Why regional parties succeed at the sub-national level in India

Abstract

Existing scholarship in comparative politics has either focused on regionalism or on political and economic decentralisation to explain the growth of regional parties. Using quantitative evidence from India, I show that a hitherto ignored explanation, the level of regional branch autonomy within polity-wide parties, also has a significant impact on the growth of regional parties. When regional branches of polity-wide parties have autonomy, regional parties find it difficult to grow. In contrast, regional parties benefit electorally when regional branches of polity-wide parties are less autonomous. To further account for endogeneity between regional party growth and regional branch autonomy, I use quantitative and qualitative evidence to show that the growth of regional parties is not positively correlated with more regional branch autonomy.

Keywords

federalism India party organisation political parties regional party system sub-national comparative method

Introduction

Why are regional parties successful in some parts of a country, but not others? The existing literature in comparative politics has either emphasised electoral institutions (Cox, 1997), political and economic decentralisation (Brancati, 2008; Chhibber and Kollman, 2004), or regionally based social cleavages (Lipset and Rokkan, 1967). However, in the Indian context these factors appear ill-equipped to explain variations in the success of regional parties. Despite following the same electoral system and experiencing similar levels of political and economic decentralisation, some Indian states produce successful regional parties while others do not. Importantly, the presence of distinct regional cleavages is not sufficient to explain the variations in the success of regional parties at the state level in India.¹

Building on existing work on party organisation in India (Chhibber et al., 2014; Ziegfeld, 2016b: 151–170), I argue that the variation in the success of regional parties in India can be partly explained by the variation in the autonomy of regional branches of national (polity-wide) parties. Regional parties are more successful in states where polity-wide parties do not give autonomy to their regional branches than in states where they do give them autonomy. This is because autonomy from the central branch allows the regional branch of the party to accommodate sub-nationalist sentiments, thereby preventing regional parties from having a ‘monopoly’ over emotional issues that are important to the voters. This helps to reduce voter-side demand for regional parties. In addition, autonomy from the central branch of a polity-wide party creates fewer incentives for regional elites to defect from polity-wide parties. This is because regional elites are not only able to have control over resources and power at the sub-national level, but also have more influence nationally through intra-party linkages.

Drawing from the literature on party organisation in multi-level settings (Thorlakson, 2009), this study contributes to the growing literature on how intra-party organisation shapes the regional party system in multi-level settings. While much of the existing scholarly work in comparative politics on party adaptation (Detterbeck and Hepburn, 2007; Fabre, 2008; Hopkin and Van Houten, 2009; Renzsch, 2001) has focused predominantly on polity-wide parties giving autonomy to their regional branches when they face competition from regional parties, this article shows that polity-wide parties are able to impede the rise of regional parties if they give autonomy to their regional branches.

To illustrate the theoretical claims, I use a multi-method approach combining statistical analysis of the major Indian states with qualitative evidence from a number of Indian states. In the section ‘Qualitative evidence,¹ I provide examples to show that the rise of regional parties did not motivate polity-wide parties like the Congress Party and the Bharatiya Janata Party (BJP) to give more autonomy to their regional branches.

Theoretical argument

When it comes to defining regional parties, I follow the typology created by Ziegfeld (2009: 87–93) and Kailash (2014). Parties will be categorised as polity-wide or regional based on the electoral performance of the parties across the country. The key criterion distinguishing polity-wide and regional parties is the geographical spread of votes. Regional parties are parties that have a concentration of votes in one or at the most two states, whereas polity-wide parties are parties that win votes and seats in most (if not all) parts of the country. I further draw a conceptual distinction between regional parties based on their political goals (Palshikar, 2004). Regional parties that champion the cause of a specific region will be referred to as regionalist parties. Centre–state relations are a key issue on the agenda of the regionalist parties. Moreover, regionalist parties focus on the interests of a particular state, and mobilise voters by appealing to the state’s regional pride, culture, language, and customs. Regionalist parties are almost always regional parties, but regional parties are not necessarily regionalist. Even when regional parties do bring up the issue of centre–state relations or regional pride, it is more in general terms rather than with respect to any particular state. In contrast, regionalist parties invoke loyalty to and emotion about a particular identified territory (Kailash, 2014). Regional parties aspire to be polity-wide parties while regionalist parties do not have such ambitions. It is important to draw a distinction between regional and regionalist parties, because the theoretical mechanisms through which regional and regionalist parties succeed are different. I elaborate on the different theoretical mechanisms below.

I argue that regional and regionalist parties are more likely to be successful if regional branches of polity-wide parties do not have autonomy from their central branch. Autonomy is understood as the degree of freedom of the regional branch to conduct its affairs without interference from the central branch. Examples include regional branches’ autonomy with respect to financing, candidate and leader selection, designing campaign platforms and/or branding, developing policy proposals, and finally, having the flexibility to form alliances and negotiate with regional parties and regionally dominant groups at the sub-national level (Fabre, 2011; Thorlakson, 2011).

