, the
ECI has proved impotent in arresting an unsavoury trend in Indian poli-
tics where candidates with reported criminal charges are elected to legisla-
tive office: as the previous section noted, one in three of members to the
16th Lok Sabha reported criminal charges against him/her. Unfortunately,
there is nothing in the MCC to prevent this trend from continuing.

vk.borooah@ulster.ac.uk
 The Anatomy of Indian Parliamentary Elections  41

Notes
1. Article 342(2) states that the Election Commission shall consist of the
Chief Election Commissioner and such number of other Election
Commissioners, if any, as the President may from time to time fix and the
appointment of the Chief Election Commissioner and other Election
Commissioners shall, subject to the provisions of any law made in that
behalf by Parliament, be made by the President.
2. See McMillan (2012) for a detailed account of the formation of the ECI.
3. See http://eci.nic.in/eci_main/MCC-ENGLISH_28022014.pdf,
accessed 5 November 2015.
4. As a consequence of employing over 2 million workers during elections,
the ECI’s observers are ubiquitous and, since they are drawn from the
ranks of those in civilian employment, cannot be easily identified. In addi-
tion, the Indian media seizes upon any infractions of the MCC and affords
them considerable publicity.
5. Indeed, it is a moot point whether the fact that ‘vote buying’ is virtually
unknown in Western countries is due more to the difficulty of doing so
than to any innate moral superiority. Wang and Kurzman (2007) detail the
planning, organisation, and sheer expenditure required for a widespread
vote buying in the 1993 elections in Taiwan. Vote buying required an
extensive network of brokers who would each control small groups of vot-
ers. In order to be effective, such a network was predicated on: detailed
local knowledge; relationships of trust between party brokers and voters; a
large budget; and legal circumspection in conjunction with, possibly, judi-
cial protection. To compound these problems, 45 percent of voters did not
deliver on their promises to vote appropriately.
6. Press Information Bureau, Election Commission of India, http://pib.nic.
in/newsite/PrintRelease.aspx?relid=104537, retrieved 6 November 2015.
7. Press Information Bureau, Election Commission of India, http://pib.nic.
in/newsite/PrintRelease.aspx?relid=104537, retrieved 6 November 2015.
8. Chevella (erstwhile Andhra Pradesh, now Telengana), North West Delhi, West
Delhi, Bangalore North, Bangalore Rural, Indore (Madhya Pradesh), Thane
(Maharashtra), Ghaziabad (Uttar Pradesh), and Unnao (Uttar Pradesh)
9. Though turnout in the UK General Elections exceeded 80 percent in the
1950 and 1951 elections and remained above 70 percent for all elections
between 1945 and 1997.
10. Though turnout in the UK General Elections exceeded 80 percent in the
1950 and 1951 elections and remained above 70 percent for all elections
between 1945 and 1997.
11. This point is developed by Praveen Chakravarty, ‘Independent
Candidates: party-poopers in disguise’, Business Standard, 28 November

vk.borooah@ulster.ac.uk
42 V.K. Borooah

2013, http://www.business-standard.com/article/opinion/praveen-
chakravar ty-independent-candidates-par ty-poopers-in-disguise-­
113112800936_1.html, accessed on 29 September 2015.
12. Before NOTA, voters wishing to reject all the candidates were required to
enter their names in a register and cast their vote on a separate paper
ballot.
13. Union of India versus Association for Democratic Reforms. In a subsequent
judgement in 2003—Union of India versus People’s Union for Civil
Liberties—the Supreme Court mandated the compulsory declaration of
candidates’ financial assets. Details in Sen (2012).
14. The difference between the two sources was that while the Golden (2014)
data was available for individual candidates, the Association for Democratic
Reforms (2014) data was available only in aggregated form and only for
winners.
15. For reasons set out in the previous note, we were unable to present state-
wise information for the Lok Sabha elections of 2014.
16. It should be cautioned that the numbers from Golden (2014) and the
Association for Democratic Reforms (2014) are not entirely consistent.
For example, according to Golden (2014), there were 129 CC members in
the 15th Lok Sabha while the Association for Democratic Reforms (2014)
put this figure at 158.
17. See Ministry of Statistics and Programme Implementation, Government of
India, http://mplads.nic.in/, accessed 7 November 2015. This facility is
also available to members of the Upper House, the Rajya Sabha.

References
Association for Democratic Reforms. (2014). Analysis of criminal background,
financial, education, gender and other details of winners, Lok Sabha elections
2014. New Delhi: Association for Democratic Reforms (www.adrindia.org).
Banerjee, M. (2014). Why India votes? New Delhi: Routledge.
Besley, T., Pande, R., & Rao, V. (2012). Just rewards? Local politics and public
resource allocation in South India. World Bank Economic Review, 26,
191–216.
Breeding, M. (2011). The micro-politics of vote banks in Karnataka. Economic
and Political Weekly, XLVI, 71–77.
Brennan, G., & Lomasky, L. (1993). Democracy and decision: The pure theory of
electoral preference. Cambridge: Cambridge University Press.
Downs, A. (1957). An economic theory of democracy. New York: Harper and Row.
Gehring, K., Kauffeldt, T. F., & Vadlamannati, K. C. (2015). Crime, incentives
and political effort: A model and empirical application for India. Discussion
paper no. 170. Germany: Georg-August-Universität Göttingen.

vk.borooah@ulster.ac.uk
 The Anatomy of Indian Parliamentary Elections  43

Geys, B. (2006). ‘Rational’ theories of voter turnout: A review. Political Studies

Review, 4, 16–35.
Golden, M. (2014). Electoral performance and criminal status of candidates con-
testing the 2004 and 2009 parliamentary elections to the Lok Sabha. Ann Arbor:
Inter-university Consortium for Political and Social Research [distributor],
2015-02-06. doi:10.3886/ICPSR35512.v1.
Hamlin, A., & Jennings, C. (2011). Expressive political behaviour: Foundations,
scope and implications. British Journal of Political Science, 41, 645–670.
Hillman, A. L. (2010). Expressive behaviour in economics and politics. European
Journal of Political Economy, 26, 403–418.
Hillman, A. L., Metsuyanim, K., & Potrafke, N. (2014). Political economy with
group identity. Tel Aviv: Bar-Ilan University.
McMillan, A. (2012). The election commission of India and the regulation and
administration of electoral politics. Election Law Journal, 11, 187–201.
Schedler, A., & Schaffer, F. C. (2007). What is vote buying? In F. C. Schaffer
(Ed.), Elections for sale: The causes and consequences of vote buying (pp. 17–30).
Boulder/London: Lynne Rienner Publishers.
Sen, A. K. (1977). Rational fools: A critique of the behavioral foundations of eco-
nomic theory. Philosophy and Public Affairs, 6, 317–344.
Sen, R. (2012). Identifying criminals and crorepatis in Indian politics: An analysis
of two supreme court rulings. Election Law Journal, 11, 216–225.
Singh, U. K. (2012). Between moral force and supplementary legality: A moral
code of conduct and the election commission of India. Election Law Journal,
11, 149–169.
Srinivas, M. N. (1955). The social structure of life in a Mysore village. In M. K.
Marriott (Ed.), Village India. Chicago: Chicago University Press.
Wang, C.-S., & Kurzman, C. (2007). The logistics: How to buy votes. In F. C.
Schaffer (Ed.), Elections for sale: The causes and consequences of vote buying
(pp. 48–61). Boulder/London: Lynne Rienner Publishers.

vk.borooah@ulster.ac.uk
 Chapter 3

Estimating the Likelihood of Winning

Parliamentary Constituencies

Abstract Borooah compares the two major protagonists in Indian elec-

tions—the Bharatiya Janata Party (BJP) and the Indian National Congress
(INC)—with respect to their respective probabilities of winning constitu-
ency battles. He provides estimates of such probabilities and shows how
these differ between the BJP and the INC both with respect to marginal
constituencies and with respect to all constituencies where the BJP and the
INC went ‘head-to-head’. Lastly, he considers the electoral performances
of the INC and the BJP separately for the Hindi-speaking and the non-
Hindi-­speaking major Indian states.

3.1   Introduction
In this chapter, we begin a comparison between the two major protago-
nists in Indian elections—the Bharatiya Janata Party (BJP) and the Indian
National Congress (INC). The comparison relates to the relative effi-
ciency of the two parties in winning constituency battles and in convert-
ing votes into seats. This chapter places emphasis on the probability of
winning elections. It provides estimates of such probabilities and shows
how these differ between the BJP and the INC. In so doing, the first port
of call is the ‘marginal constituency’: a constituency where the margin of
victory between the winner and the runner-up is so small that the result
could have been reversed with a small shift in votes from the winner to the
loser. In the context of such constituencies, we first estimate the separate

© The Editor(s) (if applicable) and The Author(s) 2016 45

V.K. Borooah, Votes, Parties, and Seats,
DOI 10.1007/978-3-319-30487-8_3

vk.borooah@ulster.ac.uk
46 V.K. Borooah

l­ikelihoods of INC and BJP candidates winning marginal seats, in post-

1984 Lok Sabha elections, after controlling for factors, like, inter alia,
incumbency and turnout.
We next consider, in this chapter, all constituencies in which the INC
and the BJP went ‘head-to-head’ in the sense that both fielded candidates
in those constituencies. In estimating the likelihoods of the INC and the
BJP winning ‘head-to-head’, we used the econometric estimation method
of bivariate probit which allowed the testing of inter-party differences.
Lastly, the chapter considers the electoral performances of the INC
and the BJP separately for the Hindi-speaking and the non-Hindi-­
speaking major Indian states. The seven Hindi-speaking states—Bihar,
Chhattisgarh, Jharkhand, Madhya Pradesh, Rajasthan, Uttarakhand, and
Uttar Pradesh—provide 204 seats of the total of 543 Lok Sabha seats, and
these are of particular importance to the BJP because a large part of its
contested constituencies are from these states: in the Lok Sabha elections
of 2014, 192 of its contested 428 constituencies were from these states.

3.2   Marginal Constituencies

A marginal constituency is one where the difference in votes received
between the winning party and the runner-up is so narrow that the result
could have been reversed with a small shift of votes away from the winner.
In this book, we adopt—admittedly on arbitrary but, it is to be hoped, not
unreasonable criteria—two definitions of a marginal constituency: (i) the
difference in vote shares between the winner and the runner-up was 10
percentage points or less so that under this definition, a shift of 5 percent
of the constituency vote away from the winner to the runner-up would
have reversed the result; and (ii) the difference in vote shares between the
winner and the runner-up was 5 percentage points or less so that under
this definition, a shift of 2.5 percent of the constituency vote away from
the winner to the runner-up would have reversed the result.
Notwithstanding the fact that elections to the Lok Sabha often have
clear winners and losers, many elections in several constituencies are
closely contested. In 2014, when the BJP won a handsome parliamentary
majority with 282 seats, there were 190 constituencies (35 percent of the
total of 540 constituencies) in which difference in vote shares between
the winning and the losing party was 10 percentage points or less and 96
constituencies (18 percent of the total of 540 constituencies) in which
difference in vote shares between the winning and the losing party was 5

vk.borooah@ulster.ac.uk
 Estimating the Likelihood of Winning Parliamentary... 47

percentage points or less. The corresponding figures for the INC “land-
slide” election of 1984 were 154 constituencies at the 10-point level and
80 constituencies at the 5-point level: respectively, 29 and 15 percent of
the total of 537 constituencies. Table 3.1 shows the number of marginal
constituencies for each of the elections between 1962 and 2014.
The important point that emerges from Table 3.1 is the growing pres-
ence of marginal constituencies in the total of constituencies. In 1971,
when the INC won 352 seats, one in four constituencies was a ‘10-point’
marginal and ‘5-point’ marginal comprised 12 percent of total constitu-
encies. By 1984, when the INC won 414 seats, ‘10-point’ marginal and
‘5-point’ marginal comprised, respectively, 28 and 15 percent of total con-
stituencies, and in the elections since 1998, marginal seats have come to
dominate reaching an apotheosis in 2009 of 63 percent of all constituen-
cies decided on a margin of 10 percent or less and 36 percent of all constit-
uencies decided on a margin of 5 percent or less. This would suggest that
targeting key groups of voters is (or should be) an increasingly important
part of the electoral strategy in India since small swings in support can,
more than ever before, make the difference between forming a govern-
ment or sitting in opposition.

Table 3.1 Number of marginal constituencies in Lok Sabha elections:

1962–2014
Year Total number of Number of marginal Number of marginal
constituencies constituencies at ≤ 10 points constituencies at ≤ 5 points
difference difference

1962 490 206 105

1967 518 221 111
1971 518 126 62
1977 542 93 49
1980 529 165 90
1984 541 154 80
1989 528 207 107
1991 537 252 131
1996 543 289 152
1998 543 327 193
1999 543 323 190
2004 543 275 151
2009 543 343 198
2014 543 190 96

Source: Own calculations from Lok Sabha election data

vk.borooah@ulster.ac.uk
48 V.K. Borooah

Table 3.2 shows the distribution of marginal constituencies, across

the major Indian states, for the 2014 Lok Sabha elections. Of the
178 marginal constituencies (at a 10 points difference) in the major
Indian states, 113 (63 percent) were located in the six states of Andhra
Pradesh, Bihar, Karnataka, Kerala, Uttar Pradesh, and West Bengal.
In Kerala, 80 percent of constituencies were ‘marginal’; in Andhra
Pradesh and Karnataka, nearly two in three constituencies were ‘mar-
ginal’; and in Bihar and West Bengal, nearly one in two constituencies
were ‘marginal’.

Table 3.2 Marginal constituencies by major Indian states in the 2014 Lok Sabha
elections
State Total number of Number of marginal Number of marginal
constituencies constituencies at ≤ 10 constituencies at ≤ 5
points difference points difference

Andhra Pradesh 42 26 15
Assam 14 6 4
Bihar 40 18 9
Chhattisgarh 11 4 4
Gujarat 26 2 0
Haryana 10 3 1
Himachal 4 1 0
Pradesh
Jammu and 6 3 1
Kashmir
Jharkhand 14 7 4
Karnataka 28 17 8
Kerala 20 16 13
Madhya Pradesh 29 5 2
Maharashtra 48 8 4
Orissa 21 7 5
Punjab 13 9 6
Rajasthan 25 5 2
Tamil Nadu 39 5 1
Uttarakhand 5 0 0
Uttar Pradesh 80 19 7
West Bengal 42 17 4
Total 517 178 90

Source: Own calculations from Lok Sabha election data

vk.borooah@ulster.ac.uk
 Estimating the Likelihood of Winning Parliamentary... 49

3.3   The Likelihood of Winning Marginal Seats

An important consideration, an important question in Indian politics—
and, indeed in electoral politics in general—is the relative strength of the
factors which determine whether or not parties win marginal seats. Using
econometric techniques, we attempt to tease out, for India’s two leading
political parties, the INC and the BJP, answers to this pressing, and com-
plex, question.
In order to do so, we estimated for the INC—across the 14 elections
between 1962 and 2014—a logit equation on data for ‘10-point’ mar-
ginal constituencies, in the 20 major Indian states (listed in Table 3.2), in
which the INC was either the winner or the runner-up. These, collectively,
yielded a total of 1989 constituency observations. A similar equation was
estimated for the BJP on data for ‘10-point’ marginal constituencies, in
the 20 major Indian states in which the BJP was either the winner or run-
ner-­up. Since the BJP only made its electoral debut in the 1984 Lok Sabha
elections, the data related to the nine elections between 1984 and 2014.
These, collectively, yielded a total of 1009 constituency observations.1, 2
In a logit model, the dependent variable, y, takes the value 1 if the
condition is present (a party wins the election from constituency i: yi = 1)
and the value 0 if the condition is absent (a party loses the election from
constituency i: yi = 0). Suppose there are N constituencies which the party
contests, so that yi = 1 for some constituencies and yi = 0 for the others. If
Pr[yi = 1] and Pr[yi = 0] represent, respectively, the probabilities of the
party winning from constituency i, i = 1…N, the logit formulation expresses
the log of the odds ratio (OR) as a linear function of K variables (indexed
k = 1…K) which take values, Xi1,Xi2 … XiK in constituency i, i = 1…N:

 Pr [ yi = 1]  K
 1 − Pr [ y = 1]  ∑
log  = β k Xik + ui = Z i (3.1)
 i  k =1

Where, βk is the coefficient associated with variable k, k = 1…K.

From Eq. 2.1, it follows that:

e zi
Pr [ yi = 1] = (3.2)
1 + e zi

where, the term ‘e’, in the above equation, represents the exponential
term.

vk.borooah@ulster.ac.uk
50 V.K. Borooah

The explanatory power of the logit equations is shown in terms of the

‘Pseudo-R2’. The ‘Pseudo-R2’ is a popular measure of the model’s perfor-
mance in binary models and compares the maximised log-likelihood value
of the full model (log L) to that obtained when all the coefficients, expect
the intercept term, are set to 0 (log L0) and is defined as 1-(log L/log L0).
The measure has an intuitive appeal in that it is bounded by 0 (all the slope
coefficients are 0) and 1 (perfect fit).3
The dependent variable in the logit equations estimated in this section
took the value 1 for a (marginal) constituency if the party (INC or BJP)
was the winner in that constituency and the value 0 if the party (INC
or BJP) was the runner-up in that constituency.4 In some of these con-
tests in marginal constituencies the INC and the BJP went head-to-head
(meaning, one was the winner and the other the runner-up), but in other
contests, the INC and the BJP went head-to-head with other opponents.5
There were seven variables which were hypothesised to play a signifi-
cant role in determining the outcome (winner or runner-up) in a marginal
constituency:

1. The share of the total votes received by the party in that

constituency;
2. Whether the party held the constituency in the previous election
(i.e., it was the ‘incumbent’ party);6
3. The percentage of the electorate voting in that election
(‘turnout’);
4. The number of independent candidates in the election;
5. The number of ‘other’ (i.e., third, fourth, etc.) party candidates in
the election;
6. The year of the election; and
7. The state in which the constituency was located.

In order to allow for non-linear effects, the squared value of ‘vote

share’, ‘turnout’, ‘the number of independent candidates’, and the num-
ber of ‘other’ parties was also included in the equations. The logit esti-
mates for the INC and the BJP equations are shown in Tables 3.3 and
3.4, respectively. The coefficient estimates shown in the second column of
Tables 3.3 and 3.4 are the estimates of the coefficients βk in Eq. 3.1. The
third column shows the standard errors associated with these estimates.
Dividing the estimates by their corresponding standard errors yields the
z-value shown in the fourth column. The value in the fifth column shows
the probability of observing the z-value under the null hypothesis that

vk.borooah@ulster.ac.uk
 Estimating the Likelihood of Winning Parliamentary... 51

Table 3.3 Logit estimates for the probability of the INC winning marginal
constituencies
1 2 3 4 5

Variable Estimated Standard z-value p-value

coefficients error [Pr > |z|]

INC vote share −0.572 0.078 −7.33 0.00

INC vote share squared 0.009 0.001 9.34 0.00
INC incumbent 0.351 0.113 3.10 0.00
Turnout −0.150 0.039 −3.88 0.00
Turnout squared 0.001 0.000 3.43 0.00
Number of independents 0.056 0.023 2.42 0.02
Number of independents −0.001 0.001 −1.90 0.06
squared
Number of ‘other’ parties 0.526 0.095 5.53 0.00
Number of ‘other’ parties −0.029 0.006 −4.51 0.00
squared
Year [Reference: 1967]
1971 −0.600 0.305 −1.97 0.05
1977 −1.415 0.334 −4.24 0.00
1980 −0.077 0.269 −0.29 0.77
1984 −0.798 0.282 −2.83 0.01
1989 −1.645 0.269 −6.11 0.00
1991 −1.412 0.289 −4.88 0.00
1996 −1.285 0.298 −4.31 0.00
1998 −0.992 0.276 −3.59 0.00
1999 −1.813 0.266 −6.83 0.00
2004 −1.308 0.272 −4.81 0.00
2009 −0.674 0.298 −2.26 0.02
2014 −1.794 0.407 −4.41 0.00
State [Reference: Andhra Pradesh]
Assam 1.087 0.347 3.13 0.00
Bihar 0.078 0.299 0.26 0.79
Chhattisgarh −2.214 0.813 −2.72 0.01
Gujarat −1.086 0.271 −4.00 0.00
Haryana 0.644 0.375 1.72 0.09
Himachal Pradesh −0.880 0.522 −1.69 0.09
Jammu and Kashmir 0.857 0.637 1.34 0.18
Jharkhand −0.828 0.832 −0.99 0.32
Karnataka −0.006 0.241 −0.02 0.98
Kerala −0.336 0.246 −1.37 0.17
Madhya Pradesh −0.810 0.256 −3.16 0.00
Maharashtra −0.261 0.240 −1.09 0.28

(continued)

vk.borooah@ulster.ac.uk
52 V.K. Borooah

Table 3.3 (continued)

1 2 3 4 5

Variable Estimated Standard z-value p-value

coefficients error [Pr > |z|]

Orissa −0.025 0.298 −0.09 0.93

Punjab −0.448 0.315 −1.42 0.16
Rajasthan −0.898 0.285 −3.15 0.00
Tamil Nadu −1.117 0.359 −3.12 0.00
Uttarakhand −0.266 0.929 −0.29 0.77
Uttar Pradesh 0.432 0.270 1.60 0.11
West Bengal −1.305 0.267 −4.88 0.00
Intercept 11.198 1.919 5.84 0.00

Source: Own calculations from Lok Sabha election data

Notes: (i) The equation was estimated on data for 1989 constituencies for major Indian states, as listed, for
all Lok Sabha elections between 1967 and 2014
(ii) Pseudo R2 = 0.1844; likelihood ratio test: χ2(40) = 508.49

the coefficient was 0. At 5 and 10 percent levels of significance, this null

hypothesis was ‘rejected’ for, respectively, p < 0.05 and p < 0.1.
Following the advice contained in Long and Freese (2014), the results
from the estimated equations in Tables 3.3 and 3.4 are presented, for
subsequent analysis, in the form of the predicted probabilities or, equiva-
lently, predicted likelihoods (the terms ‘probability’ and ‘likelihood’ are,
hereafter, used interchangeably) computed from the estimated logit coef-
ficients, from which these probabilities are derived, and not in terms of
the estimates themselves. In other words, the subsequent analysis uses the
expression in Eq. 3.2 to compute the outcome probabilities where these
are derived from the coefficient estimates of Eq. 3.1. This is because the
logit estimates themselves do not have a natural interpretation—they exist
mainly as a basis for computing more meaningful statistics, and in this
case, these are the predicted probabilities of winning under a variety of
configurations.
It should be emphasised that these predicted probabilities will, in gen-
eral, differ from the sample proportions. This is because the predicted
probabilities are computed after controlling (or adjusting) for the effects
of the conditioning variables (noted above) while the sample proportions
represent raw, unadjusted data. Since a property of the logit model is that

vk.borooah@ulster.ac.uk
 Estimating the Likelihood of Winning Parliamentary... 53

Table 3.4 Logit estimates for the probability of the BJP winning marginal
constituencies
1 2 3 4 5

Variable Estimated logit Standard z-value p-value

coefficients error

INC vote share −0.195 0.109 −1.80 0.07

INC vote share squared 0.005 0.001 3.48 0.00
INC incumbent −0.192 0.155 −1.23 0.22
Turnout 0.127 0.075 1.70 0.09
Turnout squared −0.001 0.001 −1.90 0.06
Number of independents −0.025 0.014 −1.87 0.06
Number of independents 0.000 0.000 0.40 0.69
squared
Number of ‘other’ parties 0.223 0.129 1.73 0.08
Number of ‘other’ parties −0.007 0.008 −0.95 0.34
squared
Year [Reference: 1989]
1991 0.037 0.441 0.08 0.93
1996 0.728 0.464 1.57 0.12
1998 −0.136 0.433 −0.31 0.75
1999 −0.141 0.431 −0.33 0.74
2004 0.076 0.436 0.17 0.86
2009 −0.027 0.440 −0.06 0.95
2014 0.728 0.547 1.33 0.18
State [Reference: Andhra Pradesh]
Assam 1.491 0.709 2.10 0.04
Bihar 1.311 0.646 2.03 0.04
Chhattisgarh 1.485 1.050 1.41 0.16
Gujarat −0.403 0.656 −0.61 0.54
Haryana 1.246 0.866 1.44 0.15
Himachal Pradesh −0.666 0.815 −0.82 0.41
Jammu and Kashmir 0.192 1.201 0.16 0.87
Jharkhand 1.279 0.819 1.56 0.12
Karnataka 0.750 0.624 1.20 0.23
Kerala –
Madhya Pradesh 0.384 0.627 0.61 0.54
Maharashtra 0.250 0.627 0.40 0.69
Orissa 0.336 0.808 0.42 0.68
Punjab −0.233 0.954 −0.24 0.81
Rajasthan −0.273 0.646 −0.42 0.67
Tamil Nadu −0.685 0.944 −0.73 0.47
Uttarakhand −0.901 1.127 −0.80 0.42

(continued)

vk.borooah@ulster.ac.uk
54 V.K. Borooah

Table 3.4 (continued)

1 2 3 4 5

Variable Estimated logit Standard z-value p-value

coefficients error

Uttar Pradesh 1.764 0.634 2.78 0.01

West Bengal 0.450 1.064 0.42 0.67
Intercept −4.599 2.842 −1.62 0.11

Source: Own calculations from Lok Sabha election data

Notes: (i) The equation was estimated on data for 1009 constituencies for major Indian states, as listed, for
all Lok Sabha elections between 1989 and 2014
(ii) Pseudo R2 = 0.1836; likelihood ratio test: χ2(34) = 253.58

it passes through the sample mean, the overall predicted probability, from
the logit model, of winning a marginal constituency will be the same as the
overall sample proportion of marginal constituencies in which the party
was victorious. However, while the estimated model passes through the
overall sample mean, it does not pass through the means of the different
sample subgroups. This is illustrated in Table 3.5 which compares, for
each year and in aggregate, the predicted probabilities and the sample
averages. The two quantities differ for each election (though they follow
each other closely over the elections) but are the same when aggregated
over all the elections.
The general methodology for computing the predicted probabilities
was to calculate, for each of the observations (1989 for the INC; 1009
for the BJP), the probability of winning the election under a hypothetical
situation (Scenario 1) in which some of the independent variables took
specified values (e.g., the variable ‘year’ was set to 1967), the values of the
other independent variables (turnout, etc.) being as observed. This then
yielded 69.2 percent as the predicted probability of winning a marginal
constituency in 1967.
In order to obtain the predicted probability of winning in 1971, the
variable ‘year’ was set to 1971, the values of the other independent vari-
ables being as observed. This then yielded Scenario 2. The difference in
the average probability of winning between the scenarios could then be
ascribed to the change in the value of the independent variable(s), in this
case between the years 1967 and 1971.

vk.borooah@ulster.ac.uk
 Estimating the Likelihood of Winning Parliamentary... 55

3.3.1  Election-on-Election Changes in the Probability

of Winning Marginal Constituencies
Table 3.5 showed that the predicted probability of the BJP winning a
marginal constituency was, except for the 2009 election, always greater
than that of the INC. So, there is prima facie evidence, that with respect
to marginal constituencies at least, the BJP is a more electorally efficient
party than the INC; however, we postpone, till the next section, a detailed
examination of this hypothesis. The other question raised by the results
of Table 3.5 is whether the year-on-year likelihoods of winning marginal
constituencies significantly different from each other.
Table 3.6 shows the results from testing the significance of election-on-­
election changes in the likelihood of the INC and the BJP winning mar-
ginal constituencies. This table shows that for the INC, the change in the
likelihood of winning marginal constituencies was significantly different
from 0 (hereafter, simply “significant”) between: the 1967 and 1971 elec-
tions (went down), the 1971 and 1977 elections (went down), the 1977
and 1980 elections (went up), the 1980 and 1984 elections (went down),
the 1984 and 1989 elections (went down), the 1998 and 1999 elections
(went down), the 1999 and 2004 elections (went up), the 2004 and 2009
elections (went up), and the 2009 and 2014 elections (went down).
For the BJP, the year-on-year changes were significantly different from
0 between: the 1991 and 1996 elections (went up), the 1996 and 1998
elections (went down), and the 2009 and 2014 elections (went up).
Overall, in terms of contesting marginal constituencies, the two good
elections for the INC since 1989 have been 2004 and 2009 after both of
which the INC’s predicted probability of winning marginal constituencies
rose. For the BJP, on the other hand, 1996 was a good election in terms of
contesting marginal constituencies and, of course, so was the most recent
election of 2014.

