Legal identity at the margins: the impact of violent conflict on birth registration in India

ABSTRACT

In India, armed groups characterised by a diversity of ideologies and aims have emerged and persisted even in the presence of a fairly strong state. These groups often operate in areas that have long suffered from state neglect. We examine how violent conflict influences patterns of birth registration. State-recognised documents are crucial for establishing legal identity, and accessing citizenship rights. We draw on the 2015–16 National Family and Health Survey to measure civil registration, community and household characteristics. Our conflict data come from the Uppsala Conflict Data Program, which we leverage to describe community exposure to organised violence. Our statistical analysis shows that the likelihood that a child is in possession of a birth certificate is significantly lower in areas affected by violent conflict, even when controlling for other individual and social characteristics associated with marginalisation.

KEYWORDS:

• Birth registration
• legal identity
• conflict
• violence
• India

Introduction

According to recent estimates, globally about 850 million people lack official documents that prove their legal identity (Clark, Metz, and Casher 2022). Legal identity, generally established with birth certificates documenting place of birth and parentage, is key for effective claims of citizenship. In India, estimates from the National Family Health Survey indicate that as of 2016, around 20% of children under the age of five have not been registered with Indian civil authorities, suggesting that 2.7 million children in this age group remained unregistered (International Institute for Population Sciences (IIPS) and ICF 2017; Kumar and Saikia 2021). A similar number of children have been registered but do not actually possess a birth certificate. For people without state-recognised documents, the threat to the enjoyment of citizenship rights derives not from their legal status, but from their inability to demonstrate this status (Bloom and Kingston 2021; Hunter 2019; Stevens 2017). Comparative research shows that individuals without official identity documents often come from marginalised communities (e.g. Cheva-Isarakul and Sperfeldt 2023; Liew 2021; Pearson 2021; Sardelić 2021; Sperfeldt 2020) or live in zones without effective state institutions (Koning et al. 2021; Lee and Zhang 2017; Adamczyk and Doumit 2024; Harbers 2020). This creates a vicious cycle as vulnerable individuals are rendered even more vulnerable by their lack of state-recognised documents. Because Indian citizenship is primarily determined by descent, this vulnerability can be transmitted across generations within affected communities.

The Indian state of Assam demonstrates how making citizenship rights conditional on documentary proof fosters exclusion. Assam, a poor state in the Northeast of the country, has historically had low rates of literacy and of birth registration. In 2001, only 27% of children in rural areas below the age of five had been registered (Sadiq 2008, 83). Despite substantial improvements, as recently as 2012, only 68% of children born in Assam’s rural areas had received birth certificates (Rane et al. 2020, 117). Even though this percentage has increased to 85% according to the 2015–16 National Family and Health Survey (NFHS), low rates of registration in the past imply that younger generations are unable to demonstrate the citizenship status of their parents and grandparents. While the state has been ‘standoffish’ towards the people along its Northeastern border (Slater and Kim 2015) and has failed to provide public goods like education, it has deployed the military to assert its claim to the territory. Protracted conflict between the state and armed groups seeking self-determination has led to the militarisation of the region (Sur 2021). The state’s coercive approach to governance deepened the divide between the rural poor and state institutions, which compounded the difficulties of registration arising from poverty and remoteness. In a context where many poor citizens had not been registered, the state then began to harshly police the boundaries of citizenship. Since 2015, the National Register of Citizens (NRC) has attempted to identify unauthorised Bangladeshi migrants in Assam by demanding evidence of a citizen or their family’s residence in India prior to the 1971 cut-off date. Many poor citizens, especially women in rural areas, are unable to produce state-recognised, standardised documents substantiating their claims (Sabhapandit and Baruah 2021; Sur 2021). The NRC has excluded at least 1.9 million residents of Assam from Indian citizenship, and the rights of many have been rendered precarious (Jha and Chakrabarty 2023; Punathil 2022).

Marginalisation and conflict are intertwined across India, raising the question of how conflict influences access to legal identity and citizenship. In this paper, we examine how violent conflict affects birth registration. The South Asia Terrorism Portal identifies 76 active insurgent groups in India, and an average of 2100 conflict-related fatalities per year between 2000–2022 (South Asia Terrorism Portal SATP 2023b, 2023a). Insurgent conflict tends to be concentrated in remote areas or along India’s international borders, such as the Northeast of the country and in Jammu & Kashmir, and in tribal areas that have long suffered from state neglect (South Asia Terrorism Portal SATP 2015; Staniland 2021).¹ Political science research has shown that political and socio-economic marginalisation are key contributors to conflict (Buhaug, Cederman, and Ketil Rød 2008; Cederman, Wimmer, and Min 2010; Tollefsen and Buhaug 2015). Yet, while the lack of comprehensive civil registration is often seen as a symptom of state neglect and inadequate investment in infrastructure, the absence of state-issued ID documents can also lead to further marginalisation. Access to welfare benefits across India has recently become linked to citizens’ ability to prove their identity, leaving those without state-recognised documents unable to claim subsidised food or social services to which they would otherwise be entitled (e.g. Bhatia, Donger, and Bhabha 2021; Sriraman 2018).

