Economic voting has traditionally been perceived as anathema in Indian elections, with emphasis being placed on identity-based voting—on caste, religion, or region. However, a growing body of research centered on the pan-Indian ascendance of the BJP points to the increasing salience of economic considerations, or the persistent presence of economic ideology as a second axis of political competition next to the politics of recognition. According to one strand of this research, the BJP’s pan-Indian success was the result of an electoral strategy that judiciously combined Hindu-ness (Hindutva) and nationalism—the party’s traditional ideological bases—with strategic pre-election social spending to construct a hegemonic narrative.
This was, arguably, achieved through a series of ‘smart’ welfare programmes initiated by Prime Minister Narendra Modi that sought to make a direct connection between the central government and subaltern voters. In a related vein, Tariq Thachil (2011, 2014) attributes the rising support for the BJP among India’s poor Adivasi (Indigenous) groups, to various welfare schemes and other welfare work undertaken by BJP–affiliated front organizations. Through provision of basic social services, these organisations, Thachil argues, politicise and recruit the beneficiary population.
All of these scholars offer insights on the role of the economy in Indian elections, but they do not engage systematically with the multi-level governance structures of the Indian polity. While they all delineate the BJP’s rise in states where the party controls the government or its opposition comprises of non-programmatic political parties, they cannot satisfactorily explain the party’s growth in settings, such as West Bengal and Kerala, where left and centre-left parties have historically competed around issues of redistribution and public goods provision.
Economic voting in federal contexts
Economic voters reward governments responsible for bringing about superior economic conditions by re-electing them, while casting aside governments that do not deliver. However, a crucial step in this process is voters’ ability to accurately identify who has responsibility for economic policy. Ability to attribute responsibility, in turn, is mediated by voter competence, and institutional design. In the complex institutional contexts of federal democracies, clarity of responsibility is obscured as power is dispersed among national and state governments, rendering it difficult for voters to ‘throw the rascals out’. This is a particularly vexing problem in asymmetric federations (like India), which further complicates responsibility attribution.
A related key consideration that goes into the voting calculus is whether voters evaluate personal or national economic circumstances. The dominant trend for nearly four decades had been to reject vote choice based on assessments of personal economic circumstances (egotropic vote) in favour of national economic conditions (sociotropic vote).
Our study throws light upon the voting calculus of the economic voter in settings where the nationally incumbent party is not the incumbent at the state level, focusing on the conditions under which sociotropic or egotropic voting or both may be observed in national elections.
Going by conventional wisdom, we would expect to see sociotropic voting in national elections when the state and national incumbents are from the same party, that is, voters who assess that national economic conditions have improved (worsened) over the period leading up to the election would be more (less) likely to vote for the national incumbent relative to the state incumbent. Consider now a setting of partisan incongruence, where the national and state incumbent are from different parties. In this setting, the rational state voter would surmise that changes in national economic conditions would be an inadequate yardstick for punishing or rewarding the state incumbent because the latter can have minimal impact, at best, on the macroeconomic policy choices on which national economic conditions rest.
Unlike national economic conditions, however, personal economic conditions are more likely to be affected by the policy decisions of the state incumbent. Related to this, assessments of personal economic conditions are less likely to suffer from endogeneity bias, that is, voters can better assess to what degree state policies affected their personal economic conditions than the degree to which federal policies contributed to national economic conditions. It follows, therefore, that we ought to expect egotropic voting to coexist with sociotropic voting in this setting.
The key question that remains to be asked is how voters will decide between the competing sociotropic and egotropic motivations. We hypothesise that the discerning voter will use the national election as a referendum on both the state and national incumbent. More specifically, she will choose between the two incumbents by comparing perceived household economic conditions and national economic conditions, that is, by using national economic conditions as a benchmark for assessing household economic conditions and vice-versa. This is particularly so when the parties at the state level have historically competed on the basis of programmatic appeals to provide social welfare forging ties of reciprocity with the electorate.
