Monthly Archives: June 2013

Relevance of National Parties

Let us consider large States that have over 12 Members of Parliament in the Lok Sabha. Further, in these States, let us grade the INC and BJP as electorally relevant or otherwise. A party is determined relevant(rated 1) in a given State if it’s within the first two positions electorally. Else it gets a 0. In slightly complicated situations — such as the BJP in Punjab/ Maharashtra — the party gets a 0.5. Typically, there are states where both are relevant, both are irrelevant and some where only the INC is (Andhra Pradesh, for instance). As discussed earlier, there aren’t many parts of the country where the BJP is present and the INC is not.

Once we do this simple classification of states, a picture of each party’s natural constituency and electoral alliance emerges. Let’s begin with the BJP since it has the problem of a smaller footprint across the nation. The first fact that strikes one when looking at the states where the BJP is relevant and where it isn’t is this: a reinforcement of the conventional opinion that the BJP is an upper caste party. In states where the party is relevant, the average percentage of upper caste* presence is 22.7%. In states where the party is irrelevant, the upper castes form 9.7% of the population. That’s a clear 13 percentage point divergence which can’t be dismissed even if the rather wide range of the contributing data points is taken into account. Just as crucial is the opposite trend amongst Muslims, confirming the other piece of conventional wisdom that the BJP has a problem with Muslims. Or, the Muslims have one with the BJP. In states where the BJP is relevant, the Muslims, on average, form 10.55% of the population; where the party is irrelevant, the corresponding number is 21.12%. The divergence again is 10.57 percentage points. Similarly, the divergence of the OBC population is about 5 percentage points against the BJP; that is, on average, a larger OBC population in a given state compared to the norm is detrimental to the BJP. Dalit(SC) votes are distributed evenly amongst both sets of States while the divergence of the Scheduled Tribes seem to favor the BJP.

Another simple way to measure is to measure basic correlation coefficient for the various groups with BJP’s relevance. They stack up along the lines of the above analysis — the Upper Caste population strength and the Tribal population are positively correlated while that of the Muslims and SCs are negatively correlated. The OBCs have a very slight negative correlation and can be considered as neutral given the data is unclean. If one were to compare this classification with that of the INC, a remarkably similar structure emerges. The INC too is a party that depends on higher percentage of upper castes and a slightly lower presence of both OBCs and Dalits for its relevance. The correlation coefficients of upper castes and Tribes are highly positive while that of Dalits and OBCs are highly negative with INC presence in these states. Unlike the BJP though, the Muslim vote does not impact the INC’s relevance negatively.

The significance is not just in the similarity of the relevance of these two parties but in that of their irrelevance. Both parties start losing relevance gradually when the OBCs and Dalits put together cross 60% of the population. The BJP loses it quite dramatically near the threshold while the INC starts losing it slightly more gradually owing to some Muslim presence. This analysis, as discussed previously, does not mean that the Dalits and OBCs don’t vote for the main national parties. But simply that their demographic strength beyond a certain threshold makes a caste formation that’s driven by traditional upper caste polity very difficult to sustain. There is space for OBC and Dalit driven formations — resulting in SP, RJD, JDU, BSP, DMK, AIADMK etc. It is also the case in these states of Bihar, UP and Tamil Nadu that the OBCs and Dalits when put together exceed 75% of the population. In the case of Tamil Nadu, they exceed 90%, rendering either of these parties outside the top 4.

*Upper Caste for the purpose of this analysis is the set of all population excluding Dalits, OBCs, Tribals and Muslims. Christians have been excluded in the case of Kerala.

First Past the Post & Pyrrhic Victories

The physics of a two cornered contest varies vastly from that of a multi-cornered contest. That’s common sense and quite a trivial observation. However, the way relative strengths of the competing parties in such a contest work is interesting.

Consider an election where there are only two main players — like say an Obama v Romney contest. The American polling data usually reports favorability and unfavorability numbers of both candidates. The former usually (though not always) has high correlation with the likely vote share reported for each candidate and is rarely interesting. The latter measure though is what Data Analysts in either camps use to decide where to spend their resources in trying to convince people. This convincing is usually along the lines of ‘the other guy is worse than you think.’  In such a contest the task is simple and straightforward as a concept; it’s far more difficult to implement it however, especially when both sides have unending resources as they did in 2012.

In a multi-party Parliamentary democracy where voting has high correlation with demographic data, the asymmetry in unfavorability rating can be even more debilitating. In the most basic example, if there is a three-way split and and say the leading party has a 5 percentage point lead over its nearest rival and about 10% of the voters are undecided or are willing to shift allegiance, under normal circumstances, one’d expect the leading contender to be more comfortable than in a bipolar contest. But the difference between a bipolar contest, say in the United States, and Indian elections is that the reason people swing are very different. The American swing voter is undecided because he/she is more often equally likely to vote either way. The Indian voter maybe undecided but the decision to not vote a particular way is often a given. For example, in the last General Elections of 2009, the “swing voter” of Uttar Pradesh was Muslim; swinging from SP to INC and giving the party an unexpected 20 MPs from the state. In other words, the undecided voter was not going to vote BJP anyway and therefore was only undecided within one half of the contest. To a lesser extent, the upper caste Hindu was undecided too;  swinging towards BJP from the BSP if the previous Assembly Elections were a benchmark. That partly explained the 10 seats that the BJP won. This upper caste Hindu wasn’t going to vote SP just as the Muslim wasn’t going to vote BJP.

