Abstract
Pre-election alliances have long figured prominently in Indian elections. Has the politics surrounding pre-election alliances changed with the rise of the Bharatiya Janata Party (BJP)? This article analyzes patterns of election alliance formation across India’s states and examines the extent to which these alliances affected the outcome of the 2019 election. It finds that patterns of alliance formation in the run-up to the 2019 election did not, on the whole, differ noticeably from prior elections. Though the BJP’s strong performance muted the impact of some key alliances, election alliances nevertheless had a perceptible impact on the final allocation of seats and appear to have influenced voting behaviour largely as anticipated.
Election season in India invariably means that the drama of pre-election alliances takes centre stage. Ahead of virtually every election, political parties negotiate with one another over whether to form a pre-election alliance, also known as a seat-sharing agreement. When parties conclude a pre-election alliance, they agree not to compete against each other. In a single-member district plurality electoral system like India’s, parties in a pre-election alliance divide up the seats being contested, each fielding candidates only in their designated seats. In doing so, they hope to amass multiple parties’ support behind a single candidate in each race, thereby maximizing the likelihood that the alliance’s candidates win their seats. The success or failure of alliance negotiations can determine who ultimately comes to power after the election.
Just as Indian politics has changed, so too has the politics surrounding pre-election alliances. In India’s first post-independence decades, opposition parties formed broad pre-election alliances aimed mainly at dislodging the dominant party, Congress, from power (Sharma, 1989; Ziegfeld & Tudor, 2017). As the BJP emerged as Congress’ chief rival in the 1990s, it aggressively formed pre-election alliances, often with regional parties, in its bid to expand across India and cobble together stable national-level governments (Sridharan, 2005). And, as Congress reluctantly accepted its decline, it too increasingly embraced pre-election alliances to win elections. Today, competition in many Indian states features two (or more) duelling multiparty alliances.
The 2019 Lok Sabha election took place against a backdrop of the BJP’s rise. Has this development altered how parties approach pre-election alliances or how pre-election alliances impact electoral outcomes? Or, from the perspective of pre-election alliances, was 2019 an election like any other in recent memory? This article examines four questions related to the formation and impact of pre-election alliances in 2019. The answers to these questions shed light on whether the BJP’s rise has fundamentally changed alliance politics in India.
First, were pre-election alliances more or less frequent in 2019 than in recent elections? They were not, particularly compared to the 2009 and 2014 elections. Second, was the composition of alliances dramatically different in 2019 compared to other elections over the past 20 years? Although a few noteworthy new alliances took shape in 2019, few states saw unprecedented pre-election alliances form. Rather, in most states, alliances in 2019 were similar to those formed in other recent elections. Third, did pre-election alliances have their anticipated impact on voting behaviour, impelling a party’s supporters to vote for an allied party when their own party was not contesting? The overwhelming majority of voters whose preferred party did not compete in their constituency, due to an alliance, instead voted for their preferred party’s ally. Fourth, did alliances influence the final outcome of the election? The presence or absence of alliances had a perceptible, if limited, impact on the outcome. Taken together, these findings suggest that although much about Indian politics has changed since the BJP assumed power in 2014, the politics of pre-election alliances so far exhibits much more continuity than change.
Pre-election Alliances in 2019
A number of pre-election alliances took shape ahead of the 2019 election, some new and some longstanding. Table 1 lists the main pre-election alliances in the country’s 20 large states, defined as those sending six or more MPs to the Lok Sabha. The table includes alliances formed by the BJP and Congress, as well as any alliance headed by a party winning at least 10 per cent of the statewide vote. The online appendix provides expansions for the acronyms in the table (Table A1) as well as the number of seats that each party contested (Table A2). Parties winning more than 10 per cent of the statewide vote that did not form an alliance do not appear in the table.
Pre-Election Alliances in the 2019 Lok Sabha Election
In all, at least one major-party pre-election alliance took shape in 11 of India’s large states. Of the 19 alliances involving a major party in a large state, nine included the BJP, eight included Congress and two included neither of the national parties. In some alliances, like the BJP’s alliance with the Rashtriya Loktantrik Party (RLP) in Rajasthan, the senior partner in the alliance ceded very few seats to its minor-party allies. In others, the partners divided seats much more evenly, as in the case of the BJP’s alliance with the Shiv Sena in Maharashtra. In nine large states, no major party formed an alliance. In many of these states, two parties dominate the party system, making the absence of alliances unsurprising. Others, like West Bengal, have frequently had pre-election alliances in the past.