Table 1: Sophisticated voting in state pro-poor contexts
The preceding discussion leads to the following hypotheses: Given that the choice before the voter is either to vote for the national incumbent or the state incumbent in national elections, and the state incumbent is demonstrably committed to pro-poor politics.
H1. (Sociotropic voting): The more positive (negative) the assessment of national economic conditions the higher (lower) the propensity to vote for the national incumbent relative to the state incumbent.
H2. (Egotropic Voting): The more positive (negative) the assessment of household economic conditions the lower (higher) the propensity to vote for the national incumbent relative to the state incumbent.
H3. (Sophisticated Voting): If household economic conditions are perceived to have improved (worsened) relative to national economic conditions, the lower (higher) the propensity to vote for the national incumbent relative to the state incumbent.
It is instructive to note that a poorly performing national incumbent can try to alter the economic calculus of the state voter through a well-targeted programme of social spending, particularly in the run-up to the national election. Indeed, a key element of Modi’s campaign strategy for the 2019 Indian parliamentary election was to personalise federal social spending through a slew of welfare programmes bearing the signature of the Prime Minister, leaving little room for misattribution by voters. This strategy, though, is unlikely to translate into egotropic votes in pro-poor contexts since voters in these settings may see these programmes as top-down and less relevant to their needs due to the already entrenched welfare politics at the state level. Alternatively, anticipating the central government’s motives in undercutting their support base, the incumbent political parties in these settings may simply refuse to implement these federal programmes or integrate them within existing programmes or even substitute them with new programmes. We discuss the national and state political context of the 2019 vote in West Bengal and Kerala and show that, at least in the welfare domain, both states have considerably outperformed other Indian states, leading to the fragmented economic voting patterns.
Data and methods
The West Bengal pre-election survey in Junglemahal was conducted from 22 February-13 March 2019, yielding a total of 1140 respondents. The Kerala pre-election survey in Thiruvananthapuram was carried out between 14–22 April 2019, and covered 1034 respondents.
Our primary dependent variable was collated from a question in the surveys, which asked respondents “Which political party do you intend to vote for in 2019, or which party are you leaning towards?” Among the 938 respondents who answered this question in Thiruvananthapuram, 308 reported being inclined to vote for the BJP, 342 for the INC, 272 for one of the Left parties, 14 for the Indian Muslim Union League, and 2 for Others. Among the 753 respondents who answered this question in Junglemahal, 430 reported being inclined to vote for the BJP, 306 for the TMC, 11 for the Left Front, and 6 for the INC.
Our independent variables model the core concerns of this article i.e. the effect of perceived economic conditions on party choice (H1–H3).
In addition to investigating the relationship between assessments of aggregate economic conditions and vote choice, we also examined how assessments of particular central government economic policies affected vote choice. The economic policy initiatives surveyed included policies intended to promote general welfare—demonetization, the goods and services tax, biometric identification system (Aadhaar)—as well as targeted pro-poor policies such as farm policies and Ayushman Bharat.
The control variables were classified into four categories: (i) vote choice in previous elections; (ii) partisanship; (iii) demographic characteristics; and (iv) perceptions of social policy.
In Kerala, we estimated respondents’ inclination to vote for the LDF over the BJP and the UDF over the BJP. In the case of West Bengal, we estimated the inclination to vote for TMC over the BJP.
Among the control variables for Junglemahal, vote choice in 2016, partisanship, Muslim identity, and attitudes towards the BJP’s social engineering policies have the predicted effects. We see that voting for the BJP in the 2016 assembly election lowers the average predicted inclination of voting for the TMC over the BJP in 2019 by roughly 43 percentage-points. This result extends to long-term party association as well. A one unit increase in the propensity to vote for the BJP lowers the average predicted inclination of voting for the TMC over the BJP by roughly 5 percentage-points.
The data also show clear evidence of Hindutva-based polarisation. Being a Muslim increases the average predicted inclination of voting for the TMC over the BJP by roughly 39 percentage-points. Ethnonationalism was also an important theme in the 2019 election: a one unit increase in approval of the central government’s social engineering policies lowers the average predicted inclination of voting for the TMC over the BJP by roughly 2 percentage-points.
