Final Report Summary - RIOTS AND WELFARE (A Micro Level Analysis of Communal Violence in India with a new Database on Maharashtra)
Project context and objectives
The project aims at better understanding the processes leading to and resulting from civil violence in India, by adopting a micro-level approach. Civil violence in India mainly takes the form of 'communal violence', also called riots. However, despite the common wisdom which associates riots with spontaneous outbursts of violence, there is a consensus that communal violence in India is in fact highly organised. Communal violence in India has been the focus of many social science studies. However, it remains unclear how and to what extent riots impact the economic life and choices of individuals and, in turn, how characteristics of individual vulnerability shape violence locally. The project aims to bridge this gap by implementing a large survey combining quantitative and qualitative instruments.
The approach followed in the project comprised implementing a household survey in rioting and non- rioting sites in the state of Maharashtra. Information on spatial distribution of violence within Maharashtra was gathered by combining new police data on 'ethnic riots' at the district level, careful media scans and key informant interviews within each selected district. We selected 45 sites in 10 districts. Within each of these sites, we randomly surveyed 24 households for a total sample size of 1089 respondents. Survey instruments combined quantitative and qualitative elements. The latter took the form of in-depth interviews which were conducted on a sub-sample of individuals who were identified through trigger questions in the quantitative questionnaire as involved in violence or, in non-rioting sites, involved in the community life. The quantitative questionnaire includes instruments to inform on household composition, consumption, education and employment, access to public goods, relation to police, community problems, trust and civic life and civil violence. The fieldwork, which took place in March 2010, was also used to gather extensive cartographic materials from each of the sampled sites.
Project outcomes
Three research outputs came from this: a paper authored by Jaideep Gupte, Patricia Justino and Jean-Pierre Tranchant entitled "Civil Violence in India: Micro-Level Foundations and Welfare Consequences", laying out the rationale and logic of the study; a working paper by the same authors relating household and neighbourhood characteristics to victimisation through a multilevel framework ("Households amidst Urban Riots: The Economic Consequences of Civil Violence in India"); and an ongoing research paper taking a spatial look at violence and vulnerability.
Project results
The principal results are as follows:
First, the geographic pattern of violence in Maharashtra revealed by the sampling design is at odds with firmly-held beliefs in India. Places commonly considered as affected by rioting did not necessarily prove as such while other places that are never cited as potential sites of violence appeared as highly affected. We hypothesise that this discrepancy is explained by the stigma attached to cities or neighbourhoods famed for spectacular riots. In-depth interviews during which respondents confided their acute fear of riots in places where civil violence is absent corroborated this hypothesis.
This leads to the second and related finding: riots exert a powerful psychological impact which is extremely long-lasting. This is especially true for members of local minorities who are found to fear riots twice as much as members of the local dominant group.
Thirdly, actual victimisation, on the other hand, is not strongly related to any identity marker at the regional level.
Fourth, there is a large overlap between social, economic and spatial vulnerabilities, on the one hand, and vulnerability to violence on the other. We interpret this as the inability by vulnerable populations to shield them from the disruption caused by the violence (on job and commodity markets, through fear or curfew). However, ceteris paribus, households with a higher number of visible assets are more likely to report victimisation. Civic life and trust which have been hypothesised to protect cities from civil violence seem to act at the individual level more than at the neighbourhood level. Trust is considered especially important for the more wealthy people for whom good relations with neighbours offset the effect of visible assets.
Building on this work, we considered it worthwhile to compare the vulnerability-violence nexus in urban areas with that in rural areas. Large parts of rural India form part of the red corridor where the Maoist guerrillas (Naxals) are firmly entrenched. Although violence levels caused by the guerrilla are high, for most people the main consequences of the war are likely to be indirect - through disruption to markets, governance and community relations - rather than direct - through killing. There is thus a parallel to be drawn with the civil violence case where most of the victimisation has also been found to be indirect.
This led to a research paper authored by Jean-Pierre Tranchant, Patricia Justino and Catherine Mueller entitled "Children Malnutrition, Drought, and Rebel Control: Evidence from the Naxal Insurrection in Andhra Pradesh", which looks at whether proximity to Maoist rebellion worsens socio-economic vulnerabilities. To do that, we studied the variations in child malnutrition following drought episodes in locations in Andhra Pradesh where Naxals are present. We used the fact that in 2004 a ceasefire between the rebels and the newly elected state government was in place. It turns out that the degree of child malnutrition was twice as large during droughts which occurred before 2004 than for those in 2004, where violence was absent, controlling the severity of the droughts. There is thus a coherent pattern in India whereby violence primarily affects the most vulnerable and reinforce existing vulnerabilities.
We were successful in continuing the project beyond the Marie Curie funding. We obtained a grant from DFID-ESRC on Agency and Governance in Contexts of Violence which allowed us to finance a second wave of the field survey. This is extremely beneficial as panel data enable the use of more powerful quantitative methods than a single cross-section. The second wave of data collection was organised in March-April 2012. The quantitative instruments build and expand substantially on those of the first wave. In addition, we piloted behavioural games in two sample districts (Kolhapur and Sangli) in order to study in greater depth the issues of trust, social co-operation and interactions and violence. The new data will permit us to considerably deepen the analysis of civil violence and vulnerabilty in the months to come.
Results were disseminated to several audiences: the underlying objectives and rationale of the project was presented in Manchester at the 10 Years of War Against Poverty Conference organised by the CPRC in 2010 and at the Sixth HICN workshop in Bogota in the same year. Results on determinants of victimisation were presented at the MICOCON workshop in Ghent in 2010 and at the Seventh HICN workshop in Barcelona in 2011. Finally, the research on child malnutrition, drought and Naxals was presented at the Seventh HICN workshop in Barcelona and at the Conflict, Violence and Development seminar at IDS, both in 2011.
