One of the widely commented upon features of the Indian political system is the deep fragmentation of its party system, its regionalisation and federalisation over the years. This has given rise to the necessity of running coalition governments in the centre, with regional parties playing a key role in determining who sits in the treasury benches and who is in the opposition.

Commentary in mainstream media these days, however, appear to point to a reversal of this trend with many forecasting that national issues will dominate and national parties such as the Bharatiya Janata Party will ride a wave of anti-incumbency against the ruling Congress and its coalition. Implicitly, this has reduced the discussion about the elections to one around a race between two parties (even two individuals), with the rest being mere appendages (whether or not one party or the other will support either of these fronts led by the bigger parties, before or after the polls). This form of commentary does not do justice to the diversity of India's political party system. 

The process of fragmentation, the nature of regionalisation and its impact on the political system of each state in the Indian nation have been different and variegated. So for example, in Tamil Nadu, national parties (parties defined as those with a relatively pan-India presence as opposed to those that are limited to one or two states) play a third or even fourth fiddle to regional parties which have dominated the system for years. On the other hand, Gujarat, Rajasthan and Madhya Pradesh have evolved a stable bipolar system that has endured for more than a decade.

This variegation and diversity in India’s political system has resulted in a complex arrangement of political parties across the spectrum. It is therefore a rather reductive exercise that the Indian media has generally engaged in during its coverage of the imminent parliamentary elections, projecting the same as if it were a two-horse race, with some pretenders playing spoilers or kingmakers.

The real picture

Many factors and reasons have been attributed to the fragmentation of the political system, that has brought forth a number of new political forces – mostly regional parties – to add to the variegation in this system. We shall attempt to use data tools to depict this fragmentation of the party system and why Indian national elections have increasingly become a complex agglomeration of regional political equations and variables.

The expectation of a First Past the Post System – wherein each constituency would only have one representative, whoever gets the highest number of votes – is that it would result in a bipolar two party system. In political science parlance, this is postulated as Duverger’s Law.

In India’s case, however, the realisation of this expectation varies from state to state. It would suffice to say that only some states have shown a bipolar two party system, whereas many other states have a number of viable political parties representing various identity groups, ideologies, etc.

The question is, could there be a quantitative way to demonstrate and explain the variegation in party systems across India? Enter, the elegant statistical variable - "effective number of parties". This is a revealing measure that provides us information about how many parties effectively contested a seat (on the basis of vote shares). It could also be measured in terms of how many parties effectively won seats in an assembly election in each state.

These two versions of the same variable can be used to calculate the degree of fragmentation of the party system and the level of competition in each state in the country. And as we look into this with the eyes of the "political animal", we could decipher how the electoral story for each state has been playing out across India in the Parliamentary elections. 

To explain, “Effective Number of Parties” defines the number of parties each of which received a substantive vote share in any constituency, calculated using the Laakso and Taagepara method.

Going by that measure, here is a motion chart that depicts the changes in the average effective number of parties (based on party vote shares) per state per Lok Sabha election over time. 

 

Note: To plot Vote Shares over time - select "Order-Alphabetical" in x axis, "Effective_parties_VS" [Effective Number of Parties calculated on vote share received by parties in each constituency, averaged over state] in y axis, and color: unique. To plot seat shares over time, retain x axis, "Effective_parties_SS" [Effective number of parties calculated on seat share received by parties in each state] in y axis and color: unique. To plot Seat vs Vote Shares, select, "Effective_Parties_SS" in x axis and "Effective_Parties_VS" in y axis and color: unique. The viewer is requested to consider the actual years of Lok Sabha polls in the country (1977, 1980, 1984, 1989, 1991, 1996, 1998, 1999, 2004, and 2009) for the purpose of ascertaining the effective number of parties. The rest of the years are merely used in the chart to show “trails”.

One immediate inference from the chart is that the average number of effective parties across states has more-or-less increased from 1977 to the present. In fact, if you ignore some outliers such as Assam (which saw a sudden fragmentation of the party system in 1985, following a period of intense political turmoil and insurgency), it is very clear that there has been a general increase in the number of effective parties from 1991 onwards, in particular.

But how did this fragmentation of vote shares translate into seat shares in each state?

Vote shares vs Seat shares

A look at the effective number of parties based on seat share rather than vote share presents an interesting picture: this measure based on seat shares (the proportion of seats each party receives in a state) shows even greater fragmentation of the party system than the measure that is based on vote shares, particularly after 1991. Why is this so?

It is clear that the parties have tended to form electoral coalitions knowing its inevitability and that is why at the constituency level, in many a place, parties have fielded candidates as part of united fronts or in seat sharing agreements. This has resulted in many parties winning seats in a state even if most constituencies in those state has had competition restricted to a smaller set of parties.