The Frequency of Pre-election Alliances
The 2019 election established, beyond any doubt, that the BJP has (for the foreseeable future) firmly eclipsed its long-time rival, Congress. Did this changed political environment, with an ascendant BJP and flailing Congress, alter parties’ behaviour with respect to the formation of pre-election alliances? More specifically, were alliances formed with any greater or lesser frequency in 2019 as compared to prior elections? To assess the relative frequency of pre-election alliances in 2019, I collected data on alliances formed in Lok Sabha elections from 1998 through 2019 in India’s large states. Because pre-election alliances are virtually always brokered on a state-by-state basis, the unit of analysis is a party in a particular state. In effect, I treat the state units of national parties as separate parties, thereby counting some parties multiple times. For convenience, I refer to these party–state observations as parties, even when they are state units of a party.
As in Table 1, I gathered data on alliances formed by major parties. Throughout, I define a major party as one that wins more than 10 per cent of the national-level vote (in which case it is a major party in all states, regardless of the vote share it wins in the state) or more than 10 per cent of the statewide vote (in which case, it is a major party in the state where it crossed the 10% threshold). The data go back to 1998, since that was the first election in which Congress and the BJP were unambiguously the two main poles around which national politics revolved.
In total, my data include 350 party–state observations distributed across six national elections. Of these parties, slightly more than half (197, or 56%) formed a pre-election alliance. Figure 1 breaks down by year the share of parties that formed an alliance. The light grey portions represent the share of parties that did not form an alliance of any kind. Together, the remaining segments represent the share of parties that formed some kind of alliance. I disaggregate parties that formed alliances into three groups. The first group, represented by the dark grey portions of each bar, consists of alliances between a major party and a minor party (which is any party that fails to meet the criteria for classification as a major party), in which the major party contested at least five-sixths of the seats in the state. 2 This group captures alliances that are not all that different from ones in which a party contests alone—like the BJP’s 2019 alliance with the RLP in Rajasthan in which the BJP contested 24 of 25 seats. Next, the white portions indicate the share of alliances formed between a major party and one or more minor party, but where the major party contests less than five-sixths of the seats. For example, the Communist Party of India (Marxist) (CPM) in Kerala contests far less than five-sixths of the seats because it has historically allied with multiple smaller parties. Finally, the black portions correspond to alliances between major parties, such as the long-standing BJP–Shiromani Akali Dal (SAD) alliance in Punjab.
As Figure 1 demonstrates, 2019 was not an outlier in terms of the frequency of pre-election alliances. Indeed, 2019 closely resembles 2009, which was also an election in which a one-term incumbent government sought re-election. The biggest change in alliance frequency occurred between 2004 and 2009 when the number of parties not forming an alliance increased and the number of major parties forming alliances with each other decreased.
Figure 2 examines the frequency of alliances involving the BJP and Congress separately. The bars in the top portion of the figure indicate the percentage of large states in which the BJP or Congress formed an alliance of any kind. The bottom portion indicates the percentage of large states in which the BJP or Congress formed an alliance with another major party. Here too, 2019 is not an outlier as it was not particularly different from 2009 or 2014 for either party. With respect to major-party alliances, Figure 2 shows that Congress has consistently allied less frequently with major parties than the BJP. Since Congress was, until recently, a leading party in considerably more states than the BJP, Congress often did not need the support of another large party to potentially win an election. In contrast, since the BJP was, for many years, quite marginal in much of Eastern and Southern India, winning seats in these states typically required an alliance with a much larger party.
More recently, the frequency with which the BJP has allied with major parties has declined, whether because other parties have snapped ties with the BJP, like the Biju Janata Dal (BJD) in Odisha, or because the BJP’s rise has marginalized previous allies that were once, but are no longer, major parties in their states, such as the Asom Gana Parishad (AGP) in Assam or the Indian National Lok Dal (INLD) in Haryana. Figure 2 therefore suggests that the modest decline in the frequency of alliances beginning in 2009 that is evident in Figure 1 is largely attributable to the BJP, whose expanding footprint across India has made it less reliant on regional allies.
The Composition of Pre-election Alliances
The previous section demonstrated that the 2019 election was not unusual in terms of the frequency of alliance formation. Might it nonetheless be unusual in terms of the composition of election alliances? After all, the BJP’s rising fortunes brought together long-time rivals such as the Bahujan Samaj Party (BSP) and Samajwadi Party (SP) in Uttar Pradesh. Was this alliance in Uttar Pradesh symptomatic of a broader pattern of alliances between erstwhile rivals in an effort to combat the BJP’s rise?