3.3.2  Incumbency Effects
Table 3.7 shows that the average likelihood of the INC winning marginal
constituencies was 53.5 percent if it was the incumbent party and 46.7
percent if it was the non-incumbent, and this difference was significantly
different from 0. For the BJP, on the other hand, the likelihood of win-
ning marginal constituencies was 54.5 percent if it was the incumbent
party and 58.2 percent if it was the non-incumbent, but this difference

vk.borooah@ulster.ac.uk
 56

Table 3.5 INC and BJP predicted probabilities of winning marginal constituencies
INC BJP
V.K. Borooah

Probability of Sample proportion Total number of Probability of Sample proportion Total number of
winning of marginal marginal winning of marginal marginal
marginal constituencies won constituencies marginal constituencies won constituencies
constituency contested constituency contested

1967 0.692 0.537 203

1971 0.583 0.553 92
1977 0.427 0.493 73
1980 0.678 0.595 131
1984 0.545 0.594 138
1989 0.385 0.480 175 0.543 0.553 38
1991 0.428 0.511 188 0.550 0.512 121
1996 0.451 0.450 180 0.676 0.543 140

vk.borooah@ulster.ac.uk
1998 0.507 0.515 163 0.516 0.554 168
1999 0.355 0.382 178 0.516 0.592 169
2004 0.447 0.421 164 0.557 0.558 138
2009 0.569 0.592 218 0.537 0.525 143
2014 0.358 0.302 86 0.676 0.728 92
Total 0.499 0.499 1989 0.566 0.566 1009

Source: Own calculations from Lok Sabha election data

 Estimating the Likelihood of Winning Parliamentary... 57

Table 3.6 Election-on-election changes in the likelihood of winning marginal

constituencies
INC BJP

Difference Standard z-value p-value Difference Standard z-value p-value

in the error in the error
probability probability
of winninga of winninga

1967– 0.109 0.056 1.95 0.05

71
1971– 0.156 0.072 2.15 0.03
77
1977– −0.252 0.067 −3.77 0.00
80
1980– 0.134 0.056 2.40 0.02
84
1984– 0.160 0.053 3.05 0.00
89
1989– −0.043 0.048 −0.90 0.37 −0.007 0.084 −0.08 0.93
91
1991– −0.024 0.046 −0.52 0.60 −0.127 0.056 −2.26 0.02
96
1996– −0.056 0.053 −1.05 0.29 0.160 0.065 2.47 0.01
98
1998– 0.153 0.047 3.27 0.00 0.001 0.051 0.02 0.99
99
1999– −0.092 0.046 −2.01 0.04 −0.041 0.052 −0.80 0.43
04
2004– −0.122 0.048 −2.54 0.01 0.020 0.054 0.36 0.72
09
2009– 0.211 0.057 3.67 0.00 −0.139 0.072 −1.93 0.05
14

Source: Own calculations from Lok Sabha election data

a
The difference is computed as the likelihood of winning in the earlier year minus the likelihood of win-
ning in the later year. The likelihoods are shown in Table 3.5

was not significantly different from 0. Consequently, the overall evidence,

over the elections between 1962 and 2014, was that for the INC, there
was a significant pro-incumbency effect operating in marginal constituen-
cies. On the face of it, there was an anti-incumbency effect operating in
marginal constituencies for the BJP. However, considering the elections

vk.borooah@ulster.ac.uk
58 V.K. Borooah

Table 3.7 INC and BJP predicted probabilities of winning marginal constituen-
cies as the incumbent and non-incumbent parties
INC BJP

Incumbent Non-incumbent Incumbent Non-incumbent

probability of probability of probability of probability of
winning winning winning winning

1967 0.724 0.664

1971 0.618 0.551
1977 0.462 0.396
1980 0.711 0.650
1984 0.581 0.513
1989 0.418 0.354
1991 0.462 0.396 0.529 0.565
1996 0.487 0.420 0.657 0.690
1998 0.543 0.475 0.496 0.533
1999 0.387 0.325 0.495 0.532
2004 0.482 0.415 0.536 0.573
2009 0.604 0.537 0.517 0.553
2014 0.390 0.328 0.657 0.690
All 0.535 0.467 0.545 0.582
elections

Source: Own calculations from Lok Sabha election data

between 1989 and 2014 in their entirety, the observed anti-incumbency

effect was not statistically significant.

3.3.3  Number of Candidates, Turnout, and Vote Share

Figure 3.1 shows the predicted probability, computed over all the Lok
Sabha elections between 1967 and 2014, of the INC winning a “10-point”
marginal constituency as 49.9 percent.7 In computing this probability,
the values of all the other variables—in particular, the number of inde-
pendent and ‘other’ party candidates—were ‘as observed’. If the model
was tweaked so that there were no independent candidates—the number
of ‘other’ party candidates as observed—the predicted probability of the
INC winning a marginal constituency would have fallen to 45.7 percent.
Under a different, but related, scenario in which there no ‘other’ party
candidates—the number of independent candidates as observed—the

vk.borooah@ulster.ac.uk
 Estimating the Likelihood of Winning Parliamentary... 59

Fig. 3.1 INC and BJP predicted probabilities of winning marginal constituen-
cies with no independent candidates or just two parties contesting (Source: Own
calculations from Lok Sabha election data)

predicted probability of the INC winning a marginal constituency would

have fallen to 35.8 percent. The results when the predicted sample was
restricted to the elections between 1989 and 2014 were similar. The gen-
eral conclusion is that electoral competition in marginal constituencies—
through the presence of independents and ‘other’ parties—enhances the
INC’s chances of winning marginal constituencies. These candidates split
the anti-INC vote so that in their absence, the INC’s chances of winning
would have been lower.
For the BJP, on the other hand, the predicted probability, computed
over all the Lok Sabha elections between 1989 and 2014, of its winning
a marginal constituency would have risen from 56.6 percent, when the
number of independent and ‘other’ party candidates were ‘as observed’,
to 60.4 percent under a scenario under which there were no independent
candidates. In electoral terms, the presence of independent candidates
erodes the BJP vote and reduces its chances of winning. In this respect,
the effect of independent candidates on the chances of the INC and the
BJP winning marginal constituencies are diametrically different: indepen-
dents help the INC but hurt the BJP.

vk.borooah@ulster.ac.uk
60 V.K. Borooah

Fig. 3.2 INC and BJP predicted probabilities of winning marginal constituen-
cies at different rates of voter turnout (Source: Own calculations from Lok Sabha
election data)

Figure 3.2 shows that a high turnout of voters in marginal constituen-

cies hurts the winning chances of both the INC and BJP. Ceteris paribus
the INC was predicted to have a 48.7 percent chance, computed over all
the Lok Sabha elections between 1989 and 2014, of winning a marginal
constituency when the turnout of voters was 50 percent. The correspond-
ing prediction for the BJP was 59.8 percent. As the turnout rate rose, the
likelihood of both parties winning fell: at a 65 percent turnout, the pre-
dicted probabilities of the INC and the BJP winning marginal constituen-
cies were, respectively, 43.5 and 53.1 percent.
It is a truism of electoral politics that the recipe for winning in getting
your supporters into the polling booth while, simultaneously, ensuring
that your opponents stay at home. The above findings illustrate this tru-
ism. Both the INC and the BJP have core supporters who would readily
vote for their party. A low turnout, on the other hand, ensures that many
putative voters—who may not be as enthusiastic about the INC or the BJP
as their more committed supporters—do not spoil the party by coming
out to vote.
Figure 3.3 shows the predicted probabilities of the INC and the
BJP winning marginal constituencies for different vote shares obtained.

vk.borooah@ulster.ac.uk
 Estimating the Likelihood of Winning Parliamentary... 61

Fig. 3.3 INC and BJP predicted probabilities of winning marginal constituen-
cies at different vote shares (Source: Own calculations from Lok Sabha election
data)

Computed over all the eight elections between 1989 and 2014, the pre-
dicted probability of winning marginal constituencies, at each of the three
vote shares, 35, 40, and 45 percent, was always higher for the BJP than for
the INC: with a 35 percent vote share, the INC had a 34.4 percent chance
of winning a marginal constituency compared to the BJP’s 56.3 percent.
The next section examines the relative performance of the INC and the
BJP in greater depth.

3.4   The Electoral Performance of the INC

and the BJP Compared

The previous section examined the electoral performance, in marginal con-

stituencies, of the INC and the BJP. This was, however, conducted sepa-
rately for the two parties without attempting to assess their c­ omparative
performance. So, for example, we examined, for marginal constituencies
in which the INC was the winner or the runner-up, the likelihood of it
winning the election regardless of who its closest opponent was: this could
have the BJP, or another party, or even an independent candidate. In a

vk.borooah@ulster.ac.uk
62 V.K. Borooah

similar vein, we examined the likelihood of the BJP winning in marginal

constituencies, in which it was the winner or the runner-up, regardless of
who its closest opponent was: this could have the INC, or another party,
or even an independent candidate. By contrast, in this section, we make
a head-to-head comparison of the INC and BJP by analysing all the con-
stituencies that were contested by both parties.
Table 3.8, which sets out the number of constituencies contested by
both parties, shows that the proportion of all Lok Sabha constituencies
contested by both parties has increased from 41.5 percent in 1984 to
around 85 percent in 1991 and 1996, before falling to around 66 percent
in the last three Lok Sabha elections of 2004, 2009, and 2014. This fall
had been engendered by the INC and the BJP having to bow to the exi-
gencies of coalition politics and contesting fewer seats than they were used
to in the 1990s.
As Fig. 3.4 shows, the constituencies contested by the INC in 2014
were, at 464, 62 seats fewer than the 526 contested by it in 1996. For the
BJP, the largest numbers of constituencies contested were in 1991 and
1996 when it contested well over 450 constituencies. It then drew in its
horns for the 1998, 1999, and 2004 elections, when its tally of contested
constituencies was well short of 400; since then the BJP has extended its
electoral reach, contesting 433 and 428 constituencies, respectively, in the
2009 and 2014 election.

Table 3.8 Constituencies contested by the INC and the BJP

Contested by Contested by Contested by BJP Contested by Total
INC and BJP INC but not by but not by INC neither party
BJP

1984 225 292 4 20 541

1989 221 289 4 14 528
1991 457 47 21 12 537
1996 456 70 13 4 543
1998 352 125 36 30 543
1999 307 146 32 58 543
2004 310 107 54 72 543
2009 361 79 72 31 543
2014 373 91 55 24 543
Total 3062 3669 291 439 7461

Source: Own calculations from Lok Sabha election data

vk.borooah@ulster.ac.uk
 Estimating the Likelihood of Winning Parliamentary... 63

Fig. 3.4 Total number of seats contested by the INC and the BJP: 1984–2014
(Source: Own calculations from Lok Sabha election data)

3.4.1  Econometric Methodology
In order to assess the relative electoral performance of the INC and the
BJP, we estimated a two-equation probit (bivariate probit) model over
the sample of constituencies—in the 20 major states of India (listed in
Tables 3.3 and 3.4) and over the nine Lok Sabha elections from 1984
to 2014—which were contested by both the INC and the BJP.8 The first
equation related to the INC: the dependent variable in this equation took
the value 1 (yi = 1) if the INC won the election for constituency i, i = 1,…N,
and 0 if it did not (yi = 0). The second equation related to the BJP: the
dependent variable in this equation took the value 1 (zi = 1) if the BJP won
the election for constituency i, i = 1,…N, and 0 if it did not (zi = 0).
This system of two probit equations (bivariate probit) is the discrete
choice analogue of the Seemingly Unrelated Regression Equations (SURE)
method of estimation with continuous dependent variables (Greene,
2003, pp. 710–19). Like SURE estimates, the estimates from the bivariate
probit system are more efficient than those obtained from estimating each
equation as a single equation because the correlation between the error
terms of the two equations is explicitly taken into account. In addition,
and more importantly for the purpose of this analysis, the fact that the

vk.borooah@ulster.ac.uk
64 V.K. Borooah

equations are estimated as a system allows hypotheses to be tested between

equations rather than just within individual equations. As we will see, this
allows one to arrive at an assessment of the comparative electoral efficiency
of the INC and the BJP.
The estimates from the bivariate probit equation, estimated on data
for the 2684 constituencies contested by both the INC and the BJP over
1989–2014, are shown in Table 3.9. The same conditioning variables
were used in both the probit equations—one for the INC and the other
for the BJP—and, indeed, are those used in the logit analysis of the previ-
ous section.9 To recapitulate, these were:

(i) The share of the total votes received by the party in that
constituency;
(ii) Whether the party held the constituency in the previous election
(i.e., it was the ‘incumbent’ party);
(iii) The percentage of the electorate voting in that election (‘turnout’);
(iv) The number of independent candidates in the election;
(v) The number of ‘other’ (i.e., other than the INC and the BJP) party
candidates in the election;
(vi) The year of the election; and
(vii) The state in which the constituency was located.

The comparison between the electoral performance of the INC and the
BJP, in constituencies where both parties were contestants, was made with
respect to two parameters:

1. The overall probability of winning the constituency computed with

the values of the conditioning variables taking their observed con-
stituency values; and
2. The overall probability of winning the constituency for each party
obtaining a particular vote share: 35, 40, and 45 percent. In other
words, what would be the likelihood of the INC and the BJP win-
ning a constituency if their vote shares had been (say) 35 percent,
and was this likelihood significantly different between the two
parties?

Table 3.10 compares the predicted probabilities of the INC and the
BJP winning Lok Sabha elections between 1989 and 2014 in constituen-
cies, in the 20 major Indian states, which they both contested. Aggregated
over all these elections, the first row of Table 3.10 shows that the INC

vk.borooah@ulster.ac.uk
 Table 3.9 Bivariate probit estimates for the likelihood of the INC and the BJP winning constituency elections
INC BJP

Variable Estimated Standard z-value p-value Estimated Standard z-value p-value

bivariate probit error bivariate probit error
coefficients coefficients

INC vote share 0.064 0.023 2.79 0.01 0.161 0.033 4.82 0.00
INC vote share 0.001 0.000 2.59 0.01 0.000 0.000 −1.05 0.30
squared
INC incumbent 0.358 0.084 4.24 0.00 0.050 0.087 0.58 0.57
Turnout −0.042 0.027 −1.53 0.13 0.041 0.040 1.02 0.31
Turnout squared 0.000 0.000 1.31 0.19 0.000 0.000 −1.16 0.25
Number of 0.017 0.013 1.33 0.18 −0.002 0.014 −0.12 0.90
independents
Number of 0.000 0.000 −1.24 0.22 0.000 0.000 −0.26 0.79
independents squared
Number of ‘other’ 0.102 0.055 1.87 0.06 0.143 0.054 2.64 0.01
parties

vk.borooah@ulster.ac.uk
Number of ‘other’ −0.002 0.003 −0.75 0.46 −0.005 0.003 −1.79 0.07
parties squared
Year [Reference: 1989]
1991 0.451 0.154 2.92 0.00 −0.254 0.177 −1.44 0.15
1996 0.456 0.165 2.76 0.01 0.226 0.200 1.13 0.26
1998 0.503 0.166 3.03 0.00 −0.198 0.176 −1.12 0.26
1999 0.047 0.170 0.27 0.78 −0.239 0.178 −1.35 0.18
2004 0.391 0.168 2.33 0.02 −0.193 0.175 −1.10 0.27
2009 0.753 0.166 4.53 0.00 −0.291 0.184 −1.58 0.12
2014 −0.359 0.223 −1.61 0.11 0.314 0.250 1.26 0.21
Estimating the Likelihood of Winning Parliamentary...

State [Reference: Andhra Pradesh]

65

(continued)
 Table 3.9 (continued)
66

INC BJP

Variable Estimated Standard z-value p-value Estimated Standard z-value p-value

bivariate probit error bivariate probit error
coefficients coefficients
V.K. Borooah

Assam 0.761 0.214 3.56 0.00 0.794 0.378 2.10 0.04

Bihar −0.387 0.258 −1.50 0.13 1.018 0.367 2.77 0.01
Chhattisgarh −1.747 0.589 −2.97 0.00 1.198 0.535 2.24 0.03
Gujarat −1.062 0.199 −5.35 0.00 0.114 0.358 0.32 0.75
Haryana 0.976 0.257 3.81 0.00 0.817 0.398 2.05 0.04
Himachal Pradesh −0.962 0.412 −2.33 0.02 −0.023 0.485 −0.05 0.96
Jammu and Kashmir 0.553 0.433 1.28 0.20 0.950 0.571 1.67 0.10
Jharkhand 0.366 0.346 1.06 0.29 1.256 0.437 2.88 0.00
Karnataka −0.155 0.162 −0.96 0.34 0.675 0.351 1.92 0.05
Kerala −0.317 0.179 −1.77 0.08 –
Madhya Pradesh −0.822 0.173 −4.76 0.00 0.475 0.346 1.37 0.17
Maharashtra −0.187 0.180 −1.04 0.30 0.117 0.351 0.33 0.74

vk.borooah@ulster.ac.uk
Orissa −0.160 0.206 −0.78 0.44 0.455 0.446 1.02 0.31
Punjab −0.203 0.321 −0.63 0.53 −0.114 0.602 −0.19 0.85
Rajasthan −0.689 0.213 −3.24 0.00 −0.048 0.361 −0.13 0.89
Tamil Nadu −0.490 0.277 −1.77 0.08 0.065 0.421 0.16 0.88
Uttarakhand −0.307 0.427 −0.72 0.47 −0.008 0.537 −0.01 0.99
Uttar Pradesh 0.538 0.214 2.51 0.01 1.381 0.349 3.95 0.00
West Bengal −0.745 0.218 −3.42 0.00 0.147 0.476 0.31 0.76
Intercept −3.709 0.899 −4.12 0.00 −7.626 1.294 −5.89 0.00

Source: Own calculations from Lok Sabha election data

Notes to Table 3.9: Observations on 2684 constituencies, in the 20 major Indian states, which were contested by both the INC and the BJP
 Estimating the Likelihood of Winning Parliamentary... 67

Table 3.10 INC and BJP predicted likelihood of winning by year

1 2 3 4 5 6 7

INC BJP Difference Standard z-value for p-value

likelihood of likelihood of error of testing H0:
winning winning difference Difference = 0

All 0.306 0.405 −0.099 0.010 −9.59 0.00

years
1989 0.243 0.422 −0.179 0.039 −4.58 0.00
1991 0.327 0.379 −0.052 0.030 −1.77 0.08
1996 0.328 0.459 −0.131 0.033 −3.96 0.00
1998 0.337 0.389 −0.052 0.032 −1.63 0.10
1999 0.251 0.382 −0.131 0.031 −4.18 0.00
2004 0.315 0.390 −0.075 0.033 −2.26 0.02
2009 0.387 0.373 0.014 0.032 0.45 0.65
2014 0.185 0.472 −0.288 0.041 −7.01 0.00

Source: Own calculations from Lok Sabha election data

had a 30.6 percent chance of winning an election compared to the BJP’s

40.5 percent, and furthermore, reading across the row, this difference of
nearly 10 points was significantly different from 0. For every election in
this period, except for the 2009 election, the predicted likelihood of the
BJP winning was greater than that of the INC, and for several elections
(1989, 1996, 1999, 2004, and 2014), this difference in the likelihoods
was significantly different from 0.
Table 3.11 compares the probabilities of winning for the INC and
the BJP under different scenarios for the vote share obtained. If the
INC received 35 percent of the vote, then ceteris paribus its predicted
probability of winning would be 22.6 percent; if, on the other hand, the
BJP received 35 percent of the vote, then ceteris paribus its predicted
probability of winning would be 31.5 percent. Under a 40 percent vote
share scenario, the predicted probabilities of winning would rise for both
parties, but the BJP’s probability would still be higher than that of the
INC: 50.8 percent against 39.6 percent. The pattern was repeated when
each party received a hypothetical 45 percent share of the total vote:
both likelihoods of winning would rise further, but the BJP advantage in
terms of a higher winning probability would continue (69 percent versus
60.2 percent).

vk.borooah@ulster.ac.uk
68 V.K. Borooah

Table 3.11 INC and BJP likelihood of winning at different constituency vote
shares
1 2 3 4 5 6 7

Vote INC BJP Difference Standard z-value for p-value

share likelihood of likelihood of error of testing H0:
(%) winning winning difference Difference = 0

All years
35 0.226 0.315 −0.089 0.020 −4.49 0.00
40 0.396 0.508 −0.112 0.024 −4.62 0.00
45 0.602 0.690 −0.088 0.040 −2.22 0.03
1989
35 0.152 0.333 −0.181 0.058 −3.10 0.00
40 0.301 0.530 −0.230 0.071 −3.23 0.00
45 0.506 0.712 −0.206 0.072 −2.87 0.00
1991
35 0.255 0.263 −0.009 0.044 −0.20 0.85
40 0.438 0.451 −0.012 0.053 −0.23 0.82
45 0.650 0.642 0.008 0.052 0.16 0.88
1996
35 0.256 0.400 −0.144 0.057 −2.54 0.01
40 0.440 0.600 −0.160 0.062 −2.57 0.01
45 0.651 0.767 −0.116 0.072 −1.60 0.11
1998
35 0.269 0.278 −0.010 0.049 −0.20 0.84
40 0.455 0.468 −0.013 0.057 −0.23 0.82
45 0.666 0.658 0.008 0.053 0.15 0.89
1999
35 0.161 0.267 −0.106 0.045 −2.36 0.02
40 0.314 0.455 −0.141 0.057 −2.48 0.01
45 0.521 0.646 −0.125 0.059 −2.13 0.03
2004
35 0.239 0.280 −0.040 0.049 −0.83 0.41
40 0.419 0.470 −0.051 0.060 −0.85 0.39
45 0.631 0.659 −0.028 0.058 −0.48 0.63
2009
35 0.339 0.254 0.085 0.047 1.80 0.07
40 0.536 0.439 0.097 0.056 1.74 0.08
45 0.737 0.631 0.106 0.053 2.01 0.04
2014
35 0.093 0.427 −0.334 0.064 −5.25 0.00
40 0.208 0.626 −0.419 0.071 −5.90 0.00
45 0.390 0.786 −0.396 0.100 −3.94 0.00

Source: Own calculations from Lok Sabha election data

vk.borooah@ulster.ac.uk
 Estimating the Likelihood of Winning Parliamentary... 69

3.5   Hindi-Speaking Versus Non-Hindi-Speaking

States
Of the total of 543 Lok Sabha constituencies, 204 (or 37.6 percent) are—
and have been since the 1996 Lok Sabha election—in the seven Hindi-­
speaking (HS) states, Bihar, Chhattisgarh, Jharkhand, Madhya Pradesh,
Rajasthan, Uttarakhand, and Uttar Pradesh; of these 204 constituencies,
respectively, 40 and 80 are in Bihar and Uttar Pradesh.10 The HS states are
of particular importance for the BJP since a large number of its contested
constituencies are from these states: in 2014, as Fig. 3.5 shows, nearly 45
percent (192 out of 428) of the constituencies contested by the BJP were
from the HS states. These states are also important for the INC but to a
lesser degree: as Fig. 3.5 shows, 34 percent (158 out of 464) of the con-
stituencies contested by the INC in 2014 were from the HS states.
Figure 3.6 shows that of the 373 constituencies which were contested
by both the INC and BJP in 2014, 153 constituencies (41 percent) were
from the HS states while in 2009, of the 361 constituencies which were
contested by both the INC and BJP, 149 constituencies (40 percent) were
from the HS states. Although this proportion of approximately 40 percent
of ‘head-to-head’ contests, between the INC and the BJP, in constitu-

Fig. 3.5 Seats contested by the INC and the BJP in Hindi-speaking states
(Source: Own calculations from Lok Sabha election data)

vk.borooah@ulster.ac.uk
70 V.K. Borooah

Fig. 3.6 Seats contested by both the INC and the BJP in Hindi-speaking states
(Source: Own calculations from Lok Sabha election data)

encies in the HS states has dipped from the corresponding figure of 48

percent in the 1999 and 2004 elections—145 out of 307 constituencies
in 1999 and 151 out of 310 constituencies in 2004—these constituencies
are, and likely to remain, an important battlefield for both parties.
This raises the question of whether the electoral performances of the
INC and the BJP, when they contested the same constituency, differed
according to whether the constituency was in an HS or a non-HS state.
With a view to answering this question, this section compares the electoral
performances of the INC and the BJP when they (both) contested con-
stituencies in the HS and in the non-HS states.
In order to compare the performances of the INC and the BJP, we
estimated two separate bivariate probit models (of the type described in
the earlier section): the first bivariate probit model was estimated on data
for constituencies, which were contested by both the INC and BJP, in the
13 major non-HS states; and the second bivariate probit model was esti-
mated on data for similar constituencies in the seven major HS states. In
total, over the eight elections between 1989 and 2014, there were 1456
such constituencies in the non-HS states and 1228 constituencies in the
HS states.

vk.borooah@ulster.ac.uk
 Estimating the Likelihood of Winning Parliamentary... 71

As in the previous section, the first equation in each of the two bivariate
probit models related to the INC: the dependent variable in this equation
took the value 1 (yi = 1) if the INC won the election for constituency i,
i = 1,…N, and 0 if it did not (yi = 0). The second equation related to the
BJP: the dependent variable in this equation took the value 1 (zi = 1) if
the BJP won the election for constituency i, i = 1,…N, and 0 if it did not
(zi = 0). The control variables in the non-HS and the HS models were the
same as those used in the previous section: the share of the total votes
received by the party in that constituency; whether the party held the
constituency in the previous election (i.e., it was the ‘incumbent’ party);
the turnout in that election; the number of independent candidates in the
election; the number of ‘other’ (i.e., other than the INC and the BJP)
party candidates in the election; the year of the election; and the state in
which the constituency was located.
The comparison between the electoral performance of the INC and the
BJP, in constituencies where both parties were contestants, was made—
separately for non-HS and HS major states—with respect to two parame-
ters: (i) the overall probability of winning the constituency with the values
of the conditioning variables taking their observed constituency values;
and (ii) the overall probability of winning the constituency when each
party obtained a particular vote share; 35, 40, and 45 percent.
Table 3.12 shows that in terms of the overall predicted probability of
winning a constituency—computed over all the eight elections between
1989 and 2014, with the conditioning variables taking their observed
constituency values—the electoral performances of the INC and the BJP,
in constituencies they both contested, differed according to whether
these constituencies were in non-HS or in HS states. The INC was much
stronger in the non-HS states—at 40.7 percent, its average probability of
winning in these states was significantly higher than the BJP’s 28.2 per-
cent—and the BJP, however, was much stronger in the HS states—at 55.5
percent, its average probability of winning in these states was significantly
higher than the INC’s 18.7 percent.
These probabilities of winning varied when they were computed on
an election-by-election basis. For example, the superior performance of
the INC in non-HS states withered in the 2014 election when there was
no significant difference between the INC and the BJP in their respective
probabilities of winning in the non-HS major states (26.2 percent versus
29.1 percent) but the superior performance of the BJP over the INC in
HS states was magnified (8.2 percent versus 74.3 percent). In the 1996

vk.borooah@ulster.ac.uk
72 V.K. Borooah

Table 3.12 INC and BJP likelihood of winning by year, non-Hindi and Hindi-­
speaking major states
1 2 3 4 5 6 7