We show that conflict exposure substantially increases the risk that children remain unregistered. This effect is robust even when controlling for other factors associated with the social marginalisation of conflict zones, such as poverty, remoteness and rough terrain. As the boundaries of Indian citizenship have become increasingly politicised (e.g. Jha and Chakrabarty 2023; Punathil 2022), it is crucial to recognise not only that conflict occurs in zones of state neglect, but also that it produces unequal access to legal identity and citizenship rights for individuals in these areas (Dasgupta 2021).

In the next section, we briefly outline patterns of violent conflict in India, and then highlight challenges for accessing civil registration in conflict zones. We subsequently introduce our data and describe our analysis. We close with a discussion of our findings and their implications.

The landscape of violent conflict in India

Violent conflicts tend to emerge and persist in zones where state institutions do not provide basic services, or where the quality of services such as education and public health is low. Scholars of civil war argue that states are less able to effectively pre-empt and counter challenges from armed groups in areas where they do not have the institutional infrastructure to provide public goods (Buhaug, Gates, and Lujala 2009; Fearon and Laitin 2003; Sobek 2010). Peripheries are – almost by definition – characterised by remoteness, difficult terrain and low population density, making investments in infrastructure for public goods provision expensive, and potentially unattractive for states interested in maximising revenue (Herbst 2000; Steinberg 2018).² The ethnic, linguistic or religious minorities that often inhabit the peripheries of contemporary states may also be perceived as disloyal or marginal to the state’s political project (e.g. Sur 2021 on the Northeast of India). As a result, the governance of peripheries tends to be more repressive than in other parts of the country as state actors – worried about their tenuous hold on these areas – use force to ward off real and perceived threats (Braun and Kienitz 2022, 313). The state’s ‘weak’ capacity in these areas is at least in part an outcome of its limited interest in fully incorporating these areas in the first place, rather than an inability to do so. Slater and Kim (2015) refer to the strategy that emerges from the state’s unwillingness to engage with populations perceived as challenging as ‘standoffishness’. Even during insurgencies, when we might expect the state to have a strong interest in (re-)asserting control, Slater and Kim encounter this reluctance to fully incorporate specific populations. Rather than seeking to render these populations legible, state elites hold them at bay. States may strengthen their local repressive capacity, for instance by deploying security forces, but without simultaneously investing in providing public goods such as education. State neglect, coercive engagement and inaccessibility then create fertile ground for conflicts to emerge and persist.

The landscape of conflict in India reflects this pattern. Armed groups exist primarily in peripheral and border regions. As Figure 1 shows, organised violence is reported from Kashmir (which borders Pakistan and China), Punjab (bordering Pakistan) as well as the Northeast (bordering China, Bangladesh and Myanmar). India’s postcolonial international borders have been sites of inter-state conflict -including international armed conflict under international humanitarian law- and continued contention over territory. Given secessionist claims from minorities in the periphery, state coercion, and the existence of neighbouring irredentist states, such as Pakistan, that lay claim to populations and territory along the border, Indian border regions are fertile grounds for violent conflict (Nair and Sambanis 2019; Sur 2021; Baruah 1999). Kashmir exhibits a combination of Islamist as well as secessionist armed groups (e.g. Harkat-ul-Mujahideen, and Jammu and Kashmir Liberation Front respectively), while in the Northeast ethno-nationalist secessionist claims often inter-lock with leftist ideologies (e.g. United Liberation Front of Ahom). Simultaneously, armed groups are also active in the so-called ‘red corridor’ of the country; a forested area of Central-East/East India characterised by difficult terrain, rich mineral deposits, and large tribal populations. Armed groups are widely documented to assert control over such peripheral areas given opportunities for extracting rents and generating income from natural resources, the existence of marginalised populations, as well as shelter from state security forces (Buhaug, Gates, and Lujala 2009). The marginalisation and the involuntary displacement of local tribal populations are both a result of state neglect as well as corporate interests (Government of India 2014). A report to the Planning Commission (Government of India 2008) highlighted that local populations often turn to leftist groups for support for land-reclamation and to protect forest rights (Chandra 2014; Kar 2015).

Figure 1. UCDP events, 1989–2010.

