Das, Sabyasachi, Democratic Backsliding in the World’s Largest Democracy (July 3, 2023). Available at SSRN: dx.doi.org/10.2139/ssrn.4512936
A working paper titled ‘Democratic Backsliding in the World’s Largest Democracy’, authored by Sabyasachi Das, an academic teaching economics in Ashoka University, has received much traction on social media and in academic circles. The paper merits close scrutiny as it raises worries about the ‘future of democracy’ in India by claiming to detect ‘irregular patterns’ in the 2019 Lok Sabha elections.
The framework
Based on election studies literature that draws from democracies around the world, mainly the U.S., the paper argues that the ‘irregularity patterns’ visible in the 2019 elections can be attributed either to ‘electoral manipulation’ or to ‘precise control’ theses.
The ‘precise control’ thesis refers to the incumbent party’s ability to precisely predict and then affect the win margins with the help of ‘its superior electoral machinery’ in closely contested constituencies. For this purpose, the party targets polling stations, especially the ones in urban areas with a large number of voters, as they are easier to access. This may result in a large turnout in these polling stations, resulting in high vote shares for the incumbent party. On the other hand, the ‘electoral manipulation’ thesis refers to targeted electoral discrimination at the local level, partly facilitated by the weak monitoring of election observers. This include the strategic/targeted suppression of certain community/group votes in the form of registration and/or turnout manipulation. This amounts to electoral fraud and raises serious question about electoral integrity.
The crux of the paper
Coming to India, the paper traces ‘irregularity’ in the 2019 elections in the form of ‘excess/disproportionate wins’ of the incumbent party (the BJP) in ‘constituencies that were closely contested between a candidate from the incumbent party and a rival’. Using data from the website of the Election Commission of India (ECI), the paper mentions 59 parliamentary constituencies (PCs) where the absolute win margin for the winning party was within 0.05%, out of which the BJP won 41. The party won 22 out of 27 PCs in the States where it was the incumbent party. In these PCs, which had significant Muslim electorates, the discrepancy was visible in the high turnout at the polling stations, making a case for ‘irregularity’. The paper looks at the datasets collected from ECI websites, the Centre for the Study of Developing Societies (CSDS) post-poll survey and from ‘standard sources’ like published papers, and states that this ‘pattern of irregularity’ can be explained by the ‘electoral manipulation thesis’.
In support of the contention that the data is ‘less supportive’ of the ‘precise control thesis’, the paper presents two arguments. The first argument is based on the CSDS post-poll data, which shows that the BJP and other contestant parties did not ‘campaign significantly harder’ in the closely contested PCs where the BJP managed to ‘barely’ win. Even then, these PCs, particularly the polling stations with a significant number of Muslim votes, registered high turnouts. The second argument is that the BJP did not campaign ‘harder’ in the polling stations having sizable Muslim voters, as the party did not ‘expect to get significant support from them’. As per the CSDS poll data of the 2014 and 2019 elections, only ‘8 percent of Muslims reported to have voted for the BJP’.
As for the ‘pattern of irregularity’ being more consistent with the ‘election manipulation thesis’, the paper refers to the ‘discrepancy’ that occurred between the ‘final count of EVM votes’ for each Lok Sabha constituency for the first four of the seven phases of election as released by the ECI on its website, which did not match with the ‘constituency wise number of votes in EVM’ that was released later. The ECI ‘removed the earlier figures from its website’ after the ‘media query’. The paper claims to trace ‘discontinuous change with respect to the incumbent’s win margin’ in the later official figure and that ‘the discontinuity’ was ‘concentrated in BJP ruled States’. Also, the EVM data discrepancy was ‘significantly larger’ in ‘closely contested PCs barely won by BJP’. The paper claims to find ‘the evidence on turnout discrepancy as indicative of manipulation done locally at the polling stations, rather than resulting from aggregation fraud at the constituency level’.
The paper argues that polling officials deputed in the polling stations ‘exercise significant discretion in allowing registered voters’. It further alleges that the Muslim voters due to their names being easily identified in the electoral roll, were subjected to ‘both strategic deletion and strategic discrimination’. The paper claims that this happened at polling stations having ‘significant’ Muslim voters in the BJP ruled States, as the State governments could ‘influence assignment of officials in charge’. State Civil Service officers (SCS) were deputed in these polling stations because they were ‘more likely to be politically pliable’ being from the same State and under the State government. The role of the ECI also comes under scrutiny for the ‘weak monitoring of counting of votes’. The author claims to have access to the data of 539 PCs (out of 543) containing 1, 804 counting observers to make this charge.