INC BJP Difference Standard z-value for p-value

likelihood likelihood of error of testing H0:
of winning winning difference Difference = 0

All years
NHS 0.407 0.282 0.126 0.014 9.25 0.00
HS 0.187 0.555 −0.368 0.015 −24.50 0.00
1989
NHS 0.388 0.231 0.157 0.055 2.83 0.01
HS 0.104 0.630 −0.526 0.056 −9.34 0.00
1991
NHS 0.446 0.271 0.175 0.044 4.00 0.00
HS 0.177 0.496 −0.319 0.044 −7.26 0.00
1996
NHS 0.416 0.341 0.076 0.040 1.88 0.06
HS 0.251 0.585 −0.334 0.062 −5.41 0.00
1998
NHS 0.441 0.273 0.169 0.042 4.04 0.00
HS 0.204 0.561 −0.357 0.050 −7.14 0.00
1999
NHS 0.341 0.264 0.077 0.043 1.77 0.08
HS 0.145 0.539 −0.394 0.045 −8.84 0.00
2004
NHS 0.405 0.286 0.118 0.043 2.75 0.01
HS 0.207 0.498 −0.291 0.052 −5.65 0.00
2009
NHS 0.489 0.273 0.216 0.038 5.63 0.00
HS 0.294 0.479 −0.184 0.054 −3.42 0.00
2014
NHS 0.262 0.291 −0.029 0.053 −0.55 0.58
HS 0.082 0.743 −0.660 0.063 −10.53 0.00

Source: Own calculations from Lok Sabha election data

NHS is non-Hindi-speaking major states; HS is Hindi-speaking major states

(when the INC won 139 seats to the BJP’s 161) and 1999 (when the INC
won 114 seats to the BJP’s 182) elections, too, there was no significant
difference between the two parties in their respective likelihoods of win-
ning in non-HS states.11 In general, however, the pattern of the various
elections was that in constituencies contested by both parties, the average

vk.borooah@ulster.ac.uk
 Estimating the Likelihood of Winning Parliamentary... 73

Table 3.13 INC and BJP predicted likelihood of winning at different constitu-
ency vote shares: Hindi- and non-Hindi-speaking major states
1 2 3 4 5 6 7

Vote share INC BJP Difference Standard z-value for p-value

(%) likelihood of likelihood of error of testing H0:
winning winning difference Difference = 0

All years
NHS:35 0.187 0.165 0.022 0.024 0.90 0.37
HS:35 0.292 0.482 −0.190 0.034 −5.65 0.00
NHS:40 0.349 0.346 0.003 0.039 0.09 0.93
HS:40 0.458 0.682 −0.224 0.036 −6.26 0.00
NHS:45 0.578 0.579 −0.001 0.145 −0.01 1.00
HS:45 0.628 0.833 −0.205 0.031 −6.67 0.00
1989
NHS:35 0.169 0.078 0.091 0.060 1.53 0.13
HS:35 0.137 0.582 −0.445 0.095 −4.67 0.00
NHS:40 0.327 0.208 0.120 0.103 1.16 0.25
HS:40 0.273 0.768 −0.495 0.096 −5.14 0.00
NHS:45 0.561 0.429 0.132 0.134 0.99 0.32
HS:45 0.447 0.890 −0.444 0.091 −4.86 0.00
1991
NHS:35 0.226 0.134 0.092 0.062 1.48 0.14
HS:35 0.273 0.384 −0.111 0.071 −1.56 0.12
NHS:40 0.406 0.305 0.101 0.097 1.03 0.30
HS:40 0.449 0.594 −0.146 0.079 −1.86 0.06
NHS:45 0.642 0.544 0.098 0.137 0.72 0.47
HS:45 0.631 0.769 −0.138 0.069 −1.99 0.05
1996
NHS:35 0.196 0.286 −0.090 0.078 −1.15 0.25
HS:35 0.410 0.512 −0.102 0.106 −0.97 0.33
NHS:40 0.366 0.506 −0.141 0.113 −1.25 0.21
HS:40 0.596 0.712 −0.116 0.100 −1.15 0.25
NHS:45 0.602 0.721 −0.120 0.438 −0.27 0.79
HS:45 0.761 0.854 −0.093 0.078 −1.19 0.23
1998
NHS:35 0.221 0.137 0.084 0.056 1.51 0.13
HS:35 0.323 0.477 −0.154 0.087 −1.76 0.08
NHS:40 0.400 0.311 0.089 0.079 1.13 0.26
HS:40 0.506 0.681 −0.175 0.087 −2.02 0.04
NHS:45 0.637 0.550 0.087 0.122 0.71 0.48
HS:45 0.684 0.833 −0.149 0.070 −2.14 0.03
1999
(continued)

vk.borooah@ulster.ac.uk
74 V.K. Borooah

Table 3.13 (continued)

1 2 3 4 5 6 7

Vote share INC BJP Difference Standard z-value for p-value

(%) likelihood of likelihood of error of testing H0:
winning winning difference Difference = 0

NHS:35 0.128 0.123 0.005 0.053 0.10 0.92

HS:35 0.212 0.444 −0.232 0.075 −3.10 0.00
NHS:40 0.265 0.287 −0.022 0.083 −0.26 0.79
HS:40 0.375 0.652 −0.276 0.081 −3.41 0.00
NHS:45 0.488 0.524 −0.036 0.119 −0.30 0.76
HS:45 0.558 0.812 −0.254 0.072 −3.54 0.00
2004
NHS:35 0.184 0.162 0.022 0.058 0.39 0.70
HS:35 0.329 0.387 −0.058 0.085 −0.68 0.50
NHS:40 0.349 0.348 0.002 0.085 0.02 0.99
HS:40 0.512 0.598 −0.085 0.089 −0.96 0.34
NHS:45 0.585 0.587 −0.003 0.155 −0.02 0.99
HS:45 0.689 0.772 −0.082 0.075 −1.09 0.28
2009
NHS:35 0.275 0.138 0.137 0.055 2.48 0.01
HS:35 0.489 0.360 0.129 0.084 1.53 0.13
NHS:40 0.467 0.311 0.156 0.077 2.03 0.04
HS:40 0.671 0.570 0.100 0.082 1.22 0.22
NHS:45 0.699 0.550 0.149 0.120 1.24 0.21
HS:45 0.819 0.750 0.068 0.064 1.07 0.28
2014
NHS:35 0.074 0.171 −0.097 0.064 −1.51 0.13
HS:35 0.100 0.759 −0.660 0.100 −6.61 0.00
NHS:40 0.172 0.361 −0.189 0.100 −1.90 0.06
HS:40 0.216 0.890 −0.674 0.110 −6.15 0.00
NHS:45 0.362 0.600 −0.238 0.182 −1.31 0.19
HS:45 0.377 0.958 −0.581 0.124 −4.70 0.00

Source: Own calculations from Lok Sabha election data

NHS: X is non-Hindi-speaking at X percent vote share; HS: X is Hindi-speaking at X percent vote share

likelihood of the INC winning, compared to that for the BJP, was signifi-
cantly higher in non-HS states and significantly lower in HS states.
As the results in Table 3.13 show, the thrust of these results was not
altered when the likelihood of winning was computed at different vote
shares. For a 40 percent vote share in an HS state constituency (row
HS:40 in Table 3.13), the predicted probability of a BJP victory would

vk.borooah@ulster.ac.uk
 Estimating the Likelihood of Winning Parliamentary... 75

be 68.2 percent compared to the INC’s 45.8 percent; with the same vote
share in a non-HS state (row NHS:40 in Table 3.13), however, the INC
would win with probability 34.9 percent compared to the BJP’s 34.6 per-
cent, a difference which was not statistically significant. The pattern was
repeated under a hypothetical 45 percent share of the total vote: both
likelihoods of winning would rise further, but the BJP advantage in terms
of a significantly higher winning probability in HS states would remain
(83.3 percent versus 62.8 percent in row HS:45 of Table 3.13); in non-
­HS states, the difference between the parties in their respective likelihoods
of winning remained statistically insignificant (57.8 percent for the INC
versus 57.9 percent for the BJP in row NHS:45 of Table 3.13).

3.6   Concluding Remarks

This chapter represented the first step towards the overall purpose of this
book which is to evaluate the relative electoral efficiency of India’s two
major parties—the INC and the BJP. Whether one considered the mar-
ginal constituencies in which the INC was the winner or the runner-up
(and a parallel set of constituencies in which the BJP was the winner or the
runner-up), or whether one considered the set of all constituencies which
the INC and the BJP both contested, the answer always seemed to be the
same: the average predicted probability of the BJP winning a Lok Sabha
constituency election was, except for the 2009 Lok Sabha election, always
greater than that of the INC.
When attention was narrowed to constituencies in Hindi-speaking
states and those in non-Hindi-speaking states, the advantage of the BJP
over the INC in Hindi-speaking states (in terms of the average probability
of winning constituencies in these states) was statistically significant; on
the other hand, for constituencies in non-Hindi-speaking states, the dif-
ference between the INC and the BJP in their respective probabilities of
winning was not statistically significant.
It is important to emphasise that the results presented in this chapter
are based on average probabilities: that is, the average of the predicted
probabilities of winning individual constituencies. So, the results should
not be interpreted to mean that in every constituency, the probability of
a BJP win is greater than that of the INC. There will be constituencies
where the INC was predicted to have a better chance of winning than the
BJP, but averaging over these probabilities, the BJP was better placed to
win than the INC.

vk.borooah@ulster.ac.uk
76 V.K. Borooah

Notes
1. Note that ‘observations’ are distinguished by constituency name and by
year of election: for example, Adilabad in the 1989 and 1991 Lok Sabha
election is treated in the analysis as two distinct constituencies and, there-
fore, as two separate observations.
2. The focus was on constituencies in which the winning margin was 10
points or less in order to harvest the largest number of constituencies from
the data while remaining within the ambit of marginal constituencies.
3. See Long and Freese (2014), pp. 126–28 for a discussion of measures of fit
in binary models.
4. By implication, we do not consider marginal constituencies in which the
INC or the BJP was neither the winner nor the runner-up.
5. Over the elections from 1962 to 2014, there were 113 constituencies in
total in which there was only one party candidate, the rest being indepen-
dents. The most recent of such these was Kokrajhar (Assam) in 2004 when
the INC unsuccessfully fought the seat alongside three independents.
6. The effects of incumbency are analysed in detail in the next chapter. Here
incumbency is simply used as a variable determining electoral outcome.
7. This is also the figure reported in the last row, second column, of Table 3.5.
8. The difference between a logit and a probit model, both of which deal with
binary outcomes, is in the assumption made about the distribution of the
error term. In a logit model, the error term is assumed to be logistically
distribution while in a probit model, it is assumed to follow a normal
distribution.
9. Listed in Table 3.9 which mirrors the listing in Tables 3.3 and 3.4.
10. After the creation in November 2000 of the states of Uttarakhand and
Jharkhand from, respectively, the erstwhile states of Uttar Pradesh and
Bihar, the former lost five constituencies while the latter lost 14
constituencies.
11. Notwithstanding the fact that, compared to the BJP, the likelihood of win-
ning was greater for the INC.

References
Greene, W. H. (2003). Econometric analysis (5th ed.). Upper Saddle River:
Prentice Hall.
Long, J. S., & Freese, J. (2014). Regression models for categorical dependent vari-
ables using Stata. College Station: Stata Press.

vk.borooah@ulster.ac.uk
 Chapter 4

Incumbents, Challengers, and Electoral Risk

Abstract Borooah develops a methodology, based on Bayes’ theorem,

for evaluating the electoral risk associated with being the incumbent party,
as opposed to being a challenger party, in a constituency. His overall
conclusion is that there is no obvious way of measuring the degree of
anti-­incumbency, or its obverse, pro-incumbency. There are at least four
measures based on the likelihood of winning. Which measure is appropri-
ate depends on what one is trying to establish.

4.1   Introduction
A major issue in the study of elections is whether, and to what extent, the
chances of a candidate or a party being elected from a constituency are
improved or damaged by virtue of the fact that he/she/it is the incum-
bent in that constituency (i.e., had won the previous election from that
constituency). The literature on US elections suggests that incumbents
enjoy a considerable advantage over their challenger rivals: they are not
only much more likely to be re-elected, but also their margin of victory
has increased significantly over time (Alford and Hibbing 1981; Collie
1981; Garand and Gross 1984). By contrast, a recurring theme in the
literature on Lok Sabha elections in India since the 1990s is that of ‘anti-­
incumbency’: it is alleged that at every election since 1991, voters have
cut a swathe through incumbent members of parliament and chosen to
replace many of them with a fresh set of faces.

© The Editor(s) (if applicable) and The Author(s) 2016 77

V.K. Borooah, Votes, Parties, and Seats,
DOI 10.1007/978-3-319-30487-8_4

vk.borooah@ulster.ac.uk
78 V.K. Borooah

The ‘anti-incumbency’ sentiment of Indian voters in a particular con-

stituency may be underpinned by any one of four ‘grievances’. At its
broadest, it may represent a vote against the ruling party at the centre
(‘national government incumbency’). More narrowly, but still within the
purview of a ruling party, it may represent a vote against the party of
government in the state in which the constituency is based (‘state gov-
ernment incumbency’). Alternatively, it may represent a vote against the
party which won the constituency in the previous election, regardless of
whether that party is part of the government at the centre or in the state
(‘party incumbency’). Finally, anti-incumbency might focus on the candi-
date rather than the party and represent a vote against the sitting member
of parliament (‘candidate incumbency’).
In this chapter—and, indeed, in this book—incumbency is defined in
terms of the party which won the constituency in the previous election
(‘party incumbency’), and an anti-incumbent vote is, therefore, a vote
against the incumbent party. Consequently, issues relating to ‘govern-
ment incumbency’ (Yadav 2004) or ‘candidate incumbency’ (see Linden
2003) are not addressed. Within the context of party incumbency, this
chapter draws on Bayes’ Theorem to make more precise the concept of
‘anti-incumbency’ and then, based on this concept, measures the extent
of anti-incumbency towards the Indian National Congress (INC) and the
Bharatiya Janata Party (BJP).

4.2   Bayes’ Theorem and the ‘Incumbency Effect’

The Reverend Thomas Bayes, an eighteenth century Presbyterian minister,
proved what, arguably, is the most important theorem in statistics.1 Bayes’
Theorem states that the probability of a hypothesis being true (event T),
given that the data has been observed (event A), is the probability of the
hypothesis being true, before any data has been observed, times an ‘updat-
ing factor’. The theorem is encapsulated by the well-known equation:

P(A | T )
P (T | A ) = × P (T ) (4.1)
P ( A)

where: P(T)represents the prior belief that the hypothesis is true before
the data has been observed; P(A) is the probability of observing the data,
regardless of whether the hypothesis is true or not; P(A|T) is the probability of

vk.borooah@ulster.ac.uk
 Incumbents, Challengers, and Electoral Risk  79

observing the data, given that the hypothesis is true, and P ( A | T ) / P ( A)

is the Bayesian ‘updating factor’ which translates one’s prior (i.e., before
observing the data) belief about the hypothesis’s validity into a posterior
(i.e., after observing the data) belief.2
In this chapter, we use Bayes’ ideas to analyse the question of whether
incumbents are more or less likely to win elections than challengers. As in
the preceding paragraph, let A and Ā denote the ‘data’ which, in this case,
is: (i) the party is the incumbent in that constituency, eventA; and (ii) the
party is a challenger (i.e., not the incumbent) in that constituency, event Ā.
Similarly, let T and T denote the ‘hypothesis’ which, in this case, is: (i)
the party wins the election to that constituency, event T; and (ii) the party
loses the election to that constituency, event T . Then P(T) is the prob-
ability of the party winning the election for that constituency in the absence
of any information about whether it is the incumbent or challenger party
there. The probability that the party wins the election for the constituency,
given that it is the incumbent party in that constituency is P (T | A) , and this
can be obtained by applying Bayes’ theorem as in Eq. 4.1. Similarly, the
probability that the party wins the election in the constituency, given that
it is a challenger party in that constituency, is P (T | A ) , and this can also be
obtained from Bayes’ theorem by replacing A with Āin Eq. 4.1.

4.2.1  The ‘Bayes Factor’ and the ‘Inverse Bayes Factor’

One definition of the risk, associated with being the incumbent, is the
ratio of the likelihood that the incumbent party wins an election to the
likelihood that it loses it. This ratio is, hereafter, referred to as the risk ratio
(RR) and is denoted by ρ, where:

P (T | A ) P ( A | T ) P (T ) P ( A | T ) P (T ) P (T )
ρ= = × = × =Φ (4.2)
P (T | A ) P ( A | T ) P (T ) P ( A | T ) 1 − P (T ) 1 − P (T )

P(A |T ) ρ P (T )
where: Φ = , where λ =
= is (OR) that the ratio of the
P(A |T ) λ 1 − P (T )
likelihood of winning, to the likelihood of losing, the election.
The term Φ in Eq. 4.2 is the so-called Bayes Factor (BF) applied to
incumbent parties. The BF is a measure of whether the data (A: the party
is the incumbent) is more likely to be observed under one outcome (T:

vk.borooah@ulster.ac.uk
80 V.K. Borooah

the party wins) than under the alternative outcome ( T : the party loses):
Φ > 1 ( < 1) signifies that the likelihood of being an incumbent is higher
(lower) when the party wins compared to when the party loses. It tells us
by how much we should alter our prior belief that the party will win with
probability, P(T), and lose with probability, P (T ) = 1 − P (T ) , in the light
of the data that the party is an incumbent.3

4.2.2  The Inverse Bayes Factor

The RR, ρ in Eq. 4.1, measures the odds of the null hypothesis being
‘true’ (the party wins the election from a constituency) to it being ‘false’
(the party loses the election from that constituency) under a particular
set of data which, in this case, is that the party is the incumbent party in
the constituency. In this formulation of risk, the data applicable to the
different outcomes (winning or losing the election) was the same (the
party was the incumbent). An alternative view of risk is obtained by pos-
ing the following question: given two rival scenarios: in the first, a party
is an incumbent in an election to a constituency, while, in the second, it
is a challenger—what is the ratio of its probabilities of winning in these
different situations?
In this case, the RR of being the incumbent party is the ratio of the
likelihood that the party wins the election if it was the incumbent to the
likelihood that the party wins the election if it was a challenger. Here the
outcome is the same (the party wins the election) but the data that is
input is different (incumbent or challenger). In order to answer this ques-
P (T | A )
tion, the relevant RR (represented by σ) is σ = . Hereafter, σ is
P (T | A )
referred to as the inverse risk ratio (IRR): given two different ‘pieces’ of
information—a party is an incumbent or a challenger—what is the ratio of
the party’s probabilities of winning the election?
In turn, one can expand σ so that:

P (T | A ) P ( A | T ) P (T ) P ( A)
σ= = ×
P (T | A ) P ( A) P ( A | T ) P (T )
P ( A | T ) P ( A) P ( A)
= × =Ψ (4.3)
P ( A | T ) P ( A) P ( A)

vk.borooah@ulster.ac.uk
 Incumbents, Challengers, and Electoral Risk  81

P(A |T ) σ P ( A)
where: Ψ = = where µ = is the inverse odds ratio (IOR):
P(A |T ) µ P ( A)
the ratio of the likelihood of contesting a seat as a challenger party to
that of contesting it as the incumbent party. The term Ψ in Eq. 4.3 is the
inverse Bayes Factor (IBF) applied to the party that won that constituency.
The IBF is the odds of the null hypothesis being true (the party wins)
under one set of data (the party was the incumbent), against it being
true (the party wins) under the obverse set of data (the party was a chal-
lenger). If Ψ > 1 ( < 1) then, given that the hypothesis is true (the party
wins), we are more (less) likely to observe one set data (A: the party is the
incumbent party) than the complementary set of data (A: the party is a
challenger).

Table 4.1 Winning and incumbency outcomes for the INC and the BJP:
1962–2014
INC BJP

Number of Number of Number of Number of

seats won in incumbent seats in seats won in incumbent seats in
year t year t + 1 year t year t + 1

1962:1967 357 305

1967:1971 281 279
1971:1977 352 332
1977:1980 154 146
1980:1984 353 360
1984:1989 414 410 2 2
1989:1991 197 199 85 85
1991:1996 244 246 121 121
1996:1998 139 139 161 161
1998:1999 141 141 182 182
1999:2004 114 105 182 157
2004:2009 145 120 138 103
2009:2014 206 206 116 116
2014 44 282

Source: Own calculations from Lok Sabha election data

vk.borooah@ulster.ac.uk
82 V.K. Borooah

4.3   Risk Ratio and Bayes Factor Calculations

for Lok Sabha Elections

Table 4.1 shows the winning and incumbency outcomes for seats con-
tested by the INC and the BJP in Lok Sabha elections. The INC results
pertain to the 14 successive Lok Sabha elections in India from 1962 (3rd
Lok Sabha) to 2014 (16th Lok Sabha); since the BJP only made its elec-
toral debut in the 1984 Lok Sabha election, its results pertain to the nine
Lok Sabha elections between 1984 and 2014.
If there were no constituency changes between elections, then the
number of seats won by a party (say, the INC) in one election should
be the number of seats in which it was the incumbent in the subsequent
election. However, boundary changes mean that constituencies disap-
pear between elections and, sometimes, even reappear. A case in point
is the number of changes that occurred between the 2004 and 2009
Lok Sabha elections. The INC won 145 Lok Sabha seats in the General
Election of 2004 but in the 2009 election, it was the incumbent in only

Table 4.2 Incumbency/non-incumbencya performance of the INC in Lok Sabha

elections
Contested Winning Losing Winning Losing Electoral
seats incumbent incumbent challenger challenger turnover
seats seats seats seats (%)

1967 514 176 129 105 104 46

1971 441 223 47 129 42 40
1977 492 101 224 53 114 56
1980 492 121 13 232 126 50
1984 517 297 59 117 44 34
1989 510 148 260 49 53 61
1991 504 148 50 96 210 29
1996 526 92 151 47 236 38
1998 477 68 70 73 266 30
1999 453 52 88 62 251 33
2004 417 45 57 100 215 38
2009 440 70 46 136 188 41
2014 464 34 170 10 250 39

Source: Own calculations from Lok Sabha election data

Electoral turnover: percentage of seats contested by INC which changed hands between INC and other
parties
a
Incumbency refers to seats held by the INC

vk.borooah@ulster.ac.uk
 Incumbents, Challengers, and Electoral Risk  83

Table 4.3 Incumbency/non-incumbencya performance of the BJP in Lok Sabha

elections
Contested Winning Losing Winning Losing Electoral
seats incumbent incumbent challenger challenger turnover
seats seats seats seats (%)

1989 225 1 1 84 139 38

1991 478 45 39 76 318 24
1996 469 86 34 75 274 23
1998 388 108 53 74 153 33
1999 339 115 65 67 92 39
2004 364 79 76 59 150 37
2009 433 52 50 64 267 26
2014 428 103 13 179 133 45

Source: Own calculations from Lok Sabha election data

Electoral turnover: percentage of seats contested by BJP which changed hands between BJP and other
parties
a
Incumbency refers to seats held by the BJP

119 ­constituencies. Similarly, the BJP won 138 Lok Sabha seats in the
General Election of 2004, but, in the 2009 election, it was the incumbent
in only 103 constituencies.4
Tables 4.2 and 4.3 show, respectively, the losses and gains by the INC
and the BJP depending upon whether they were the incumbent or a chal-
lenger party. So, when the 2009 elections were announced, the INC, as
Table 4.2 shows, was the incumbent in 120 constituencies. However, in the
2009 elections, it decided to contest only 116 of its ‘incumbent’ constitu-
encies, and of these, it won 70 and lost 45.5 In the constituencies where
it was not the incumbent party, it won 136 and lost 189. Consequently, a
total of 181 seats changed hands between the INC and the other parties
(45 INC incumbents lost and 136 INC challengers won)6 which repre-
sented an ‘electoral turnover’ for the INC of 41 percent of the 440 seats
it contested in 2009.
Similarly, as Table 4.3 shows, in 2009, the BJP, as the incumbent party,
won and lost, respectively, 52 and 50 seats while, as a challenger party, it
won and lost, respectively, 64 and 267 seats. As a consequence of this, a
total of 114 seats changed hands in 2009 between the BJP and the other
parties7 (50 BJP incumbents lost and 64 BJP challengers won) which was
an ‘electoral turnover’ for the BJP of 26 percent of the 433 seats it con-
tested in 2009.

vk.borooah@ulster.ac.uk
84 V.K. Borooah

A high volume of trade between a party and other parties suggests

either, or both, of two things: (i) a soft ‘centre’, so that core voters exit
easily; and (ii) a strong ‘periphery’, so that non-traditional voters enter
easily. In ‘trading’ between itself and other parties, a party can either have
an ‘electoral deficit’: out-migration exceeds in-migration (the number of
losing incumbents exceeds the number of winning challengers); or it can
have an ‘electoral surplus’: in-migration exceeds out-migration (the num-
ber of winning challengers exceeds the number of losing incumbents).
Figure 4.1 shows the net migration of seats (i.e., exits less entries) for
the two parties for parliamentary elections between 1989 and 2014. This
shows that it was only in the 2004 and 2009 elections that there was a
net inflow of seats into the INC. On the other hand, except for the 2004
election—when there was a small net outflow from the BJP—the BJP has
always been able to attract a net inflow of seats. This is evidence to suggest
that since 1989, INC fortunes have on a downward trend, and this has
been mirrored in an upward trend in the fortunes of the BJP.