Given shifting and contested claims to authority, civilians living in conflict zones must navigate an ever-changing set of relations and practices (Suykens 2015; Waterman 2023). Patterns of rebel governance in India range from projects of deliberative and democratic governance to extortionist protection rackets. Maoist rebel groups in India may provide health services to civilians by either running their own health drives or bringing health workers into their territories to deliver healthcare (Sahay, Devkota, and van Teijlingen 2016). They may also lobby local elites to expand services to the population, and provide mechanisms of conflict resolution (Shah 2021). To fund their activities, armed groups extract resources from civilians and businesses, such as contractors operating in their areas of influence. This extraction of resources ranges from regular collection of ‘taxes’ – including the issuance of tax receipts – to confiscation and extortion (Shah 2021; Suykens 2015; Thakur 2023). The Janathana Sarkar [people’s government] of the Maoist insurgency also keeps a record of land redistribution and registration, and endows land-use rights to tribal people accordingly (Kunnath 2022). Yet, while some rebel groups issue documents such as tax receipts and land titles, there is no evidence that these groups maintain alternative civil registries and issue life-cycle documents, such as birth certificates. Birth registration, even in India’s conflict zones, thus remains the purview and responsibility of the state.

Barriers to legal identity and birth registration

The Sustainable Development Goals – adopted by UN member states in 2015 as a call for action to build peaceful and inclusive societies – emphasise the commitment to guarantee ‘legal identity for all, including birth registration’ (SDG Target 16.9). Civil registration – the recording of all vital events like births, marriages and deaths in a permanent record – is a key state task, and incomplete civil registries are generally attributed to state weakness. Countries that are poor, or those at war, are seen as unable to provide their citizens with legal identity documents. Arguments about inability are unconvincing in the context of India, however. On the World Bank’s statistical capacity indicator, which measures a country’s ability to ‘collect, analyse, and disseminate high-quality data about its population and economy’, India scores in the top quartile of the 145 countries for which data are available. The country has not missed a census since 1872, and successfully drew up comprehensive voter rolls during the tumultuous period of independence, partition, and widespread ethnic violence from 1947 to 1950 (Shani 2018).³ The Indian state continues to be committed to holding elections, and to making polling accessible even in hard-to-reach areas, including conflict zones (Harbers, Richetta, and van Wingerden 2023; Singh and Roy 2019). The introduction of Aadhaar, the world’s largest biometric identification system, is heralded as a model for other countries in the Global South (World Bank 2019). Nevertheless, while enrolment in Aadhaar – which does not record legal status – has become near universal, birth registration has lagged (Bhatia, Donger, and Bhabha 2021; Manby 2021).⁴ This suggests that the Indian state is capable of reaching and ‘seeing’ its population, but selective in its engagement, which is in line with research showing that even strong states – those with ‘sufficient resources for the maintenance of an effective bureaucratic and administrative state’ (Steinberg 2018, 225) – do not deploy these resources evenly.

A substantial body of literature demonstrates that state-society relations in India are uneven, and characterised by a ‘fragmented and diverse’ set of institutions, practices, and interactions (e.g. Naseemullah 2022). Rather than a binary between state and non-state control, peripheries are often characterised by shifting and contested claims of authority over territory and populations. The Indian state’s response to challengers operating in its territory varies depending on its interest in an area, the cost of governing there, and the perceived threat posed by a group (armed or civilian). The political relationships between the Indian central government and nonstate armed groups range from total war to containment, limited cooperation, and even alliance (Staniland 2021, 3). Therefore, de jure and de facto governance of Indian conflict zones is often blurry, making civil registration challenging (see Fortin, Klem, and Sosnowski 2021 for a comparative perspective). In the following paragraphs, we draw on comparative literature about civil registration and conflict as well as evidence from India to highlight why registration in conflict zones is challenging, how this can create vulnerability, and to identify some of the mechanisms that lead to low rates of registration in conflict zones.

Research shows that conflict exacerbates the challenges of civil registration for citizens, while also raising the stakes (Fortin 2021; Fortin, Klem, and Sosnowski 2021; Hampton 2019; Adamczyk and Doumit 2024; UNICEF and Innocenti Research Center 2009). As outlined above, identity documents are vital to access state-provided services, such as cash transfers, public education and health. Since eligibility for such services generally depends on age, citizenship and/or residential status, identity documents are essential to monitor whether individuals meet such requirements. Citizens unable to produce state-recognised documents proving eligibility are often excluded even where conflict increases poverty and marginalisation. The inability to demonstrate age may also leave children unable to effectively claim the protections to which they are entitled. In India, instances of child recruitment into violent conflicts have been reported in certain conflict areas (Asian Legal Resource Centre n.d.; Shah 2019). Without birth certificates, proving age is almost impossible, and the lack of clear evidence that recruits are minors affords the state and armed groups some deniability. Further, civil registration is indispensable for post-conflict reconciliation, resettlement of civilians and the re-integration of combatants, who need to prove their legal identity and citizenship status (Abboud 2020). Conflict thus simultaneously exacerbates the challenge to, as well as the need for, obtaining identity documents.