There are a number of reasons why regional branch autonomy may make it harder for regionalist parties to succeed. In politically decentralised regions, regionalist parties can take advantage of prevalent regionalist sentiments to carve out a strong support base amongst the voters. This is particularly the case if the region was created on the basis of a regionalist identity. However, autonomous regional branches of polity-wide parties will be able to assert their independence from the party’s central branch and counter a regionalist party’s claim of ‘monopoly’ over the region’s interests. Autonomous regional branches of polity-wide parties will be better equipped to adapt the party message to regionally specific issues and contexts. In doing so, they will absorb any regionalist sentiments present amongst the voters, thereby depriving regionalist parties of their core support base. In sum, autonomous regional branches of polity-wide parties will be able to present themselves as de facto regionalist parties with the added benefit of having greater access to national-level power through intra-party linkages (Chhibber, 1999; Kohli, 1990; Singh, 2015a: 103–107; Ziegfeld, 2016b: 155–169).

From a polity-wide party’s perspective, autonomy for its regional units creates greater opportunities to keep diverse networks of regional political elites with contrasting political preferences together under the same party. It can allow specific regional branches to present regionalist messages without necessarily following a regionalist message at the central level. Here, it is important to note that autonomy within a single polity-wide party at the state level can change the structure of political competition, and in turn affect the whole party system. A single autonomous regional branch of a polity-wide party can successfully accommodate regionalist issues, and deny regionalist parties their raison d’être. Accordingly, I test the following hypothesis:

H1: Support for regionalist parties among voters will be lower in states with an autonomous regional branch of a polity-wide party than in states with no autonomous regional branches of polity-wide parties.

Second, in politically decentralised regions, if polity-wide parties give autonomy to their regional branches, then regional political elites are more likely to stay within polity-wide parties. This is because they are able to have control over resources and power at the sub-national level, thereby discouraging regional elites from forming their own regional parties. While it is true that regional political elites could have control over the sub-national level by forming and joining regional parties, this would then make it difficult for these elites to maintain influence at the central level.² Even when the sub-national level has control over important policy arenas, the central level in India still constitutionally holds more power (Ziegfeld, 2016b). Regional political elites would not want to lose influence over the central level for the sake of gaining autonomy over the sub-national level. Autonomy for their regional branches within existing polity-wide parties therefore presents an ideal opportunity for regional political elites to have control over the decision-making network at the sub-national level, but at the same time also maintain influence over the central level through intra-party linkages.³

Accordingly, I argue that autonomous regional branches of polity-wide parties will reduce the likelihood of political elites leaving the polity-wide party to form their own party, or join an existing regional or regionalist parties. Preventing defection is particularly important in the Indian context, because voters are much more candidate focused than voters in other majoritarian voting systems (Jensenius and Suryanarayan, 2018). This means that when elites defect from a polity-wide party to a regional or regionalist party they bring many of their supporters with them. However, an autonomous regional branch of a polity-wide party will decrease the incentives of political elites to defect, which in turn will decrease the support for regional or regionalist parties. Accordingly, I test two related hypotheses:

H2a: Autonomous regional branches of polity-wide parties will decrease the likelihood of political elites defecting from a polity-wide party to a regional or regionalist party.

H2b: In doing so, the presence of autonomous regional branches will decrease support for regional and regionalist parties since polity-wide party defectors boost support of the parties that they join.

The distinction between H1 and H2 highlights the necessity of distinguishing between regional and regionalist parties. At the individual level, an autonomous regional branch of a polity-wide party will decrease the incentives of regionalist voters to support regionalist parties (but will not influence their support for regional parties, which does not depend upon regionalist sentiment).⁴ This is because in principle an autonomous regional branch will be able to cater to the regionalist demands of the voters. In contrast, the elite level (H2) hypothesis works for both regional and regionalist parties. An autonomous regional branch of a polity-wide party will decrease the incentives of elites from the polity-wide party to defect to either regional or regionalist parties. This is because autonomy from the central branch gives regional and regionalist elites more power over regional resources and policy-making influence at the sub-national level.

Data and methods

To test H1, I draw on data from the 1996 and 1999 National Election Study (NES) conducted by the Centre for the Study of Developing Societies (CSDS)-Lokniti.⁵ To measure the main dependent variable, regionalist party vote, I use the voting intention questions in the NES 1999 and 1996. I classify respondents who voted for a regionalist party⁶ as 1 and others as 0.⁷

To measure the main independent variable of regional branch autonomy I use Chhibber et al. (2014) (hereafter, CJS) measure of ‘party organisation’ at the state level in India. I use their ‘party organisation’ measure as a proxy for measuring intra-party autonomy. CJS collected data on the organisational structure of all Indian parties that received at least 5% of the vote share during the state elections held between 1967 and 2004, treating each state party branch as an observation. They include data from all the 15 major Indian states.⁸ In their dataset, state branches of polity-wide parties can take different values in different states at the same time. They code a party unit as less organised

when there was no clear succession plan within the party, where party functionaries’ roles were fluid and election-focused, and where opportunities for upward mobility were either limited or prone to the whims of a few leaders. In addition, a less organised party depends on the charisma of a single leader and decision-making within the party was referred to as ad hoc.