Fig. 4.1 Net migration of seats (losses less gains) from the INC and the BJP
positive values represent a net outflow; negative values a net inflow (Source: Own
calculations from Lok Sabha election data)

vk.borooah@ulster.ac.uk
 Incumbents, Challengers, and Electoral Risk  85

Table 4.4 RR and BF calculations for the INC in Lok Sabha elections
Number of Number of RRa Number Number ORb BF c (Φ)
seats won by seats lost by (ρ) of seats seats lost (λ)
incumbent incumbent won

1967 176 129 1.36 281 233 1.21 1.12

1971 223 47 4.74 352 89 3.96 1.20
1977 101 224 0.45 154 338 0.46 0.98
1980 121 13 9.31 353 139 2.54 3.67
1984 297 59 5.03 414 103 4.02 1.25
1989 148 260 0.57 197 313 0.63 0.90
1991 148 50 2.96 244 260 0.94 3.15
1996 92 151 1.80 139 387 0.36 5
1998 68 70 0.97 141 336 0.42 2.31
1999 52 88 0.59 114 339 0.34 1.74
2004 45 57 0.79 145 272 0.54 1.46
2009 70 46 1.56 206 234 0.88 1.77
2014 34 170 0.2 44 420 0.10 2

Source: Own calculations from Lok Sabha election data

Incumbency refers to seats held by the INC
RR = the ratio of the number of seats won by incumbents to seats lost by incumbents (see Eq. (3.2))
a

b
OR = Number of seats won to number of seats lost, by the INC
BF = RR/OR
c

Table 4.4 shows that the INC’s RR (ρINC)—defined, as

ρ = P ( T | A ) / P ( T | A ) in Eq. 4.2—as the ratio of the number of seats
contested by the INC incumbents that were won and lost—for all the Lok
Sabha elections from 1967 to 2014; Table 4.5 does the same for the BJP’s
RR (ρBJP) for all the Lok Sabha elections from 1989 to 2014.8 For four of
the five elections between 1967 and 1984, the RR for the INC was greater
than unity (meaning that the chance of the INC winning a seat in which
it was an incumbent was greater than that of losing it) but in the four of
the five elections held after 1996, the RR for the INC was less than unity
(meaning that the chance of the INC winning a seat in which it was an
incumbent was less than that of losing it): in the INC massacre of 2014,
the chances of incumbents winning their seats were just 20 percent of their
chances of losing them.
For the BJP, on the other hand, the RR was always positive. Even in the
difficult elections of 2004 and 2009, both of which led to INC-led coali-

vk.borooah@ulster.ac.uk
86 V.K. Borooah

Table 4.5 RR and BF calculations for the BJP in Lok Sabha elections
Number of Number of RRa Number Number ORb BF c (Φ)
seats won by seats lost by (ρ) of seats seats lost (λ)
incumbent incumbent won

1989 1 1 1 85 140 0.61 1.64

1991 45 39 1.15 121 357 0.34 3.38
1996 86 34 2.53 161 308 0.52 4.87
1998 108 53 2.04 182 206 0.88 2.32
1999 115 65 1.77 182 157 1.16 1.53
2004 79 76 1.04 138 226 0.61 1.70
2009 52 50 1.04 116 317 0.37 2.81
2014 103 13 7.92 282 146 1.93 4.10

Source: Own calculations from Lok Sabha election data

Incumbency refers to seats held by the BJP
RR = the ratio of the number of seats won by incumbents to seats lost by incumbents (see Eq. (3.2))
a

b
OR = Number of seats won to number of seats lost, by the BJP
BF = RR/OR
c

Fig. 4.2 The RR for the INC and the BJP compared (Source: Own calculations
from Lok Sabha election data)

vk.borooah@ulster.ac.uk
 Incumbents, Challengers, and Electoral Risk  87

tion governments, the RR for the BJP was slightly over unity, meaning
that the chance of the BJP winning a seat in which it was an incumbent
was just greater than that of losing it. In other elections, the RR for the
BJP was comfortably over unity, and most spectacularly, the BJP’s RR in
the 2014 election was 7.92: the chances of the BJP winning an incumbent
seat were eight times than that of losing it.9 Figure 4.2 compares the RRs
for the INC and the BJP from the 1991 election onwards.
The OR, λ, is the ratio of the total number of seats won, to the total
number of seats lost, by the INC and is the empirical equivalent of the
term P (T ) / P (T ) in Eq. 4.2. Figure 4.3 compares the ORs for the INC
and the BJP from the 1991 election onwards with the lowest and high-
est ORs being recorded for the 2014 elections: in this election, the INC
and the BJP won 0.1 and 1.9 seats, respectively, for every seat that they
lost.
The RR (ρ) when compared to the OR (λ) yields the BF defined as the
term Φ = P ( A | T ) / P ( A | T ) in Eq. 4.2.10 If the RR is greater than the OR
ρ
(BF = > 1), it means that in the light of the information that the party is
λ
an incumbent, we should revise upwards—by the amount suggested by the
BF—our prior belief that the party will win with probability, P(T), and lose

Fig. 4.3 The OR for the INC and the BJP compared (Source: Own calculations
from Lok Sabha election data

vk.borooah@ulster.ac.uk
88 V.K. Borooah

with probability, P (T ) = 1 − P (T ) . Conversely, if the RR is less than the OR

ρ
(BF = < 1), it means that in the light of the information that the party is
λ
an incumbent we should revise downwards—by the amount suggested by
the BF—our prior belief that the party will win with probability,P(T), and
lose with probability, P (T ) = 1 − P (T ) .11
Table 4.4 shows that except for 1977 and 1989, the RR was always
greater than the OR for the INC (meaning that the BF value, entered in the
last column of Table 4.4, was greater than one): even in the 1996, 1999, and
2014 elections, when it was very ‘risky’ standing as an INC incumbent,12 it
was not as risky as standing as an INC challenger. Consequently, in 1999,
the likelihood of an INC win being an incumbent victory was almost twice
as likely (RR/OR = 1.91) as an INC loss being an incumbent defeat. Only in
the 1977 and 1989 elections, both of which were characterised by a strong
anti-INC sentiment, was it more risky being an INC incumbent compared
to being an INC challenger: BF = (risk/OR) < 1 implied that the likelihood
of an INC loss being an incumbent defeat was greater (by 2 percent in

Fig. 4.4 BF values for the INC and the BJP compared (Source: Own calculations
from Lok Sabha election data)

vk.borooah@ulster.ac.uk
 Incumbents, Challengers, and Electoral Risk  89

1977 and 10 percent in 1989) than the likelihood of an INC win being an
incumbent victory.
For the BJP, too, the RR was always greater than the OR (meaning that
the BF value entered in the last column of Table 4.5 was greater than one).
In the 2014 election, the likelihood of a BJP win being an incumbent vic-
tory was more than four as likely (RR/OR = 4.1) as of a BJP loss being
an incumbent defeat. However, as Fig. 4.4 shows, the BF was generally
higher for the BJP than for the INC. For both parties, a win was more
likely to signal an incumbent victory than a defeat was to signal an incum-
bent loss, but this gap was larger for the BJP than the INC.

4.4   The Inverse Risk Ratio and the Inverse Bayes

Factor Calculations for Lok Sabha Elections
The proportion of incumbent seat wins to the proportion of incumbent
seat losses (from the seats contested by, respectively, the INC and BJP
as incumbent parties) were compared in Tables 4.4 and 4.5 to yield the
RR. This then led, through a comparison of the RR with the OR, to
the BF.
Tables 4.6 and 4.7 compare, for respectively, the INC and the BJP,
the ratio of the proportion of incumbent wins (from the seats contested
by, respectively, the INC and BJP as incumbent parties) to the propor-
tion of challenger wins (from the seats contested by, respectively, the INC
and BJP as challenger parties). This ratio is the IRR defined by the term
σ = P ( T | A ) / P ( T | A ) in Eq. 4.3. The IOR—the term µ = P ( A ) / P ( A )
in Eq. 4.3—represents the ratio of the likelihood of contesting as a
challenger to the likelihood of contesting as an incumbent. A compari-
son of the IRR with the IOR then results in the IBF. This is the term
Ψ = P ( A | T ) / P ( A | T ) in Eq. 4.3. If Ψ > 1 ( < 1) , the probability of win-
ning as an incumbent is greater (less) than the probability of winning as
a challenger.
Table 4.6 shows that except for the elections of 1977 and 1989, the
IRR was always greater than 1, meaning that the INC had a greater chance
of winning from where it was the incumbent than from where it was the
challenger: indeed, since 1991, the chances of winning as an incumbent
have been more than twice that of winning as a challenger. Table 4.7
tells a similar story with respect to the BJP’s IRR: the likelihood of an

vk.borooah@ulster.ac.uk
90 V.K. Borooah

Table 4.6 IRR and IBF calculations for the INC in Lok Sabha elections
Proportion Proportion IRRa Number Number of IORb IBF c
of seats of seats (σ) of incumbent (μ) (Ψ)
contested by contested by challenger seats
INC INC seats contested
incumbents challengers contested
that were that were
won by them won by
(%) them (%)

1967 57.7 50.2 1.15 209 305 0.69 1.67

1971 82.6 75.4 1.10 171 270 0.63 1.75
1977 31.1 31.7 0.98 167 325 0.51 1.92
1980 90.3 64.8 1.39 358 134 2.67 0.52
1984 83.4 72.7 1.15 161 356 0.45 2.56
1989 36.3 48.0 0.76 102 408 0.25 3.04
1991 74.8 31.4 2.38 306 198 1.55 1.54
1996 37.9 16.6 2.28 283 243 1.16 1.97
1998 49.3 21.5 2.29 339 138 2.46 0.93
1999 37.1 19.8 1.87 313 140 2.23 0.84
2004 44.1 31.8 1.39 315 102 3.09 0.45
2009 60.3 42.0 1.44 324 116 2.79 0.52
2014 16.7 3.9 4.28 260 204 1.27 3.37

Source: Own calculations from Lok Sabha election data

Incumbency refers to seats held by the INC
a
IRR = the ratio of the proportion of contested seats won by incumbents to the proportion of contested
seats won by challengers (see Eq. (3.3))
b
IOR = Number of challenger seats contested to number of incumbent seats contested by the INC
IBF = IRR/IOR
c

incumbency win was always greater than that of a non-­incumbency win.

Figure 4.5 brings together the values of the IRR for the INC and BJP
from the 1991 election onwards.
The values of the IOR, μ, shown in Tables 4.6 and 4.7 are the ratios of
the total number of seats which the parties—respectively, INC and BJP—
contested as challengers to the total number of seats they contested as
incumbents. Since 1991, there has been a sharp reduction in the number
of seats won by the INC in Lok Sabha elections notwithstanding the fact

vk.borooah@ulster.ac.uk
 Incumbents, Challengers, and Electoral Risk  91

Table 4.7 IRR and IBF calculations for the BJP in Lok Sabha elections
Proportion Proportion IRRa Number Number of IORb IBF c
of seats of seats (σ) of incumbent (μ) (Ψ)
contested by contested by challenger seats
INC INC seats contested
incumbents challenger contested
that were that were
won by them won by
(%) them (%)

1989 50 37.7 1.33 223 2 111.5 0.01

1991 53.6 19.3 2.78 394 84 4.69 0.59
1996 71.7 21.5 3.33 349 120 2.91 1.14
1998 67.1 32.6 2.06 227 161 1.41 1.46
1999 63.9 42.1 1.52 159 180 0.88 1.73
2004 51.0 28.2 2.64 209 155 1.35 1.96
2009 51.0 19.3 2.64 331 102 3.25 0.81
2014 88.8 57.4 1.55 312 116 2.69 0.58

Source: Own calculations from Lok Sabha election data

Incumbency refers to seats held by the BJP
a
IRR = the ratio of the proportion of contested seats won by incumbents to the proportion of contested
seats won by challengers (see Eq. (3.3))
b
IOR = Number of challenger seats contested to number of incumbent seats contested by the BJP
IBF = IRR/IOR
c

that the number of constituencies contested by the INC has not fallen
commensurately.
Consequently, post-1991, the INC emerges as a challenger party in the
majority of the seats contested by it, and in 2004, it contested three times
as many constituencies where it was a challenger compared to where it was
the incumbent. For the BJP, the three elections of 1996, 1998, and 1999
were ‘good’ elections when it won, respectively, 161, 182, and 182 seats,
and consequently, it built up a stock of seats in which it was the incumbent
party. This stock, combined with the fact that it contested far fewer seats
than the INC (in 1999 the BJP contested only 339 constituencies com-
pared to the INC’s 453), meant that it had a smaller ratio of challenger to
incumbent seats. Figure 4.6 brings together the values of the IBF for the
INC and BJP from the 1991 election onwards.

vk.borooah@ulster.ac.uk
92 V.K. Borooah

Fig. 4.5 The IRR for the INC and the BJP compared (Source: Own calculations
from Lok Sabha election data)

Fig. 4.6 The IOR for the INC and the BJP compared (Source: Own calculations
from Lok Sabha election data)

vk.borooah@ulster.ac.uk
 Incumbents, Challengers, and Electoral Risk  93

Fig. 4.7 IBF values for the INC and the BJP compared (Source: Own calcula-
tions from Lok Sabha election data)

When the IRR was greater than the IOR ( σ / µ > 1 ), the
chance of a party win being an incumbent victory was greater than
the chance of it being a challenger victory, and this is reflected
in the fact that the IBF, Ψ > 1, implying P ( A | T ) > P ( A | T ) .
When the IRR was less than the IOR ( σ / µ < 1 ), the chance of a party
win being a challenger victory was greater than the chance of it being
an incumbent victory, and this is reflected in the fact that the IBF, Ψ < 1,
implying P ( A | T ) < P ( A | T ) .13 Figure 4.7 brings together the values of
the IBF for the INC and BJP from the 1991 election onwards.
Figure 4.7 shows that the INC’s IBF value was 3.4 for the Lok Sabha
election of 2014. Even though the INC only won 44 seats in this elec-
tion, its constituency victories, as and when they did occur, were 3.4 times
more likely to have been as the incumbent, than as a challenger, party. On
the other hand, the BJP which went into the 2014 election with only 116
incumbent constituencies but ended up winning 282 seats. In the event
of a BJP’s victory in this election, the likelihood that the party was the
incumbent in a constituency was only 60 percent of the likelihood that it
was a challenger.

vk.borooah@ulster.ac.uk
94 V.K. Borooah

4.5   Concluding Remarks

This chapter’s contribution lay in developing a methodology, based on
Bayes’ theorem, for evaluating the electoral risk associated with being the
incumbent party, as opposed to being a challenger party, in a constituency.
The first concept was that of ‘the risk ratio’—the likelihood of a party
winning, compared to the likelihood of a party losing, a constituency as its
incumbent party. On this measure, for the five Lok Sabha elections after
1996 of 1998, 199, 2004, 2009, and 2009, the likelihood of the INC
losing an incumbent seat was larger than its likelihood of winning it; on
the other hand, for the same elections, the likelihood of the BJP losing an
incumbent seat was smaller than its likelihood of winning it. So, on this
measure—for the five Lok Sabha elections of 1998, 1999, 2004, 2009,
and 2014—there was an anti-incumbency effect for the INC but a pro-­
incumbency effect for the BJP.
The second concept was that of the BF. If BF > 1, the party was more
likely to have been the incumbent in a constituency if it won from that
constituency compared to losing from it. The fact that BF > 1 for the INC
tells us that while it was ‘risky’ for the INC to contest an election as the
incumbent—in the sense that the probability of winning was greater than
that of losing—it was not as ‘risky’ as the INC contesting the election as
the challenger. So, on this interpretation, there was a pro-incumbency
effect for the INC. For the BJP, too, BF > 1 and its value were larger for
the INC. For both parties, a win was more likely to signal an incumbent
victory than a defeat was to signal an incumbent loss, but this gap was
larger for the BJP than the INC.
The third concept was that of the IRR: the likelihood of a party win-
ning a constituency as the incumbent compared to the likelihood of a
party winning a constituency as the challenger. Both INC and the BJP
had a greater chance of winning as the incumbent party compared to
winning as the challenger party (the IRR was greater than 1) so that
on this interpretation, there was a pro-incumbency effect towards both
parties.
The fourth concept was that of the IBF: when IBF > 1, a party win
was more likely to be as the incumbent than as a challenger; conversely,
when IBF < 1, a party win was more likely to be as a challenger than as the
incumbent. On this interpretation, there was a pro-incumbent effect for
the INC in the Lok Sabha election of 2014 (in the event of an INC’s vic-

vk.borooah@ulster.ac.uk
 Incumbents, Challengers, and Electoral Risk  95

tory in this election, the likelihood that the party was the incumbent in a
constituency was 3.4 times the likelihood that it was a challenger) but an
anti-incumbency effect for the BJP (in the event of a BJP’s victory in this
election, the likelihood that the party was the incumbent in a constituency
was only 60 percent of the likelihood that it was a challenger).
The overall conclusion of this chapter is that there is no obvious way of
measuring the degree of anti-incumbency, or its obverse, pro-­incumbency.
There are at least four measures based on the likelihood of winning.
Which measure is appropriate depends on what one is trying to establish.
As Huckleberry Finn advised (in Chap. 28 of Mark Twain’s eponymous
novel): ‘you pays your money and you takes your choice’.

Notes
1. See “In Praise of Bayes”, The Economist, 28 September 2000.
2. The updating factor is the ratio of the probability of observing the data
when the theory is true, to that of observing the data regardless of whether
the theory is true or false: P ( A) = P ( A | T ) P (T ) + P A | T P T ,( ) ( )
T being the event that the theory is false.
3. See Matthews (2000).
4. For example, in Delhi: Sadar, Outer Delhi, and Karol Bagh which were
2004 Lok Sabha constituencies disappeared in 2009.
5. The four constituencies in 2009 which the INC did not contest, even
though it was the incumbent party in these, were: Bombay North East,
Hatkanangale, Namakkal, and Nilgiris.
6. ‘Parties’ include independent candidates.
7. ‘Parties’ include independent candidates.
8. Equation (4.2) is defined in terms of the proportion of contested incum-
bent seats won to the proportion of contested incumbent seats lost, but
NWincum / NWincum + N Lincum NWincum
since the denominators are equal, ρ = =
N Lincum / NWincum + N Lincum N Lincum
where NWincum and NLincum are the number of seats won and lost by a party
as an incumbent.
9. In the 2014 election, the BJP contested 116 seats in which it was the
incumbent party and won 103 of them.
P (T | A ) P ( A | T ) P (T ) P(A |T ) ρ
10. Proof: ρ = = ⇒ = =Φ
P (T | A ) P ( A | T ) P (T ) P(A |T ) λ

vk.borooah@ulster.ac.uk
96 V.K. Borooah

11. In other words, if BF > 1, it means that a party is more likely to have been
the incumbent in a constituency if it won, than if it lost, from
( )
there: P ( A | T ) > P A | T . Conversely, if BF < 1, it means that a party is
more likely to have been the incumbent in a constituency if it lost, than if
(
it won, from there: P ( A | T ) < P A | T . )
12. In 1999, for example, INC incumbents lost 88 of the 140 seats they
contested.
P (T | A ) P ( A | T ) P ( A) P(A |T ) σ
13. Proof: σ = = ⇒ = =Ψ
P (T | A ) P ( A | T ) P ( A) P(A |T ) µ

References
Alford, J., & Hibbing, J. R. (1981). Increasing incumbency advantage in the
house. The Journal of Politics, 43, 1042–1061.
Collie, M. (1981). Incumbency, electoral safety, and turnover in the house of rep-
resentatives, 1952–1976. The American Political Science Review, 75, 119–131.
Garand, J. C., & Gross, D. A. (1984). Changes in the vote margins for congres-
sional candidates: A specification of historical trends. The American Political
Science Review, 78, 17–30.
Linden, L. L. (2003). Are candidates really advantaged? The preference for non-­
incumbents in Indian elections. Massachusetts Institute of Technology (mimeo),
Cambridge (Mass).
Matthews, R. J. (2000). Facts versus factions: The use and abuse of subjectivity in
scientific research. In J. Morris (Ed.), Rethinking risk and the precautionary
principle (pp. 247–282). Oxford: Butterworth Heniemann.
Yadav, Y. (2004). The elusive mandate of 2004. Economic and Political Weekly, 39,
5383–5398.

vk.borooah@ulster.ac.uk
 CHAPTER 5

Analysis of Vote Shares

Abstract Borooah examines the effects of incumbency on vote share: are

incumbent parties in a constituency, on average, more likely to get a larger
vote share than parties that are challenging? Aggregated over all eight
elections between 1989 and 2014, the average predicted vote share for
INC incumbents was higher than that for INC challengers and, similarly,
the average predicted vote share for BJP incumbents was higher than that
for BJP challengers. However, compared to the average predicted vote
shares for the BJP when it was the incumbent party, the INC did not do
as well as the incumbent party but the INC did better than the BJP when
both were challenger parties.

5.1 INTRODUCTION
The analysis in the previous chapter focused on the effects of being an
incumbent or a challenger on the probability of winning. An alternative
mode of analysis would be to examine the effects of incumbency on vote
share: regardless of whether they win or lose, are incumbent parties in a
constituency, on average, more likely to get a larger vote share than parties
that are challenging? In order to examine this hypothesis, we estimated,
using constituency data, two equations: the first had as its dependent vari-
able the vote share of the INC in a constituency (i.e., the votes received by
the INC in the constituency as a percentage of the total votes cast in that

© The Editor(s) (if applicable) and The Author(s) 2016 97

V.K. Borooah, Votes, Parties, and Seats,
DOI 10.1007/978-3-319-30487-8_5

vk.borooah@ulster.ac.uk
98 V.K. BOROOAH

constituency) and the second had as its dependent variable the vote share
of the BJP in that constituency.

5.2 ESTIMATING VOTE SHARES

The equations were estimated as a system of equations, using the method
of Seemingly Unrelated Regression Equations (SURE) (Greene 2003,
Chap. 14). SURE estimates are more efficient than those obtained from
estimating each equation as a single regression equation because the
correlation between the error terms of the two equations is explicitly
taken into account. In addition, and more importantly for the purpose of
this analysis, the fact that the equations are estimated as a system allows
hypotheses to be tested between equations as well as within individual
equations.
The conditioning variables in the two equations were as follows:

1. whether the party was the winner in that constituency,

2. whether the party held the constituency in the previous election (i.e., it
was the ‘incumbent’ party),
3. the percentage of the electorate voting in that election (‘turnout’),
4. the number of independent candidates in the election,
5. the number of ‘other’ (i.e., other than the INC and the BJP) party
candidates in the election,
6. the year of the election, and
7. the state in which the constituency was located.

The coefficient on the ‘incumbent’ variable (item 2, above) was allowed

to vary according to the year of the election: this meant that the ‘incum-
bent coefficient’ would be different according to the election that was
being considered. The equations were estimated on data for the 2684
constituencies contested by both the INC and the BJP in the eight Lok
Sabha elections: 1989, 1991, 1996, 1998, 1999, 2004, 2009, and 2014.
Table 5.1 shows the results from estimating the (two equation) SURE
model with the equation statistics shown in Table 5.2.

vk.borooah@ulster.ac.uk
 ANALYSIS OF VOTE SHARES 99

Table 5.1 Seemingly unrelated regression estimates for vote shares of the INC
and the BJP

INC
INC won 13.64 0.41 32.88 0.00
INC incumbent −4.92 2.01 −2.45 0.01
Year [Reference: 1989]
1991 −4.40 1.98 −2.22 0.03
1996 −7.81 2.00 −3.91 0.00
1998 −13.39 1.98 −6.77 0.00
1999 −8.52 2.00 −4.27 0.00
2004 −10.29 1.98 −5.18 0.00
2009 −8.25 1.96 −4.20 0.00
2014 −15.59 1.99 −7.82 0.00
Incumbent × year
1991 5.56 2.27 2.45 0.01
1996 2.77 2.21 1.26 0.21
1998 9.92 2.28 4.35 0.00
1999 5.18 2.33 2.23 0.03
2004 6.91 2.32 2.98 0.00
2009 5.96 2.26 2.63 0.01
2014 11.39 2.15 5.30 0.00
Turnout 0.43 0.15 2.97 0.00
Turnout squared 0.00 0.00 −2.96 0.00
Number of independents −0.18 0.03 −5.67 0.00
Number of independents squared 0.00 0.00 4.46 0.00
Number of ‘other’ parties −1.20 0.25 −4.87 0.00
Number of ‘other’ parties squared 0.05 0.01 3.46 0.00
State [Reference: Andhra Pradesh]
Assam −1.01 1.18 −0.85 0.39
Bihar −9.62 1.07 −8.98 0.00
Chhattisgarh 10.51 1.59 6.63 0.00
Gujarat 6.30 1.05 6.01 0.00
Haryana −2.88 1.33 −2.16 0.03
Himachal Pradesh 8.16 1.88 4.33 0.00
Jammu and Kashmir −3.92 2.30 −1.71 0.09
Jharkhand −4.67 1.86 −2.51 0.01
Karnataka 3.33 0.90 3.68 0.00
Kerala 5.33 1.02 5.24 0.00
Madhya Pradesh 4.92 0.92 5.34 0.00
Maharashtra 3.36 0.97 3.44 0.00
Orissa 0.61 1.06 0.58 0.57
Punjab 4.66 1.86 2.51 0.01
Rajasthan 4.55 0.99 4.59 0.00
Tamil Nadu −3.22 1.23 −2.62 0.01

(continued)

vk.borooah@ulster.ac.uk
100 V.K. BOROOAH

Table 5.1 (continued)

Uttarakhand 3.66 2.45 1.49 0.14
Uttar Pradesh −16.06 0.89 −17.99 0.00
West Bengal −4.54 1.01 −4.49 0.00
Intercept 33.93 4.87 6.97 0.00
BJP
BJP won 16.27 0.39 41.20 0.00
BJP incumbent 17.84 6.07 2.94 0.00
Year [Reference: 1989]
1991 −2.12 0.83 −2.56 0.01
1996 −2.40 0.84 −2.87 0.00
1998 7.72 0.89 8.68 0.00
1999 8.71 0.97 9.00 0.00
2004 4.64 0.92 5.06 0.00
2009 0.44 0.86 0.50 0.61
2014 10.06 0.92 10.92 0.00
Incumbent × year
1991 −10.18 6.20 −1.64 0.10
1996 −12.71 6.17 −2.06 0.04
1998 −16.20 6.15 −2.63 0.01
1999 −16.62 6.15 −2.70 0.01
2004 −14.17 6.14 −2.31 0.02
2009 −12.82 6.15 −2.08 0.04
2014 −13.33 6.12 −2.18 0.03
Turnout 0.26 0.14 1.90 0.06
Turnout squared 0.00 0.00 −3.08 0.00
Number of independents 0.11 0.03 3.54 0.00
Number of independents squared 0.00 0.00 −2.60 0.01
Number of ‘other’ parties −1.20 0.23 −5.20 0.00
Number of ‘other’ parties squared 0.05 0.01 4.37 0.00
State [Reference: Andhra Pradesh]
Assam 9.89 1.10 9.02 0.00
Bihar 9.27 0.97 9.54 0.00
Chhattisgarh 13.58 1.52 8.92 0.00
Gujarat 19.52 1.00 19.55 0.00
Haryana 8.88 1.24 7.14 0.00
Himachal Pradesh 20.76 1.77 11.74 0.00
Jammu and Kashmir 6.65 2.14 3.10 0.00
Jharkhand 6.58 1.74 3.79 0.00
Karnataka 12.49 0.86 14.58 0.00
Kerala −5.06 0.95 −5.35 0.00
Madhya Pradesh 16.14 0.89 18.21 0.00
Maharashtra 14.85 0.92 16.15 0.00
Orissa 5.79 0.98 5.93 0.00
(continued)

vk.borooah@ulster.ac.uk
 ANALYSIS OF VOTE SHARES 101

Table 5.1 (continued)

Punjab 12.20 1.74 7.02 0.00
Rajasthan 17.73 0.93 19.06 0.00
Tamil Nadu −1.39 1.14 −1.23 0.22
Uttarakhand 18.10 2.29 7.91 0.00
Uttar Pradesh 5.75 0.82 7.01 0.00
West Bengal 4.24 0.92 4.60 0.00
Intercept 13.55 4.31 3.15 0.00

Source: Own calculations from Lok Sabha election data

Notes to Table 5.1: Observations on 2684 constituencies, in the 20 major Indian states, which were
contested by both the INC and the BJP

Table 5.2 Equation statistics from the SURE equation

Equation Constituencies Parameters RMSE R2 χ2 p-value for χ2

INC vote share 2684 41 8.86 0.676 5697.4 0.0

BJP vote share 2684 41 8.26 0.784 9915.65 0.0

5.2.1 Predictive Performance of the SURE Model

The coefficients reported in Table 5.1 were used to predict the vote shares
of the INC and the BJP, collectively over all the eight elections between
1989 and 2014 and, also, for each individual election. The predictions
were made by applying the coefficient estimates (shown in Table 5.1) to
the values of the variables as observed in the 2684 constituencies, in the
20 major states, which were contested by both the INC and the BJP. This
yielded 2684 predicted vote shares (one for each constituency) for the
INC and another 2684 predicted vote shares for the BJP. The average of
these 2684 predicted vote shares for the INC and the BJP is reported in
this chapter.
A property of the estimated SURE equation—which is a general prop-
erty of linear regressions—is that it ‘passes through the mean’. As a con-
sequence of this property, the average predicted vote shares for the INC
and the BJP was identical to the mean of their actual vote shares in the
2684 constituencies in the 20 major states which were contested by both
parties.1

vk.borooah@ulster.ac.uk
102 V.K. BOROOAH

The predictions for the vote shares for the INC and the BJP, for an
individual election (say, the 2014 election), were made under the hypothet-
ical scenario that all the 2684 observations related to the 2014 election.
In other words, in computing the predicted INC and BJP vote shares for
this prediction, the coefficient pertaining to the 2014 election (shown in
Table 5.1) was applied to all 2684 observations, the coefficients pertaining
to the other elections being ignored. Similarly, in computing the predicted
INC and BJP vote shares for another election (say, the 2009 election),
the coefficient pertaining to the 2009 election (shown in Table 5.1) was
applied to all 2684 observations, the coefficients pertaining to the other
elections being ignored. The difference between the predictions, of the
INC and BJP vote shares, represents the election effect on vote shares:
since these two sets of predictions differ only in the fact that the first set
of predictions related to the 2014 election and the second set related to
the 2009 election, without any change in the values of the explanatory
variables underpinning the two sets of predictions, the difference between
them must be entirely due to the effect of differences between the 2014
and 2009 elections, that is to the ‘election effect’.
The predictions from the SURE model are compared to actual out-
comes in Table 5.3. When the elections are considered in their entirety
(row: ‘All years’ in Table 5.3), the predictions and the outcomes are iden-
tical since the regression ‘passes through the mean’.2 The predicted and

Table 5.3 The predictive performance of the SURE Model, 1989–2014

Observed Predicted

Number of INC BJP Difference INC BJP Difference

constituenciesa

All years 2684 32.2 31.5 0.7 32.2 31.5 0.7

1989 212 39.4 27.5 11.9 39.6 31.4 8.2
1991 434 37.7 22.2 15.5 37.4 26.4 11.0
1996 434 29.7 23.9 5.8 32.9 25.4 7.5
1998 331 31.3 36.0 −4.7 30.1 34.6 −4.5
1999 288 34.4 39.8 −5.4 33.1 35.5 −2.4
2004 294 32.9 34.5 −1.6 32.0 32.1 −0.1
2009 340 36.3 25.4 10.9 33.7 28.3 5.4
2014 351 24.4 39.6 −15.2 28.4 37.7 −9.3

a
Constituencies relate to those in the 20 major states (listed in Table 5.1) which were contested by both
the INC and the BJP

vk.borooah@ulster.ac.uk
 ANALYSIS OF VOTE SHARES 103

observed outcomes differ in terms of the individual elections, but not

substantially. The predicted INC vote share, in constituencies in the major
Indian states contested by both the INC and the BJP, falls from a high of
39.6 percent (observed: 39.4 percent) in 1989 to a low of 28.4 percent
(observed: 24.4 percent), while the predicted BJP vote share, in the same
2684 constituencies, rises from a low of 25.4 percent (observed: 23.9 per-
cent) in 1996 to a high of 37.7 percent (observed: 39.6 percent) in 2014.