It was not. Between the 2014 and 2019 elections, alliances in many states changed, though the extent of these changes was not particularly unusual, as described in Discussion A in the online appendix. Few of these changes, however, resulted in alliances that were unprecedented; many instead involved parties that had allied before the BJP’s rise. These changes represented a mix of both strategic advantages and disadvantages for the BJP, as India witnessed neither a rush of smaller parties eager to join the BJP camp, nor a serious attempt to form a comprehensive anti-BJP front. Indeed, anti-BJP alliances failed to form in most states. In short, the BJP’s rise has not prompted a radical change in the composition of pre-election alliances.
To begin with, about half of India saw little change in alliances in the run-up to 2019, suggesting that the election did not bring about a widespread reordering of alliances. Table A3 in the online appendix documents all changes in alliances in large states between the 2014 and 2019 elections. In six states that together account for about a quarter of Lok Sabha seats (Chhattisgarh, Delhi, Gujarat, Madhya Pradesh, Punjab, West Bengal), there were no changes at all in the alliances formed by major parties between these two elections. In six more states, accounting for another quarter of Lok Sabha seats (Assam, Haryana, Kerala, Maharashtra, Odisha, Rajasthan), changes in alliances were relatively minor, involving a major party and one or more minor parties. For example, in Odisha, Congress had contested alone in 2014 but contested in 2019 with the CPM, Communist Party of India (CPI), and Jharkhand Mukti Morcha (JMM), each of which contested only one seat. Though support from small parties can sometimes matter greatly for election outcomes as they provide a large party with the added votes needed to eke out victory in many seats, Congress’ alliance with these small parties did not materially alter political competition in Odisha. Just as in 2014, elections in Odisha in 2019 were largely three-cornered contests featuring the BJD, BJP and Congress (or its small allies).
In India’s remaining nine states, which account for a bit less than half of all Lok Sabha seats, the composition of alliances changed more substantially, involving the formation or termination of an alliance between two or more major parties. Alliance changes involving more than one large party can dramatically alter the competitive landscape. For instance, in Bihar, most constituencies featured three-way competition between the BJP, Rashtriya Janata Dal (RJD) and the Janata Dal (United) (JD[U)] (and each of these parties’ smaller allies) in 2014; but, the rapprochement between the BJP and the JD(U) meant that 2019 saw largely two-way competition.
Even in the nine states where there were substantial changes in alliances between 2014 and 2019, only three states witnessed the formation of alliances that were hitherto unprecedented: Jharkhand, Karnataka and Uttar Pradesh. In all three, alliances formed between major parties that had never before allied—BSP and SP in Uttar Pradesh; Congress and the Janata Dal (Secular) (JD[S]) in Karnataka; and the Jharkhand Vikas Morcha (Prajatantrik) (JVP[M]) with Congress and the JMM in Jharkhand—largely in an effort to combat the BJP.
Elsewhere, major changes in the composition of alliances either involved the absence of alliances altogether or the formation of alliances that had occurred in the past, though not in 2014. In Andhra Pradesh, Jammu and Kashmir and Telangana the absence of alliances was noteworthy. 3 Though all major parties in Andhra Pradesh and Telangana had contested alone at some point since 1998, in 2019 they all contested without alliance partners. Similarly, in Jammu and Kashmir, all four main parties—the BJP, Congress, Jammu & Kashmir National Conference (JKNC) and Jammu & Kashmir People’s Democratic Party (JKPDP)—contested alone, which had never occurred since the JKPDP’s inaugural election in 2004. However, because the JKPDP’s presence in the state was marginal in 2019, competition in Jammu and Kashmir largely resembled the 1998 and 1999 elections when the BJP, Congress and JKNC all contested separately prior to the JKPDP’s founding.
Finally, in Bihar and Tamil Nadu the reordering of alliances in 2019 represented a return to prior patterns. In Bihar, the resumption of the BJP–JD(U) alliance, which competed against an RJD–Congress alliance, signalled a return to the alliance configuration present in 1998, 1999 and 2004. 4 The novel feature of 2019 pre-election alliances in Bihar was the alliance between the JD(U) and the much smaller Lok Jan Shakti Party (LJP), which had never previously occurred. In Tamil Nadu, competition between the BJP–All India Anna Dravida Munnetra Kazhagam (AIADMK) and Congress–Dravida Munnetra Kazhagam (DMK) alliances harkens back to 2004 when these two alliances similarly faced off. Additionally, the BJP and AIADMK were allied in 1998, as were the DMK and Congress in 2009.