In contrast, they code a party unit as more organised

when career decisions for party activists and succession issues were more transparent and routinized and the party did not depend only on the personalities of individuals. Also, in these units politicians had clarity about their role in the organization, such as the qualifications required to fill certain positions, the process for upward mobility in the party, the rules of succession planning, the organization’s tolerance for intra-party factionalism and, finally, the extent to which party decisions are taken based on clearly understood institutional norms as opposed to the whims of leaders. (Chhibber et al., 2014: 493)

It is clear from the above quotations that Chhibber et al.’s (2014) ‘party organisation’ measure does not directly measure intra-party autonomy. Instead, their ‘party organisation’ measure is more akin to intra-party institutionalisation.⁹ I acknowledge this limitation. However, theoretically speaking, it is still a good proxy for autonomy since it would be difficult to find an institutionalised regional branch of a polity-wide party with low levels of autonomy. In this sense strong institutionalisation allows the regional party branch to develop cohesive regional agendas, and to cultivate a strong position to bargain with the central branch. Furthermore, where regional branches of polity-wide parties are autonomous, ambitious regional elites are more likely to be given space within the regional unit of the polity-wide party. Ziegfeld (2016a: 117) also supports this theoretical link, in that he argues that a high level of intra-party decentralisation is incompatible with high levels of autocracy. Moreover, in the comparative context, Randall and Svåsand (2002) also discuss how intra-party institutionalisation implies some degree of autonomy from the external environment. It is true that autonomy from the external environment is different from autonomy from the party’s central branch,¹⁰ but extending the logic of autonomy from the external environment to the intra-party dimension, we could theoretically assume that high-level institutionalisation does include a high level of autonomy for the state branch.

The literature on democratic decentralisation offers yet another theoretical prism through which to understand the relationship between intra-party institutionalisation and intra-party autonomy. Studies on democratic decentralisation highlight a principal agent problem at the heart of democratic decentralisation. They argue that competitive government and council elections at the local level generate incentives for local politicians to act on their own, at times, posing political challenges to their party leaders in central government. When this agency dilemma is severe, central branch leaders hesitate to devolve autonomy to local politicians. However, as argued by Sadanandan (2017: 45–50), in well-disciplined or institutionalised regional branches, local politicians are reliable partisan colleagues, where the central branch leaders can devolve power without fearing indiscipline by local politicians. In this way, party institutionalisation or discipline not only allows the regional branch to develop strong regional agendas and better negotiate with the central branch, but also alleviates some of the risks of central branch leaders hesitating over devolving autonomy. In addition to the above arguments, please see the Supplemental appendix for a further discussion on why I use CJS’s ‘party organisation’ dataset to test my hypothesis.

CJS provide coding for all the 138 regional elections years between 1967 and 2004, with state-party branches receiving a score of 1, 2, or 3, with 3 being the most organised and 1 being the least, with 2 as the intermediate category. Using the three categories as proxies, I understand regional branches with a score of 1 as having low levels of autonomy, and 2 or 3 as having high levels of autonomy. This is consistent with how other scholars (Ziegfeld, 2016a: 117) have coded CJS’s data.

To measure the other main independent variable of regionalism, I use the following survey question:¹¹ ‘We should be loyal to our region first and then to India’. The respondents had three response choices: Agree, Don’t know/No opinion, Disagree. I drop those respondents that chose Don’t Know/No opinion. I rescale the variable where Agree = 1 and Disagree = 0.

Regional branch autonomy, regionalist sentiment, and support for regionalist parties

In order to test H1, I run two logistic regression models, one for each wave: NES 1999 and NES 1996.¹² For both waves, I regress regionalist party vote on the above measures of regionalism, polity-wide party organisation, an interaction between regionalism and polity-wide party organisation, and a number of individual control variables.¹³ The primary variables of interest are regionalism and the interaction between regionalism and polity-wide party organisation. Polity-wide party organisation is a measure that classifies states with a single autonomous regional branch of a polity-wide party during the 1999 or 1996 general elections. In other words, polity-wide party organisation is assigned a value of 1 for those states which have a single autonomous regional branch of a polity-wide party during the 1999 or 1996 general elections.¹⁴ Those states which do not have a single autonomous regional branch of a polity-wide party during the 1999 or 1996 general elections are captured by a value of 0 in the same variable. Importantly, I use CJS data to classify only those states which have a presence of a viable regionalist and non-regionalist party.¹⁵ For the NES 1999 sample or Model 1, I understand only the Congress Party and the BJP as polity-wide parties. For the NES 1996 sample or Model 2, I understand the Congress Party, the BJP, and the Janata Dal as polity-wide parties.¹⁶

In Table 1, Model 1, we see that the interaction term, Regionalism 99 * Party Organisation 99 is statistically significant and negatively correlated with the dependent variable. This shows that voters with a regionalist sentiment are less likely to vote for regionalist parties in those states where there exists a single autonomous regional branch of a polity-wide party. By contrast, voters with regionalist sentiments are more likely to vote for regionalist parties in states where there exist non-autonomous regional branches of polity-wide parties. This is captured by the positive coefficient of the stand-alone variable of Regionalism 99.¹⁷

Table 1.