5.3 INCUMBENT AND CHALLENGER VOTE SHARES

The SURE model also predicts the INC and BJP vote shares when they
are incumbents and challengers. In making these predictions, the method-
ology that was used took the following form. Using the coefficients of the
INC equation (shown in the first panel of Table 5.1), the predicted INC
vote share in each of these constituencies was computed and, then, aver-
aged, first when it was supposed that all the INC candidates in the 2684
constituencies were incumbents and next when it was supposed that all the
INC candidates in the 2684 constituencies were challengers. Since these
two sets of predictions differ only in the fact that in the first prediction, the
INC was assumed to be the incumbent, and in the second, it was assumed
to be the challenger, the difference between these two predictions must be
due entirely to the incumbent-challenger effects for the INC. An identical
exercise was then conducted for the BJP.
These predicted vote shares with respect to incumbents and challeng-
ers are shown in Tables 5.4 and 5.5. (It is important to emphasise that
by the structure of the SURE equations from which they were derived,
these predictions are made with respect to only those 2684 constituencies
that were contested by both the INC and the BJP.) Table 5.4 compares
the vote shares of INC incumbents and challengers and BJP incumbents
and challengers: columns 2 and 3 of Table 5.4 show the predicted vote
shares of, respectively, INC incumbents and challengers—and, in the row
below, the predicted vote shares of BJP incumbents and challengers—first,
aggregated over all elections and, then, for individual elections; the dif-
ference between the incumbent and challenger vote shares is recorded in
column 4; column 5 shows the standard error of the difference; column
6 shows the z-value (computed as the difference divided by the standard
error); and column 7 records the probability of obtaining, under the null
hypothesis that the difference is zero, a value greater than the observed z.

vk.borooah@ulster.ac.uk
104 V.K. BOROOAH

Table 5.4 Vote shares of incumbents and challengers: BJP and INC
1 2 3 4 5 6 7

Incumbent vote Challenger vote Difference Standard z-value p-value

share (%) share (%) error

All years
INC 33.5 31.8 1.7 0.4 4.2 0.0
BJP 35.5 30.4 5.2 0.6 9.4 0.0
1989
INC 36.6 41.5 −4.9 2.0 −2.5 0.0
BJP 44.3 26.4 17.9 6.1 2.9 0.0
1991
INC 37.8 37.1 0.6 1.1 0.6 0.5
BJP 32.0 24.3 7.7 1.3 6.0 0.0
1996
INC 31.6 33.7 −2.2 1.0 −2.3 0.0
BJP 29.2 24.0 5.1 1.1 4.8 0.0
1998
INC 33.2 28.1 5.0 1.1 4.6 0.0
BJP 35.8 34.2 1.7 1.0 1.7 0.1
1999
INC 33.3 33.0 0.3 1.2 0.2 0.8
BJP 36.4 35.1 1.2 1.0 1.2 0.2
2004
INC 33.2 31.3 2.0 1.2 1.7 0.1
BJP 34.8 31.1 3.7 1.0 3.7 0.0
2009
INC 34.3 33.3 1.1 1.0 1.0 0.3
BJP 31.9 26.9 5.0 1.0 5.2 0.0
2014
INC 32.4 25.9 6.5 0.8 8.0 0.0
BJP 41.0 36.5 4.5 0.9 5.3 0.0

Source: Own calculations from Lok Sabha election data using SURE estimates of Table 5.1

Table 5.4 shows that considering all the eight elections between 1989
and 2014 collectively, the average predicted vote shares of INC incum-
bents and challengers were, respectively, 33.5 and 31.8 percent, and read-
ing across the columns of that row, this difference of 1.7 points (column
4) was significantly different from zero.3 The next row does the same for
the BJP: the average predicted vote shares of BJP incumbents and chal-
lengers were, respectively, 35.5 and 30.4 percent, and using the preceding
argument, this difference, too, was significantly different from zero.

vk.borooah@ulster.ac.uk
 ANALYSIS OF VOTE SHARES 105

Table 5.5 Differences in vote shares between BJP and INC incumbents and
challengers
1 2 3 4 5 6 7

INC BJP Difference Standard z-value p-value

vote vote error
share share

All years incumbents 32.2 31.5 0.8 0.3 3.0 0.0

and challengers (I + C):
1989 (I + C) 39.6 31.4 8.3 2.2 3.8 0.0
1991 (I + C) 37.4 26.4 11.0 0.8 13.9 0.0
1996 (I + C) 32.9 25.5 7.5 0.8 9.0 0.0
1998 (I + C) 30.1 34.6 −4.5 0.8 −5.7 0.0
1999 (I + C) 33.1 35.5 −2.4 0.9 −2.6 0.0
2004 (I + C) 32.0 32.1 −0.1 0.8 −0.1 0.9
2009 (I + C) 33.7 28.3 5.4 0.7 7.3 0.0
2014 (I + C) 28.5 37.7 −9.3 0.9 −10.7 0.0
All years: incumbent 33.5 35.5 −2.0 0.6 −3.4 0.0
1989 incumbent 36.6 44.3 −7.7 6.1 −1.3 0.2
1991 incumbent 37.8 32.0 5.8 1.4 4.0 0.0
1996 incumbent 31.6 29.2 2.4 1.3 1.9 0.1
1998 incumbent 33.2 35.8 −2.7 1.2 −2.2 0.0
1999 incumbent 33.3 36.4 −3.1 1.3 −2.5 0.0
2004 incumbent 33.2 34.8 −1.5 1.3 −1.2 0.2
2009 incumbent 34.3 31.9 2.4 1.2 2.0 0.1
2014 incumbent 32.4 41.0 −8.6 1.2 −7.5 0.0
All years: challenger 31.8 30.4 1.4 0.3 4.4 0.0
1989 challenger 41.5 26.4 15.1 2.0 7.4 0.0
1991 challenger 37.1 24.3 12.8 0.9 14.5 0.0
1996 challenger 33.7 24.0 9.7 0.9 10.3 0.0
1998 challenger 28.1 34.2 −6.0 0.9 −6.6 0.0
1999 challenger 33.0 35.1 −2.1 1.0 −2.1 0.0
2004 challenger 31.3 31.1 0.2 0.9 0.2 0.9
2009 challenger 33.3 26.9 6.4 0.8 7.9 0.0
2014 challenger 25.9 36.5 −10.6 0.9 −11.2 0.0

Source: Own calculations from Lok Sabha election data using SURE estimates of Table 5.1

In terms of the individual elections, the vote share for BJP incumbents
always exceeded that for BJP challengers, except in the 1998 and 1999
elections (in both of which the BJP did particularly well, relative to the
INC, winning 182 seats in each election to the INC’s 141 in 1998 and
114 in 1999) when the difference in vote shares between incumbents and

vk.borooah@ulster.ac.uk
106 V.K. BOROOAH

challengers was not significantly different from zero. On the other hand,
the vote share of INC incumbents was significantly smaller than that of
INC challengers in the 1989 and 1996 elections and significantly larger
in the 1998 and 2014 Lok Sabha elections. This would suggest that in
elections which went against the INC (e.g., 1998 and 2014) it was left to
the incumbent constituencies to produce the votes while, in elections that
went in favour of the INC (e.g.,1991 and 2009), incumbents and chal-
lengers were on an equal footing. A similar picture emerges with respect to
the BJP: when it did well, as in 1998 and 1999, incumbents and challeng-
ers got similar vote shares; when it did badly, as in 1991 and 2009, BJP
incumbents obtained a larger vote share than BJP challengers.
Table 5.5 compares the vote shares of INC and BJP candidates. In
particular, Table 5.5 compares: (i) the predicted vote shares of all INC
candidates with all BJP candidates (labelled I + C in Table 5.5); (ii) the
predicted vote shares of INC and BJP incumbents; and (iii) the predicted
vote shares of INC and BJP challengers. Columns 2 and 3 show, respec-
tively, the relevant INC and BJP vote shares with the difference in column
4 and the its standard error in column 5; column 6 shows the z-value
(computed as the difference divided by the standard error), and column
7 records the probability of obtaining, under the null hypothesis that the
difference is zero, a value greater than the observed z.
Aggregating over all candidates, incumbents and challengers, and over
all eight elections from 1989 to 2014, there was no significant difference
between the vote shares obtained by INC (32.3 percent) and BJP candi-
dates (31.5 percent) in the 2684 constituencies, in the 20 major states,
contested by both parties. In terms of the individual elections, however,
the vote share of INC, compared to that of BJP, candidates was signifi-
cantly larger in the 1989, 1991, and 2009 elections and significantly lower
in the 1998, 1999, and 2014 elections.
In terms of comparing the INC and BJP in terms of the vote shares
of their incumbent and challenger candidates, aggregating over all eight
elections from 1989 to 2014, with respect to the 2684 constituencies (in
the 20 major states) that were contested by both the INC and the BJP, the
vote share of INC incumbents (33.5 percent) was significantly lower than
that of BJP incumbents (35.5 percent); however, compared to that of BJP
challengers (30.4 percent), the vote share of INC challengers (31.8 per-
cent) was significantly higher. In terms of individual elections, BJP incum-
bents had a significantly higher vote share than INC incumbents.

vk.borooah@ulster.ac.uk
 ANALYSIS OF VOTE SHARES 107

5.4 VOTE SHARES IN HINDI-SPEAKING AND

NON-HINDI-SPEAKING STATES
Chapter 3 pointed to the importance of Hindi-speaking (HS) states to the
relative electoral fortunes of the INC and the BJP. To recapitulate: of the
total of 543 Lok Sabha constituencies, 204 (or 37.6 percent) are—and have
been since the 1996 Lok Sabha election—in the seven HS states of Bihar,
Chhattisgarh, Jharkhand, Madhya Pradesh, Rajasthan, Uttarakhand, and
Uttar Pradesh, and of these 204 constituencies, respectively, 40 and 80 are
in Bihar and Uttar Pradesh. The HS states are of particular importance for
the BJP since a large number of its contested constituencies are from these
states: in 2014, nearly 45 percent (192 out of 428) of the constituencies
contested by the BJP were from the HS states. These states are also impor-
tant for the INC but to a lesser degree: 34 percent (158 out of 464) of
the constituencies contested by the INC in 2014 were from the HS states.
The relative importance of these states to the INC and the BJP is
reflected in the vote shares that the two parties obtained from these two
parts of India. In 2014, the INC had an average all-India vote share of
19.3 percent which was composed of 17.5 percent of the vote in the HS
states and 20.4 percent of the vote in the non-HS states; the BJP, on the
other hand, with a national vote share of 31 percent, obtained 44 percent
of the total vote in the HS states, but only 23.5 percent of the total vote
in the non-HS states. To put it differently, two-thirds of the nearly 107
million votes obtained by the INC in 2014 were from the 306 constituen-
cies it contested in the non-HS states, and one-third came from the 158
constituencies it contested in the HS states; for the BJP, on the other
hand, 48 percent of its total vote in 2014 was from the 236 constituen-
cies it contested in the non-HS states, and 52 percent came from the 192
constituencies it contested in the HS states.
Table 5.6, which shows the vote shares of the INC and the BJP in HS
and non-HS states, in each of the eight elections between 1989 and 2014,
reinforces the point made earlier about the imbalance in support for the
INC and the BJP between constituencies in the HS and non-HS states.
For example, in the 1989 Lok Sabha election, when the INC obtained a
national vote share of nearly 40 percent, its vote share in the HS and non-
HS states were, respectively, 33 and 44 percent; in the 2014 Lok Sabha
election, when the BJP obtained a national vote share of 31 percent, its
vote share in the HS and non-HS states were, respectively, 44 and 24 per-
cent. These are, of course, vote shares pertaining to all constituencies in

vk.borooah@ulster.ac.uk
108 V.K. BOROOAH

Table 5.6 INC and BJP vote shares in Lok Sabha constituencies in Hindi-
speaking and non-Hindi-speaking states
Vote share in Hindi- Vote share in non-Hindi- All India vote
speaking states (%) speaking states (%) share (%)

INC BJP INC BJP INC BJP

1989 32.9 18.2 43.6 8.0 39.9 11.5

1991 27.8 30.0 41.2 14.7 36.5 20.0
1996 17.7 32.2 34.6 13.9 28.7 20.3
1998 17.8 35.6 30.6 19.6 25.8 25.6
1999 22.6 32.5 31.6 18.8 28.3 23.8
2004 19.7 29.8 30.3 18.0 26.5 22.2
2009 25.2 25.7 30.2 15.4 28.6 18.8
2014 17.5 44.0 20.4 23.5 19.3 31.0

Source: Own calculations from Lok Sabha election data

the HS and non-HS states, some of which were not contested by one or
both of the two parties: in the 2014 Lok Sabha election, the INC contested
158 of the 204 constituencies (77 percent) in HS states and 306 of the
339 constituencies (90 percent) in non-HS states; by contrast, in that same
election, the BJP contested 192 of the 204 constituencies (94 percent) in
HS states and 236 of the 339 constituencies (70 percent) in non-HS states.
In order to compare the performances of the INC and the BJP in
the HS and non-HS states, in respect of the vote shares of incumbents
and challengers, we estimated two separate SURE models (of the type
described in the earlier section and controlling for the variables, noted in
Table 5.1): the first SURE model was estimated on data for constituencies,
which were contested by both the INC and BJP, in the 13 major non-HS
states, and the second SURE model was estimated on data for similar con-
stituencies in the seven major HS states. In total, over the seven elections
between 1989 and 2014, there were 1456 such constituencies in the non-
HS states and 1228 constituencies in the HS states.4
Table 5.7 shows that considered over all the elections between 1989 and
2014, for constituencies in HS states which were contested by both the INC
and the BJP, the vote shares of INC and BJP incumbents (respectively, 29
and 40 percent) were significantly larger than that of their correspond-
ing challengers (respectively, 25 and 37 percent). However, as Table 5.8
shows, for similar constituencies in non-HS states, INC incumbents did
not have any advantage, in terms of significantly higher vote shares, over

vk.borooah@ulster.ac.uk
 ANALYSIS OF VOTE SHARES 109

Table 5.7 Vote shares of incumbents and challengers: BJP and INC in Hindi-
speaking states
1 2 3 4 5 6 7

Incumbent Challenger vote Difference Standard z-value p-value

vote share (%) share (%) error

All years
INC 29.1 25.1 4.0 0.8 5.4 0.0
BJP 40.0 37.1 2.9 0.5 5.8 0.0
1991
INC 32.2 29.1 3.1 2.2 1.4 0.2
BJP 38.9 32.0 6.9 1.5 4.6 0.0
1996
INC 21.2 20.3 0.9 1.6 0.6 0.6
BJP 38.3 36.9 1.4 1.3 1.1 0.3
1998
INC 28.8 22.3 6.5 1.8 3.6 0.0
BJP 41.1 38.0 3.1 1.3 2.4 0.0
1999
INC 31.6 29.1 2.5 1.8 1.4 0.2
BJP 38.7 38.3 0.4 1.4 0.3 0.8
2004
INC 31.2 24.9 6.3 1.8 3.5 0.0
BJP 37.2 34.3 2.9 1.3 2.2 0.0
2009
INC 29.9 26.2 3.7 1.9 2.0 0.0
BJP 36.2 32.2 4.0 1.3 3.1 0.0
2014
INC 28.6 22.6 5.9 1.2 5.1 0.0
BJP 45.5 43.3 2.2 1.1 1.9 0.1

Source: Own calculations from Lok Sabha election data using SURE estimates of equation for non-Hindi-
speaking states

INC challengers; however, compared to BJP challengers, BJP incumbents

continued to have a significantly higher vote share in constituencies in
non-HS states contested by both the INC and the BJP.
Tables 5.9 and 5.10 compare the vote shares of INC and BJP candi-
dates in constituencies which were contested by both the INC and the
BJP in, respectively, HS and non-HS states. In particular, Tables 5.9 and
5.10 compare: (i) the predicted vote shares of all INC candidates with all
BJP candidates (labelled I + C in Tables 5.9 and 5.10); (ii) the predicted
vote shares of INC and BJP incumbents; and (iii) the predicted vote shares

vk.borooah@ulster.ac.uk
110 V.K. BOROOAH

Table 5.8 Vote shares of incumbents and challengers: BJP and INC in non-
Hindi-speaking states
1 2 3 4 5 6 7

Incumbent Challenger vote Difference Standard z-value p-value

vote share (%) share (%) error

All
years
INC 37.4 36.8 0.6 0.5 1.2 0.2
BJP 31.5 25.8 5.7 0.7 8.2 0.0
1991
INC 42.5 42.2 0.3 1.3 0.2 0.8
BJP 26.5 19.7 6.9 2.2 3.1 0.0
1996
INC 36.4 42.6 −6.3 1.2 −5.3 0.0
BJP 25.1 18.1 7.0 1.9 3.6 0.0
1998
INC 37.2 33.3 3.9 1.3 2.9 0.0
BJP 32.0 30.6 1.4 1.5 0.9 0.4
1999
INC 36.0 35.5 0.4 1.5 0.3 0.8
BJP 36.2 32.3 3.9 1.5 2.7 0.0
2004
INC 35.8 37.1 −1.3 1.5 −0.9 0.4
BJP 33.2 28.2 5.1 1.4 3.6 0.0
2009
INC 38.8 38.1 0.7 1.2 0.6 0.6
BJP 28.7 22.2 6.5 1.4 4.5 0.0
2014
INC 36.8 29.4 7.5 1.1 7.0 0.0
BJP 37.2 30.2 7.0 1.2 5.8 0.0

Source: Own calculations from Lok Sabha election data using SURE estimates of equation for Hindi-
speaking states

of INC and BJP challengers. Columns 2 and 3 show, respectively, the rel-
evant INC and BJP vote shares with the difference in column 4 and the
its standard error in column 5; column 6 shows the z-value (computed as
the difference divided by the standard error); and column 7 records the
probability of obtaining, under the null hypothesis that the difference is
zero, a value greater than the observed z.
Aggregating over all candidates, incumbents and challengers, and over
all eight elections from 1989 to 2014, Table 5.9 shows that the average
predicted vote share of BJP candidates was significantly larger than that

vk.borooah@ulster.ac.uk
 ANALYSIS OF VOTE SHARES 111

Table 5.9 Differences in vote shares between BJP and INC incumbents and
challengers, Hindi-speaking states
1 2 3 4 5 6 7

INC vote BJP vote Difference Standard z-value p-value

share share error

All years incumbents 25.6 38.2 −12.7 0.4 −35.4 0.0

and challengers (I + C):
1991 (I + C) 29.7 34.8 −5.1 1.2 −4.4 0.0
1996 (I + C) 20.4 37.3 −16.9 1.4 −12.2 0.0
1998 (I + C) 23.6 39.1 −15.5 1.1 −13.9 0.0
1999 (I + C) 29.5 38.3 −8.7 1.1 −7.8 0.0
2004 (I + C) 26.2 35.3 −9.2 1.1 −8.7 0.0
2009 (I + C) 26.9 33.7 −6.8 1.1 −6.5 0.0
2014 (I + C) 23.8 44.1 −20.3 1.2 −16.3 0.0
All years: incumbent 28.8 40.0 −11.2 0.7 −16.1 0.0
1991 incumbent 32.1 38.7 15.7 2.3 6.8 0.0
1996 incumbent 21.1 38.1 11.0 2.1 5.3 0.0
1998 incumbent 28.7 40.9 5.0 1.7 2.9 0.0
1999 incumbent 31.5 38.5 −0.5 1.8 −0.3 0.8
2004 incumbent 31.1 37.0 2.3 1.6 1.4 0.2
2009 incumbent 29.8 36.0 9.9 1.7 5.9 0.0
2014 incumbent 28.5 45.3 −0.6 1.6 −0.4 0.7
All years: challenger 24.5 37.1 −12.6 0.4 −28.5 0.0
1991 challenger 29.0 31.8 −6.6 2.5 −2.7 0.0
1996 challenger 20.2 36.7 −17.0 2.0 −8.7 0.0
1998 challenger 22.2 37.8 −12.1 1.9 −6.4 0.0
1999 challenger 29.0 38.1 −7.0 1.8 −3.8 0.0
2004 challenger 24.8 34.1 −5.8 2.0 −3.0 0.0
2009 challenger 26.1 32.0 −6.1 2.0 −3.0 0.0
2014 challenger 22.5 43.1 −16.8 1.6 −10.7 0.0

Source: Own calculations from Lok Sabha election data using SURE estimates of equation for Hindi-
speaking states

of INC candidates in constituencies in HS states which were contested by

both parties (38.2 percent versus 25.6 percent); this result was also true
for a comparison of BJP and INC incumbents and BJP and INC chal-
lengers. In terms of the individual elections, too, BJP candidates were
predicted to receive, on average, a higher vote share than their INC rivals
in constituencies, in the HS states, which they both contested.
On the other hand, aggregating over all candidates, incumbents and
challengers, and over all eight elections from 1989 to 2014, Table 5.10
shows that the average predicted vote share of INC candidates was signifi-

vk.borooah@ulster.ac.uk
112 V.K. BOROOAH

Table 5.10 Differences in vote shares between BJP and INC incumbents and
challengers, non-Hindi-speaking states
1 2 3 4 5 6 7

INC BJP Difference Standard z-value p-value

vote vote error
share share

All years incumbents 36.2 27.0 9.3 0.4 25.6 0.0

and challengers (I + C):
1991 (I + C) 42.1 21.1 20.9 1.1 19.4 0.0
1996 (I + C) 39.7 19.6 20.1 1.1 19.1 0.0
1998 (I + C) 34.8 30.9 3.9 1.1 3.5 0.0
1999 (I + C) 35.5 33.1 2.4 1.3 1.8 0.1
2004 (I + C) 36.3 29.3 7.0 1.2 5.8 0.0
2009 (I + C) 38.2 23.5 14.7 1.0 15.0 0.0
2014 (I + C) 32.4 31.7 0.7 1.2 0.6 0.5
All years: incumbent 37.0 31.5 5.5 0.7 7.7 0.0
1991 incumbent 42.2 26.6 15.7 2.3 6.8 0.0
1996 incumbent 36.1 25.2 11.0 2.1 5.3 0.0
1998 incumbent 37.0 32.0 5.0 1.7 2.9 0.0
1999 incumbent 35.7 36.2 −0.5 1.8 −0.3 0.8
2004 incumbent 35.6 33.3 2.3 1.6 1.4 0.2
2009 incumbent 38.6 28.7 9.9 1.7 5.9 0.0
2014 incumbent 36.6 37.3 −0.6 1.6 −0.4 0.7
All years: challenger 36.0 25.8 10.2 0.4 23.3 0.0
1991 challenger 41.9 19.7 22.2 1.3 17.7 0.0
1996 challenger 42.4 18.1 24.3 1.2 20.1 0.0
1998 challenger 33.1 30.6 2.4 1.3 2.0 0.1
1999 challenger 35.3 32.3 3.0 1.5 2.0 0.0
2004 challenger 36.9 28.2 8.6 1.4 6.4 0.0
2009 challenger 37.9 22.2 15.7 1.1 14.6 0.0

Source: Own calculations from Lok Sabha election data using SURE estimates of equation for non-Hindi-
speaking states

cantly larger than that of BJP candidates in constituencies in non-HS states

which were contested by both parties (36.2 percent versus 27 percent);
this result generally held for a comparison of BJP and INC incumbents
and was always true for a comparison of BJP and INC challengers. For the
individual elections as well, except for the 2014 Lok Sabha election, INC
candidates were predicted to receive, on average, a higher vote share than
their BJP rivals in constituencies, in the non-HS states, which they both
contested.

vk.borooah@ulster.ac.uk
 ANALYSIS OF VOTE SHARES 113

5.5 CONCLUDING REMARKS

The central contribution of this chapter was to undertake an economet-
ric investigation of the vote shares of the INC and BJP, in constituen-
cies, which they both contested, in the 20 major states. This investigation
paid particular attention to whether, in these constituencies, the two par-
ties were incumbents or challengers. Aggregated over all eight elections
between 1989 and 2014, the average predicted vote share for INC incum-
bents was higher than that for INC challengers, and similarly, the average
predicted vote share for BJP incumbents was higher than that for BJP
challengers. So, measured in terms of vote shares, there would appear to
be a pro-incumbency effect towards both the INC and the BJP.
Compared to the average predicted vote shares for the BJP when it was
the incumbent party, the INC did not do as well as the incumbent party.
However, in terms of average predicted vote shares, the INC did better than
the BJP when both were challenger parties. So, on this interpretation, in
terms of a cross-party comparison, there would appear to be an anti-incum-
bency effect towards the INC but a pro-incumbency effect towards the BJP.