Given India’s ever-shifting party landscape, nearly every election features a handful of alliances that are, in some way, unprecedented. The 2019 election was no different; but it was only in a few states that major parties that had never before allied came together. Might the alliance changes between 2014 and 2019 still represent a broader trend of parties either joining hands with the BJP or forming anti-BJP alliances? In fact, alliance changes between 2014 and 2019 represent a mix of developments, some of which benefited and others of which hurt the BJP. In two of the eight states where there were major alliance changes ahead of the 2019 election, the BJP gained major-party allies, namely the JD(U) in Bihar and the AIADMK in Tamil Nadu. In a third, Jharkhand, the BJP gained a minor ally—the All Jharkhand Students Union Party (AJSUP). Meanwhile, in five states, the BJP either lost a major ally—the Telugu Desam Party (TDP) in Andhra Pradesh and Telangana, which contested alone—or faced a new, explicitly anti-BJP alliance (Jharkhand, Karnataka, Uttar Pradesh). Finally, in Jammu and Kashmir, the end of the Congress–JKNC alliance aided the BJP indirectly by splitting the non-BJP vote.
Though changes from 2014 to 2019 do not reveal a uniform pattern of parties deserting the BJP and forming explicitly anti-BJP alliances, there were far more cases of this type of behaviour than there were of parties (successfully) trying to join hands with the BJP to ride its coat-tails to victory. Nevertheless, the formation of comprehensive anti-BJP fronts was, in fact, quite rare. Of course, the formation of broad anti-BJP alliances including all major non-BJP parties is impossible in states where the BJP is one of only two major parties (Chhattisgarh, Gujarat, Madhya Pradesh, Rajasthan), and implausible in states where the BJP cannot credibly vie to win elections (Andhra Pradesh, Kerala, Punjab, Tamil Nadu, Telangana). But, in the remaining 11 large states, there was little evidence of broad-based efforts to counter the BJP’s rise. In Assam and Bihar, the BJP gained allies in 2019; and in five other states, major parties chose to contest alone rather than ally against the BJP (Delhi, Jammu and Kashmir, Odisha, Punjab, West Bengal). In four of these five states, some of the major parties in question had either allied with each other in the past (Jammu and Kashmir, West Bengal) or openly discussed, but failed to conclude, an alliance in 2019 (Delhi, Punjab), meaning that anti-BJP alliances were realistic possibilities. In a sixth state, Haryana, Congress could potentially have allied with a number of smaller parties, including the INLD, Aam Aadmi Party (AAP), or the newly formed Jannayak Janta Party (JJP). Finally, in both Uttar Pradesh and Maharashtra, the major non-BJP alliances failed to include smaller but still sizable BJP opponents, namely Congress in Uttar Pradesh and the Vanchit Bahujan Aghadi (VBA) in Maharashtra.
Most regional parties in India would have benefited from a hung verdict and the formation of a non-BJP government comprising a weak Congress and a host of regional parties. Yet, as noted above, relatively few of these parties engineered comprehensive alliances aimed at taking on the BJP. The most obvious explanation for small parties’ inability or unwillingness to politically isolate the BJP in alliance formation is the primacy of state-level electoral considerations. In much the same way that a party’s participation in national-level cabinets in India ‘is mediated through the logic of politics of the state in which it is located’ (Kailash & Arora, 2016, p. 68), the formation of pre-election alliances continues to reflect the logic of state-level competition far more than it does a national-level story about the BJP’s rise. For example, in Punjab, Congress gambled (correctly) that it could win Punjab alone, without an alliance with AAP. In eschewing an alliance with AAP, Congress risked losing the election but also ensured that it did nothing to arrest AAP’s decline in the state. Meanwhile, in Delhi, where Congress is fighting to avoid obsolescence, it drove a hard bargain in negotiations with AAP, clearly preferring the possibility of losing the election to relegating itself to being AAP’s minor ally. In other words, though the BJP has improved its electoral fortunes across much of India and was widely expected to emerge as the largest party in the Lok Sabha in 2019, alliance formation in 2019 did not appear to reflect national-level concerns about the BJP’s rise. Rather, state-level considerations continued to drive alliance formation. As a result, although the 2019 election brought about a few noteworthy and unprecedented alliances, the composition of most pre-election alliances was rarely all that surprising.