How does conflict influence birth registration? On the one hand, the state may be unable to protect local officials from insurgent violence, leaving it unable to staff the required bureaucracy. Registrars are often not prioritised in conflict areas (see Hunter 2019 on Peru and Colombia), so that the infrastructure for civil registration may be inaccessible. On the other hand, as outlined above, the state may have limited interest in ‘seeing’ populations in peripheries in the first place (see Sur 2021 on India’s Northeast). Armed groups often operate in remote and inaccessible areas. The presence of armed groups then further undermines the state’s ability to construct infrastructure, which is perceived by insurgents as imposing state authority and aimed at eroding rebel control. In India, insurgent groups such as the Naxalites have targeted infrastructure projects in order to prevent state access and intervention (Shah 2021). Public health services, which are crucial for promoting birth registration, may also be unavailable. The lack of vital infrastructure renders registration inaccessible for citizens, and security concerns amongst public officials and citizens further dampen registration.

Often states cannot (or do not) distinguish between civilians and insurgents. Shah (2021) recounts numerous instances where state security forces have used violence indiscriminately against populations in India’s ‘red corridor’. She writes:

Everyone was now afraid whenever the security forces crested the hills, terrified of getting caught in a crossfire with the Maoists. Moreover, they said, the guerrillas could escape into the woods, but those left behind in the villages had to face the brunt of police brutality. Those who could fled to the houses of their relatives in other villages when the state forces came to the villages. Others considered joining the guerrilla columns. Still others want to abandon the area entirely. (Shah 2021, 144)

As the quote illustrates, the violence associated with conflict can lead to displacement; and internally displaced people need documents to demonstrate that – as citizens – they are entitled to settle elsewhere in the country. In situations of displacement, birth certificates are essential to document family relations to prevent separation and the trafficking of children. Yet, individuals often lose documents during displacement and expulsion, making it difficult for parents to prove their identity and to obtain birth or death certificates for their children (Hampton and Petkova Khan 2024).⁵

Citizens considering obtaining legal identity documents from the state may have additional security concerns when such documents could be interpreted as signs of disloyalty. Fortin (2021) describes that the information registered on civil documents and identity cards, such as surnames, often carries information about social identifiers like ethnicity, gender or religion. These identifiers may be interpreted by rebels or the state in exclusionary terms. For instance, in Northeast India, the heavily contested politics of registration and legal recognition have been associated with ethnic violence and the targeting of Muslims of East-Bengali origin, where their identities were often marked by highly politicised surnames (Baruah 1999). This culminated in the Nellie massacre of around 1,800 to 2,000 Bengali Muslims in 1984. Given the conflictual legacy and the sheer number of people potentially rendered stateless, civil registration in the context of conflict in India merits greater attention. The case of Assam, discussed above, indicates how demanding proof of citizenship in areas which have historically had low rates of registration, can foster exclusion. Our paper seeks to contribute to the understanding of this issue with a quantitative analysis of how violent conflict shapes rates of birth registration.

Analysis

Empirical approach and data

To examine the relationship between conflict and birth registration, we draw on two main data sources: the 2015–16 wave of the National Family Health Survey (NFHS) (International Institute for Population Sciences IIPS and ICF 2017) for our dependent variable and the Uppsala Conflict Data Program (Davies, Pettersson, and Öberg 2022; Sundberg and Melander 2013) for our main independent variable. The NFHS survey data can be accessed through the Demographic and Health Surveys (DHS) platform.⁶ The survey asks caregivers of children younger than five years whether a child has been registered with civil authorities and possesses a birth certificate. In our analyses, we focus on the actual possession of a birth certificate (rather than registration without a birth certificate), since a document-in-hand provides the strongest protection of rights. The NFHS also includes data on the age and gender of the child as well as on relevant household characteristics, such as poverty, location and social group membership. The survey is nationally representative, and the sample size is sufficient to produce indicators at the state and at the district level. It includes data for 28,522 primary sampling units (PSUs), which consist of local communities, specifically villages in rural areas and census enumeration blocks (CEBs) in urban areas. Overall, the survey, which was conducted between 20 January 2015 and 4 December 2016, provides data on birth registration for 269,137 young children from 601,509 sampled households.⁷

The Uppsala Conflict Data Program Georeferenced Event Dataset (UCDP GED) provides geographically and temporally disaggregated data on organised violence. Specifically, UCDP GED reports incidents where ‘armed force was used by an organised actor against another organised actor, or against civilians, resulting in at least 1 direct death at a specific location and a specific date’ (Högbladh 2022, 4). For the period between 1989 and 2010, that is the period prior to the births of the cohort included in NFHS’ registration data, UCDP reports 11,674 events of organised violence in India, which belong to 49 distinct conflicts, and 59 conflict dyads.⁸ Figure 1 maps the location of all events reported during the time period. It illustrates the spatially uneven distribution of conflict violence described above with events clustering in the Northwest, the Northeast, and the ‘red corridor’ in the Interior. Overall, the figure indicates that exposure to organised violence varies sharply based on where a family happens to live.