NOTES
1. Note that these are weighted means, the weights being the proportions of
the constituencies’ vote to the total vote.
2. Note that these predictions relate only to those constituencies, in the 20
major states, contested by both the INC and the BJP.
3. Since dividing this difference by its standard error of 0.41 (column 6)
yielded a z-value of 4.2 (column 6), the p-value of column 7 shows that the
probability of observing a z-value of this magnitude, under the null hypoth-
esis that the difference was zero, was absurdly small, and so, this hypothesis
could be ‘rejected’.
4. In presenting the results, the 1989 election was omitted since there were
only two BJP incumbents in this election, and both constituencies were in
non-HS states: Hanamkonda in Andhra Pradesh (won by C.J. Reddy) and
Mehsana in Gujarat (won by A.K. Patel).

REFERENCE
Greene, W. H. (2003). Econometric analysis (5th ed.). Upper Saddle River:
Prentice Hall.

vk.borooah@ulster.ac.uk
 Chapter 6

The Relationship Between Votes and Seats

Abstract Borooah develops the concept of the amplification coefficient

which assesses its ability to convert votes into seats. In this respect, his
major finding was that the BJP, in electoral terms, was much more effi-
cient than the INC. The central result is that, averaging over the eight Lok
Sabha elections between 1989 and 2014, in order to win 200 seats in the
Lok Sabha the INC would have had to receive 35 votes for every 100 votes
received by the collective of non-INC parties (that is, 26 percent of the
total vote) but the BJP would have had to receive only 27 votes for every
100 votes received by the collective of non-BJP parties (that is, 21 percent
of the total vote).

6.1   Introduction
As noted earlier, in Lok Sabha elections, a single representative for each of
543 constituencies is elected—on the basis of obtaining the largest num-
ber of votes of all the candidates contesting that constituency—as a mem-
ber of the Lok Sabha for that constituency. This system of election is called
the First-Past-the-Post (FPTP) system. The disjoint, under this system,
between the votes obtained and the seats won by a party frequently causes
consternation. Unlike a proportional electoral system, in which a party’s
share of the total vote is a good predictor of its share of parliamentary
seats, the relation between seats and votes in an FPTP system often works
in mysterious ways.

© The Editor(s) (if applicable) and The Author(s) 2016 115

V.K. Borooah, Votes, Parties, and Seats,
DOI 10.1007/978-3-­319-30487-8_6

vk.borooah@ulster.ac.uk
116 V.K. Borooah

For example, in the Lok Sabha elections of 2014, the Bharatiya Janata
Party (BJP) won 282 seats with 31 percent of the vote while the INC with
nearly 20 percent of the vote could manage only 44 seats; in the same elec-
tion, the All India Anna Dravida Munnetra Kazhagam (AIADMK) won 37
seats with just 3.3 percent of the vote. Nor is this anomaly between votes
and seats confined to India: in the UK General Election of May 2015, the
Scottish National Party won 56 seats in the House of Commons on the
back of just under 1.5 million votes, while in the same election, the UK
Independence Party received nearly 4 million votes and were rewarded
with just one seat.
In this chapter, we attempt to understand some of these mysteries in
respect of General Parliamentary Elections in India. This chapter is con-
cerned with analysing the fortunes of India’s two largest political parties,
the INC and the BJP, and in the context of the above remarks, this chapter
examines, in some detail, the relationship between the votes obtained and
the seats won by the INC and the BJP.
The starting point of the analysis is the Law of the Cubic Proportion
according to which, in a two-party contest, ‘the proportion of seats won
by the victorious party varies as the cube of the proportion of votes cast
for that party over the country as a whole’ (Kendall and Stuart, 1950,
p. 183).1 In this chapter, we take a different approach to this ‘law’ by
separately computing for the INC and BJP the coefficient which equates
the proportion of their votes to the proportion of their seats. We term this
the amplification coefficient and show that its value is very different for the
two parties. Since the BJP gained political traction only from the 1989
General Election—when it won 85 seats, having won just two seats in the
previous General Election of 1984—the analysis in this chapter is confined
to eight General Elections: 1989, 1991, 1996, 1998, 1999, 2004, 2009,
and 2014.

6.2   The Electoral Fortunes of the INC

and the BJP

Table 6.1 presents a summary account of election outcomes for the INC
and the BJP for the eight Indian Lok Sabha elections held between 1989
and 2014. This highlights two features of the electoral performance of
the INC and the BJP. First, the BJP always contested fewer seats than the

vk.borooah@ulster.ac.uk
 Table 6.1 BJP and INC election results for eight Lok Sabha elections: 1989–2014
BJP INC

Year Lok Seats Constituencies Vote Votes Votes per Seats Constituencies Vote Votes Votes per
Sabha contested share seat won contested share seat won

2014 16 282 428 31.3 171,657,552 608,715 44 464 19.5 106,938,240 2,430,415
2009 15 116 433 18.8 78,435,352 676,167 206 440 28.6 119,110,824 578,207
2004 14 138 364 22.2 86,371,560 625,881 145 417 26.5 103,408,952 713,165
1999 13 182 339 23.8 86,562,208 475,617 114 453 28.3 103,120,328 904,564
1998 12 182 388 25.6 94,266,192 517,946 141 477 25.8 95,111,128 674,547
1996 11 161 469 20.3 67,697,336 420,480 139 526 28.8 96,034,448 690,895

vk.borooah@ulster.ac.uk
1991 10 121 478 20.0 55,953,668 462,427 244 504 35.7 102,059,792 418,278
1989a 9 85 225 11.4 34,171,476 402,017 197 510 39.5 118,894,704 603,526

Source: Own calculations from Election Commission of India data

a
Elections were not held in Assam because electoral rolls were incomplete
The Relationship Between Votes and Seats 
117
118 V.K. Borooah

INC though, with the INC’s acceptance of the exigencies of seat adjust-
ment under coalition government, the number of seats it contested fell
from a high of 526 in 1996 (97 percent of the total of 543 Lok Sabha
seats) to a low of 417 in 2004 (77 percent of Lok Sabha seats). Second,
except for the 1991 and 2009 elections, when it won over 200 seats, the
INC has always paid a higher ‘price’ in terms of votes for the seats that
it did win: for example, in 2004, when both parties won roughly the
same number of seats—145 for the INC to the BJP’s 138—the votes
per seat for the INC, at 713,165, was considerably higher than the BJP’s
625,881.
The corollary is that compared to the BJP, the INC is relatively ineffi-
cient in terms of converting votes into seats: in 2004, it won 26.5 percent
of the vote compared to the BJP’s 22.2 percent, but only won seven more
seats on the strength of this four point advantage; in 1996, it obtained a
larger vote share (28.8 percent compared to 20.3 percent), but won fewer
seats (139 compared to the BJP’s 161); and in 1998, the BJP won 41
more seats than the INC (182 compared to 141) with the same share of
the vote as the INC (26 percent).
Given our interest in the two leading protagonists, the INC and the
BJP, the focus of the analysis was those constituencies in which there
was an INC and/or a BJP candidate so that constituencies in which
there was neither an INC nor a BJP candidate were excluded from the
analysis. Table 6.2 shows that of the total of 4323 constituencies in the
eight Lok Sabha elections between 1989 and 2014, there were only 245

Table 6.2 Constituencies contested by the INC and the BJP in Lok Sabha elec-
tions: 1989–2014
1989 1991 1996 1998 1999 2004 2009 2014 Total

Neither INC nor 14 12 4 30 58 72 31 24 245

BJP contested
INC contested 289 47 70 125 146 107 79 91 954
but BJP did not
BJP contested 4 21 13 36 32 54 72 55 287
but INC did not
Both INC and 221 457 456 352 307 310 361 373 2837
BJP contested
Total 528 537 543 543 543 543 543 543 4323

Source: Own calculations from Election Commission of India data

vk.borooah@ulster.ac.uk
 The Relationship Between Votes and Seats  119

c­ onstituencies which neither party contested (5.7 percent of the total) and
2837 constituencies which were contested by both parties (65.6 percent
of the total).2

6.3   The ‘Amplification Coefficient’ and the Law

of the Cubic Proportion

We assume a two-party system (parties A and B), in which the representa-

tive in each constituency is elected under an FPTP system. Let VAt and VBt
represent the votes obtained, and SAt and SBt won, by parties A and B,
( )
respectively, at an election held in time t such that VAt < VBt . If VAt / VBt < 1
is the ratio of votes accruing to parties A and B, then for some real number
α, we must have:

±
 VAt  S At
 V t  = (6.1)
B SBt

We refer to the term α as the amplification coefficient because it ampli-

fies the votes ratio into a seats ratio. For example, if α = 3 and the vote ratio
is 40/60 (i.e., 1:1.5, meaning that for every vote obtained by party A,
party B obtains 1.5 votes), then the seats ratio will be (40)3/(60)3, that is
1:3.4, meaning that for every seat won by party A, party B would win 3.4
seats. So, if there were 100 seats contested, parties A and B would win 23
and 77 seats, respectively.3
The value α = 3 embodies the ‘Law of the Cubic Proportion’ of election
results under an FPTP system (Kendall and Stuart, 1951; Rajagopalan,
1959; Curtice and Steed, 1986; Norris and Crewe, 1994). In general, of
course, the value of α will be different from 3. Indeed, for any given elec-
tion, it is possible to solve for, α, the amplification coefficient associated
with that election as:

±=
(
log S At / SBt ) (6.2)
log (V t
A /V )t
B

Various values of α are possible:

vk.borooah@ulster.ac.uk
120 V.K. Borooah

1. If α > 1, then the votes of party A—which is trailing in terms of votes—

are amplified or exaggerated in terms of the number of seats by which
it trails party B. In other words, the seats ratio, SAt/SBt, is smaller than
the votes ratio, VAt/VBt.
2. If α < 1, then the votes of party A—which is trailing in terms of votes—
are de-amplified or dampened in terms of the number of seats by which
it trails party B. In other words, the seats ratio, SAt/SBt, is larger than
the votes ratio, VAt/VBt.
3. If α = 1, then the FPTP system mimics a proportional system: seats are
won in the same ratio that votes were obtained.
4. There could be the outcome when parties A and B obtain the same
number of votes ( VAt = VBt ) but win different numbers of seats ( S At ≠ SBt ).
In this case, the denominator of equation (6.2) is zero, and α will not
be defined.
5. There could be the outcome in which party A obtains fewer votes than
party B but wins more seats: VAt VBt but S At SBt . In this situation, the
numerator in Eq. (6.2) is positive, with the denominator negative, so
that α < 0.4

Case 5, above, represents the (not unusual) situation where party A

obtains a majority in parliament without securing a majority of votes. It is
important to point out that since, by assumption, party A trails party B in
terms of votes, the smaller the value of α, the smaller will be party A’s rela-
tive disadvantage in terms of seats. When α = 0, S At = SBt implying VAt < VBt ,
and when α < 0, S At > SBt implying VAt < VBt .

6.4   Calculating Amplification Coefficient

the
for the INC and the BJP

We compute the value of the amplification coefficient, from data for eight
Lok Sabha elections, held between 1989 and 2014 (9th to the 16th Lok
Sabha) for the two main protagonists in these elections—the BJP and the
INC—by mimicking a two-party system. In the first instance, we compare
the INC (party A in the above analysis) with the collective of non-INC
parties, including independent candidates (party B in the above analysis);
in the second instance, we compare the BJP (party A in the above analysis)
with the collective of non-BJP parties, including independent candidates
(party B in the above analysis).

vk.borooah@ulster.ac.uk
 The Relationship Between Votes and Seats  121

 = 0.239 and S INC / S INC

Table 6.3 shows that in 2014, VINC / VINC  = 0.088
for the INC (where, depending on the context, ^ over the party name
represents ‘non-INC’ or ‘non-BJP’). Consequently, using the expression
in Eq. 6.2, the value of the amplification coefficient, α, in the 2014 Lok
Sabha elections was 1.7 for the INC and −0.09 for the BJP.
Since VINC / VINC  = 0.088 ⇒ ( VINC / 0.239 ) ,
 = 0.239 and S INC / S INC

 and ( S INC / 0.088 ) = S INC

= VINC 

for every vote won by the INC, the non-INC collective won 4.2 votes
(=1/0.239), but for every seat won by the INC, the non-INC col-
lective won 11.4 seats (=1/0.088). On the other hand, for the BJP,
 = 0.449 and S BJP / S BJP
VBJP / VBJP  = 1.08
implying that for every vote won by
the BJP, the non-BJP collective won 2.2 votes (= 1/0.449), but for every
seat won by the BJP, the non-BJP collective won 0.93 seats.5 Furthermore,
V / V = 0.239 ⇒ VINC = 0.239 × VINC
since INC INC  ⇒ VINC / VINC + VINC

,( )
= 0.239 / (1.239 )

and since
 = 0.088 ⇒ S INC = 0.088 × S INC
SINC / SINC (
 ⇒ S INC / S INC + S INC
 ),
= 0.088 / (1.088 )
it follows that in 2014, the INC received 19.3 percent (= 0.239/1.239) of
total votes while winning only 8 percent of seats (= 0.088/1.088), while

Table 6.3 Vote and seat ratios and values of the amplification coefficient in Lok
Sabha elections: 1989–2014
INC/non-INC BJP/non-BJP

Votes Seats Amplification Votes Seats Amplification

ratio ratio coefficient (α) ratio ratio coefficient (α)
(VA/VB) (SA/SB) (VA/VB) (SA/SB)

2014 0.239 0.088 1.70 0.449 1.08 −0.09

2009 0.400 0.611 0.54 0.232 0.272 0.89
2004 0.361 0.364 0.99 0.285 0.341 0.86
1999 0.395 0.266 1.43 0.312 0.504 0.59
1998 0.348 0.351 0.99 0.344 0.504 0.64
1996 0.403 0.344 1.18 0.254 0.421 0.63
1991 0.576 0.833 0.33 0.250 0.291 0.89
1989 0.663 0.595 1.26 0.129 0.192 0.81

Source: Own calculations from Election Commission of India data

vk.borooah@ulster.ac.uk
122 V.K. Borooah

the BJP, with 31 percent ( = 0.449 / 1.449 ) of total vote, won 52 percent
( = 1.08 / 2.08 ) of the seats.
One can verify that the amplification coefficients are correctly calculated
by computing the total number seats that a party would have won and
comparing these with the numbers actually won: these should be identical
if α has been correctly computed. In order to do so, define Á = (VA / VB ) ,
α
^

where ±̂ represents the value of the amplification coefficient computed

using Eq. 6.2; then, from Eq. 6.1, the computed number of seats won by
party A is: Sˆ A =  Á / (1 + Á) × ( S A + SB ) , where S A + SB represents the total
number of elected seats in parliament. This should equal, SA, the actual
number of seats won by party A.
The fact that in 2014, α > 1 for the INC reflects the fact that the INC
won 19.3 percent of the popular vote, but only 8 percent of Lok Sabha
seats: as noted in point 1, above, this means that INC’s disadvantage, rela-
tive to the non-INC parties, in terms of votes were amplified in terms of
seats. The fact that in the same election, α < 0 for the BJP reflects the fact
that the BJP won a majority in the Lok Sabha without winning a majority
of the popular vote (point 5, above).6
In 2004, the INC’s votes and seats ratios were almost equal—respec-
tively, 0.361 and 0.364—yielding a value of the INC amplification coeffi-
cient close to unity (α = 0.99). This illustrates point 3 above: in the context
of the INC, the electoral system yielded a proportional outcome. However,
in the same election, α = 0.89, meaning that the BJP’s disadvantage, rela-
tive to the non-BJP parties, was dampened with respect to seats.
Remembering that a lower value of α is more desirable than a higher
value, Table 6.3 shows that the value of α was smaller for the BJP than the
INC for six—1989, 1996, 1998, 1999, 2004, and 2014—of the eight elec-
tions studied. Equally worrying for the INC was that α > 1 for four of the
eight elections between 1989 and 2014. This meant that in the 2014, 1999,
1996, and 1989 elections, the INC vote disadvantage, relative to the non-
INC parties, was amplified in terms of a seats disadvantage. For example, in
1999, for every vote won by the INC, the non-INC parties won 2.5 votes,
but for every seat won by the INC, the non-INC parties won 3.8 seats.
The BJP did not have this problem: as Table 6.3 shows, in each of the eight
elections, the value of its amplification coefficient was less than one meaning
that is was able to neutralise some of its vote disadvantage, relative to the
non-BJP parties, in terms of its seat disadvantage. For example, in 2009, with
α = 0.89, for every vote won by the BJP, the non-BJP parties won 4.3 votes,

vk.borooah@ulster.ac.uk
 The Relationship Between Votes and Seats  123

but for every seat won by the BJP, the non-BJP parties won 3.7 seats. Thus,
the essential difference between the BJP and the INC was that compared to
the INC, the BJP was more efficient in translating votes into seats.
In order to understand this, measure of this inefficiency consid-
ers a party which targets, say, 200 (out of a total of 543) seats in the
Lok Sabha. Then, from Eq. 6.2, the vote ratio which will deliver this is:
VA / VB = ( S A / SB ) = ( 0.368) . Excluding the election of 2014—which
1/ ± 1/ ±

was a landslide victory for the BJP and, arguably, an outlier on a scale
unlikely to be repeated—the average of the amplification coefficient over
the seven elections between 1989 and 2009 was 0.96 for the INC and
0.76 for the BJP. Applying these averages, in order to win 200 seats in the
Lok Sabha, the INC and BJP would have required a vote ratio of, respec-
tively, 0.353 and 0.267.
In other words, to win 200 seats in the Lok Sabha, the INC would have
had to receive 35 votes for every 100 votes received by the collective of
non-INC parties (i.e., 26 percent of the total vote), but the BJP would have
had to receive only 27 votes for every 100 votes received by the collective of
non-BJP parties (i.e., 21 percent of the total vote). The value of the amplifi-
cation coefficient, α, is thus a measure of electoral efficiency—the smaller its
value, the greater the ease with which votes are converted into seats.

6.5   Hindi-Speaking States

In earlier chapters, we had drawn attention to the importance of Hindi-
speaking (HS) states to the relative electoral fortunes of the INC and
the BJP. To recapitulate: of the total of 543 Lok Sabha constituencies,
204 (or 37.6 percent) are—and have been since the 1996 Lok Sabha elec-
tion—in the seven HS states of Bihar, Chhattisgarh, Jharkhand, Madhya
Pradesh, Rajasthan, Uttarakhand, and Uttar Pradesh, and of these 204
constituencies, respectively, 40 and 80 are in Bihar and Uttar Pradesh. The
HS states are of particular importance for the BJP since a large number
of its contested constituencies are from these states: in 2014, nearly 45
percent (192 out of 428) of the constituencies contested by the BJP were
from the HS states. These states are also important for the INC but to a
lesser degree: 34 percent (158 out of 464) of the constituencies contested
by the INC in 2014 were from the HS states. We can pursue the analy-
sis of electoral efficiency (meaning the relative ease with which votes are
converted into seats) by computing the INC and BJP amplification coef-
ficients for constituencies in the seven Hindi-speaking states.

vk.borooah@ulster.ac.uk
124 V.K. Borooah

Table 6.4 Vote and seat ratios and values of the amplification coefficient in Lok
Sabha elections for Hindi-speaking statesa: 1989–2014
INC/non-INC BJP/non-BJP

Votes Seats Amplification Votes Seats Amplification

ratio ratio coefficient (α) ratio ratio coefficient (α)
(VA/VB) (SA/SB) (VA/VB) (SA/SB)

2014 0.212 0.036 2.15 0.784 5.375 −6.92

2009 0.338 0.437 0.76 0.347 0.417 0.83
2004 0.245 0.159 1.31 0.425 0.581 0.63
1999 0.291 0.200 1.30 0.481 0.907 0.13
1998 0.217 0.193 1.08 0.552 1.22 −0.33
1996 0.215 0.152 1.22 0.475 1.147 −0.18
1991 0.385 0.291 1.29 0.429 0.658 0.49
1989 0.489 0.152 2.63 0.221 0.378 0.64

Source: Own calculations from Election Commission of India data

Bihar, Chhattisgarh (2004 and after), Jharkhand (2004 and after), Madhya Pradesh, Rajasthan,
a

Uttarakhand (2004 and after), and Uttar Pradesh

Table 6.4, which reproduces the all-India results of Table 4.3 for the
collective of HS states, shows that the BJP’s electoral efficiency was far
greater than that of the INC in the HS states. Except for the 2009 elec-
tion, the INC’s seat ratio was always smaller than its vote ratio resulting
in an amplification coefficient that was greater than one. By contrast, the
BJP’s seat ratio was always larger than its vote ratio resulting in an ampli-
fication coefficient that was less than one. Indeed, in the 1996, 1998, and
2014 elections, the BJP secured a majority of the Lok Sabha seats from the
HS states on a minority vote (point 5 above).
In contrast, as Table 6.5 shows, the INC performed much better,
relative to the BJP, in the non-HS states. Its votes and seats ratios (i.e.,
VINC / VINC  ) were both higher in the non-HS states than in
 and S INC / S INC
the HS states, and its amplification coefficient was more favourable: bear-
ing in mind that as discussed earlier, a lower value of the amplification
coefficient is more desirable than a higher value; the amplification coeffi-
cient for the INC was always lower in the non-HS, than in the HS, states;
and conversely, the amplification coefficient for the BJP was always lower
in the HS, than in the non-HS, states.
In order to gain an appreciation of differences in inter-party elec-
toral performance between the HS and non-HS states, suppose that
the INC and the BJP each targets one in three of the seats from the

vk.borooah@ulster.ac.uk
 The Relationship Between Votes and Seats  125

Table 6.5 Vote and seat ratios and values of the amplification coefficient in Lok
Sabha elections for non-Hindi-speaking statesa: 1989–2014
INC/non-INC BJP/non-BJP

Votes Seats Amplification Votes Seats Amplification

ratio ratio coefficient (α) ratio ratio coefficient (α)
(VA/VB) (SA/SB) (VA/VB) (SA/SB)

2014 0.253 0.114 1.58 0.301 0.395 0.773

2009 0.415 0.683 0.434 0.177 0.181 0.987
2004 0.430 0.505 0.810 0.213 0.211 1.01
1999 0.456 0.299 1.54 0.225 0.284 0.844
1998 0.432 0.469 0.901 0.237 0.219 1.06
1996 0.524 0.476 1.15 0.154 0.157 0.99
1991 0.690 1.44 −0.977 0.170 0.114 1.23
1989 0.77 1.06 −0.252 0.085 0.084 1.01

Source: Own calculations from Election Commission of India data

a
Andhra Pradesh, Assam, Gujarat, Haryana, Himachal Pradesh, Jammu and Kashmir, Karnataka, Kerala,
Maharashtra, Orissa, Punjab, Tamil Nadu, and West Bengal

HS states (total 204 constituencies) and a similar proportion from the

non-HS states (total 313 constituencies). Excluding the 2014 elec-
tion, the average of the amplification coefficients over the last three
elections (1999, 2004, and 2009) was, in the HS state, 1.12 for the
INC and 0.53 for the BJP, and in the non-HS states, it was 0.93 and
0.95 for, respectively, the INC and the BJP. Then, the vote ratios that
would deliver half of the constituencies, in the HS states, to the INC
and the BJP are, respectively, VA / VB = ( S A / SB ) = ( 0.5)
1/1.12 1 /1.12
= 0.54 and

VA / VB = ( S A / SB ) = ( 0.50 )
1/ 0.53 1/ 0.53
= 0.27. In other words, to win one-third
of the seats in the HS states, the INC and the BJP would have required
vote shares of, respectively, 35 (=0.54/1.54) and 21 percent (=0.27/1.27)
in the HS states. On the other hand, to win one in three seats in the non-
­HS states, both the INC and the BJP would have needed a vote ratio—of,
respectively, INC to non-INC votes and BJP to non-BJP votes, of 0.48 or
32 percent (=0.48/1.48) of the vote in the non-HS states.7

vk.borooah@ulster.ac.uk
126 V.K. Borooah

6.6   Concluding Remarks

The contribution of this chapter was to develop the concept of the ampli-
fication coefficient which, when applied to the votes received and seats won
by a party, could be used to assess its ability to convert votes into seats.
In this respect, this chapter’s major finding was that the BJP, in electoral
terms, was much more efficient than the INC. This can be encapsulated
in our finding that averaging over the eight Lok Sabha elections between
1989 and 2014, in order to win 200 seats in the Lok Sabha, the INC
would have had to receive 35 votes for every 100 votes received by the
collective of non-INC parties (i.e., 26 percent of the total vote), but the
BJP would have had to receive only 27 votes for every 100 votes received
by the collective of non-BJP parties (i.e., 21 percent of the total vote).
This places the INC at a considerable disadvantage vis-à-vis the BJP in
terms of contesting elections and suggests that this is an issue that the
INC’s managers could usefully address.

Notes
1. Kendall and Stuart (1950) draw attention to the fact that the law was first
proposed by James Parker Smith—who, in turn, attributed it to
P.A. MacMahon—in evidence before the Royal Commission on Systems of
Elections (2010).
2. Bear in mind that ‘observations’ were distinguished by constituency name
and by election year: so, for example, Adilabad in the 1989 Lok Sabha elec-
tion represented a separate observation from Adilabad in the 1991 Lok
Sabha election.
3. Party A: (1/4.4) × 100 = 22.7 and party B: (3.4/4.4) × 100 = 77.3.
4. Another perverse outcome would be when party A obtains more votes than
party B, but wins fewer seats: VAt > VBt but S At < S Bt . In this situation, the
numerator in equation (2) is negative, with the denominator positive, so
that α < 0. This is a situation in which where the party A’s majority in votes
fails to translate into a parliamentary majority.

 = 0.239 ⇒ VINC = 0.239 × VINC

5. VINC / VINC 

 + 0.239 × VINC − 0..239 × VINC

= 0.239 × VINC
( )
 = 0.239 / (1.239 )
⇒ VINC / VINC + VINC

This implies that in 2014, the INC received 19.3 percent of total votes while
winning only 8 percent of seats while the BJP, with 31 percent of total vote,
won 52 percent of the seats. For the INC, [ 0.239 / 1.239] = 0.193 ; and for
the BJP, [ 0.449 / 1.449] = 0.31 and [1.08 / 2.08] = 0.52

vk.borooah@ulster.ac.uk
 The Relationship Between Votes and Seats  127

6. If the BJP, on a minority vote, had won the same number of seats as the col-
lective of non-BJP parties (i.e., 271 or 272 seats), the value of α would have
been zero.
7. Vote ratios of 0.47 (= (0.5)1/0.93) and 0.48 (= (0.5)1/0.95) which amounts to
32 percent of the vote in the non-HS states

References
Curtice, J., & Steed, M. (1986). Proportionality and exaggeration in the British
electoral system. Electoral Studies, 5(3), 209–228.
Kendall, M. G., & Stuart, A. (1950). The law of the cube proportion in election
results. The British Journal of Sociology, 1(3), 183–196.
Norris, P., & Crewe, I. (1994). Did the British marginals vanish? Proportionality
and exaggeration in the British electoral system revisited. Electoral Studies,
13(3), 201–221.
Rajagopalan, K. R. (1959, December 12). The law of cubic proportions. The
Economic Weekly, 1669–1670.

vk.borooah@ulster.ac.uk
 Chapter 7

The Concentration and Distribution

of Votes

Abstract Borooah examines the topics of vote concentration (differences

between the INC and the BJP in the concentration of their votes and seats
in the various states) and vote distribution (unevenness in the distribution
of the INC and BJP vote across the constituencies). Combining these top-
ics the central question is what proportion of the seats won by the INC
and the BJP was due to a high average vote and what proportion was the
result of a favourable distribution of votes? The message is that even if the
INC received the same number of total votes as the BJP it would still win
fewer seats. For the INC to nullify the effects of its distributional disadvan-
tage it must raise its vote substantially above the BJP vote.