The project aims at better understanding the processes leading to and resulting from civil violence in India, by adopting a micro-level approach. Civil violence in India mainly takes the form of 'communal violence', also called riots. However, despite the common wisdom which associates riots with spontaneous outbursts of violence, there is a consensus that communal violence in India is in fact highly organised. Communal violence in India has been the focus of many social science studies. However, it remains unclear how and to what extent riots impact the economic life and choices of individuals and, in turn, how characteristics of individual vulnerability shape violence locally. The project aims to bridge this gap by implementing a large survey combining quantitative and qualitative instruments.
The approach followed in the project comprised implementing a household survey in rioting and non- rioting sites in the state of Maharashtra. Information on spatial distribution of violence within Maharashtra was gathered by combining new police data on 'ethnic riots' at the district level, careful media scans and key informant interviews within each selected district. We selected 45 sites in 10 districts. Within each of these sites, we randomly surveyed 24 households for a total sample size of 1089 respondents. Survey instruments combined quantitative and qualitative elements. The latter took the form of in-depth interviews which were conducted on a sub-sample of individuals who were identified through trigger questions in the quantitative questionnaire as involved in violence or, in non-rioting sites, involved in the community life. The quantitative questionnaire includes instruments to inform on household composition, consumption, education and employment, access to public goods, relation to police, community problems, trust and civic life and civil violence. The fieldwork, which took place in March 2010, was also used to gather extensive cartographic materials from each of the sampled sites.
Project outcomes
Three research outputs came from this: a paper authored by Jaideep Gupte, Patricia Justino and Jean-Pierre Tranchant entitled "Civil Violence in India: Micro-Level Foundations and Welfare Consequences", laying out the rationale and logic of the study; a working paper by the same authors relating household and neighbourhood characteristics to victimisation through a multilevel framework ("Households amidst Urban Riots: The Economic Consequences of Civil Violence in India"); and an ongoing research paper taking a spatial look at violence and vulnerability.
Project results
The principal results are as follows:
First, the geographic pattern of violence in Maharashtra revealed by the sampling design is at odds with firmly-held beliefs in India. Places commonly considered as affected by rioting did not necessarily prove as such while other places that are never cited as potential sites of violence appeared as highly affected. We hypothesise that this discrepancy is explained by the stigma attached to cities or neighbourhoods famed for spectacular riots. In-depth interviews during which respondents confided their acute fear of riots in places where civil violence is absent corroborated this hypothesis.
This leads to the second and related finding: riots exert a powerful psychological impact which is extremely long-lasting. This is especially true for members of local minorities who are found to fear riots twice as much as members of the local dominant group.
Thirdly, actual victimisation, on the other hand, is not strongly related to any identity marker at the regional level.
Fourth, there is a large overlap between social, economic and spatial vulnerabilities, on the one hand, and vulnerability to violence on the other. We interpret this as the inability by vulnerable populations to shield them from the disruption caused by the violence (on job and commodity markets, through fear or curfew). However, ceteris paribus, households with a higher number of visible assets are more likely to report victimisation. Civic life and trust which have been hypothesised to protect cities from civil violence seem to act at the individual level more than at the neighbourhood level. Trust is considered especially important for the more wealthy people for whom good relations with neighbours offset the effect of visible assets.
Building on this work, we considered it worthwhile to compare the vulnerability-violence nexus in urban areas with that in rural areas. Large parts of rural India form part of the red corridor where the Maoist guerrillas (Naxals) are firmly entrenched. Although violence levels caused by the guerrilla are high, for most people the main consequences of the war are likely to be indirect - through disruption to markets, governance and community relations - rather than direct - through killing. There is thus a parallel to be drawn with the civil violence case where most of the victimisation has also been found to be indirect.
This led to a research paper authored by Jean-Pierre Tranchant, Patricia Justino and Catherine Mueller entitled "Children Malnutrition, Drought, and Rebel Control: Evidence from the Naxal Insurrection in Andhra Pradesh", which looks at whether proximity to Maoist rebellion worsens socio-economic vulnerabilities. To do that, we studied the variations in child malnutrition following drought episodes in locations in Andhra Pradesh where Naxals are present. We used the fact that in 2004 a ceasefire between the rebels and the newly elected state government was in place. It turns out that the degree of child malnutrition was twice as large during droughts which occurred before 2004 than for those in 2004, where violence was absent, controlling the severity of the droughts. There is thus a coherent pattern in India whereby violence primarily affects the most vulnerable and reinforce existing vulnerabilities.
We were successful in continuing the project beyond the Marie Curie funding. We obtained a grant from DFID-ESRC on Agency and Governance in Contexts of Violence which allowed us to finance a second wave of the field survey. This is extremely beneficial as panel data enable the use of more powerful quantitative methods than a single cross-section. The second wave of data collection was organised in March-April 2012. The quantitative instruments build and expand substantially on those of the first wave. In addition, we piloted behavioural games in two sample districts (Kolhapur and Sangli) in order to study in greater depth the issues of trust, social co-operation and interactions and violence. The new data will permit us to considerably deepen the analysis of civil violence and vulnerabilty in the months to come.
Results were disseminated to several audiences: the underlying objectives and rationale of the project was presented in Manchester at the 10 Years of War Against Poverty Conference organised by the CPRC in 2010 and at the Sixth HICN workshop in Bogota in the same year. Results on determinants of victimisation were presented at the MICOCON workshop in Ghent in 2010 and at the Seventh HICN workshop in Barcelona in 2011. Finally, the research on child malnutrition, drought and Naxals was presented at the Seventh HICN workshop in Barcelona and at the Conflict, Violence and Development seminar at IDS, both in 2011.