Therein lies the paradigm of recent Indian elections for the two major parties: the INC may face significant corruption charges and have low favorability numbers owing to poor governance but the BJP possibly has a basic level of high unfavorability rating among sections of population that is far more difficult to surmount. That is, it’s more difficult for the BJP to win any election where there isn’t a bipolar contest than it is for the INC. Even worse for the BJP, it’s easier to lose what it holds given the above structure. This hypothesis also explains another phenomenon: by definition, in such a scenario, those who support the BJP are more committed and ideologically inclined. To lesser degree, it could be argued, it’s true of the candidates themselves. That is to say, the conviction threshold required of the BJP cadre and its office bearers in their own ideology and in the collective outpouring of their animosities towards their opponents is far higher than is required for the INC. Conversely, therefore, the INC cadre and its office bearers are less likely to be as ideologically committed, more likely to view the politics as a means and suffer the million issues that crop with it.

If the above analyses were true, the certainty of ideological commitment of one party will alienate even the genuinely undecided: such as  those who aren’t voting exclusively on caste lines mostly because they are upper caste themselves. And if they aren’t part of the choir already, the aggression of the zealous can only be counter-productive. The reason the undecided are that is not because they cannot decide but that they are unsure how to weight a complicated moral choice; in which the silence of the mercenary is a better electoral weapon than the aggression of a bigot.

Quarterly Results vs Election Cycles

Consider three large states: Maharashtra, Bihar and Uttar Pradesh. In all these states there are four major parties in contention. Among the top four contenders in each state, the INC and the BJP feature as national parties. There are two regional parties in each of these states. However, the difference in the type of those regional parties and the way in which the BJP and the INC engage in these states is quite revealing.

The INC, almost across all of India attempts to build a rainbow coalition. Typically its vote base includes some Dalits and Upper Castes in varying degrees and to a lesser extent some OBCs (likely the non-dominant OBCs in that region) when the party is doing well. The BJP’s main appeal is also similar — with the mix of OBCs increasing in case of a Hindu consolidation. The INC when under such pressures on its core vote, typically leans on Muslims to counter the BJP. The equilibrium shifts to one side or another based on local factors and may throw wildly different results. A problem arises for the two national parties when one coalition bloc in the above arithmetic is a bit too strong rendering their overall appeal vulnerable to dominant caste formations.

Typically, it begins with an OBC mobilization in almost all of India and that’s particularly true in the Gangetic plain. In Bihar there are two OBC formations lead by regional players that are dominant while the INC and BJP add to them. The Bihar Dalit formation is not as well organised as it should be given the percentage of Dalits in that State is one of the highest among large states. In Uttar Pradesh, however, the two main parties happen to be OBC and Dalit formations — rendering coalitions impossible without hurting their own longterm prospects. The structure also makes one wonder if the Dalits in Bihar have ceded space to the JDU or if the other OBCs in UP have ceded space to the BSP. However, a cursory look at the demographics makes it plain: Bihar’s Yadav population is 11% while Muslims form 16%; UP’s Yadav population is 16% while Muslim population is 18%

Though the numbers are similar, the Bihar numbers by virtue of falling slightly short give room for another OBC dominated party much more readily than in UP. That is, the SP can afford to slip a bit more among Muslims than the RJD can afford to. And even when the slip is steep, the natural social order gives the SP enough chances to comeback in UP than when RJD gets usurped in Bihar. 

The BSP because of its social base cannot ally at all with either the SP or the INC — the former has an antagonistic base and the latter an overlapping and hence competing one. The INC by having flashes in the pan such as winning 20 MPs in 2009 General Elections is still not completely resigned to being a bit player and therefore does not want to surrender its Muslim votes and ally with the SP. That leaves the BJP is a four cornered contest by default in UP — a far better proposition for the party than what it faces in Bihar. The INC by virtue of its lack of strength and therefore fungible alliance ability in Bihar, poses a far more significant short term problem to the BJP.

Maharashtra on the other hand, offers a completely different picture. The presence of Dalits and Muslims in the State is slightly lower as a percentage of population when compared to Bihar and UP. The INC, interestingly, takes a completely different role in the state — it has a strong base among OBCs. And because of the OBC base and the party’s ability to not completely antagonize either the Dalits or the Muslims, the life of every other formation seems short lived and difficult. That also explains the long tradition of Dalit activism in the state that has remained at a certain level of electoral success (or the lack thereof, depending on one’s viewpoint). Even more surprising is that its alliance partner is the NCP, another party with a strong OBC base. The two parties on the other side of the coalition — BJP and SS — have limited specific caste reach beyond the upper castes and urban voters. The BJP-SS alliance in the state therefore is a pretty poor one for both the parties in the absence of a wave against the INC or Muslims or someone else. It’s further complicated by the fact that the five/six regions of Maharashtra are so different from each other. If one were to look at Maharashtra from the BJP HQ, praying of an NCP-INC break up and a social upheaval seems the best approach. Or, maybe the BJP should look to understand the long history of Dalit movements in Maharashtra’s hinterland.