Regionalism and regionalist party vote, conditional on national party organisation.

	Regionalist party vote 99	Regionalist party vote 96
	(1)	(2)
Regionalism 99	0.330*
	(0.194)
Party organisation 99	−0.169
	(0.283)
Regionalism 99 * Party Org 99	−0.672**
	(0.323)
Regionalism 96		0.605**
		(0.237)
Party organisation 96		−0.564**
		(0.289)
Regionalism 96*Party Org 96		−0.601**
		(0.325)
Constant	−0.018	−0.759**
	(0.324)	(0.370)
Controls	Yes	Yes
Observations	1145	1193
Log likelihood	−741.894	−721.904

Logistic regression models with robust standard errors.

p < 0.1.

p < 0.05.

***

p < 0.01.

These results are replicated in Model 2, which estimates the same empirical model on the NES 1996 sample. Once again, we see that the interaction term is statistically significant and negatively correlated with the dependent variable. This shows that voters with a regionalist sentiment are less likely to vote for regionalist parties in states with an autonomous regional branch of a polity-wide party. By contrast, regionalist sentiments amongst voters are positively correlated with voting for regionalist parties in states with non-autonomous regional branches of polity-wide parties. This effect is captured by the stand-alone variable of regionalism, which is positive and statistically significant. These results are therefore robust across samples.

To illustrate the substantive impact of these interactions, Figures 1 and 2 plot the predicted vote share for regional parties across different levels of regional branch autonomy by voters’ regionalist sentiment. From both figures we can clearly see that support for regionalist parties among voters with a regionalist sentiment is much higher in states where there exist non-autonomous regional branches of polity-wide parties than in states where polity-wide parties’ regional branches have autonomy. By contrast, among voters who do not have regionalist sentiments, regional branch autonomy doesn’t influence support for regionalist parties and the confidence intervals overlap.

Figure 1.

Party organisation, regionalist sentiment, and regionalist party vote, NES 1999.

Figure 2.

Party organisation, regionalist sentiment, and regionalist party vote, NES 1996.

Regional branch autonomy and elite defection

To test H2, I use two main data sources. First, I use CJS’s ‘party organisation’ data, and second, I use incumbency-mapping data compiled by the Trivedi Centre for Political Data (TCPD) at the Ashoka University (Jensenius and Verniers, 2017). The researchers at the TCPD developed a name-matching algorithm that identifies identical names in order to facilitate the coding of individual politicians’ careers. I use their algorithm to compile a dataset of Turncoats (politicians who defect from a polity wide party to a regional or regionalist party). I merge this Turncoat data with CJS’s party organisation data. Here, it is important to note that TCPD’s algorithm only includes Turncoat data for 8 out of the 15 states included in CJS’s dataset. Data for important states like Assam, Bihar, Kerala, Orissa, Maharashtra, Tamil Nadu, and West Bengal are missing. This missing data creates further problems for testing my argument for the Janata Dal and the Janata Party. Therefore, for the first part of this section, I restrict my analysis to the Congress Party and the BJP.

To provide empirical evidence for H2, I conduct separate analyses for the Congress Party and the BJP. The main dependent variables are Turncoats from the Congress Party and the BJP, that is, the number of members of political elites that left the Congress Party or the BJP for regional or regionalist parties.¹⁸ I exclude those political elites that left the Congress Party or the BJP for other polity-wide parties, that is, switching between these two parties or joining the Janata Party or the Janata Dal.¹⁹ The main independent variable is CJS ‘party organisation’ data, which I use as a proxy for the autonomy of the Congress and BJP regional branches.²⁰ As before, I recode this variable where a CJS score of 1 is equivalent to having a low level of autonomy, and a CJS score of 2 or 3 to having a high level of autonomy. The expectation is that higher levels of autonomy would mean lower numbers of Turncoats from the Congress Party and the BJP.

In addition to the main independent variable of intra-party autonomy, I add a number of control variables. First, I control for the respective party’s vote share in order to account for any increase in defections simply because politicians are deserting an unpopular party.²¹ Second, I control for the presence of coalition governments, since regional political elites may be more likely to leave the Congress Party or the BJP when they have an opportunity to participate in government without first rising through the ranks of a polity-wide party. India has experienced the rise of coalition governments at the national level following the 1989 general election (Ziegfeld, 2012), and therefore I have included a dummy variable (Post.1989) to capture its effect. Third, I control for Turnout since increasing voter participation could motivate higher numbers of Turncoats from the polity-wide party (Jaffrelot, 2003; Yadav, 2000). Fourth, I control for state-specific factors like political and economic decentralisation at the state level (Brancati, 2008; Chhibber and Kollman, 2004), (2) state’s dependence on the centre for economic help (Diwakar, 2010), level of sub-nationalism (Singh, 2015b), and state-specific ethnic fractionalisation, by including state fixed effects.²² I estimate the models using ordinary least squares (OLS) estimation technique, but provide Poisson regression models in the appendix where I do not weight the raw number of Turncoats from the Congress Party and the BJP.²³

In Table 2, we can see that there is a statistically significant negative relationship between autonomy for the Congress regional branch (INC organisation) and Turncoats from the Congress Party. This shows that on average there are fewer regional political elites who abandon the Congress Party for regional or regionalist parties in states where the Congress regional branch has autonomy compared to states where the Congress regional branch has less autonomy.

