Introduction

The coverage and content errors identified in Chapter 3 are not randomly distributed across Jammu and Kashmir (J&K). Content errors are mostly confined to the data on non-scheduled languages and Scheduled Tribes (STs). The nature of content errors, though, changes across the three regions of the erstwhile state. Coverage errors, on the other hand, mostly affect the headcount of Kashmir and the STs. Since conventional explanations fail to account for the errors, we will have to explore the larger context, including the politicisation of census, that may have affected the reported headcount.

Panandiker and Umashankar (1994: 96–97) argue that in states where Hindus are a minority – for example, Muslim-majority J&K and Sikh-majority Punjab – the majority views population control as a measure to reduce their political strength. They add that this is not true of Kerala because of the higher literacy rate and per capita income. However, it is not true that all such provinces are suspicious of population control. Christian-majority states of the north-east are cases in point. Further, in Kerala, the population is divided among Muslims, Hindus and Christians of various castes and sects, with only Muslims having a clear geographical concentration in the north. The fractionalisation and geographical dispersion of communities and strong linguistic ties transcending religions check persistent communal polarisation in Kerala. Even in Punjab there is no region exclusively populated by only one religious community, and despite militancy both the communities share a linguistic and regional identity. Polarisation is further mitigated by the presence of heterodox sects and Scheduled Castes (SCs) among both Hindus and Sikhs in Punjab.

J&K is different. It has two major geographically separable religious groups (Tables 4.1a–4.1c). The exodus of Hindus from Kashmir in the 1990s further hardened this divide. The strongholds of Kashmiri-speaking Muslims and Dogri-speaking Hindus do not overlap even though they share the old Doda district of the Jammu division with other communities. STs are spread across all districts of J&K. They are almost entirely Muslim except in Ladakh and parts of Jammu. On the other hand, SCs are entirely Hindu and confined to Jammu.

Introduction

More than three decades ago, when Amartya Sen flagged the problem of missing women (Sen 1990), India's overall sex ratio and child sex ratio (CSR) were 927 and 945, respectively. Since then sex-determination technology that ‘permits couples to resort to sex selection’ has played a major role in skewing sex ratio amidst ‘declining fertility and entrenched son preference’ (Guilmoto 2009: 524). Son preference also contributes to the neglect of the health of the girl child, which further aggravates the skewed sex ratio by increasing the girl child mortality. The union government introduced the Pre-Conception and Pre-Natal Diagnostic Techniques Act, 1994, to prohibit the misuse of diagnostic techniques to identify the sex of the foetus. Yet, in 2001, ‘the child sex ratio (CSR) first dropped below that of the overall sex ratio’, and the problem spread beyond the north-western states (John 2011: 11). A decade later, the overall sex ratio increased to 943, even as the CSR further dropped to 918. The CSR has, in fact, steadily declined from 976 to 918 between 1961 and 2011, even as the overall sex ratio fluctuated between 927 and 943 (Figure 3.1).

Very low CSRs were first reported in some of the north-western states, but Jammu and Kashmir (J&K) was not among them. However, defying conventional wisdom, the state, which was above the all-India CSR until 2001, reported a sharp decline in 2011 that placed it among the worst-performing states in the country (Figure 3.2). This development has not received adequate attention in academic and public debates, possibly because insurgency-hit J&K is widely seen as an exceptional state, and, in any case, it accounts for about 1 per cent of the country's population and less than 1 per cent of the national income. In other words, J&K does not have a large impact on national figures and is not among the major states that are indispensable for national (policy) debates and academic analyses.

Researchers and policymakers have uncritically used the results of the 2011 census for J&K. In fact, census officials, too, did not critically examine the data (see, for instance, Government of India [GoI] 2011d: 81–84). The joint director of census operations (Jt DCO) observed that ‘child sex ratio was equally worrisome as it has dropped by 100 points from 963 in 1981 to 863 in 2011.

Introduction

We examined coverage and content errors in the census data for Jammu and Kashmir (J&K) and discussed their administrative, legal and political contexts at different levels of aggregation. We found that the over-reporting of children (particularly male children), a large increase in the slum population and a large increase in the number of households between the houselisting and household phases of census in Kashmir explain several interconnected anomalies in the 2011 census of J&K, including the drop in the child sex ratio (CSR), the rise in the share of child population, the rise in the population share of Kashmir within the state and the drop in the corresponding population shares of the Jammu division and groups concentrated almost entirely in Jammu such as the Scheduled Castes (SCs). Further, errors in data on non-scheduled languages, dialects of scheduled languages and tribes were also examined and attributed to political mobilisation, unintentional misclassification and a large increase in the population of Generic Tribes. Even conservative estimates of politically motivated over-reporting of the headcount, which do not account for (a) the over-reporting in 2001 that manifested in the abnormal increase in mean household size of Kashmir and (b) the over-reporting of the population aged 10–14 years in 2011, suggest that the 2011 census overestimated Kashmir's population by about 10 per cent.

We argued that the inability of the government to conduct reliable censuses in Kashmir – reflected in the cancellation of the exercise on two occasions, ad hoc changes to the reference date in other censuses and the contested nature of the data – can be explained from three different perspectives.