The Transfer of Votes to Allies
The previous two sections showed that patterns of pre-election alliance formation were not particularly unusual in 2019. But, what about the impact of pre-election alliances? When parties form pre-election alliances, each party hopes that its supporters will vote for its allies in constituencies where it is not contesting and that its allies’ supporters will vote for it in the constituencies where it is competing. Did votes actually transfer in this way from one ally to another in 2019? Using data from the 2019 Indian National Election Study, I examine the frequency with which respondents residing in constituencies where their preferred party was not competing because of an alliance reported that they instead voted for their party’s ally. Discussion B in the online appendix provides additional details on this analysis.
I determine a respondent’s preferred party by taking advantage of a series of questions asking respondents how they feel about specific parties. Respondents were asked to indicate whether they like a lot, somewhat like, like very little or don’t like at all several large parties active in their state. I code a respondent as preferring a party when he or she rates a party more favourably than any of the other parties about which he or she was asked. Of the 9,744 respondents in states with major-party pre-election alliances, 55 per cent preferred one party to all others. Discussion C uses a different survey question to determine a respondent’s preferred party. The results are similar to those presented below.
Among those whose responses reveal a party preference, I exclude those residing in constituencies where their preferred party was competing. Of the remaining 1,135 respondents, 82 per cent reported voting for their preferred party’s ally. (This figure includes a significant number of respondents who report voting for their preferred party, even though they reside in a constituency where it did not compete. I presume that most of these respondents report having voted for their preferred party because they treat a vote for their preferred party and its ally as largely interchangeable. Discussion D in the online appendix shows that excluding these respondents does not substantially alter the results presented here.) To put this rate of vote transfer in context, among those who could vote for their preferred party, 89 per cent did so. Thus, the proportion of votes that transferred to allies was not that much lower than the proportion of votes that a party won from those respondents who preferred it to all other major parties.
Figure 3 breaks these data down by state. It includes states for which there were at least 75 respondents who could not vote for their preferred party because it was in an alliance. The grey bars indicate the share of respondents in a state who voted for their preferred party (because they voted in constituencies where they could do so), while the black bars indicate the share who voted for their preferred party’s ally (because they could not vote for their preferred party). Because the samples for each state are relatively small, the patterns in Figure 3 should be treated with caution, particularly for Jharkhand and Maharashtra, where there were fewer than 100 respondents.
Votes transfer at very high rates in all states except Jharkhand. In five of the six states in Figure 3, a two-sample test of proportions indicates that the share of respondents whose votes ‘transferred’ is effectively indistinguishable from the share of respondents who voted for their preferred party. The only state where the difference in proportions is statistically significant is Jharkhand, where the share of respondents voting for their preferred party’s ally (when they could not vote for their preferred party) is much lower than the share voting for their preferred party (when they could).
Tamil Nadu and Uttar Pradesh are the only two states for which it is possible to break the data down by party. In both states, rates of vote transfer were similar between the two main parties for which there are ample respondents. In Tamil Nadu, 78 per cent of AIADMK and 81 per cent of DMK supporters reported voting for their preferred party’s ally, while in Uttar Pradesh, 86 per cent of Samajwadi Party supporters and 91 per cent of BSP supporters voted for an ally. The similar—and very high—rates of vote transfer for both BSP and SP supporters contrasts with much of the post-election reporting claiming that a failure of vote transfers doomed the alliance (Dikshit, Mathur, & Lalchandi, 2019; IANS-English, 2019).
Although Jharkhand stands out as a potential exception worthy of further investigation, the main takeaway from this analysis is that most voters who supported a party voted for that party’s ally when their preferred party was not on the ballot. Of course, only a little more than half of all respondents in states where there were major-party pre-election alliances revealed a clear preference for one party over all others. For the remaining respondents, the question of vote transfers is, in effect, irrelevant since they do not have a preferred party from whom to take cues. Instead, the presence of an alliance may have shaped their voting behaviour in different ways, beyond those explored here.
The Impact of Pre-election Alliances on Seat Shares
Although most respondents in the Indian National Election Study reported having voted for their party’s ally when they could not vote for their preferred party—just as parties hope they will when forming pre-election alliances—post-mortem election analyses frequently referred to pre-election alliances as having ‘failed’ in 2019. In particular, many have offered especially damning assessments of the alliances that took shape in Karnataka and Uttar Pradesh with the aim of defeating the BJP (Beg, Pandey, & Kare, 2019; Shastri, 2019).