Table 2.

Turncoats from the Congress Party and the BJP.

	Turncoats INC	Turncoats BJP
	(1)	(2)
INC organisation	−3.460**
	(1.623)
INC vote share	−0.058
	(0.086)
BJP organisation		0.242
		(0.786)
BJP vote share		0.176**
		(0.080)
Post 1989	2.455*	2.16***
	(1.308)	(0.637)
Turnout	−0.024	0.038
	(0.042)	(0.033)
Constant	7.223***	–2.361
	(2.516)	(3.248)
State FE	Yes	Yes
Observations	71	52
Adjusted R²	0.434	0.411

BJP: Bharatiya Janata Party; OLS: ordinary least squares; FE: fixed effects; INC: Indian National Congress.

OLS regression models with robust standard errors.

p < 0.1.

p < 0.05.

***

p < 0.01.

In the BJP’s case, there is no evidence that regional branch autonomy reduces the number of Turncoats from the BJP. This could be down to several reasons, not least because of the small sample size, and the loss of degrees of freedom after including state fixed effects. More substantively speaking, the null effect of regional branch autonomy on Turncoats from the BJP could be down to something more systematic. Recent evidence marshalled by Nellis (2018) illustrates the differences in the overall nature of factionalism between the Congress Party and the BJP. A strong unitary culture within the BJP organisation could be preventing regional elites from defecting despite the lack of intra-party autonomy. In contrast, the only way to avoid regional elites from defecting from a factionalised party like the Congress Party is regional branch autonomy.

In both the models, I find evidence for Ziegfeld’s (2012) coalition government thesis, that defection from the Congress and the BJP to regional and regionalist parties is greater in the post-1989 period than it was in the pre-1989 period, when single party government was the norm.

The second part of H2 argues that more Turncoats from polity-wide parties lead to more support for regional and regionalist parties. Below I provide empirical evidence for the final part of H2. The main dependent variable is Regional party vote share (%)²⁴ and the main independent variables are Turncoats from polity-wide parties (Turncoats INC and Turncoats BJP). I use OLS regression technique to estimate two separate models, one for the Congress Party (Model 1) and one for the BJP (Model 2). In both the models, I control for Post.1989, Turnout, and the respective party’s vote share. Both the models also include state fixed effects.²⁵

In Table 3, Model 1, we see that the Turncoats INC variable is positive and statistically significant. This shows that a higher number of Turncoats from the Congress Party is positively correlated with greater regional party vote share.²⁶ In Model 2, for the BJP’s case, we also see that the primary variable of interest, Turncoats BJP is positive and statistically significant. Once again, this means that a higher number of Turncoats from the BJP is positively correlated with greater regional party vote share. In sum, both Models 1 and 2 illustrate that a higher rate of defections from polity-wide parties do give an electoral boost to regional or regionalist party vote shares.

Table 3.

Regional party vote share, party organisation, and Turncoats.

	Regional party vote share (%)	Regional party vote share (%)
	(1)	(2)
Turncoats INC	0.370***
	(0.171)
INC vote share	−0.224
	(0.178)
Turncoats BJP		0.410*
		(0.230)
BJP vote share		−0.809***
		(0.197)
Constant	59.110***	38.621***
	(15.583)	(16.220)
Controls	Yes	Yes
State FE	Yes	Yes
Observations	71	52
Adjusted R²	0.548	0.643

BJP: Bharatiya Janata Party; OLS: ordinary least squares; FE: fixed effects; INC: Indian National Congress.

OLS regression models with robust standard errors.

p < 0.1.

p < 0.05.

***

p < 0.01.

Endogeneity

As mentioned above, existing work on party adaptation in comparative politics (Fabre and Swenden, 2013) has argued that polity-wide parties give autonomy to their regional branches when they face competition from regional or regionalist parties. In contrast, I argue that polity-wide parties are able to impede the rise of regional and regionalist parties if they give autonomy to their regional branches. In this section, I present quantitative and qualitative evidence to refute the argument in favour of endogeneity, that is, that polity-wide parties give autonomy to their regional branches when they face competition from regional or regionalist parties.

Quantitative evidence

If the party adaptation literature were to hold true in India, we should expect to see a positive correlation between increasing lag regional party vote share (%) and regional branch autonomy within the two party organisations. To investigate this possibility, I estimate two separate models using logistic regression. The main dependent variables are INC.organisation and BJP.organisation. The main independent variable is lagged regional party vote share (%).²⁷ I also include control variables like Post.1989 and Turnout.²⁸ In the models presented, I do not include state-fixed effects to show that the null effect is not simply an artefact of lower degrees of freedom. However, the effects are null even after including state-fixed effects.

In Table 4, we see that increasing regional party vote share does not increase the probability of higher regional branch autonomy within the Congress Party and the BJP in the subsequent regional election year. The effect is null even after including a number of control variables and state-fixed effects.²⁹ This provides quantitative evidence against the party adaptation literature, at least in the Indian case.