Introduction

The quality of census data for Jammu and Kashmir (J&K) is circumscribed by uncertain weather, armed insurgency, forced migration, regional and communal polarisation, uneven accounting of armed forces, international conflicts, bureaucratic moral hazard, ineffective legal and administrative measures to check manipulation and changes in borders, reference dates, demarcation of snowbound territory and distribution of mobile population groups. Our ability to understand the impact of these factors on data quality is limited by the poor quality of metadata. It is, however, not clear why successive governments have failed to pay attention to the growing data deficit of J&K. Most recently, the anomalous headcounts of 2001 and 2011 did not prompt a closer scrutiny. We found that the over-reporting of children (Chapter 3) and an unusually large growth of the slum population (Chapter 4) in Kashmir account for several intertwined coverage errors in the reported population of the state. In this chapter, we will show that in 2011 the number of households was also inflated in Kashmir. These anomalies inflate the population shares of both Kashmir and Muslims, and the effect is compounded in the case of the latter by a very sharp rise in the population of Generic Tribes (Chapter 3). We found that conventional demographic and non-demographic factors cannot explain these anomalies and argued that in both 2001 and 2011 enumeration was affected by communal propaganda in Kashmir that fanned fears of demographic marginalisation. Chapters 4 and 5 discussed the political and legal–administrative contexts of the census. The present chapter relates the indifference of New Delhi towards anomalous population statistics of J&K to strong priors that suggest over-enumeration is unlikely, and that too in conflict zones, and to an older tendency to read population growth as a measure of well-being. It then discusses coverage and content errors in census data on Punjab and J&K. It argues that some of the major anomalies in the headcount of J&K are products of deliberate intervention rather than being mere aggregations of random errors at the grassroots.

Introduction

Located in the north-west extreme of India's ethno-geographic periphery, Jammu and Kashmir (J&K) was initially a Category B state. Unlike other princely states belonging to Category B and Category C that were absorbed into larger states or reconstituted as union territories (UTs), J&K was directly granted the status of a separate state. The state had two capitals: Srinagar (summer) and Jammu (winter). Before being reorganised into two UTs on 9 August 2019, the state comprised three administrative divisions: Jammu (26,293 square kilometres), Kashmir (15,948 square kilometres) and Ladakh (59,146 square kilometres). The last was until recently a part of the Kashmir division and was constituted as a separate administrative division in February 2019, less than six months before the reorganisation of the state. The reorganisation marked the end of the seven-decade-old constitutional arrangement under Article 370 that guaranteed autonomy to the state to accommodate the restive Kashmir Valley amidst domestic and international challenges to the Muslim-majority princely state's accession to India (Timeline 2.1). International and domestic conflicts spawned by the unsettled political status have affected the whole range of government statistics in J&K, compounding the challenge posed by difficult weather and terrain.

In this chapter, we will first discuss the less-than-satisfactory accounting of the area of J&K. This will be followed by a discussion of how political instability, uncertain weather, mobile population groups and shifting reference dates affect headcounts and how our ability to understand their impact on the quality of census data is constrained by the paucity of metadata.

Indeterminate Borders

The territory of J&K under the Indian administration has remained largely unchanged after tribal militia backed by the Pakistani army seized parts of the princely state in 1947 and China occupied parts of Leh in the 1950s and early 1960s. The only officially recognised change on the western front happened under the Simla Agreement of 2 July 1972 after the 1971 Indo-Pakistan war that froze the ‘line of control [LoC] resulting from the cease-fire of December 17, 1971’ (GoI n.d.2). Census reports are deficient in information on international territorial changes, though.

In the first two decades after independence, the union government's role as the facilitator of interstate redistribution was closely linked to decennial population censuses. The allocation of seats in the parliament and federal funds tracked the most recent census data. In the mid-1970s, the growing concern over a rapidly increasing population amidst food and other scarcities forced a hasty uncoupling of the census and key federal policies to make room for more aggressive population control measures. This was, perhaps, necessary to protect the interests of states that had already achieved relatively lower levels of fertility. In the following decades, the census could not be conducted in Assam (1981) and Jammu and Kashmir (J&K) (1991) due to political disturbances. (In 1994, substantive changes were introduced in the Census Act, 1948, to expand the scope of punitive measures, among other things.) The year 2001 was therefore very important for the Census of India because the government was trying to enumerate the whole country once again after a gap of three decades, and there was an expectation that interstate redistribution of resources and power could be recoupled with the headcount. There was also a hype around the first census of the new millennium.

While the census managed to cover the entire country in 2001, it was marred by the politicisation of the headcount. The government had to postpone interstate delimitation to until after the first census taken after 2026. Six states – Arunachal Pradesh, Assam, J&K, Jharkhand, Manipur and Nagaland – could not even use the latest data for intrastate delimitation. In some of these states, the civil society and political parties alleged that the process of enumeration was subverted by vested interests and moved courts to challenge delimitation based on a flawed census. At least five others – Chhattisgarh, Meghalaya, Sikkim, Tripura and Uttarakhand – conducted intrastate delimitation under (political) constraints.

Soon after the 2001 census, Radhabinod Koijam, a former chief minister of Manipur, flagged the politically contentious nature of cartographic and demograhic statistics in his neighbourhood. In 2005, in an interview with Sanjoy Hazarika, chief minister Neiphiu Rio admitted that Nagaland's headcount was highly inflated due to political competition among tribes. As a student of engineering and later economics, I found it intriguing that a ‘simple’ measure such as headcount could be so deeply contested.