To capture this uneven exposure to conflict, we transform events mapped in Figure 1 with the kernel density tool in ArcGIS. Kernel density smoothing allows us to generate a measure of conflict exposure for each primary sampling unit, because the technique converts events (or points in GIS terms) in Figure 1 into a continuous variable that has a value for each location (or cell in a raster dataset) on the map. This variable indicates the density of events nearby. For each event, locations close to the event receive higher values derived from the event than those further away. For each location, the continuous variable reflects the cumulative effect of the proximity to specific events and the frequency of nearby events.⁹ We obtained the coordinates of primary sampling units from DHS, and extracted the values for our two conflict exposure variables at each PSU location.¹⁰ Figure 2 visualises the conflict exposure scores for PSUs. In addition to the spatially uneven exposure to conflict, the map also illustrates the broad geographic coverage of the NFHS.

Figure 2. Conflict intensity at primary sampling units.

Colours reflect the intensity of conflict exposure, with darker colours indicating higher values.

Beyond our main variables, we include controls for risk factors commonly associated with low registration rates and marginalisation (Bhatia et al. 2019; Bhatia, Kim, and Subramanian 2021; Kumar and Saikia 2021; Mohanty and Tesfaye 2018). At the individual level, we control for the child’s gender, the child’s age, and the mother’s level of education. At the household level, we include urban/rural residence and household wealth, measured with an index that groups households into asset-based quintiles. We also control for the head of household’s religion and whether the head of household is a member of the so-called categories of Scheduled Castes, Scheduled Tribes or Other Backward Classes (OBC).¹¹ In addition to conflict exposure, we include the altitude of the primary sampling unit, the standard deviation in altitude across the district, and average forest cover in the district. These geographic variables are intended to tap into the ruggedness of the terrain (Tollefsen and Buhaug 2015). As outlined above, geography shapes the costs of building infrastructure, including infrastructure for the provision of public health and the bureaucracy required for civil registration. Rough and difficult terrain is therefore often associated with lower levels of public goods provision. In addition, remote and inaccessible areas often become sites of conflict as armed groups retreat into areas that are difficult for security forces to access. We expect rugged terrain to have a negative impact on birth registration. Data for the altitude of the primary sampling unit comes directly from the NFHS. We calculated the standard deviation in altitude at the district level based on a digital elevation model provided by the U.S. Geological Survey’s Center for Earth Resources Observation and Science (EROS), and district-level shapefiles shared through SHRUG, the Socioeconomic High-resolution Rural-Urban Geographic Data Platform for India (Asher et al. 2021).¹² Average forest cover at the district-level in 2010 has been calculated on the basis of Vegetation Continuous Field (VCF) data originally provided by Dimiceli et al. (2015) and available through SHRUG (Asher et al. 2021). We report descriptive statistics for all variables in the analysis in the appendix.

Analysis

Our observations are children in households, that live in local communities (i.e. primary sampling units), in districts and in states. Factors at all these levels likely influence whether a young child has been registered and possesses a birth certificate. Our analysis therefore includes variables at different levels. Since the dependent variable is binary, we analyse the relationship between conflict exposure and possession of a birth certificate with a series of logistic regression models.¹³ Model 1 pools observations across states, whereas Model 2 includes state fixed effects to capture state-specific differences. Model 3 is a multilevel mixed-effects logistic regression model with state-fixed effects and districts as the second level. Results of all three models are reported in Table 1.

Table 1. Models predicting likelihood that a child is registered and in possession of a birth certificate.