7.1   Introduction
The previous chapter analysed the electoral efficiency of the INC and
BJP in terms of their ability to convert votes into seats. A large part of
this ability depends upon the geographical distribution of their vote. An
excessive concentration of the party’s vote in a small area leads to a small
of seats with large majorities. On the other hand, if spread too thinly,
electoral support dissipates resulting in many ‘near misses’, but few elec-
toral successes. This observation leads, in this chapter, to an analysis of
the distribution of the party vote between constituencies and between
states. Within this broad theme, we pursue two topics. Firstly, there is

© The Editor(s) (if applicable) and The Author(s) 2016 129

V.K. Borooah, Votes, Parties, and Seats,
DOI 10.1007/978-3-319-30487-8_7

vk.borooah@ulster.ac.uk
130 V.K. Borooah

the question of concentration. Borrowing an analogy from industrial eco-

nomics, are there differences between the INC and the BJP in the degree
to which their votes and seats are concentrated in the various states and
union territories which produce their votes and seats? The second ques-
tion relates to the unevenness in the distribution of the INC and BJP vote
across the constituencies. In terms of their seat tally, to what extent would
the two parties benefit, or suffer, from a more equal distribution of their
vote? To put matters differently, what proportion of the seats won by the
INC and the BJP was due to a high average vote and what proportion was
the result of a favourable distribution of votes?
The first question is answered in terms of measures of concentra-
tion popular in the industrial economics literature, in particular, the
Herfindahl- Hirschman index (HHI) of concentration. The issue that is
addressed here is vote supply: how much of a party’s total vote is sourced
from different states/constituencies? Then, there is the different, and
conceptually separate, issue of vote shares: what proportion of the total
vote in a state/constituency does a party obtain? We show how the two
issues of vote supply and vote shares are related and arrive at measures
of vote concentration (issue 1) and vote distribution (issue 2). Lastly,
the chapter, using electoral simulations, shows how differences in their
respective vote distributions affect the electoral fortunes of the INC and
the BJP very differently.

7.2   Where the Votes Come from:

The Concentration of Votes by State
One can think of the total number of votes obtained by a party (V) as
being produced by K (k = 1…K) states, with each state producing Vk votes
for the party. If V is excessively concentrated in a few states—that is, the
production of votes for the party is characterised by oligopolistic tenden-
cies—then it will win few seats but with large majorities. On the other
hand, if V is fairly evenly spread over the states—that is, the production
of votes for the party is characterised by competitive tendencies—then it
may again win few seats, this time with small majorities. The optimal geo-
graphical distribution of the total national vote of a party must, therefore,
take regard of both having enough supporters in a state’s constituencies to
comprise a plurality of voters while, at the same time, avoiding concentra-
tion of its total support in just a few states.

vk.borooah@ulster.ac.uk
 The Concentration and Distribution of Votes 131

These considerations raise the question of the degree to which the votes
of the INC and BJP are concentrated in the states. A popular measure of
concentration, used in the industrial economics literature, to measure the
degree of competition in a market, is the HHI.1 Applied to the concen-
tration of a party’s votes across the Indian states, the HHI for party j is
represented by HHIj and defined as:

K
HHI j = ∑ vkj ( )
2
(7.1)
k =1

where: Vj is total number of votes obtained by party j; Vkj is total number

of votes obtained by party j in state k; and vkj = Vkj / V j is state k’s share in
party j’s total vote (k = 1…K). At one extreme, if state k produces all the
votes for party j, then vkj = 1 and HHI j = 1 , which is the maximum value
of the index. At the other extreme, if all the states have an equal share in
the total vote for party j, HHI j = 1 / K which is the minimum value of the
index. Consequently, 1 / K ≤ HHI j ≤ 1 .
Table 7.1 shows: (i) the shares of India’s major states in the total vote
produced by these states; (ii) the shares of India’s major states in the
total INC vote produced by these states; and (iii) the shares of India’s
major states in the total BJP vote produced by these states. The last lines
of Table 7.1 compare the total Lok Sabha vote in the major states with the
total all-India Lok Sabha vote: this shows, for example, that in 2014, the
former was nearly 97 percent of the latter; similarly, the total INC and
BJP vote in the major states were, respectively, 96 and 97 percent of the
corresponding all-India vote.
Of the total vote emanating from the major states in the Lok Sabha
election of 2014, Uttar Pradesh produced 15.1 percent, followed by
West Bengal with 9.6 percent, Maharashtra with 9.1 percent, and Andhra
Pradesh with 9 percent. Table 7.1 also shows that in the 2014 Lok Sabha
election, the BJP did particularly well, and the INC did particularly badly,
in Uttar Pradesh: 20.6 percent of the BJP vote, but only 5.9 percent of
the INC vote, came from this state which produced over 15 percent of the
total (major states) vote. By contrast, in the same election, the INC did
particularly well, and the BJP did particularly badly, in Orissa: 5.4 percent
of the INC vote, but only 2.8 percent of the BJP vote, came from this
state which produced 4 percent of the total (major states) vote.

vk.borooah@ulster.ac.uk
 Table 7.1 The contributions of the major states to the total Lok Sabha votea
2004 2009 2014

Total INC BJP Total INC BJP Total INC BJP

contribution contribution contribution contribution contribution contribution contribution contribution contribution

Andhra 9.4 14.9 3.6 10.4 14.5 2.1 9.0 5.4 2.5
Pradesh
Assam 2.7 3.6 2.9 3.0 3.8 2.6 2.8 4.3 3.3
Bihar 7.7 1.3 5.1 6.0 2.2 4.5 6.7 2.9 6.3
Chhattisgarh 1.9 2.9 4.1 2.1 2.8 5.1 2.3 4.6 3.6
Gujarat 4.0 6.7 8.6 4.3 6.7 10.8 4.8 8.3 9.2
Haryana 2.1 3.4 1.7 2.0 3.0 1.3 2.1 2.6 2.4
Himachal 0.7 1.3 1.3 0.7 1.1 1.8 0.6 1.2 1.0
Pradesh
J&K 0.6 0.6 0.6 0.6 0.6 0.6 0.7 0.8 0.7
Jharkhand 2.5 2.0 3.7 2.3 1.2 3.3 2.4 1.7 3.1
Karnataka 6.6 9.3 10.5 6.1 8.2 13.5 5.8 12.3 8.0
Kerala 4.0 4.9 1.9 4.0 5.7 1.3 3.4 5.4 1.1

vk.borooah@ulster.ac.uk
Madhya 4.9 6.3 10.6 4.8 6.9 11.2 5.5 10.1 9.6
Pradesh
Maharashtra 9.0 8.2 9.3 9.2 6.4 8.9 9.1 8.6 8.0
Orissa 4.5 6.9 3.9 4.4 5.2 4.0 4.0 5.4 2.8
Punjab 2.7 3.5 1.3 2.9 4.7 1.6 2.6 4.5 0.7
Rajasthan 4.6 7.2 10.2 4.4 7.5 8.7 5.1 8.0 9.0
Tamil Nadu 7.6 4.1 1.7 7.5 4.1 0.9 7.6 1.7 1.3
Uttar 14.1 6.4 14.1 13.7 9.0 12.8 15.1 5.9 20.6
Pradesh
Uttarakhand 0.7 1.0 1.3 0.8 1.2 1.4 0.8 1.5 1.5
West Bengal 9.8 5.4 3.6 10.6 5.1 3.5 9.6 4.8 5.2
 Total 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0
Total vote 379,076,480 99,695,696 83,489,256 403,313,440 112,695,216 75,491,856 536,140,896 102,771,824 166,309,568
from major
states
Total 389,779,776 103,408,952 86,371,560 417,158,656 119,110,824 78,435,352 553,801,792 106,938,240 171,657,552
all-India vote

Source: Own calculations from Election Commission of India data

a
The ‘major states’ are: Andhra Pradesh, Assam, Bihar, Chhattisgarh, Gujarat, Haryana, Himachal Pradesh, Jammu and Kashmir, Jharkhand, Karnataka, Kerala, Madhya
Pradesh, Maharashtra, Orissa, Punjab, Rajasthan, Tamil Nadu, Uttarakhand, Uttar Pradesh, and West Bengal

vk.borooah@ulster.ac.uk
 134

Table 7.2 Values of the Herfindahl, entropy, and dissimilarity indices for the vote contribution of major statesa
V.K. Borooah

Total vote in major states Total INC vote in major states Total BJP vote in major states

Year Number HHI Entropy Dissimilarity HHI Entropy Dissimilarity HHI Entropy Dissimilarity
of major
states

1989 17 0.085 2.562 0.262 0.086 2.554 0.268 0.145 2.159 0.498
1991 17 0.086 2.574 0.261 0.083 2.596 0.239 0.115 2.378 0.368
1996 17 0.084 2.609 0.285 0.086 2.617 0.245 0.119 2.395 0.399
1998 17 0.085 2.606 0.274 0.089 2.571 0.292 0.11 2.458 0.356
1999 17 0.085 2.601 0.281 0.083 2.619 0.263 0.098 2.514 0.341
2004 20 0.075 2.743 0.292 0.072 2.777 0.263 0.081 2.702 0.334

vk.borooah@ulster.ac.uk
2009 20 0.075 2.745 0.286 0.072 2.782 0.253 0.085 2.66 0.361
2014 20 0.076 2.744 0.285 0.069 2.801 0.245 0.094 2.627 0.36

Source: Own calculations from Election Commission of India data

a
Note that the states of Chhattisgarh, Jharkhand, and Uttarakhand did not exist prior to the 2004 Lok Sabha election
 The Concentration and Distribution of Votes 135

Table 7.2 shows values of the HHI (defined by Eq. 7.1), for each Lok
Sabha election since 1989, with the major states as the vote-generating
units. The values are shown with respect to: (i) the total vote emanating
from the major states; (ii) the total INC vote emanating from the major
states; and (iii) the total BJP vote emanating from the major states. This
table shows that for every election between 1989 and 2014, the BJP had
associated HHI values which were greater than the corresponding HHI
values for the INC: this implied that in the context of the major states,
the BJP’s votes were more concentrated than those of the INC’s. This is
a reflection of the fact that the INC, as the older party, has a significant
presence in parts of India—like Assam, Kerala, and Jammu and Kashmir—
where the BJP, until recently, has been all but invisible.
Also shown are the values with respect to two other indices. The first of
these is Shannon’s entropy index defined as:

( )
E = − ∑vkj log vkj (7.2)
k =1

And the second of these is the dissimilarity index defined as:

1  K  j 1 
D= ∑  vk − K  
2  k =1 
(7.3)


If a state’s vote share, with respect to party j, is equal to 1 (meaning that

party j gets all its votes from that state) so that, say, v1j = 1, v2j … = vKj = 0 ,
then E = 0, which is its minimum value, and D = K-1 which is its maximum
value; on the other hand, if all the states have equal shares in party j’s
total vote so that, v1j = v2j = .... = vKj = 1 / K , then E = log (1 / K ) , which
is its maximum value, and D = 0, which is its minimum value. The values
of these indices confirm the fact that in the context of the major Indian
states, the BJP vote is more concentrated than that of the INC: for every
election between 1989 and 2014, the value of the entropy index (E in
Eq. 7.2) is higher—and the value of the dissimilarity index (D in Eq. 7.3)
is lower—for the INC compared to its value for the BJP.

vk.borooah@ulster.ac.uk
136 V.K. Borooah

7.2.1  The Effective Number of States

As Table 7.1 shows, over 536 million votes cast in the 2014 Lok Sabha elec-
tion emanated from the 20 major Indian states. The contributions from
the different states, however, varied considerably, from Uttar Pradesh’s
15.1 percent of the total vote to Himachal Pradesh’s 0.6 percent. This
makes the obvious point that in terms of ‘producing’ votes, not all states
are equal; it also raises, by way of corollary, a query about the effective
number of states in the political system when the 20 major states were
adjusted by their vote shares.
This concept of an ‘effective number’ was first applied to political par-
ties (see Dunleavy and Bouchek 2003). Suppose there are N political par-
ties in the system with each party receiving different vote shares. Some of
these vote shares might be so small, and others so large, that effectively,
there are fewer than N political parties in the electoral system. Laakso
and Taagepera (1979) suggested that the effective number of parties, N*,
could be computed as the inverse of the HHI as:

N * = 1 / HHI (7.4)

where: HHI is the HHI computed from the vote shares of the N par-
ties. If all the N parties received the same share of the total vote, 1/N,
HHI = 1/N, and N* = N: the effective number of parties is same as the
total number of parties. If one party obtained the entire vote, HHI = 1 and
N* = 1: effectively, the electoral system consists of a single party. In general,
the greater the concentration of votes (larger the HHI value), the smaller
will be the number of effective parties.
These ideas can equally be applied to the Indian states which contribute
unequally to the total amount of votes they generate. Consequently, the
effective number of (major) states is smaller than the actual number, 20,
of states. How much smaller can be determined by applying the Laakso
and Taagepera (1979) formula of Eq. 7.4? So, in the 2014 election, the
HHI values for the INC and the BJP were, respectively, 0.069 and 0.094:
consequently, the effective number of states for the INC and the BJP
were, respectively, 14.5 (=1/0.069) and 10.6 (=1/0.094). The effective
number of states differs between the two parties because the concentra-
tion of their votes, within the major states, is different: the number of
effective states was larger for the INC—with a smaller concentration of its
vote—than for the BJP with a greater vote concentration.

vk.borooah@ulster.ac.uk
 The Concentration and Distribution of Votes 137

7.3   Inequality in the Inter-Constituency

Distribution of Party Vote Shares
The previous section addressed the question of vote supply in the context
of the major Indian states with particular reference to the concentration of
the national vote. In that section, the key variables were the proportions of
the total vote, the total INC vote, and the total BJP vote that were sourced
from the different states: for example, in the 2014 Lok Sabha election, 20.6
percent of the BJP vote was sourced from Uttar Pradesh while Madhya
Pradesh supplied 10 percent of the INC vote. This section turns to the
separate, but related, question of party vote shares, namely, the proportion
of the total vote in a particular geographical area (state or constituency)
that accrued to a particular political party. It is relatively straightforward
to show that vote supply (previous section) and vote shares (this section),
though conceptually different, are, in fact, empirically related.
Suppose that of the total of Vk votes in an area k (e.g., state or constitu-
ency), Vkj is in favour of party j. Then the vote share of party j in area k
K K
(k = 1…K) is represented by vkj = Vkj / Vk . Let V = ∑Vk and V j = ∑Vkj rep-
k =1 k =1

resent, respectively, the total (national) vote (over all parties) and party j’s
total (national) vote. Then the vote share can be decomposed in the terms
of vote supply as follows:

( )(
vk = Vkj / Vk = Vkj / Vk V / V j V j / V )( )
(
= Vk / V
j j
) (V / V ) (V
k
j
) (
/ V = g / nk v j
j
k ) (7.5)

where: gkj = Vkj / V j is the proportionate contribution that area k makes to

the party j’s national vote; nk = Vk / V is the proportionate contribution
that area k makes to the total (national) vote; and v j = V j / V is party j’s
share of the national vote.
By way of a numerical example, suppose that k represents Uttar Pradesh
and that j represents the BJP. Now, from Table 7.1, for the Lok Sabha elec-
tion of 2014, gkj = 20.6, nk = 15.1,and vkj = 31 , implying, from Eq. 7.5 that
the BJP obtained 42.3 percent of the total vote in Uttar Pradesh.
In this section, we measure inequality in the distribution of inter-­
constituency vote shares of the INC and the BJP in the major states, and
having done that, in the section following, we decompose inter-­constituency

vk.borooah@ulster.ac.uk
138 V.K. Borooah

inequality by the states to which the constituencies belong. The first

exercise, of inequality measurement, will suggest a relationship between
electoral popularity and electoral inequality while the second exercise,
of inequality decomposition, will evaluate how much of overall inter-­
constituency inequality in vote shares can be explained by the aggregation
of constituencies by the state.
The inequality measure used, both for measurement and for decompo-
sition, belongs to the family of entropy measures. The logic of the entropy
measure is taken from information theory. Suppose that a party’s vote
share, v, is a random variable which takes values, v1, v2, …, vN, over N con-
N
stituencies, with probabilities, p1 , p2 ,…, pN , 0 ≤ pi ≤ 1, ∑ pi = 1 . Now the
i =1
information content of a message that the random variable v has taken an
unusual value is greater than that of a message that is has taken a more
commonly observed value. Hence, the information content, hi, of v tak-
ing a specific value, vi, is a decreasing function of pi, the probability of
observing that value, so that hi = h ( pi ) is a decreasing function of the
pi. Also, since the values assumed by v are assumed independent of each
other, the information content of the joint occurrence of two values,
say, v = vr and v = vs , is the sum of the individual information contents:
h ( vr ,vs ) = h ( vr ) + h ( vs ) . A decreasing function that satisfies this property
is h ( pi ) = log (1 / pi ) = − log ( pi ) .
A measure of the expected amount of information or entropy in a sys-
tem, defined by the values of a random variable v and the associated prob-
abilities, is given by (Renyi, 1965):

N N
E = ∑ pi h ( pi ) = − ∑ pi log ( pi ) (7.6)
i =1 i =1

The maximum value of E in Eq. 7.6—and also in Eq. 7.2—occurs when

the values are equally likely so that p1 = p2 = … = pN = 1 / N and a measure
of the disorder of the system is the extent to which the expected value falls
below this maximum:

N N N
I = ∑ (1 / N ) × h (1 / N )− ∑ pi × h ( pi ) = ∑  log ( pi ) − log (1 / N ) (7.7)
i =1 i =1 i =1
 
maximum value observed valuue

vk.borooah@ulster.ac.uk
 The Concentration and Distribution of Votes 139

The larger the value of I in Eq. 7.7, the greater will be the disorder or
inequality in the system. If we set the probabilities to the observed vote
shares, so that pi = vi , i = 1…N, and let v represent the mean vote share,
we can obtain Theil’s (1967) Mean Logarithmic Deviation (MLD) index
as:

 N 
MLD =  ∑ log(v / vi  / N (7.8)
 i =1 

Table 7.3 shows the MLD and Gini values for the INC and the BJP
in respect of the inter-constituency distribution of their vote shares for
every Lok Sabha election between 1989 and 2014 with higher values of
both indices representing higher inequality levels.2 These values show that
inequality in the distribution of INC vote shares was at a low in 1989;
thereafter it rose steadily, reaching a peak in 1998; it fell in 1999, remained
fairly steady till 2009 but then rose sharply in 2014. By contrast, the inter-­
constituency distribution of the BJP vote was highly unequal in 1989 after
which it fell reaching a low in 1999; then it peaked in 2009 before falling
back in 2014.
Set alongside the values of the MLD and Gini indices are the party vote
shares. These make clear (see Figs. 7.1 and 7.2) that in general, when-
ever overall support for a party was high, inequality in the distribution of
the party’s vote share, between the constituencies, was low: the INC in
1989, 1991, and 2009 and the BJP in 1998, 1999, and 2014. Conversely,

Table 7.3 Inequality in the distribution of constituency vote shares

INC BJP

MLD Gini Vote share MLD Gini Vote share

1989 0.042 0.155 39.5 0.563 0.449 11.4

1991 0.110 0.216 35.7 0.381 0.412 20
1996 0.302 0.299 28.8 0.500 0.425 20.3
1998 0.399 0.353 25.8 0.131 0.211 25.6
1999 0.190 0.252 28.3 0.085 0.187 23.8
2004 0.250 0.267 26.5 0.133 0.238 22.2
2009 0.142 0.211 28.6 0.513 0.433 18.8
2014 0.457 0.394 19.5 0.135 0.233 31.3

Source: Own calculations from Election Commission of India data

vk.borooah@ulster.ac.uk
140 V.K. Borooah

Fig. 7.1 INC vote shares and vote share inequality: 1989–2014 (Source: Own
calculations from Lok Sabha election data)

Fig. 7.2 BJP vote shares and vote share inequality: 1989–2014 (Source: Own
calculations from Lok Sabha election data)

vk.borooah@ulster.ac.uk
 The Concentration and Distribution of Votes 141

­ henever overall support for a party was low, inequality in the distribution
w
of the party’s vote share, between the constituencies, was high: the INC in
1998 and 2014 and the BJP in 1989, 1991, 1996, and 2009.

7.4   The Decomposition of Inequality in Vote

Shares
Inequality in a party’s vote share across the different constituencies leads
one to ask: what ‘explains’ such inequality? Is it due to the fact that con-
stituencies are segmented into states, with different states embodying dif-
ferent ‘political’ cultures? In that case, we would expect that some of the
observed inequality can be explained by differences between states because
constituencies in some states offer, on average, a lower vote share to that
party compared to constituencies in other states. But not all of inequality
in vote shares can be explained by differences between states—some of the
observed (overall) inequality will be due to the fact that there is inequality
in constituencies within the same state because the party does not receive
the same vote share from all constituencies within a particular state.
Of course, one need not subdivide constituencies by state—one could,
equally well, have subdivided them by region (e.g., North, South, East,
West, and Central) or by their level of income (e.g., low-income, medium-­
income, and high-income states). Whenever, and however, one subdivides
households, there are always two sources of inequality: between-group and
within-group. The method of inequality decomposition attempts to sepa-
rate (or decompose) overall inequality into these two constituent parts:
between-group inequality and within-group inequality. When the decom-
position is additive, overall inequality can be written as the sum of within-­
group inequality and between-group inequality:

I = A + B
  
overall ineqality within group inequality between group inequality

When inequality is additively decomposed, then one can say that the
basis on which the constituencies were subdivided (say, by state) contrib-
uted [(B/I) × 100] percent to overall inequality in a party’s vote shares,
the remaining inequality, [(A/I) × 100] percent, being due to inequality
within the states. If one subdivided the constituencies by income (say,
three groups) and by state (20 major states), so that one had 60 c­ ategories,

vk.borooah@ulster.ac.uk
142 V.K. Borooah

then by additively decomposing inequality, as above, one could say that

income and state collectively accounted for [(B/I) × 100] percent of over-
all inequality in the vote shares of a party, the remaining inequality being
due to inequality within the 60 categories. So, inequality decomposition
provides a way of analysing the extent to which inter-constituency inequal-
ity in a party’s vote share can be ‘explained’ by a constellation of factors.
More formally, suppose that the total of N constituencies is divided
into M mutually exclusive states groups with Nm (m = 1…M) constituen-
cies in each state. Let v = {vi } and v m = {vi } represent the vector of vote
shares for a party in, respectively, all the constituencies (i = 1…N) and in
the constituencies in state m. Then an inequality index I(v; N) defined
over this vector is said to be additively decomposable if:

M
I ( v;N ) = ∑I ( v m ;N m ) wm + B = A + B (7.9)
m =1

where: I(v; N) represents the overall level of inequality; I(vm; Nm) repre-
sents the level of inequality within state m; A—expressed as the weighted
sum of the inequality in each state, wm being the weights—and B repre-
sent, respectively, the within-group and the between-group contribution to
overall inequality.
If, indeed, inequality can be ‘additively decomposed’ along the lines of
Eq. 7.9 above, then, as Cowell and Jenkins (1995) have argued, the pro-
portionate contribution of the between-group component (B) to overall
inequality is the income inequality literature’s analogue of the R2 statistic
used in regression analysis: the size of this contribution is a measure of the
amount of inequality that can be ‘explained’ by the factor (or factors) used
to subdivide the sample.
Only inequality indices which belong to the family of Generalised
Entropy Indices are additively decomposable (Shorrocks, 1980). These
indices are defined by a parameter θ, and when θ = 0, the weights are the
constituency shares of the different states (i.e., wm = N m / N ); since the
weights sum to unity, the within-group contribution A of Eq. 7.9 is a
weighted average of the inequality levels within the groups. When θ = 0,
the inequality index is Theil’s MLD, defined in Eq. 7.8 of the previous
section, which, because of its attractive features in terms of the interpreta-
tion of the weights, is used in this chapter to decompose inequality in a
party’s vote shares.

vk.borooah@ulster.ac.uk
 The Concentration and Distribution of Votes 143

Figure 7.3 shows the within-state and between-state contributions to

inter-constituency inequality in INC and BJP vote shares. This shows that
for several elections—1989, 1996, 1999, 2009, and 2014—the between-­
state contribution to inequality in the distribution of BJP vote shares
exceeded 60 percent, and this contribution was not less than 50 percent
for any election. Except for the Lok Sabha election of 1998, the between
state contribution to inequality in the INC vote share was always lower
than the corresponding contribution for the BJP. The overall consensus
from this decomposition is that for both parties, over half of inequality in
the distribution of inter-constituency vote shares could be explained by
the location of the constituencies in different states.

7.5   The Effect of the Distribution of Votes

on the Number of Seats Won

We hypothesise that the number of seats won (S) by a party at a Lok Sabha
election, given the number of seats contested, depends upon its mean vote
(μ) and the degree of inequality (I) in the distribution of its vote both

Fig. 7.3 Between income inequality as a proportion of total inequality, decom-

position by major states (Source: Own calculations from Lok Sabha election data)

vk.borooah@ulster.ac.uk
144 V.K. Borooah

computed over the constituencies in which it fielded candidates.3 More

formally:

 + ?
S = f µ,I  (7.10)
 

7.5.1  Simulation A: The Equal Distribution of Votes

It is impossible to specify a priori the distribution of votes (as encapsu-
lated in the value of I in Eq. 7.10) which, in the face of a given total of
votes, will maximise the number of seats won. So, in order to investigate
the separate contributions of μ and I to the number of seats won by the
BJP and INC, we ran a simulation in which, with the mean vote of each
party unchanged, the inter-constituency vote distribution of the two par-
ties was rendered the same; we then examined the number of seats each
party would have won under this scenario. The simplest distributional uni-
formity was to assume that each party’s total vote was equally distributed
between the constituencies it contested, and Table 7.4 shows the number

Table 7.4 BJP and INC seats for eight Lok Sabha elections, 1989–2014, under
an equally distributed scenario
BJP INC

Year Seats Seats under Votes per Seats Seats under Votes per seat
won equal seat won equal contested
distribution contested distribution

2014 282 278 401,075 44 18 230,465

2009 116 40 181,144 206 183 270,707
2004 138 131 237,285 145 130 247,983
1999 182 165 255,345 114 109 227,638
1998 182 147 242,954 141 50 199,394
1996 161 72 144,884 140 110 183,375
1991a 120 56 116,821 244 276 200,628
1989b 85 34 151,873 197 207 233,126

Source: Own calculations from Election Commission of India data

Including delayed elections in Punjab held in 1992
a

Elections were not held in Assam because electoral rolls were incomplete.
b

vk.borooah@ulster.ac.uk
 The Concentration and Distribution of Votes 145

of seats the party would have won or lost under this ‘equally distributed’
scenario.
In the six elections after (and including) the 1996 Lok Sabha elections,
Table 7.4 shows that both parties would have lost seats under an equal
distribution scenario. In the 2014 Lok Sabha elections, the BJP, with its
31.3 percent share of the vote (which translated to 401,075 votes per con-
stituency contested), won 282 seats. If it had received exactly 401,075 votes
in each of the 428 constituencies it contested in 2014, it would have won
278 seats, or in other words, the unequal distribution of its vote across the
428 seats it contested enabled it to win an additional four seats. Similarly,
in the 2014 Lok Sabha elections, the INC, with its 19.3 percent share of
the vote (which translated to 230,465 votes per constituency contested),
won 44 seats. If it had received exactly 230,465 votes in each of the 464
constituencies it contested in 2014, it would have won only 18 seats, or
in other words, the unequal distribution of its vote across the 464 seats it
contested enabled it to win an additional 26 seats.
The effect of distribution, on the number seats won, varied by elec-
tion. As Table 7.4 shows, the effect of inequality in the inter-constituency
distribution of the BJP vote, on the number of seats it won, was greatest
in 1989, 1991, 1996, and 2009. In these elections, the distribution of its
vote helped it win a large number of additional seats: 51 seats in 1984;
64 seats in 1991; 89 seats in 1996; and 76 seats in 2009. By contrast, the
effect of distribution on the number of seats won by the INC was more
muted. The most marked effect was in 1998 when its vote distribution
across the constituencies helped it win an additional 91 seats; apart from
this particular election, the INC vote distribution, compared to the BJP
vote distribution, added far fewer seats to what it would have won with an
equal distribution of votes across the constituencies.