Table 4.

Lag regional party vote share (%) and polity-wide party organisation.

	INC.organisation	BJP.organisation
	(1)	(2)
Lag regional vote share	−0.015	0.029
	(0.014)	(0.038)
Post 1989	−0.077	2.840***
	(0.479)	(0.985)
Turnout	0.049	−0.125*
	(0.035)	(0.055)
Constant	−3.336*	6.163
	(1.775)	(2.332)
State FE	No	No
Observations	120	53
Log likelihood	–37.995	–21.90

BJP: Bharatiya Janata Party; FE: fixed effects; INC organisation: Congress regional branch.

Logistic regression models with robust standard errors.

p < 0.1.

p < 0.05.

***

p < 0.01

Qualitative evidence

I next provide examples from a number of Indian states to show that the growth of different types of regional and regionalist parties does not necessarily motivate polity-wide parties to give more autonomy to their regional branches. The first example comes from Andhra Pradesh. The rise of a regionalist party such as the Telugu Desam Party (TDP) is one of the most interesting phenomena of party politics in India. The party was formed by N.T. Rama Rao (NTR), a famous Telugu cinema actor, in March 1982, and it came to power in the subsequent regional elections in January 1983, all within a matter of 9 months. One of the crucial factors behind the TDP’s success was its use of Telugu sub-nationalism, and articulation of messages presenting Congress state leaders as factionalist and as being puppets in the hands of the Congress central branch leaders (Kohli, 1988). As documented by Suri (2004: 1482), NTR called upon the people to hold high Telugu jati gowravam (national honour) and atma gowravam (self-respect) by defeating the Congress Party and teaching a lesson to its leaders. However, the rise of the TDP on the back of Telugu sub-nationalism did not motivate the Congress Party’s central branch to give more autonomy to its sub-national branch. CJS consistently award a score of 1 (low autonomy) for the Congress state branch in Andhra Pradesh during the 1980s.

The second example I wish to highlight is that of Bihar. Bihar was at the centre stage of regional politics in the late 1990s. Laloo Prasad Yadav, one of the foremost leaders of the Janata Dal unit of Bihar, formed his own party, the Rashtriya Janata Dal (RJD) in the late 1990s. The RJD contested the 2000 regional elections under its regional party label, separate from a second splinter group of the Janata Dal, which contested as the Janata Dal (United) (JD (U)). Despite the entry of two regional parties, the RJD and the JD (U), prior to the 2000 regional elections, the Congress Party did not give more autonomy to its regional branch. CJS award a score of 1 (low autonomy) for the Bihar Congress state branch during the 2000 regional elections.

The third example is that of the Bahujan Samaj Party (BSP). The BSP was formed in 1984 by Kanshi Ram to mobilise Scheduled Castes and other minority groups to gain political power (Chandra, 2004). Although its target groups were Scheduled Castes and minority groups throughout the country, it was competitive mainly in the state of Uttar Pradesh, from the early 1990s. If the party adaptation literature was applicable in the Indian case, we would have seen the entry of the BSP in Uttar Pradesh motivate the Congress Party to give more autonomy to its regional branch. However, the Congress Party state branch consistently receives a score of 1 during the 1990s as per the CJS coding scheme, reflecting a low level of autonomy for the Congress state branch during the 1990s.

The fourth example is that of the Trinamool Congress Party (TMC) in Bengal. Mamata Banerjee, a grassroots leader of the Congress Party, formed the TMC in the late 1990s as a result of being consistently denied the post of leader of the Congress Bengal regional branch (Nielsen, 2011). The example of Mamata Banerjee highlights how lack of regional branch autonomy for important state-level leaders incentivises them to split and form breakaway regional parties. However, despite the rise of the TMC in Bengal, the Congress Party did not give more autonomy to its regional branch. CJS award a score of 1 (low autonomy) throughout the 1990s for the Congress Bengal branch.

The BJP is no different. For example, the rise of the Biju Janata Dal (BJD) in Odisha did not prompt the BJP to give more autonomy to its regional branch. Throughout the late 1990s and going up to the mid-2000s, the Odisha regional branch of the BJP has a low autonomy score. In Maharashtra and Punjab, the BJP has been in coalition with regionalist parties like the Shiv Sena and the Shiromani Akali Dal (SAD), respectively. Existing comparative research also tells us that regional branch autonomy expands when polity-wide parties are in coalition with regionalist parties (Renzsch, 2001), but the case of the BJP in Maharashtra and Punjab does not reflect that. The changes in the CJS scoring of the BJP regional branch in both these states does not reflect the entry and the rise of regionalist parties like the Shiv Sena and the SAD.