Furthermore, to substantiate the ‘possibility of strategic deletion of Muslim voters’, the paper calculates the growth rate of the voters in each Lok Sabha constituency taking the figures of the 2014 and 2019 elections. The paper finds that the growth rate fell discontinuously by 5 percentage points (compared to the mean of 0.09) in PCs barely won by the BJP, and the fall was ‘concentrated in PCs with higher share of Muslim electorate’. Given the Muslim voters apathy towards the BJP, the paper questions both the ‘high turnout’ and the ‘high’ BJP vote share in the polling stations having significant Muslim presence, which was ‘large in magnitude and statistically significant’ in the PCs ‘barely’ won by the party. The paper claims to base its contention on data generated on the basis of the study of 8,50,000 polling stations drawn from 3,098 assembly constituencies from 475 Lok Sabha constituencies.
A critique of the paper
Regarding the claim that the BJP was not campaigning ‘hard’ enough in the PCs where it had a narrow win, the paper itself mentions that a party having ‘the superiority of electoral machinery’ could be ‘able to accurately predict and affect win margins in closely contested constituencies’. And the BJP being the incumbent party at the national level and in some States, did have a distinctive superiority in terms of campaigning intensity, backed by human and material resources at its command. Therefore, the party did not need to campaign unusually ‘harder’.
While accepting the CSDS poll data as a credible source which shows that the BJP did not canvass Muslim houses as much as other parties, the counter-argument can be that BJP would have actually targeted non-Muslim homes, belonging to the two large disadvantaged caste groups among Hindus, namely the Scheduled Castes (SCs) and the Other Backward Classes (OBCs). These groups, the paper argues, have emerged as the ‘new’ supporting voters of the party. This might have accounted for the increase in the turnout in the identified polling stations.
Additionally, the CSDS data has its limitations due to its small sample size. In a State like U.P. with a huge population, the sample size was only 2,705, with actual interviews being much lesser. As the paper observes there are on an average seven Assembly constituencies (ACs) in each PC but in the CSDS, 2019 survey only two ACs were selected from each PC ; and three polling stations were selected from each selected AC.
As for the claim that SCS officers are ‘politically pliable’ and therefore complicit in turnout manipulations, is not an argument, but a conjecture. ‘Political pliability’ cannot be considered the hallmark of all SCS officers. That argument is unfair.
There were 36 PCs where Muslims constituted more than 40% of the total electorates in the 2014 and 2019 elections. Contesting on 21 seats, the BJP received 28.92% of the vote. The BJP won only 5 seats in the two elections, though not from the same PCs. Interestingly, of the 59 PCs listed in the paper that show a narrow margin of win either for the BJP or any other party, there are only three PCs having more than 40% of Muslim voters. If one goes by the CSDS poll data findings based on the 2014 and 2019 elections that Muslim voters generally do not vote for the BJP, then the electoral verdicts in most PCs where they have ‘significant presence’ negate the ‘election manipulation thesis’ of the ‘suppression’ of the Muslim vote. In any case, as Hilal Ahmed, a faculty at CSDS argues, there is no ‘credible proof’ to argue that Muslims constitute a ‘single homogenous’ voting community and whose electoral choice is ‘self-evident’, (that they vote en masse against the BJP).
Interestingly, the paper itself concedes that ‘the absolute difference between the two official versions of EVM turnout data could be due to ‘administrative errors during counting of votes’. The paper also absolves the ECI of engaging in ‘direct tempering of turnout data ex-post’. Moreover, barring only one case, as per the data gathered, the magnitude of data revision is smaller than BJP’s absolute margin of victory. As for the question mark over the credibility of the ECI based on the allegations of ‘weak monitoring’ of ‘arbitrary deletion of names of registered Muslim voters’ at local levels, and also showing ‘bias in scheduling of elections’, the paper cites a set of articles. Ironically, another set of articles and even CSDS and global surveys are cited in the paper which show peoples’ high level of trust in the ‘independence and institutional strength’ of the institution.
The paper also concedes that the tests/evidence/datasets ‘ are … not proof of fraud, nor does it suggest that manipulation was widespread’ as proving ‘electoral manipulation in a robust democracy is a significantly harder task that would require detailed investigation of electoral data in each constituency separately’. Even then, it goes on to raise serious questions about the wellbeing of India’s electoral democracy. And in support of its contention about ‘democratic backsliding/democratic reversal’ in India, the paper quotes from the Freedom House Index, 2021, the Democracy Report, 2020, V-Dem Annual Report 2021. However, the paper itself refers to another world-wide Gallup Poll survey which rates India higher than older democracies like the U.K. and France in terms of the confidence of the electorates in India.
Ashutosh Kumar teaches in the Department of Political Science at Panjab University. The views expressed are personal