	(Model 1)	(Model 2)	(Model 3)
	Logit	Logit with state fixed effect	Multi-level logit with state fixed effects
Conflict Exposure	−.076***	−.051***	−.024*
	(.003)	(.005)	(.011)
Mother’s Education
No education (reference)
Primary	.477***	.336***	.299***
	(.013)	(.015)	(.015)
Secondary	.756***	.603***	.548***
	(.011)	(.013)	(.013)
Higher	.772***	.861***	.802***
	(.020)	(.022)	(.023)
Don’t know	.363*	.119	.091
	(.161)	(.175)	(.176)
Child’s Gender: Female	.066***	.068***	.072***
	(.009)	(.009)	(.010)
Child’s Age (years)	.069***	.072***	.073***
	(.003)	(.003)	(.003)
Urban Residence	.291***	.232***	.254***
	(.012)	(.014)	(.014)
Household Wealth
Poorest (reference)
Poorer	.369***	.286***	.299***
	(.012)	(.014)	(.014)
Middle	.635***	.477***	.494***
	(.014)	(.016)	(.016)
Richer	.848***	.625***	.651***
	(.016)	(.018)	(.019)
Richest	1.169***	.949***	.982***
	(.020)	(.023)	(.024)
SC/ST Status
Scheduled Caste	−.010	−.041*	−.059***
	(.015)	(.017)	(.017)
Scheduled Tribe	.097***	−.315***	−.304***
	(.017)	(.019)	(.021)
Other Backward Class	−.106***	−.056***	−.056***
	(.013)	(.014)	(.015)
None of Above (reference)
Don’t Know	.161**	−.197**	−.187**
	(.055)	(.060)	(.061)
Religion of Head of Household
Hindu (reference)
Muslim	−.273***	−.113***	−.051**
	(.014)	(.015)	(.016)
Christian	−1.076***	−.171***	−.168***
	(.026)	(.033)	(.037)
Sikh	1.594***	.608***	.521***
	(.060)	(.074)	(.079)
No Religion	−1.618***	−.461**	−.199
	(.167)	(.168)	(.178)
Other	−.546***	.010	−.041
	(.030)	(.034)	(.037)
Geographic Variables
Standard Deviation in Altitude (within district)	−.001***	.000	.000
	(.000)	(.000)	(.000)
Altitude in meters (PSU)	−.000***	−.000***	−.000***
	(.000)	(.000)	(.000)
Average Forest Cover (within district)	.025***	.007***	.004
	(.001)	(.001)	(.003)
State Fixed Effects
Andaman and Nicobar Islands (reference)			.
Andhra Pradesh		−1.983***	−1.923***
		(.181)	(.350)
Arunachal Pradesh		−2.875***	−2.756***
		(.188)	(.398)
Assam		−.705***	−.559
		(.181)	(.342)
Bihar		−2.728***	−2.743***
		(.177)	(.331)
Chandigarh		−1.249***	−1.243*
		(.283)	(.606)
Chhattisgarh		−1.480***	−1.381***
		(.178)	(.342)
Dadra and Nagar Haveli		−.213	−.213
		(.254)	(.592)
Daman and Diu		−.675**	−.593
		(.253)	(.499)
Goa		.184	.297
		(.355)	(.562)
Gujarat		−.477**	−.359
		(.180)	(.337)
Haryana		−1.297***	−1.026**
		(.179)	(.340)
Himachal Pradesh		−2.111***	−1.964***
		(.184)	(.368)
Jammu and Kashmir		−2.210***	−2.492***
		(.188)	(.377)
Jharkhand		−2.625***	−2.612***
		(.178)	(.336)
Karnataka		−.998***	−.914**
		(.180)	(.336)
Kerala		−1.645***	−1.508***
		(.191)	(.374)
Lakshadweep		−.340	−.350
		(.302)	(.618)
Madhya Pradesh		−1.446***	−1.344***
		(.177)	(.329)
Maharashtra		−1.833***	−1.688***
		(.178)	(.333)
Manipur		−3.449***	−3.393***
		(.182)	(.378)
Meghalaya		−2.529***	−2.475***
		(.183)	(.391)
Mizoram		.636**	.888*
		(.200)	(.413)
Nagaland		−2.573***	−2.508***
		(.184)	(.390)
Delhi		−1.964***	−1.969***
		(.186)	(.362)
Odisha		−2.409***	−2.367***
		(.178)	(.334)
Puducherry		−.709**	−.590
		(.218)	(.423)
Punjab		−.683***	−.569
		(.190)	(.346)
Rajasthan		−2.547***	−2.521***
		(.177)	(.332)
Sikkim		−.377	.057
		(.222)	(.472)
Tamil Nadu		−.452*	−.367
		(.183)	(.338)
Tripura		−.851***	−.759
		(.198)	(.419)
Uttar Pradesh		−3.495***	−3.482***
		(.177)	(.327)
Uttarakhand		−2.240***	−2.105***
		(.181)	(.370)
West Bengal		−.395*	−.333
		(.183)	(.345)
Telangana		−1.876***	−1.858***
		(.182)	(.356)
Constant	−.892***	1.664***	1.690***
	(.018)	(.177)	(.322)
var(_cons[shdistri])			.213***
			(.014)
Pseudo R2	0.0949	0.2112
Groups (Districts)			640
ICC (Districts)			.06
N (observations)	249157	249157	249157

*p < 0.05, **p < 0.01, ***p < 0.001.

Overall, our findings highlight how social marginalisation undermines birth registration. This is in line with previous research. At the individual level, the mother’s education plays a significant role, with children whose mother has a higher level of education significantly more likely to possess a birth certificate. At the household level, urban residence and household wealth increase the likelihood of registration. When the head of household belongs to a minoritised or marginalised group, this significantly decreases the likelihood that a child possesses a birth certificate. Specifically, our results show that children born into a so-called ‘Scheduled Caste’ or ‘Other Backward Class’ (OBC) are significantly less likely to possess birth certificates. Religious group identity also plays an important role. Muslim and Christian children are less likely to be registered than Hindu children, while children in Sikh households are more likely to be registered. As expected, geographic factors also matter. Altitude, which taps into the remoteness of a community, is negatively associated with registration across models. The standard deviation in altitude at the district level also has a negative, though not consistently significant, effect. Forest cover, somewhat surprisingly, has a positive effect on birth registration, though this is not consistently significant. State fixed-effects in models 2 and 3 highlight the variation in registration rates across states even when controlling for many household and individual-level factors.