Is this conclusion about the ‘failure’ of alliances warranted? At first glance, it would seem so. After all, if alliances were formed to defeat the BJP, then clearly they failed; the BJP’s seat tally increased in Karnataka and declined only modestly in Uttar Pradesh. However, a closer look reveals little evidence that alliances failed in any meaningful way. When parties failed to defeat the BJP, pre-election alliances do not appear to have been the culprit. Rather, pre-election alliances in 2019 did what they were intended to do—increase the allies’ likelihood of winning—even if the BJP’s strong showing in states such as Karnataka and Uttar Pradesh muted the alliances’ impact on the election outcome. Indeed, pre-election alliances across India had a noticeable, if less than seismic, impact on the distribution of legislative seats.
In India, parties form pre-election alliances primarily to avoid splitting the vote against a common adversary and losing seats to a party that wins less than the allies’ combined support. Even if parties in an alliance attract no additional support than they otherwise would have won if they contested separately, the formation of an alliance can dramatically enhance the likelihood that the allies’ candidates win the seats they contest. But, if the allied parties’ opponent wins a majority of the vote in a constituency, there is nothing that an alliance can do to materially alter the translation of votes into seats for the benefit of the alliance partners. Thus, the BJP’s ability to win most seats in Uttar Pradesh and nearly all seats in Karnataka does not necessarily mean that the anti-BJP alliances did not work, only that an alliance alone was insufficient to defeat the BJP. If candidates from the alliance were closer to winning with the alliance than they would have been without the alliance, then the alliance did what it was intended to do.
To examine the impact of pre-election alliances in 2019, I begin by looking at the two states where high-profile pre-election alliances formed with the aim of diminishing the BJP’s seat tally: Karnataka and Uttar Pradesh. In both states, the parties in the respective alliances won combined vote shares in 2014 that exceeded those won by the BJP, leading many BJP opponents to hope that these alliances could dramatically reduce the BJP’s seat tally in these states.
Figure 4 presents changes from 2014 to 2019 in the constituency-level vote share won by the leading candidates from among the 2019 alliance partners in Karnataka and Uttar Pradesh. For example, in the Mysore parliamentary constituency in Karnataka, the Congress candidate in 2014 won about 41 per cent of the vote and the JD(S) candidate won about 12 per cent. The Congress candidate was therefore the leading candidate from among the parties that formed the 2019 anti-BJP alliance. In 2019, the Congress candidate, this time contesting in alliance, won about 42 per cent of the votes. Thus, the difference between the leading alliance candidate’s vote shares in 2019 and 2014 was about 1 per cent (42% minus 41%). Figure 4 presents these vote share differences from each constituency, indicating the percentage of constituencies in a state whose vote share differences fall into a particular range of values. To illustrate, the far left bar in the Karnataka plot indicates that 7.1 per cent of the constituencies (or two constituencies) saw a decline of 10–15 per cent in the vote share won by the leading alliance candidate between 2014 and 2019, while the far right bar in the same plot indicates that 3.6 per cent of the constituencies (that is, one constituency) saw an increase of 15–20 per cent. The dashed vertical line in both sub-plots indicates no change between 2014 and 2019.
In all, just under half of the constituencies in Karnataka saw an improvement in the vote share won by the alliance’s candidates between 2014 and 2019. If an alliance is supposed to improve its members’ chances of winning, then the decline in vote share witnessed by many Congress and JD(S) candidates in Karnataka would, at first blush, seem to suggest that the alliance did not work. However, after accounting for the BJP’s improved performance in 2019, the data suggest that the Congress–JD(S) alliance largely operated as expected.
In 2014, the average constituency-level vote share won by the leading 2019 alliance candidate in Karnataka was about 41 per cent. Meanwhile, the average vote share won by Congress and the JD(S) candidates combined in 2014 was about 51 per cent, suggesting that an alliance between the two parties would potentially yield, on average, about a 10 per cent increase in the vote share won by the alliance’s candidates. However, between 2014 and 2019, the combined statewide vote share for Congress and the JD(S) in Karnataka declined by about 9 per cent, just as the BJP’s increased by about 8 per cent. In other words, the gains made by consolidating Congress and JD(S) supporters behind a single candidate in each constituency were more or less cancelled out by the parties’ losses (mainly to the BJP) between 2014 and 2019. In light of these losses (and the BJP’s gain), most Congress and JD(S) candidates should have seen little improvement in their constituency-level vote share in 2019. Indeed, in most constituencies, the change in the leading alliance partner’s vote between 2014 and 2019 was relatively modest, somewhere between a 5 per cent decline and a 5 per cent increase. Thus, the data suggest that the alliance in Karnataka likely worked as intended; the alliance consolidated both parties’ support behind one candidate in each constituency but was thwarted by the BJP’s increase in popularity.
