Statistical mirage: Bad data to fail governance

Prime Minister Narendra Modi’s election promises are under intense scrutiny these days. But one of the election promises that has significant implications for policymaking has received hardly any attention. The BJP’s Lok Sabha election manifesto hinted at the need to overhaul the government’s data collection machinery. It promised to setup “an institute of big data and analytics for studying the impact of big data across sectors for predictive science.”

The manifesto presented real time data and big data as means to solve problems as diverse as agricultural distress and intelligence gathering. But not much has been heard on the statistical front since the elections. Two developments stand out, though. First, India emerged as the fastest growing large economy after the revision of both the base year and the methodology used to estimate national income.

Second, while the BJP accused the UPA government of withholding the publication of the 2011 Census data on religion, the Modi government has also not published the data that was collected almost half a decade ago. It would not be wrong to call this a statistical U-turn of the Modi government.

The electoral enthusiasm for real time data and big data notwithstanding, there are deeper reasons why there are no easy solutions to the unavailability of good quality data for policymaking.

A country’s ability to generate and use information depends both on its administrative and technological capacities as well as the structure of its polity and economy. We will highlight the structural constraints with the help of examples from India’s northern periphery, the northeastern states and Jammu and Kashmir.

India is among the very few developing countries with a relatively robust government statistical machinery. Yet, its peripheral states suffer from data deficit, that is, unavailability of data and/or poor quality of available data. Most national level surveys either do not cover the North-East and Jammu and Kashmir or cover them irregularly. Some of the surveys that cover these regions do not have sufficiently representative samples to generate reliable estimates at the state level.

In some states, the areas surveyed vary across years. And, in extreme cases, survey samples have been drawn using old (like Assam during the 1980s and J&K during the 1990s) or even faulty (Nagaland during 1991-2011) census data. The reliability of census with regard to states such as Nagaland, J&K, Manipur and Assam is moot. In Nagaland, errors accumulated in census over more than two decades. In 2001, when the error peaked, the census overestimated Nagaland’s population by more than 25 per cent.

A similar problem affected the population estimates for parts of northern Manipur in 2001 and 2011. In J&K, the 2011 Census seems to have overestimated the male child population of the Kashmir division. The data deficit affects both academic research and policymaking because academic studies and government reports are either restricted to “major” states or they end up using flawed or incomplete data for “minor” states.
The peripheral states are also known for democracy and development deficits in addition to data deficit and the three deficits are interlinked.

Consider the democracy and data deficits first. Political disturbance, which limits both the physical reach and legitimacy of the state, also affects the state’s capacity to collect information. In extreme cases, the government may not be able to collect statistics. For instance, census could not be conducted in Assam and J&K in 1981 and 1991, respectively, due to disturbed conditions.

Lack of trust
Lack of trust in government institutions contributes to non-cooperation with the government’s data collection machinery or even the instrumental manipulation
of government data. Faulty electoral rolls and incorrect community-level headcounts, in turn, reduce faith in the state as well as democratic processes and encourage manipulation in future.

While it is widely acknowledged that good quality data is crucial for development planning, lack of development also affects data quality. Ethnically divided societies having weak state institutions, which cannot mediate between communities, and poor economies, where the government is the primary employer, invariably face the prospect of inter-community competition over scarce resources.

Ethnic competition often takes the form of manipulation of census and electoral rolls to secure better representation in legislature that in turn helps cornering public resources for the legislator’s community or constituency. Manipulation affects both the quality of statistics and their public acceptability.

Development and democracy deficits are also interrelated. Lack of development fuels insurgency and insurgency impedes development. Insurgents block development projects to secure their support base that might otherwise be weaned away by New Delhi.

Insurgents try to disrupt democratic processes as they equate participation in them with submission to the mainland’s hegemony and the idea of India as a nation state.

The disruption is used as an excuse by the state to suspend civil liberties, which pushes democracy further into the background and the resultant conflict impedes development.
The conflict over delimitation highlights the inter-relationship among the three deficits discussed here. In 2002, in most peripheral states, people opposed census-based delimitation. In some of these states, the dominant community is not prepared to accept census data that do not uphold their numerical dominance and for the same reason the smaller communities do not trust government data.

In short, the data deficit is both a cause and consequence of ‘development’ and ‘democracy’ deficits along India’s northern periphery. Any reform that addresses only one of these deficits in isolation is bound to fail as the underlying structural problems will remain unaddressed.

In fact, without strengthening political institutions and attending to the development deficit, recourse to administrative and technological solutions might even aggravate the data deficit.

(Kumar and Agrawal teach economics at Azim Premji University, Bengaluru and IIT Delhi, respectively)