Conclusion

In this article, I have put forward an overlooked explanation as to why regional and regionalist parties grow in some regions of a country, but not others. I argue that regional and regionalist parties are more likely to succeed when polity-wide parties do not give autonomy to their regional branches. While a number of studies focusing on the European case studies have used the rise of regional or regionalist parties to examine the level of intra-party autonomy within polity-wide parties, this article argues that it is in fact the level of intra-party autonomy within polity-wide parties that partly explains the success of regional and regionalist parties. I exploit the striking variations in the success of regional and regionalist parties found at the sub-national level in India to support my argument empirically. I use CSDS-Lokniti voting behaviour data alongside Turncoat data provided by the TCPD to provide empirical evidence in support of my hypotheses. Furthermore, I use case study evidence to show that the rise of regional and regionalist parties does not necessarily compel polity-wide parties in India to give autonomy to their regional branches. Why some regional branches of polity-wide parties have autonomy while others do not is a promising area for future research, well worth investigating. This is particularly important because if autonomous regional branches of polity-wide parties are able to stem the rise of regional and regionalist parties, then the next logical research step is to investigate why some regional branches of polity-wide parties have autonomy where others do not.

In addition, for researchers of Indian politics, and especially for those interested in party organisation, one of the most pressing tasks this study leaves is to collect primary data that directly measures regional branch autonomy. Since there exist no primary data on regional branch autonomy within polity-wide parties in India, I have used CJS’s dataset on intra-party institutionalisation as a proxy for measuring regional branch autonomy. As discussed before, CJS’s party organisation dataset is not a perfect proxy for measuring regional branch autonomy. Future researchers should seriously consider collecting primary data that directly measures regional branch autonomy.

Finally, in the Indian case, scholars have long argued that the geographical concentration of certain groups, in particular caste and language groups, is important in explaining why regional and/or regionalist parties succeed. If most Indian states have a specific geographical concentration of caste and linguistic groups, then why do we still observe variations in the success of regional and regionalist parties at the sub-national level in India? In this article, I have shown how intra-party autonomy partly helps explain the success of regional and regionalist parties even after controlling for political and economic decentralisation, the advent of coalition politics at the national level, and sociological factors, such as the role of social cleavages at the sub-national level. By concentrating on the party’s internal organisation, this article directly speaks of the growing literature on the role of party organisation in shaping party systems and electoral outcomes in India.

Supplemental Material

BPI912665_Appendix – Supplemental material for Why regional parties succeed at the sub-national level in India

Supplemental material, BPI912665_Appendix for Why regional parties succeed at the sub-national level in India by Dishil Shrimankar in The British Journal of Politics and International Relations

Footnotes

Acknowledgements

The author thanks Katharine Adeney, Pranav Gupta, Oliver Heath, Francesca Jensenius, Diego Maiorano, Kyriaki Nanou, Carole Spary, Siim Truum, Wilfried Swenden, Rahul Verma, Gilles Verniers, the Editors, and two anonymous reviewers for their comments and feedback in improving this manuscript. The author also thanks Lokniti-CSDS and TCPD, and their fantastic research teams, for making the NES and TCPD data available.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The author acknowledges financial support from the Asia Research Institute, University of Nottingham and the Economic and Social Research Council [Grant Number ES/T007451/1] towards the publication of this article.

ORCID iD

Dishil Shrimankar

Supplementary Information

Additional supplementary information may be found with the online version of this article.

Contents

Table A1: Party types in NES 1999

Table A2: Party types in NES 1996

Table A3: Regionalism and regionalist party vote, conditional on national party organisation, with JD as regional party in the 1996 NES wave

Table A4: Regionalism and regionalist party vote, conditional on national party organisation, with Congress splinters as regionalist

Table A5: Regionalism and regionalist party vote, conditional on national party organisation, independents as regional parties

Table A6: Regionalism and regionalist party vote, conditional on national party organisation, with Gujarat dropped in the NES 1996 wave.

Table A7: Regionalism and regional party vote i.e. DV = regional parties

Table A8: Regionalism and regionalist party vote, controlling for partisanship.

Table A9: Turncoats from the Congress Party and the BJP, Poisson models

Table D1: States and Election years included in the CJS dataset

Table D2: Descriptive statistics for Table 1, NES 1999 wave

Table D3: Descriptive statistics for Table 2, NES 1996 wave

Table D4: Descriptive statistics for Turncoats from INC, Table 2

Table D5: Descriptive statistics for Turncoats from BJP, Table 2

Table D6: Descriptive statistics for Regional party vote share (Model 1), Table 3

Table D7: Descriptive statistics for Regional party vote share (Model 2), Table 3

Table D8: Descriptive statistics for Table 4, INC organisation

Table D9: Descriptive statistics for Table 4, BJP organisation

Discussion regarding the use of CJS party organisation as a proxy for regional branch autonomy.

Discussion regarding the distinction between external and intra-party autonomy

Reference List

Notes

References

Brancati

(2008) The origins and strengths of regional parties. British Journal of Political Science 38: 135–159.

Chandra

(2004) Why Ethnic Parties Succeed: Patronage and Ethnic Head Counts in India. Cambridge: Cambridge University Press.

Chhibber

Kollman

(2004) The Formation of National Party Systems: Federalism and Party Competition in Canada, Great Britain, India and the United States. Princeton, NJ: Princeton University Press.

Chhibber

Jensenius

Suryanarayan

(2014) Party organization and party proliferation in India. Party Politics 20: 489–505.