One possibility, suggested by Shastri (2019), is that the alliance itself was responsible for the decrease in overall support for Congress and the JD(S) in 2019. If voters opted for the BJP because they opposed the idea of a Congress–JD(S) alliance, then the alliance would have failed to achieve its intended aim of increasing the alliance partners’ likelihood of winning. Unfortunately, this hypothesis is difficult to test, particularly since there are so few respondents from Karnataka in the vote transfer analysis. However, given that the BJP made substantial gains in most large states in India, and that the individual-level data suggest that most supporters transferred their votes as expected, it seems very plausible that the BJP was destined to see gains in Karnataka, with or without the Congress–JD(S) alliance.
The right panel of Figure 4 presents a comparable set of analyses for Uttar Pradesh, comparing the performance of the leading party from among the parties of the Mahagathbandhan (or Grand Alliance)—BSP, SP and Rashtriya Lok Dal (RLD)—in 2014 with the candidate from the alliance in 2019. In 2014, the average constituency-level vote for the leading candidate from among the allied parties was about 26 per cent, while the average combined constituency-level vote for all the alliance partners was about 43 per cent, indicating that an alliance could potentially yield an average improvement of about 17 per cent. However, in 2019, the parties of the Mahagathbandhan combined to win about 5 per cent less of the overall vote as compared to 2014. Unlike Karnataka, statewide vote losses were far less than the expected constituency-level gains from the alliance. Back-of-the-envelope calculations would suggest about a 12 per cent average gain at the constituency level in 2019 (17% gain from the alliance minus 5% loss in popularity from 2014 to 2019).
In fact, Figure 4 indicates that the greatest percentage of constituencies in Uttar Pradesh falls exactly in the 10–15 per cent range. Most candidates from the Mahagathbandhan fared noticeably better in 2019 as compared to the leading party from the alliance in 2014, when the parties contested separately. This aggregate-level finding corroborates the individual-level data showing that the BSP and SP supporters largely transferred their votes to the other party when their preferred party was not competing in their constituency. However, because most BJP candidates in Uttar Pradesh improved their performances between 2014 and 2019, the Mahagathbandhan could do little to deprive most of these BJP candidates of victory. Indeed, 41 BJP candidates won more than 50 per cent of the vote, meaning that no amount of opposition coordination in those constituencies could have defeated the BJP. Nevertheless, the Mahagathbandhan had an impact on the election result. The BJP won a healthy majority of seats from Uttar Pradesh in 2019, but fewer than in 2014. This modest decline in seat share occurred alongside an increase in the BJP’s vote share. Absent the Mahagathbandhan, the BJP would likely have won most of the seats won by the BSP and SP in 2019 and seen its already enormous seat tally from 2014 increase further. The formation of the Mahagathbandhan likely deprived the BJP of 10–15 seats, a conclusion substantiated in Discussion E in the online appendix.
Pre-election alliances formed in Karnataka and Uttar Pradesh failed to make a major dent in the BJP’s seat shares in those states. Does this mean that alliances had no impact on the overall distribution of seats in the 2019 Lok Sabha? No, the formation of (or failure to form) pre-election alliances had a perceptible effect on the election’s outcome, helping to enhance the BJP’s legislative majority. Alliances between Congress and a handful of regional parties could have plausibly reduced the BJP’s seat tally to about what it was in 2014, though it would have required a very unlikely collection of alliances and broken alliances to deprive the BJP of a legislative majority in 2019.
The BJP won 157 seats in states where there were multiparty systems and either the BJP had an alliance with another party or there were parties competing against the BJP that were not in an alliance. In these 157 seats, the election outcome could potentially have been different if the BJP had failed to form the alliances that it did or if other parties had formed alliances that did not ultimately materialize. To get a more precise sense of the BJP’s vulnerability in these seats, I count the number of seats that meet one of the following three criteria:
The BJP candidate won with less than 50 per cent of the vote in a state where there were multiple sizable non-BJP parties that were not in alliance (Jammu and Kashmir, Odisha, Telangana, Uttar Pradesh, West Bengal); The BJP candidate won by a margin of less than 5 per cent and the BJP was the senior ally in an alliance with much smaller parties (Assam, Jharkhand); The BJP candidate won by a margin of less than 15 per cent and the BJP was in an alliance with a larger party or party of roughly equal size (Bihar, Maharashtra, Punjab).
In states where the BJP’s allies were small, the added votes from these alliance partners were likely not needed for the BJP candidate to win, unless it won by a slim margin. However, when the BJP’s ally was larger, even wider margins of victory might not have guaranteed a BJP win if its candidates had not received votes from allied parties’ supporters. For these reasons, to identify seats where the BJP was vulnerable, I employ different margins of victory for BJP candidates based on the size of their alliance partners.