7.5.2  
Simulation B: Equal Number of Votes Received
In the second simulation (simulation B), it was assumed that the INC and
the BJP received the same number of votes nationally—which was the
average of their respective national vote—but that the distribution of the
vote across the constituencies remained unchanged for both parties. So,
for example, for the 2014 Lok Sabha election, it was assumed that both
the INC and the BJP received 139,297,888 votes—which was an aver-
age of the INC’s 106,938,240 and the BJP’s 171,657,552 votes—and

vk.borooah@ulster.ac.uk
146 V.K. Borooah

that in each of the constituencies contested by them, their respective votes

increased or decreased proportionately to the change in their national votes.
In other words, in the Lok Sabha elections of 2014, the INC vote was
marked up by multiplying the number of votes it received, in each of the
464 constituencies it contested, by1.3, and the BJP vote was marked down
by multiplying the number of votes it received, in each of the 428 constit-
uencies it contested, by 0.81.4 The implication of this was that the distri-
bution of the INC and the BJP vote remained unchanged: any inequality
index like the Gini or the MLD would yield the same value on both the
old and new set of INC—and on the old and new set of BJP—constitu-
ency votes.
Table 7.5 shows that in 2014, even with the INC and the BJP receiv-
ing the same number of votes—with the INC increasing its votes by 32.4
million votes with the BJP’s vote falling by an equal amount—the INC
would have won 122 seats compared to the BJP’s 229. As a result of these
extra 32.4 million votes, the INC would have gained won only 78 seats
(44 to 122) while the loss of 32.4 million votes would have deprived the
BJP of only 53 seats.
Suppose that if distribution did not matter, the two parties, which
shared the vote equally between them, would have also won an equal
number of seats: for the 2014 Lok Sabha election, this would have been
163 seats each.5 So, in the 2014 Lok Sabha election, for reasons of vote
distribution, the INC, which under this simulation was predicted to win
122 seats (see Table 7.5), under-performed by 41 seats, or by 25 percent
of its equal division of 163 seats, and the BJP, which under this simulation
was predicted to win 229 (see Table 7.5) seats over-performed by 66 seats,
or by 40 percent of its equal division of 163 seats.
In the 2009 Lok Sabha election, the equal division of votes was
98,773,088 which represented a shortfall for the INC (which received
119,110,824 votes in this election), and a bonus for the BJP (which
received 78,435,352 votes in this election) of 20,337,736 votes. Under
this scenario, we might have expected both parties to each win 161 seats.
However, it turned out that the INC would have won only 100 seats (61
fewer than expected 161 seats), and the BJP would have won 184 seats
(23 more than the expected 161 seats). So, in the Lok Sabha election of
2009, for reasons of vote distribution, the INC, which (under this simula-
tion) was predicted to win 100 seats under-performed by 61 seats, or by 38
percent of its equal division of 161 seats, and the BJP, which (again under

vk.borooah@ulster.ac.uk
 Table 7.5 BJP and INC seats for eight Lok Sabha elections, 1989–2014, under an equal voters’ scenario
Actual votes Equal votes Equal INC BJP
seats

INC BJP INC & BJP INC & Seats Predicted Seats Predicted seats
BJP won seats won won won

1989 118,894,704 34,171,476 76,533,088 141 197 55 85 133

1991 102,059,792 55,953,668 79,006,728 183 244 144 121 204
1996 96,034,448 67,697,336 81,865,888 150 139 80 161 211
1998 95,111,128 94,266,192 94,688,656 162 141 140 182 182
1999 103,120,328 86,562,208 94,841,264 148 114 84 182 210

vk.borooah@ulster.ac.uk
2004 103,408,952 86,371,560 94,890,256 142 145 122 138 161
2009 119,110,824 78,435,352 98,773,088 161 206 100 116 184
2014 106,938,240 171,657,552 139,297,888 163 44 122 282 229

Source: Own calculations from Election Commission of India data

The Concentration and Distribution of Votes
147
148 V.K. Borooah

this simulation) was predicted to win 184 seats over-performed by 24 seats,

or by 15 percent of its equal division of 161 seats.
Figure 7.4 shows the under- and over-performance rates of the INC
and the BJP for every Lok Sabha election since 1989. The important point
that emerges from this figure is that the INC has always under-performed
as a party: it has always failed to translate an equal division of votes between
it and the BJP into an equal division of seats. By contrast, except for the
1989 and 2004 Lok Sabha elections, the BJP has always over-performed: it
has succeeded in translating an equal division of votes between it and the
INC into a (favourable) unequal division of seats.

7.6   Concluding Remarks

This chapter highlighted the importance of the distribution of a party’s
votes in determining the number of seats it wins under a FPTP system.
The ominous message that the results of this chapter contain for the INC

Fig. 7.4 INC and BJP under- and over-performance with respect to seats when
each received an equal number of votes* (*Negative and positive values represent,
respectively, under- and over-performance). Source: Own calculations from
Election Commission of India data

vk.borooah@ulster.ac.uk
 The Concentration and Distribution of Votes 149

is that even it received the same number of total votes as the BJP, it would
still, because of differences between them in their vote distributions, win
fewer seats. For the INC to nullify the effects of its distributional disad-
vantage, it must raise its electoral popularity substantially above that of
the BJP.
Or else, it must improve its vote distribution. As the previous chapters
have pointed out, the BJP enjoys a considerable advantage over the INC
in the 204 constituencies in the Hindi-speaking states while the INC does
not enjoy, to the same degree, advantage over the BJP in the non-Hindi-­
speaking states. This is an area that the INC needs to redress, either on its
own or, more plausibly, with strategic alliances with like-minded parties.

Notes
1. See Hirschman (1964).
2. The Gini coefficient is defined in Chap. 2.
3. While the number of consistencies a party contests sets an upper limit to the
number of seats it can win, it does not follow that that the more seats it
contests, the larger will be the number of seats it wins: in the Lok Sabha elec-
tion of 1999, the INC contested 453 constituencies but won only 114 seats,
while the BJP contested 339 constituencies and won 182 seats.
4. 1.3 = 139, 297,888/106,938,240 and 0.81 = 139, 297,888/171,657,552.
5. The average of the 282 and 44 seats won, respectively, by the BJP and INC.

References
Cowell, F. A., & Jenkins, S. P. (1995). How much inequality can we explain? A
methodology and an application to the United States. Economic Journal, 105,
421–430.
Dunleavy, P., & Bouchek, F. (2003). Constructing the number of parties. Party
Politics, 9, 291–315.
Hirschman, A. O. (1964). The paternity of an index. The American Economic
Review, 54(5), 761.
Laakso, M., & Taagepera, R. (1979). Effective number of parties: A measure with
application to West Europe. Comparative Political Studies, 12, 3–27.
Renyi, A. (1965). On the foundations of information theory. Review of the
International Statistical Institute, 33, 1–14.
Shorrocks, A. F. (1980). A class of additively decomposable measures. Econometrica,
50, 613–625.
Theil, H. (1967). Economics and information theory. Amsterdam:
North-Holland.

vk.borooah@ulster.ac.uk
 CHAPTER 8

Conclusions

Abstract Borooah concludes that although India’s experience of coali-

tion governments, or minority governments with outside support, has
been largely unfavourable, such governments are likely to be the main
feature of Indian politics in future years. Coalition government in India
is provided in the context of the FPTP system allied to strong regional
parties. A combination of the electoral system and the strength of the
regional parties means that such parties can exercise inordinate influence
on national politics, through the strength of their presence in the Lok
Sabha, even if this strength is based on a localised vote which is but a sliver
of the national vote. Consequently, while one might be sanguine about
the prospects for Indian democracy, the prospect of effective government
is less certain.

The foundations of this book lay in a set of data which recorded the details
of the election result for each candidate, for all the constituencies, in every
Lok Sabha General Election from 1962 to 2014. The edifice built upon
this foundation, and discussed in the preceding chapters, was the result
of interrogating these data. The central purpose of this interrogation was
to give shape to the notion of ‘electoral efficiency’ by which is meant the
capacity of a party to convert votes into parliamentary seats. Parliamentary
elections in India—and also elections to its state Assemblies—are con-
ducted under the FPTP system: a single representative for each of 543
constituencies is elected—on the basis of obtaining the largest number of

© The Editor(s) (if applicable) and The Author(s) 2016 151

V.K. Borooah, Votes, Parties, and Seats,
DOI 10.1007/978-3-319-30487-8_8

vk.borooah@ulster.ac.uk
152 V.K. BOROOAH

votes of all the candidates contesting that constituency—as a member of

the Lok Sabha for that constituency. The disjoint, under the FPTP elec-
toral system, between the votes obtained by a party and the seats won by
it frequently causes surprise, sometimes bordering on consternation. The
primary purpose of this book was to throw light on this relationship for
Indian parliamentary elections.
Given that India’s two main political parties—the INC and the BJP—
receive, between them, over half the national parliamentary vote, the
analysis in this book is restricted to a comparison of the relative electoral
efficiencies of these two parties. This leads to a further constraint. The BJP
made its electoral debut in the 1984 Lok Sabha elections, winning just two
seats, but really got into its stride in the 1989 Lok Sabha elections when it
won 85 seats. Consequently, a great deal, but not all, of the analysis in this
book is a comparison of the INC and the BJP and, consequently, restricted
to the eight Lok Sabha elections of 1989, 1991, 1996, 1998, 1999, 2004,
2009, and 2014.
The tenor of this book was analytical, based upon a rigorous exami-
nation of the data. In the process, it drew upon the methodology of
economics and statistics to shed light on electoral outcomes in India.
Chapter 3 used systems estimation techniques to predict the probabilities
of the INC and BJP winning elections—and Chap. 5 used systems esti-
mation techniques to predict the vote shares of the INC and the BJP—in
constituencies contested by both parties; Chap. 4 used Bayesian methods
to analyse the issue of anti-incumbency; Chap. 6 refined the concept of
the ‘Cubic Law of Elections’ to develop the concept of the ‘amplifica-
tion coefficient’ which amplifies votes into seats; and Chap. 7 measured
vote concentration and vote inequality with particular reference to the
decomposition of inequality and carried out two significant simulations
with regard to the inter-constituency distribution of the INC and BJP
vote.
A consistent feature of the results was that the BJP was better at con-
verting votes into seats than the INC. Ceteris paribus, it was more likely
to win marginal seats (Chap. 3); its predicted vote share was higher than
that for the INC in seats contested by both parties (Chap. 5); it required
a smaller vote share than the INC to win the same number of seats (Chap.
6); and because it dominated the Hindi-speaking states, its vote distribu-
tion was much more favourable to winning seats than the INC: if the two
parties received the same number of votes nationally, the BJP would win
more seats than the INC.

vk.borooah@ulster.ac.uk
 CONCLUSIONS 153

None of the analysis presented in this book is intended to imply that the
INC can never win more seats than the BJP. What the analysis does sug-
gest is that compared to the BJP, it will have to do much better at the polls
to obtain a comparable result in terms of seats. The key to this result lies
in the 204 constituencies in the seven Hindi speaking (HS) states of Bihar,
Chhattisgarh, Jharkhand, Madhya Pradesh, Rajasthan, Uttarakhand, and
Uttar Pradesh. The advantage that the BJP has over the INC in these
states is not nullified by the advantage that the INC has over the BJP in
constituencies in the non-HS states. To put it in perspective, two-thirds of
the nearly 107 million votes obtained by the INC in 2014 were from the
306 constituencies it contested in the non-HS states and one-third came
from the 158 constituencies it contested in the HS states; for the BJP,
on the other hand, 48 percent of its total vote in 2014 was from the 236
constituencies it contested in the non-HS states and 52 percent came from
the 192 constituencies it contested in the HS states.
And yet, on this book’s analysis, single-party majority governments are
unlikely to be seen frequently. The height of the ‘Modi wave’ in 2014
was considerably lower than that of the ‘Indira wave’ in 1971 or the
‘Rajiv wave’ of 1984. The strength of regional parties—in Tamil Nadu,
Andhra Pradesh, West Bengal, and Uttar Pradesh—ensures that elections
are no longer a zero-sum game between two parties. So, notwithstanding
the current BJP majority government, it is difficult not to concur with
Rudolph and Rudolph (2002) when they suggest that coalition govern-
ment has come to stay in India.
It is unlikely that single-party government will be provided by any party
other than the INC or the BJP. For either to win a majority in the Lok
Sabha, they would need to win at least 30 percent of the national vote,
probably more for the INC. Since, between them, they received, on aver-
age, 50–55 percent of the post-1984 national vote, this would require the
vote of the losing party to collapse to about 20 percent (which is close to
the INC’s vote share of 19.5 percent in the 2014 Lok Sabha elections).
This, in turn, implies a large dose of disillusionment with the losing party
which, given that the INC vote share of 19.5 percent, alluded to above,
followed on the heels of its shambolic United Progressive Alliance (UPA)
government of 2009-14, is unlikely to be repeated.
An anxiety to which this prognostication gives rise is that India’s expe-
rience of coalition governments, or minority governments with outside
support, has been largely unfavourable. In terms of stability, only three
governments—The National Democratic Alliance (NDA) government

vk.borooah@ulster.ac.uk
154 V.K. BOROOAH

formed after the Lok Sabha elections of 1999, and the UPA-I and UPA-II
formed after the Lok Sabha elections of, respectively, 2004 and 2009—
completed their five-year parliamentary terms. The others have been
short-lived, either because of rivalries for the job of prime minister (the
1977 Janata Party government) or through constituent parties withdraw-
ing support over ‘issues’ (the BJP withdrawing its support to the 1989
‘National Front’ government and AIADMK withdrawing its support to
the 1998 [BJP-led] NDA government). At the same time, stability per se
has not been a guarantor of good government: the UPA-II government
(2009–2014) was, in its later stages, mired in scandal as coalition ministers
exploited their ministerial positions to make money.
The worrying feature of coalition government in India is that it is pro-
vided in the context of the FPTP system allied to strong regional parties. A
combination of the electoral system and the strength of the regional par-
ties means that such parties can exercise inordinate influence on national
politics, through the strength of their presence in the Lok Sabha even if
this strength is based on a localised vote which, furthermore, is but a
sliver of the national vote. After the 2014 Lok Sabha elections, six regional
parties—the All India Anna Dravida Munnetra Kazhagam, the All India
Trinamool Congress, the Biju Janata Dal, the Shiv Sena, the Telugu Desam
Party, and the Telangana Rashtra Samithi—between them held 136 of the
543 (25 percent) seats in the 16th Lok Sabha on the strength of just 14
percent of the national vote. Under plausible scenarios, any one of these
parties, with such parliamentary strength, would be in a position to topple
a government. So, while one might be sanguine about the prospects for
democracy in India, the prospect of effective government following on the
heels of Lok Sabha elections is less certain.

“King rules or barons rule

The strong man strongly and the weak by caprice
They have but one law, to seize the power and keep it,
And the steadfast can manipulate the greed and lust of others,
The feeble is devoured by his own”.
(T.S. Eliot, Murder in the Cathedral)

REFERENCE
Rudolph, S. H., & Rudolph, L. I. (2002). New dimensions in Indian democracy.
Journal of Democracy, 13, 52–66.

vk.borooah@ulster.ac.uk
 INDEX

A below poverty line (BPL), 20

Aam Aadmi Party, 11 Besley, T., 26
All India Anna Dravida Munnetra Bharatiya Janata Party (BJP), 4, 22
Kazhagam (AIADMK), 6, 116, amplification coefficient, 120–3
154 Bayes factor calculations, 86, 88
All India Majlis-e-Ittehadul electoral performance, 61–8
Muslimeen, 11 equal voters’ scenario, 147
All India United Democratic Front, 11 incumbency/non-incumbency
amplification coefficient, 119–20 performance, 83
Anderson, P., 11, 12 vs. INC, vote shares, 98–101
anti-incumbency, 77, 78 inverse Bayes factor calculations, 91
Armed Forces (Special Powers) Act, inverse risk ratio (IRR), 91, 92
12, 14n18 8th Lok Sabha, 117
Article 324(2), Election Commission 13th Lok Sabha, 5
of India, 18, 40, 41n1 main opposition, 1991 elections, 6
Association for Democratic Reforms, 33 net migration of seats, 84
odds ratios, 87
rise of, 5
B risk ratio, 86
Bahujan Samaj Party (BSP), 13n8, 22 seemingly unrelated regression
Banerjee, M., 11, 28 equations (SURE), 98–101
Bayes’ theorem, 78 vote share inequality, 140
incumbency effect, 78–9 winning and incumbency outcomes,
and inverse Bayes factor, 79–81 81

© The Editor(s) (if applicable) and The Author(s) 2016 155

V.K. Borooah, Votes, Parties, and Seats,
DOI 10.1007/978-3-319-30487-8

vk.borooah@ulster.ac.uk
156 INDEX

Bharatiya Lok Dal (BLD), 4 D

BJP. See Bharatiya Janata Party democracy, 8–12
(BJP) democratic achievement, 2
BLD. See Bharatiya Lok Dal (BLD) Desai, Morarji, 3, 4
BPL. See Below poverty line (BPL) Distribution of votes
Breeding, M., 21 equal distribution of votes, 144–5
Brennan, G., 27 equal number of votes received, 145–8
BSP. See Bahujan Samaj Party (BSP) number of seats won, 143–4
Downs, A., 26
Dravida Munnetra Kazhagam (DMK),
C 6, 13n6
candidates
criminal charge status of, 33–6
with criminal histories, 32–9 E
criminal status, 33 ECI. See Election Commission of India
face criminal charges, 32–9 econometric methodology, 63–8
incumbency, 78 election campaign, conduct of, 18–19
independent, 28–32 Election Commission of India (ECI),
CC. See Criminal charges (CC) 17–21
ceteris paribus, 152 elections, substantive democracy, 2
Chandra, K., 13n8 electoral deficit, 84
Communist Party of India (CPI), electoral efficiency, 151
22 electoral performance, INC and BJP,
Communist Party of India (Marxist) 61–8
(CPM), 22 electoral surplus, 84
constituency Electronic Voting Machines (EVM), 40
average number of candidates, emergency, 3
1962-2014, 28 entropy index, 134–5
average number of candidates, 2014 EVM. See Electronic Voting Machines
Lok Sabha elections, 28 (EVM)
The Constitution (73rd Amendment) expressive voting, 27
Act of 1993, 2
Corbridge, S., 2
Cowell, F. A., 142 F
criminal charges (CC) First-Past-The-Post (FPTP) system, 17
14th Lok Sabha, 37 Freese, J., 52
15th Lok Sabha, 34, 38
candidates, 2004 and 2009 Lok
Sabha elections, 33 G
16th Lok Sabha, 39, 40 Gandhi, Indira, 3
by state, 2004 and 2009 Lok Sabha Gandhi, Rajiv, 4
elections, 35–6 Gandhi, Varun, 40

vk.borooah@ulster.ac.uk
 INDEX 157

Gehring, K., 37, 38 Independent members, Lok Sabha,

General Election 1962–2014, 31
of 1957, 3 Indian National Congress (INC), 22
of 1962, 3 amplification coefficient, 120–3
of 1967, 3 Bayes factor calculations, 85, 88
of 1971, 3, 4 vs. BJP, vote shares, 98–101
of 1984, 4, 116 electoral performance, 61–8
of 1989, 4, 5, 116 equal voters’ scenario, 147
of 1991, 5, 6, 116 1957 General Election, 3
of 1996, 5, 6, 116 1962 General Election, 3
of 1998, 5, 116 1967 General Election, 3
of 1999, 5, 6, 116 1971 General Election, 3
of 2004, 83, 116 1977 General Election, 4
of 2009, 116 1989 General Election, 5, 117,
of 2014, 22, 116 118
Geys, B., 27 incumbency/non-incumbency
Gini coefficient, 24, 25 performance, 82
Golden, M., 33 inverse Bayes factor calculations, 90,
Gowda, Deve, 6 93
Gram Pradhan, 26 inverse risk ratio (IRR), 90
Gram Sabha, 2 risk ratio, 85, 86
grievances, 77, 78 net migration of seats, 84
group identity voting, 26 odds ratios, 87
Gujral, Inder, 6 seemingly unrelated regression
equations (SURE), 98–101
vote share inequality, 140
H vote shares, Hindi-speaking and
Hamlin, A., 27 non-Hindi-speaking states, 108
Herfindahl-Hirschman index (HHI), winning and incumbency outcomes,
131, 136 81
Hillman, A. L., 26–7 winning marginal constituencies,
Hindi-speaking (HS) states 51–2
and non-Hindi-speaking states; vote Indian parliamentary elections
shares, 107–12; winning candidates face criminal charges,
elections, probability of, 69–75 32–9
vote and seat ratios, 123–4 candidates with criminal histories,
32–9
Election Commission of India,
I 18–21
INC. See Indian National Congress electorates and turnout, 22–8
(INC) independent candidates, 28–32
incumbency effects, 55–8, 78–9 India’s Political Parties, 21–2

vk.borooah@ulster.ac.uk
158 INDEX

India’s Constituent Assembly, 1 M

India’s Political Parties, 21–2 Maintenance of Internal Security Act,
inter-constituency size and turnout, 14n18
24 Mandal Commission, 7
Inter-University Consortium for Marginal constituency, 45–8
Political and Social Research Member of Parliament Local Area
(ICPSR), 33 Development (MPLAD) Scheme,
inverse Bayes factor (IBF), 79–81, 37, 38
93 Model Code of Conduct (MCC),
18–20
Modi, Narendra, 5, 7
J Mozaffar, 40
Jaffrelot, C., 2 Muslim vote, 10
Jain Commission, 13 Muslim Women (Protection of Rights
Janata Party, 4 on Divorce) Act, 1986, 10
Jenkins, S. P., 142
Jennings, C., 27
N
Nanda, Gulzarilal, 12n2
L National Democratic Alliance (NDA),
Laakso, M., 136 6
Laïcité, principle of, 8 Nationalist Congress Party (NCP), 22
Law of cubic proportion, 119–20 National Security Act, 14n18
Law of the Cubic Proportion, 116 National Unity, 11–12
8th Lok Sabha, 25 NDA. See National Democratic
9th Lok Sabha, 25 Alliance (NDA)
13th Lok Sabha Nehru, Jawaharlal, 1
BJP, 5 1990s, political instability, 5–7
14th Lok Sabha, 34, 37 None of the Above (NOTA), 17, 31–2
15th Lok Sabha, 34 NOTA. See None of the Above
criminal charges members, 38 (NOTA)
16th Lok Sabha, 25, 34
criminal charges members, 39
political party composition, 22, 23 O
Lok Sabha elections. See also General Other Backward Classes, 7
Election
Bayes factor calculations, 82–9
marginal constituencies in, 47 P
risk ratio, 82–9 Panchayat, 2
Lomasky, L., 27 parliamentary activity, 37
Long, J. S., 52 Parliamentary General Election of
lower castes, rise of, 7–8 1951, 1. See also General Election

vk.borooah@ulster.ac.uk
 INDEX 159

party incumbency, 78 Samajwadi Party, 11

party vote shares, 137–41 Satanic Verses (Rushdie, Salman), 9
political instability, 1990s, 5–7 Sawhney Vs. The Union of India, 7
predicted likelihoods, 52 Schaffer, F. C., 21
predicted probabilities, 52 Schedler, A., 21, 40
Prevention of Insults to National Scheduled Castes, 7
Honour Act, 14n18 Scheduled Tribes, 7
Prevention of Terrorism Act, 14n18 Secular country, 13n5
Preventive Detention Act (1950), Secularism, 8–11
14n18 Seemingly Unrelated Regression
Prime Ministers, in India Equations (SURE), 98
Desai, Morarji, 3 equation statistics, 101
Gandhi, Indira, 3 predictive performance of, 101–3
Gandhi, Rajiv, 4 vote shares, INC vs. BJP, 98–101
Gowda, Deve, 6 Sen, A. K., 27
Gujral, Inder, 6 Shastri, Lal Bahadur, 12n2
Modi, Narendra, 5, 7 Shekhar, Chandra, 5
Nehru, Jawaharlal, 1 silent revolution, 11
Rao, Narasimha, 6 Singh, Charan, 4
Shastri, Lal Bahadur, 12n2 Singh, U. K., 20
Shekhar, Chandra, 5 Singh, V.P., 5
Singh, Charan, 4 Srinivas, M. N., 20
Singh, V.P., 5 substantive democracy, 2
Vajpayee, Atal Bihari, 5, 6 SURE. See Seemingly Unrelated
Regression Equations (SURE)

R
Rao, Narasimha, 6 T
Rashtriya Janata Dal, 10–11 Taagepera, R., 136
Recognised national parties, 22 Terrorism and Disruptive Activities
Representation of the People Acts of (Prevention) Act, 14n18
1950 and 1951, 18 Theil, H., 139
reservation, 7 Turnout, Indian states in 2014 Lok
reserved category, 7 Sabha elections, 26
Rudolph, L. I., 11, 153
Rudolph, S. H., 11, 153
Rushdie, Salman, 9 U
UK General election, 2015, 22, 24,
116
S United Progressive Alliance (UPA), 7
Sachar Committee, 10 UPA-I, 7
The Sachar Committee, 10 UPA-II, 7

vk.borooah@ulster.ac.uk
160 INDEX

Unlawful Activities (Prevention) Act, vote shares

14n18 estimation, 98–101
Unlawful Activities (Prevention) Hindi-speaking and non-Hindi-
Amendment Act, 14n18 speaking states, 107–12
unrecognised parties, 21, 22 incumbents and challengers, 104–5
untouchable castes, 12n1 inequality in, 141–3
winning elections, probability of,
58–61
V
Vajpayee, Atal Bihari, 5, 6
Vishnu, U., 13n14 W
voter turnout, 25 winning elections, probability of, 45
votes election-on-election changes, 55
concentration and distribution, Hindi-speaking vs. non-Hindi-
130–5 speaking states, 69–75
equal distribution of votes, incumbency effects, 55–8
144–5 number of candidates, 58–61
equal number of votes received, turnout, 58–61
145–8 vote share, 58–61
number of seats won, 143–4 winning marginal seats, likelihood of,
states, effective number of, 136 49–54

vk.borooah@ulster.ac.uk