With regard to conflict, our main variable of interest, our results show that children in areas affected by violent conflict are significantly less likely to possess birth certificates. This effect is robust across model specifications. Figure 3 illustrates the negative effect of conflict on the likelihood that a child has been registered across our three models. Even though the size of the effect is diminished in the models that take fuller account of state and district effects, it remains significant. The negative effect of conflict thus remains robust even when taking into account the social marginalisation that characterises many conflict zones, such as poverty and remoteness.

Figure 3. Coefficient plot of the effect of conflict exposure on birth registration across models.

To make these effects more concrete, consider how the characteristics of children and their environment increase or decrease the chance that the child possesses a birth certificate. For all of our examples, we draw on socially marginalised groups that are likely to live in conflict zones. Caregivers report that 58% of all children included in the survey are registered and in possession of a birth certificate. Now compare this with the predicted probability that a girl born into a Hindu household belonging to a ‘Scheduled Caste’ possesses a birth certificate. If her mother has no education, the household belongs to the poorest segment of the population, and she lives in a rural area not affected by conflict, there is a 38% chance that she possesses a birth certificate. This is considerably lower than the average predicted probability of 58% for all children. If the same girl is born in one of the communities most affected by conflict, however, her chance of possessing a birth certificate drops to 13%.¹⁴ Let’s also consider the likelihood that a boy born into a Christian household belonging to a ‘Scheduled Tribe’ in the Northeastern state of Manipur possesses a birth certificate. His family belongs to the poorest segment of the population and lives in a rural area, but his mother has received some education. Registration rates in Manipur are considerably lower than in other parts of the country, with only 30% of children in possession of a birth certificate. The state has long been affected by violent conflict. Even if the boy is born in one of the most peaceful areas of the state, his chance of possessing a birth certificate is only 14%. If he lives in the communities of the state most affected by conflict, however, his chance of possessing a birth certificate drops with an additional point to 13%.¹⁵ The state with the highest conflict scores in our data is Jammu and Kashmir.¹⁶ As an additional example, let’s consider the chance that a Muslim boy from a ‘Scheduled Tribe’ in Jammu and Kashmir possesses a birth certificate. The boy lives in a rural area, his family is poor, and his mother did not attend school. Only 41% of the children in the state possess a birth certificate, which again is considerably below the national average of 58%. The boy’s chance of having a birth certificate is 27% if he lives in one of the more peaceful areas of the state, but only 17% if he lives in one of the communities most affected by violent conflict.¹⁷ Overall, our results show how conflict compounds the vulnerability of groups that are already at the margins.

Conclusion

People in conflict zones are in particular need of protection and support, as conflict exacerbates existing vulnerabilities. Yet, children born in conflict zones face difficulties obtaining state-recognised identity documents. Our analysis documents a strong negative effect of conflict on possession of birth certificates. The vulnerability of children in conflict zones is compounded by the fact that they are often left out of official records. In interactions with the state, they will be at a disadvantage when proving their status as citizens, and when claiming state-provided benefits. Because eligibility for Indian citizenship depends largely on descent, issues arising from a lack of birth registration may persist across generations. Our findings indicate that civil registration, and the provision of identity documents protecting citizenship rights, need to be part of post-conflict peacebuilding programs. The ‘peace through health’ framework (e.g. Vass 2001; Christensen and Edward 2015) gains relevance in these contexts due to the potential of maternal and antenatal care to facilitate timely registration and, consequently, the safeguarding of vulnerable children. Given the importance of documents for claiming citizenship rights, practitioners need to remain vigilant about birth registration even in conflicts that may not rise to the level of ‘armed conflict’ under international humanitarian law, and even in situations where there are no alternative authorities seeking to govern populations by issuing life cycle documents (see Ganohariti 2024; Grant-Brook 2024).

India has a strong track record with regard to collecting information about populations. That we find a negative effect of conflict on birth registration even in a country with a fairly strong state is noteworthy. Low registration rates have often been attributed to a lack of state capacity. Our results suggest that scholars and practitioners need to think beyond state weakness in evaluating civil registration (or lack thereof). States may have a strategic incentive to exclude populations residing in contested territories and to avoid providing citizenship-based protections and rights. When state actors perceive populations as challenging or disloyal, they may limit engagement leading to low rates of registration. Citizens, who have experienced mostly coercive state power, may also be reluctant to engage with the state directly. In such instances, peacebuilding programs need to include confidence-building measures and mitigate perceived-risk to ensure the registration of civilian births and therefore the protection of vulnerable communities. Moreover, as the boundaries of Indian citizenship become increasingly politicised and the ability to conclusively prove belonging with documentary evidence emerges as key to protect citizenship rights (Punathil 2022; Jha and Chakrabarty 2023), it is crucial to highlight uneven access to legal identity across India. Many with a legitimate claim to citizenship, but historically at the margins, have a harder time meeting the burden of proof.

