Arresting the spread of Covid-19 requires collective action by communities, which is likely to be more challenging in settings with high diversity. Using district-level data from India, this article examines the effects of caste and religious fragmentation and economic inequality, on the growth rate of reported Covid-19 cases. It finds strong positive effects of caste homogeneity, particularly during the nationwide lockdown, and especially in the rural areas.
As the administration prepares to roll out the Covid-19 vaccines, continued adherence to social distancing measures is, by and large, recommended by health experts to arrest the course of the spread of the Virus. Given that adoption of distancing norms and hygiene practices require collective effort, the success with which they are adopted by different communities is likely to vary. As suggested by the literature on collective action, less homogeneous communities might find it more difficult to agree upon a common norm and enforce mutually beneficial behaviour (Habyarimana et al. 2007). Existing evidence suggests that individuals affiliated to common groups, empathise and care more for in-group members than outsiders (Vigdor 2004, Alesina and Ferrara 2005). Also, homogenous communities, through higher social interactions, reciprocity, credibility of social sanctions, and interpersonal trust possess greater levels of social capital that facilitates community coordination (Miguel and Gugerty 2005). Given the highly contagious nature of Covid-19, flouting social distancing norms might make the entire community vulnerable and undermine government efforts to address the pandemic. In this article, we analyse whether communities that are more homogeneous – and hence likely to have stronger social cohesion – are better protected from Covid-19 (Rathore, Das and Sarkhel 2020).
Assessing the role of diversity in the battle against Covid-19
We investigate the link between diversity and Covid-19 spread in India – a country with the second highest Covid-19 case load in the world, which is also characterised by highly stratified societies in terms of caste and religion, and sharp economic inequalities. In particular, we study the implications of social, religious, and economic diversity on the growth in the number of reported infections starting from the beginning of the nationwide lockdown (25 March 2020), until almost two months after these restrictions were lifted (26 July 2020).1
We make use of a novel district-level dataset that records daily Covid-19 infection and mortality data for 719 districts, sourced from Development Data Lab’s Covid-19 India database (Asher et al. 2020). This is mapped with National Sample Survey data (2018) on district-level indicators of caste group, religious, and economic diversity, along with characteristics such as relevant hygiene practices, for 661 of the 719 districts.. We also account for past mobility and residential patterns of the households (NSS, 2018). In addition to a range of household-level characteristics aggregated at the district level, we include supply-side data on physical and human health infrastructure and other amenities (Census, 2011) as well as share of population with chronic ailments in the district (NSS, 2018). To the best of our knowledge, the dataset is one of the first in the Covid-19 context that combines multiple sets of information at the district level in India.
To compute district-level social and religious homogeneity, we use the Herfindahl Hirschman Index (HHI)2 (Rhoades 1993). To measure district-level economic inequality, we compute the Gini Index (GI).3 For caste diversity, we use the broad caste groups that include the historically disadvantaged groups of Scheduled Caste (SCs), Scheduled Tribes (STs) and Other Backward Classes (OBCs) in addition to the privileged castes aggregated under ‘Others’.4 For religion, representation across eight major religious groups is accounted for.5 The HHI ranges from 0 to 1 with a higher value implying higher homogeneity or concentration. GI is based on monthly household consumption expenditure, which is also computed on a scale of 0 to 1 with a higher value implying higher inequality.6
Our empirical framework allows us to examine the influence of district-level caste group, and religious homogeneity along with economic inequality7 on the growth of the number of Covid-19 cases across the lockdown as well as the unlocking phase. Gauging the effects of homogeneity here necessitates isolating multiple possibilities that may confound the relationships of interest. For example, in the first few months of the pandemic in India, spread of Covid-19 cases had a distinct urban character with the infection transmitted through international travellers, and initial hotspots comprising cities with superior international air connectivity. At the same time, these cities are also likely to be major centres for job opportunities, something that may attract people from different parts of the country and give these cities a less homogeneous character. To attribute growth in cases purely to measures of social, religious, and economic diversity, we attempt to isolate these channels through a battery of empirical checks.
Positive effects of caste homogeneity
Our findings are presented in Figure 1, where the plotted estimates from left to right provide the effects of one unit increase in caste-group homogeneity (panel a), and religious homogeneity (panel b), and economic inequality (panel c) on the growth of Covid-19 cases across the eight phases of the study (the first four of these were during the lockdown, the end of which is indicated by the vertical red line).8 Our findings in panel (a) reveal that social homogeneity in terms of caste plays a significant role in ensuring lower growth in Covid-19 cases in the Indian context. In particular, this influence of social cohesion via caste group homogeneity was of vital importance in arresting the growth of infection during the nationwide lockdown. However, the marginal gains from this channel seem to gradually dampen after the initiation of the unlocking process that potentially led to higher mobility.