Moving on to our hypotheses of interest, we see that, consistent with H1 and H2, voters attributed national economic conditions to the national incumbent and household economic conditions to the state incumbent. However, assessments of national economic conditions were clearly contingent on partisanship, confirming the endogeneity bias in these assessments predicted by extant research.
Moreover, our sophisticated voting hypothesis (H3) was also confirmed. Voters who perceived that household economic conditions had improved relative to national economic conditions were more likely to vote for the TMC. Conversely, voters who perceived that household economic conditions had worsened relative to national economic conditions were more likely to vote for the BJP. However, only the latter effect was statistically significant. In terms of predicted probabilities, the average inclination to vote for the TMC over the BJP was roughly 12 percentage points lower among respondents who perceived that household economic conditions had worsened relative to national economic conditions than among respondents in the residual category.
Finally, with regards to our measures of perceptions of social spending by the BJP prior to the election, the results show that these variables did indeed have an effect in conditioning the economic vote. In particular, we see that positive assessments of the BJP’s subsidised medical insurance policy actually increased the inclination to vote for the TMC over the BJP, indicating that voters had difficulty attributing the policy correctly. Most importantly, however, in spite of the ‘noise’ introduced by these policies, we observe that the economic vote remains robust across both specifications.
There are two main results that stand out. First, the effect of ethnonationalism in boosting support for the BJP turns out to be less robust. Second, and more importantly, Junglemahal’s voters clearly used the national election as a referendum on the economic performance of both the state and national incumbents, though they were more inclined to punish the TMC for relative deterioration in their economic conditions than to reward the party for relative improvement in the same.
As in the case of Junglemahal, we find past vote choice and partisanship was a significant predictor of vote choice in Thiruvananthapuram, Kerala – where the Left Democratic Front was the incumbent – in 2019. Similarly, ethnonationalism was a vote-getter as well. However, contra Junglemahal, we find little evidence for Hindu-Muslim polarisation. Nevertheless, we find the same pattern of economic voting. Voters who perceived that their household economic conditions had improved relative to national economic conditions were more likely to vote for the LDF. Conversely, respondents who perceived that household economic conditions had worsened relative to national economic conditions were more likely to vote for the BJP. Noticeably, however, only the former effect was statistically significant. The average inclination to vote for the LDF versus not voting for the LDF was roughly 20 percentage-points higher among respondents who perceived that their household economic conditions had improved relative to national economic conditions.
We see another interesting contrast with Junglemahal. Thiruvananthapuram voters had no difficulty in attributing welfare programmes to the appropriate level of authority. Nonetheless, as in the case of Junglemahal, the economic vote remains robust across both these specifications.
To summarise, both our sub-state and state-level analyses reinforce our hypothesis that the voter in state pro-poor contexts is able to ‘cut the clutter’ created by federalism and hold state and national incumbents accountable for their economic policies. Our analyses also revealed that differences in perceived household and national economic conditions had asymmetric effects on party choice. Specifically, we find that, barring in Junglemahal – where the TMC’s ‘development as counterinsurgency’ approach had seeded ambiguous attitudes towards the party’s welfare policies – voters in both states showed a greater inclination to reward superior economic performance by the state incumbent than to punish inferior economic performance, indicating that the welfare regimes in either context continue to command substantial political and cultural legitimacy.
Subhasish Ray is a faculty member at the Jindal School of Government & Public Policy, O.P. Jindal Global University, Haryana; and an editor for the Journal of Genocide Research. He tweets @subhasish_ray75.
Anil Varughese is a faculty member at the School of Public Policy & Administration, Carleton University, Ottawa, Canada. He tweets @anilvaru1.
This article is an edited excerpt from the authors’ paper Economic voting in multi-level contexts: Evidence from Kerala and West Bengal in India, first published in the Electoral Studies journal. Read the full paper here.
(Edited by Prashant)