The moral of the story seems to be: when a national party gets in an alliance with a regional party in a state, the probability of the national party ceding ground to the regional party is far higher than the reverse case. Except when this does not happen. And it seems to not happen only when the source of strength is the mid-section of the caste pyramid. Did someone say Mr Modi was an OBC leader?

Numbers: Stacked against the BJP

The BJP is 32 years old; understandably, it has a smaller footprint across the country compared to the INC that has been in the politics business for four times as long. For example, there are large swathes of India where the BJP does not even have a single MP representing it in Parliament. The contiguous states of Kerala, Tamil Nadu, Andhra Pradesh and Orissa did not elect even one BJP member of Parliament in the 2009 General Elections. Another large state sharing a boundary with Orissa, West Bengal, has just one BJP MP — Jaswant Singh — who contested almost  as a Gorkha Janmukti Morcha candidate in Darjeeling. That makes the BJP irrelevant to the extent of drawing a blank in geographically contiguous 165 constituencies. Things may change and the BJP may indeed do well in 2014 in these States, but it’s unlikely to translate into more than 10 MPs even upon improving dramatically, given the organisational structure of the party in those states and First Past the Pole(FPTP) System of the elections.

Further, the BJP has been in fourth position or worse in at least 28% of the constituencies in Uttar Pradesh. It is reasonable to assume the probability of anti-incumbency benefiting Parties in second and third positions is higher than that of the contender in fourth position accruing gains sufficient to win. If one adds the other States where the BJP is not likely to win significantly — such as Jammu & Kashmir, Tripura, Nagaland, Manipur — roughly 200 seats have a very low probability of less than 0.15(more on the method later) for a BJP victory. That makes the party competitive (P(V) > 0.3) in 342 constituencies. A hit rate of 50% in these constituencies, an unbelievably  difficult task in itself, would still give the party only 170 seats. Whether that’s sufficient to draw allies is a question that the remainder of this post and the ensuing series of posts will attempt to explore.

The long term problem of the BJP, however, is that its appeal is in conflict with the FPTP in a multi-party democracy. Especially one in which the multiple parties and their appeal have high correlation with the demographic data. Or stated in an attention grabbing non academic way, it’s not an accident of history or the Party’s relative recent birth compared to its biggest rival that has rendered the BJP geographically limited. This post’s hypothesis is that there are structural reasons for this limitation. Before examining and illustrating the hypothesis — here are a couple of pointers,

  1. Among large states (Parliamentary Seats of 13 or higher) where the BJP is either relevant or a dominant player, the population of Muslims is between 6% and 14%. Something strange happens at either ends of this band.
  2. In these States, there is almost always a direct contest with the INC. There isn’t a single large State where the BJP is a dominant player when the other dominant player happens to be a regional party.

The hypothesis that one is tempted to put forward is as follows: when the population of Muslims in a large state exceeds 14%, it becomes increasingly likely that a regional party surfaces with enough cache from one of the dominant castes to offset the balance that the BJP and Congress are otherwise able to achieve between caste combinations within Hindus and across Hindus and Muslims in FPTP system.

Consider Bihar and Uttar Pradesh, with over 16% and 18% Muslims respectively to demonstrate this effect. In these states, the natural equilibrium seems to be a four cornered split among five groups — Muslims, dominant OBC, other OBC castes, upper castes and Dalits. Typically, the Dalits and the dominant OBC group are antagonistic on the ground and hence their alliance is unnatural. The upper caste groups are typically smaller compared to the other groups. The other OBCs are most likely to be antagonistic to the dominant OBCs and amongst themselves in varying degrees. With the internal fracture set this way, the Muslim population at about 14% seems to reach a certain criticality where it cannot be treated as another OBC group and attains stature tending to either the dominant OBC or Dalits. This allows space for a regional party to become viable state-wide. The combinations can be any one of the many possible and that’s the raison d’etre for RJD, SP, JDU and to a lesser degree, the BSP. That’s also the reason why parties similar to these are not as prominent in the contiguous state of Madhya Pradesh, which has a Muslim population of 6.37% — one that can be treated as that of another caste group.

The problem for the BJP is that while all Hindu caste sub-groups in the 4 categories are neatly divisible into competing groups, the Muslim population stands as a single separate entity that’s unlikely to vote for them. So, the asymmetry of caste formation introduced by virtue of this criticality renders the BJP in a losing proposition under normal circumstances. The 90s Hindu mobilization were abnormal times where possibly antagonistic groups on the ground — such as competing OBCs — voted BJP. But that sort of a flash in the pan will be rare and takes either a social upheaval or an enlightened electorate to happen. Neither of those seem likely in 2014.