In total, 61 seats meet one of these three criteria. In many of these, the likelihood of an alliance changing the outcome was low. But, the inclusion of Congress either in statewide alliances (with the BJD in Odisha, Trinamool Congress in West Bengal, BSP and SP in Uttar Pradesh) or in ad hoc adjustments in a few seats in Assam, Jammu and Kashmir, Punjab and Telangana could plausibly have diminished the BJP’s seat tally by 23 seats, bringing the BJP down to almost exactly the number of seats that it won in 2014 and reducing it to a very slim majority. Additionally, the inclusion of the VBA in the Congress–NCP alliance in Maharashtra could have potentially defeated the BJP in another five seats. In all, the formation of more extensive pre-election alliances could have cost the BJP as many as 28 seats. 5 Discussion F in the online appendix provides further details on how I arrive at these numbers. Of course, there is no guarantee that candidates from these hypothetical alliances would have defeated BJP candidates, but the electoral arithmetic highlights the role that the absence of pre-election alliances potentially played in providing the BJP with its enhanced legislative majority in 2019.
Finally, there were seven more seats where the BJP might have lost without its ally's support. These were seats where the BJP won with relatively narrow margins of victory that were plausibly provided by votes from a sizable alliance partner. One complicating factor when considering the impact of the termination of an alliance in states like Bihar and Maharashtra is that in the absence of an alliance, the BJP would likely have contested more seats, some of which it might have won. Thus, the loss of some seats thanks to the end of an alliance might have been partially or wholly compensated for by gains in seats that the BJP did not contest. Estimating the number of seats that the BJP might have lost in the event that some of its alliances failed to materialize is therefore difficult. But, the possibility remains that it would have lost a handful of the seats that it won in 2019 had it not concluded alliances with other parties, the JD(U) in particular.
All told, the formation or absence of pre-election alliances almost certainly had a noticeable impact on the distribution of seats in 2019. However, because a different pattern of alliances in any one state could only have influenced the outcome in a few seats, a dramatic change in the BJP’s seat total could not have occurred unless parties in many different states made different choices with respect to alliance formation. It would have taken an extraordinary confluence of events—such as an India-wide effort among non-BJP parties to coalesce against the BJP—for a different set of alliances to have deprived the BJP of its legislative majority. This conclusion does not mean, however, that alliances ‘failed’—only that there is little that a pre-election alliance can do, even when it functions as intended, to defeat candidates who enjoy the support of a majority of their constituency’s voters.
Conclusion
This article has shown that patterns of pre-election alliance formation in 2019 were not dramatically different from patterns in other national elections held since the late 1990s. In 2019, pre-election alliances were neither noticeably more nor less common than before, and the composition of alliances was not all that unusual. Additionally, the pre-election alliances that took shape operated more or less as expected. Most voters who could not vote for their preferred party voted instead for its ally; alliances seem to have enhanced their candidates’ likelihood of victory (even if they did not ultimately win); and the presence or absence of alliances likely had a noticeable, if modest, impact on the election outcome. Together, these findings suggest that alliance politics in 2019 looks much the same as it has in prior elections.
The recent past is littered with examples of pre-election alliances that failed to produce their hoped-for effects: the adjustment between Congress and the Left Front in West Bengal in 2017; the Mahakootami, which included Congress and the TDP, in Telangana in 2018; and the alliances in 2019 in Uttar Pradesh and Karnataka. In all cases, the party winning the election won by too large a margin for pre-election alliances to keep it out of power. But, declaring an end to the usefulness of pre-election alliances is wholly premature. Indeed, as the BJP establishes itself as the leading party in a growing number of states, alliances may be some parties’ only chance to retain or gain power. In states such as Jharkhand, Karnataka and Uttar Pradesh, the BJP would only have to experience minor declines in its vote share before a united opposition could reap considerable dividends from electoral cooperation. The same is true in Delhi, where the much talked about alliance between Congress and AAP failed to materialize. Pre-election alliances may also prove necessary if the Trinamool Congress and BJD wish to remain in power in West Bengal and Odisha, respectively. Although future elections may feature many more cases of alliances that fail to live up to their members’ expectations, pre-election alliances are likely to remain as important as ever as Congress and India’s many regional parties struggle to win elections amid the BJP’s rise.
Footnotes
Declaration of Conflicting Interests
The author declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.
Funding
The author received no financial support for the research, authorship and/or publication of this article.