Chhibber

(1999) Democracy without Associations: Transformation of the Party System and Social Cleavages in India. Ann Arbor, MI: University of Michigan Press.

Cox

(1997) Making Votes Count: Strategic Coordination in the World’s Electoral Systems. Cambridge: Cambridge University Press.

Detterbeck

Hepburn

(2007) Party politics in multi-level settings. In: Erk

Swenden

(eds) Exploring New Avenues in Comparative Federalism Research. Helsikni, ECPR Joint Sessions: 13.

Diwakar

(2010) Party aggregation in India: A state level analysis. Party Politics 16: 477–496.

Fabre

(2008) Party organization in a multi-level system: Party organizational change in Spain and the UK. Regional and Federal Studies 18: 309–329.

10.

Fabre

(2011) Measuring party organization: The vertical dimension of the multi-level organization of state-wide parties in Spain and the UK. Party Politics 17: 343–363.

11.

Fabre

Swenden

(2013) Territorial politics and statewide parties. Regional Studies 47: 342–355.

12.

Hainmueller

Mummolo

(2018) How much should we trust estimates from multiplicative interaction models? Simple tools to improve empirical practice. Political Analysis 27: 163–192.

13.

Hopkin

Van Houten

(2009) Decentralization and state-wide parties: An introduction. Party Politics 15: 131–135.

14.

Jaffrelot

(2003) India’s Silent Revolution: The Rise of the Lower Castes in North India. Hyderabad, India: Orient Blackswan.

15.

Jensenius

Suryanarayan

(2018) Party institutionalisation and economic voting: Evidence from India. Working paper.

16.

Jensenius

Verniers

(2017) Studying Indian politics with large-scale data: Indian Election Data 1961-Today. Studies in Indian Politics 5: 269–275.

17.

Kailash

(2014) Regional parties in the 16th Lok Sabha elections. Economic and Political Weekly 49: 64–71.

18.

Kohli

(1988) The NTR phenomenon in Andhra Pradesh: Political change in a South Indian state. Asian Survey 28: 991–1017.

19.

Kohli

(1990) From elite activism to democratic consolidation. In: Frankel

Rao

(eds) Dominance and State Power in Modern India. New Delhi, India: Oxford University Press, pp.367–415.

20.

Lipset

Rokkan

(1967) Party Systems and Voter Alignments: Cross-National Perspectives. New York: The Free Press.

21.

Nellis

(2018) The Fight Within: Intra-Party factionalism and Incumbency Spillovers in India. Available at: http://www.garethnellis.com/wp-content/uploads/2015/09/The-Fight-Within-2016d.pdf (accessed 3 March 2020).

22.

Nielsen

(2011) Congress factionalism revisited: West Bengal. In: Madsen

(ed.) Trysts with Democracy: Political Practice in South Asia. London: Anthem Press, pp.157–192.

23.

Palshikar

(2004) Revisiting state level parties. Economic and Political Weekly 39: 1477–1479.

24.

Randall

Svåsand

(2002) Party institutionalization in new democracies. Party Politics 8: 5–29.

25.

Renzsch

(2001) Bifurcated and integrated parties in Parliamentary Federations: The Canadian and the German Cases. IIGR working paper no. 4. Kingston, ON, Canada: Queen’s University.

26.

Sadanandan

(2017) Why Democracy Deepens: Political Information and Decentralisation in India. Cambridge: Cambridge University Press.

27.

Singh

(2015a) How Solidarity Works for Welfare: Subnationalism and Social Development in India. Cambridge: Cambridge University Press.

28.

Singh

(2015b) Subnationalism and social development: A comparative analysis of Indian states. World Politics 67: 506–562.

29.

Suri

(2004) Telugu Desam Party: Rise and prospects for future. Economic and Political Weekly 39: 1481–1490.

30.

Thorlakson

(2009) Patterns of party integration, influence and autonomy in seven federations. Party Politics 15: 157–177.

31.

Thorlakson

(2011) Measuring vertical integration in parties with multi-level systems data. Party Politics 19: 713–734.

32.

Yadav

(2000) Understanding the second democratic upsurge: Trends of Bahujan participation in electoral politics in the 1990s. In: Franke

Hasan

Bhargava

, et al. (eds) Transforming India: Social and Political Dynamics of Democracy. New Delhi, India: Oxford University Press, pp.120–145.

33.

Ziegfeld

(2009) Rule of law and party systems-a study of regional political parties in India. PhD Thesis, Massachusetts Institute of Technology, Cambridge, MA.

34.

Ziegfeld

(2012) Coalition government and party system change: Explaining the rise of regional political parties in India. Comparative Politics 45: 69–87.

35.

Ziegfeld

(2016a) Dynasticism across Indian political parties. In: Chandra

(ed.) Democratic Dynasties: State, Party and Family in Contemporary Indian Politics. New Delhi, India: Cambridge University Press, pp.105–135.

36.

Ziegfeld

(2016b) Why Regional Parties? Clientelism, Elites, and the Indian Party System. Cambridge: Cambridge University Press.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.06 MB