A closer look at the motion chart for elections from 1977 to 2009 reveals how coalitions or their lack have affected seat and vote shares of parties in several states. [States with seats lesser than five seats have been omitted to allow for a minimum threshold of seats; this has been done arbitrarily but without bias].

Thus, for example, in states such as Kerala and Tamil Nadu, coalitions have resulted in parties going in for seat sharing for a number of constituencies, with the effective number of parties (based on vote shares) remaining close to 2 in most seats. But since coalitions guarantee that the parties comprising them will each receive higher seat shares (as they contest in seat sharing arrangements), the effective number of parties seat-wise is much higher as years progress from 1977 onwards.

The era of coalitions set in early in Kerala maturing into a two-front system that was dominated by two parties as the 1990s passed, while it truly arrived in Tamil Nadu in 1998. A similar story is visible in Maharashtra where an alliance system – one involving the Congress and its erstwhile splinter Nationalist Congress Party (and other minor allies) and the other comprising the Shiv Sena and the Bharatiya Janata Party – has persisted since the 1996 elections. It is a similar story in West Bengal as well.

Select State to view graphs of Effective Parties - Seat vs Vote Shares

 
 
 
 

Red - Effective No. of Parties based on Vote Share; Blue - Effective No. of Parties based on Seat Share

Contrast this with the comparison of effective number of parties in terms of seat versus vote shares in states like Uttar Pradesh and Bihar. Here, despite the fragmentation of the party system and the viability of a number of parties, coalition-building has been a rather difficult exercise.

In Uttar Pradesh in particular, as years go by from 1991 onwards, there are effectively four parties contesting every seat and almost all of them are viable seat winners across the state! The unique social coalitions supporting the Samajwadi Party, the Bahujan Samaj Party, Bharatiya Janata Party and the Congress party has resulted in these electoral outcomes.

Bihar, on the other hand, had some degree of success with coalition building in the 1990s. This explains the lessening of the effective number of parties contesting each constituency as parties learnt to share seats before elections, in the mid-1990s. But as alliances began to fray, the effective number of parties contesting each constituency increased and so did the number of parties winning seats!

Interestingly, in states where party support bases are not necessarily determined by clear cut social biases, the presence of a fragmented party system could result in one party garnering a large chunk of seats despite having only marginally higher support. A fragmented party system could fragment opposition to a relatively dominant party, which could thus garner a disproportionate set of seats as compared to its vote share. This is the case with Haryana where a three party system has ensured that a single party dominates in terms of seats. Thus, the effective number of parties based on vote share compared to the same based on seat share tells a unique story for each state.

The reason for the increase in party fragmentation post 1991 has been a well-studied phenomenon – what some political scientists have called the effect of the three Ms, the Mandal commission recommendations and their outcomes, the Mandir (Ramjanmabhoomi/ Babri Masjid demolition issue) and the related mobilization resulting in Hindutva-based politics coming to the fore and finally, market reforms (the onset of economic reforms that changed India’s political economy drastically and created a new and thriving middle class).

Each of these phenomena has had its effect on the party system, with several parties forming and consolidating themselves on the basis of caste and community identities. Some were buoyed by the success of their agitations demanding implementation of the Mandal Commission recommendations, some by the rise of the BJP as principal opposition and later the ruling party, and still others by the adoption of economic reforms resulting in newer articulations and positioning by political parties across the spectrum.

The last phenomenon in particular has possibly seen the arrival of yet another aspiring pan-national force in the Aam Aadmi Party. India’s changing economy and its contradictions, such as the rise to extreme affluence of the traditional big capital-industrialists, a new middle class with newer aspirations and a persisting strata of the indigent has fuelled newer forms of discontent, that has been channelled into support for newer political forces, further augmenting the fragmentation of the political system.

In the above scheme of things, even if little attention is focussed on the complexity and diversity of India’s political system by the mainstream media, it seems that coalitions – both prior and post the elections – are again going to be inevitable. Such appears to be the nature of India’s fragmented and expanding political system, as we delve into data to reveal it in the full glory of its numbers over the years.

Note on Methodology

1)      Data for all the national elections were downloaded from the Election Commission website and scrapped into csv files by members of the online forum, “Datameet”. Link: https://github.com/datameet/india-election-data

2)      Using spreadsheet functions, effective number of parties based on vote share and seat share for each constituency in every election for every state was calculated and averaged out for every state for all elections since 1977. Spreadsheet is hosted here.

3)    Google Spreadsheets and Google Chart APIs were used to generate motion charts for the data downloaded above. View source to view the code.