In addition, we find that the more economically unequal districts are associated with a higher number of reported cases, though this relationship remains strong only until the end of the third phase of the lockdown and erodes completely by end of the fourth phase of the lockdown. Over the unlocking period (phase 5 to 8), the marginal gains from higher economic equality remain statistically insignificant.9
Figure 1. Effect of caste-group homogeneity, religious homogeneity, and economic inequality on growth of Covid-19 cases in India across different periods of lockdown and unlocking
Notes: (i) The marginal effects with confidence intervals10 of 95% are presented. (ii) The vertical line at x=4 signifies end of nationwide lockdown (3rd May 2020). (iii) HHI stands for Herfindahl Hirschman Index.
To examine if the dampening effect of unlocking on the social cohesion channel persists across areas with different degrees of social ties, we categorise districts on the basis of the proportion of urban households in total households. We find that for the more rural districts, the effects of caste-group homogeneity on growth of Covid-19 cases remains significant, even during the unlocking phase. No such effect is observed for religion and economic inequality.
Policy implications: Prioritising socially diverse areas and promoting social cohesion
Based on our study, we arrive at two important policy prescriptions. First, until a comprehensive supply-side response can be mounted, government should prioritise areas that are more socially diverse in caste-group composition as they are more vulnerable and possess lower social capital to tackle infection growth. Our analysis provides a case for identifying and zoning local areas into blocks on the basis of social homogeneity.
Second, in relatively homogenous areas, as the channel of existing social ties is observed as weakening with the opening up of the economy, strategies to strengthen community networks can be undertaken. One possible way could be decentralising health interventions through community channels, as these involve greater community participation. With vaccines becoming available, distribution mechanisms will gain importance and pose policy challenges. Our results – in this context – offer potential policy prescription to identify the priority zones for vaccination in terms of strength of social cohesion. Importantly, our study underscores the importance of promoting social cohesion and community harmony, as opposed to divisive political agendas.
I4I is now on Telegram. Please click here (@Ideas4India) to subscribe to our channel for quick updates on our content
- The lockdown in India was implemented in four phases spanning two months: Phase 1: 25 March to 14 April, Phase 2: 15 April to 3 May, Phase 3: 4 May to 17 May, and Phase 4: 18 May till 31 May. In addition, we consider four phases after the lockdown was lifted on 31 May 2020, each consisting of 14 days spanning about two months. These include Phase 5: 1 June to 14 June, Phase 6: 15 June to 28 June, Phase 7: 29 June to 12 July, Phase 8: 13 July to 26 July, which coincides with initiation of unlocking procedures.
- The HHI index is a generalised measure of concentration with a higher value implying higher homogeneity. For robustness check, we also use Shannon’s Index of Diversity and get statistically similar findings.
- The Gini index measures the extent to which the distribution of income (or, in some cases, consumption expenditure) among individuals or households within an economy deviates from a perfectly equal distribution.
- The social networks within the Indian caste system are splintered into various sub-castes or jatis. We restrict ourselves to the broad caste groupings as district-level representative data is only available for this level for period as recent as 2018.
- These include Hindu, Muslim, Christian, Sikh, Jain, Buddhist, Zoroastrian, and Others. We assume that intra-household – as against inter-household – coordination and cooperation is relatively much easier to achieve.
- We compute the variables of interest (HHI and GI) using data at the household level, which is then aggregated at the district level using weights. We do however check for the robustness of our findings using individual-level variables that are qualitatively similar.
- Standard deviation is a measure that is used to quantify the amount of variation or dispersion of a set of values from the average of that set.
- We also present the confidence bands for each of the coefficients, which when even touching the horizontal red line denotes statistical insignificance.
- Additionally, to account for any reporting errors, we also consider daily cases as rolling averages for three and seven days. Our findings remain statistically robust to this specification.
- A 95% confidence interval is a way of expressing uncertainty about estimated effects. Specifically, it means that if you were to repeat the experiment over and over with new samples, 95% of the time the calculated confidence interval would contain the true effect.
- Alesina, Alberto and Eliana La Ferrara (2005), "Ethnic diversity and economic performance", Journal of Economic Literature, 43(3): 762-800.
- Asher, S, L Tobias, R Matsuura and P Novosad (2020), ‘The Socioeconomic High-resolution Rural-Urban Geographic Dataset on India (SHRUG) COVID platform’.
- Asher, S, T Lunt and P Novosad (2020), ‘The SHRUG: A new high-resolution data platform for research on India’, Ideas for India, 13 March.
- Habyarimana, James, Macartan Humphreys, Daniel N Posner and Jeremy M Weinstein (2007), "Why does ethnic diversity undermine public goods provision?", American Political Science Review, 101(4): 709-725.
- Miguel, Edward and Mary Gugerty (2005), "Ethnic diversity, social sanctions, and public goods in Kenya", Journal of Public Economics, 89(11-12): 2325-2368.
- Rathore, U, U Das and P Sarkhel (2020) ‘Birds of a feather flock together? Diversity and spread of COVID-19 cases in India’, arXiv:2011.05839, Working Paper.
- Vigdor, Jacob L (2004), "Community composition and collective action: Analyzing initial mail response to the 2000 census", Review of Economics and Statistics, 86(1): 303-312.