Financial support

This project was supported by the European Research Council (ERC Starting Grant #949795).

Acknowledgments

An earlier version of this manuscript was presented at the workshop “Legal Identity under Insurgencies and Unrecognised States” hosted by the German Institute for Global and Area Studies (GIGA) in Hamburg in 2023, and at an online workshop with other contributors to this special issue. We gratefully acknowledge helpful comments from participants, especially Gehad Abaza, Jenny Hedström, Bart Klem and Marika Sosnowski. We also thank Gulshan Banas and Vy Tran, as well as the anonymous reviewers, for their feedback on the article, and Victor Hartman for research assistance.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by the European Research Council [949795].

Notes

1. Note that only some of these conflicts can be classified as ‘armed conflict’ under international humanitarian law. The Rule of Law in Armed Conflicts (RULAC) initiative classifies the conflict between the Indian state and the Naxalites as a non-international armed conflict, and clashes between India and Pakistan over Kashmir, and between India and China over Aksai Chin, parts of Arunachal Pradesh and along the Sino-Indian border as international armed conflicts. See https://www.rulac.org/browse/countries/india [accessed July 9, 2023].

2. We use the term periphery in the same vein as Rokkan (1999) to denote areas distinct from the administrative and economic ‘centres’ that dominate the process of state formation. Peripheries are a result of state practice in the sense that the centre attempts to incorporate them in its own political project, and on the centre’s terms.

3. The 2021 census had first been delayed by Covid, and subsequently by contention around the nationwide rollout of a National Register of Citizens (Daniyal 2022). This is a historic first.

4. Biometric identification based on finger prints and iris scans can be a powerful tool to improve the functioning of welfare systems. In India, Aadhaar has played a role in reducing corruption, and it has provided adults without birth certificates with a way to identify themselves in interactions with the state and other citizens. Aadhaar is available to residents of India, and as such does not prove citizenship.

5. Civil registries may also be intentionally destroyed, rendering the replacement of lost documents virtually impossible. While we have not seen evidence of this in India, see for example Sardelić (2021) on the Balkans and Sperfeldt (2020) on Cambodia.

6. Data are available at https://dhsprogram.com/methodology/survey/survey-display-355.cfm [accessed 23-03-2023]. Replication files for our analysis and the online appendix are available through the Harvard Dataverse at https://doi.org/10.7910/DVN/RGDGKM.

7. DHS distinguishes between ‘de facto’ children, i.e. those that slept in the household the night before the interview, and ‘de jure’ children, i.e. children ‘who are usual residents of the selected households, whether or not they stayed in the household the night before the interview’. We include both de jure and de facto children in our analysis as displacement may account for some of the difference between the two categories.

8. We focus on the period prior to the birth of the children to ensure that conflict exposure did not occur after the children were born (and possibly registered).

9. All reported events are weighted equally in the calculation.

10. We use Spatial Analyst in ArcGIS and specifically the ‘Extract Multi Values to Points’ tools. Coordinates for 131 PSUs appear to be incorrect, as their location is reported far outside the international boundaries of India. We drop those units from our analysis.

11. These terms are state classifications. The Indian Constitution recognizes the entitlement of certain marginalized communities to special protections, and maintains lists of communities who qualify for this status.

12. The digital elevation model for Asia is available through ArcGIS Online, but can also be accessed here: https://databasin.org/datasets/366a1bef53344c02bcd7d7611d5f61f7/.

13. We report additional models that probe the robustness of our results in the online appendix.

14. This calculation of predicted probabilities is based on Model 1 with the continuous variables in the model, such as child’s age, ruggedness, altitude and forest cover, held at their means. The conflict exposure variable is set at 0 and 18. Empirically, such high values of conflict exposure are found only in Jammu and Kashmir. If conflict exposure increases by one standard deviation (from 0 to 2.8), which is a more typical value, the predicted probability still drops to 33%.

15. This calculation of predicted probabilities is based on Model 2 with the continuous variables in the model, such as child’s age, ruggedness, altitude and forest cover, held at their means. The conflict exposure variable is set at .375 and 3.106, the actual range of the variable in Manipur.

16. As of 31^st October 2019, the state has been split into two (Union) Territories.

17. This calculation of predicted probabilities is based on Model 3 with the continuous variables in the model, such as child’s age, ruggedness, altitude and forest cover, held at their means. The conflict exposure variable is set at 0 and 24.91, the actual range of the variable in JK.