Why India needs to invest in a strong and reliable statistical system — Lessons from a pandemic

As the Covid-19 pandemic continues its ominous, uncertain journey worldwide, the need for accurate, regular information on its spread and containment has been felt by all. But most of us do not readily realise that uninterrupted flow of accurate, reliable information is not the default, but involves a careful, planned design.

Information is power. Information that flows from a structured system is equally powerful and useful. Both for those who are expected to make use of the information — think decision-makers in government and researchers in the academia — as well as for those such information pertains, common citizens for example. Robust information systems are synonymous with better transparency, accountability and effectiveness of public policy. They are also a natural characteristic of modern citizen societies and welfare states. Weak information systems either follow archaic models of collecting, compiling and disseminating data, or lack critical qualities of credibility and legitimacy in the procedures followed. In some cases, both.

Brilliant statisticians but a neglected statistical system

Evolution of the Indian Statistical System — In fits and starts

Consider the Indian example. In the pre-independence times, a rudimentary statistical system of collecting and compiling information was cobbled together by personal efforts of some pioneering British statisticians and military officers. But it largely remained at best an annexure to the correspondences flowing between London and Kolkata (and later Delhi) on state of affairs[i] in Britain’s richest colony. These were mostly catalogue of businesses conducted by different departments with an economic relevance such as agriculture, land revenues, trade and commerce and later, of a nascent manufacturing industrial sector. Devoid of integrating into any system, collection and compilation of the reported figures were carried out by different officials and branches of the colonial administration in silos[ii]. There were sporadic attempts to set up a Statistical Office, again due to personal efforts of some officers e.g. the Bowley-Robertson Report (1936) for establishing a Central Statistical Organization and a separate Economic Intelligence Organization (Subramanian, 1960[iii]). But, reminiscing of present-day administrative bickering of which department has jurisdiction over certain technical functions, debates and arguments within the government ensured that little synthesis took place, and the Office functioned with a skeletal staff with little technical training, and little funding.

The Indian Statistical Institute (ISI) in Kolkata (then Calcutta), growing out of a statistical laboratory at the Presidency College (now University) established by the doyen of modern Indian statistics, Prasanta Chandra Mahalonobis in 1931, was one of the first academic institute dedicated to the study of statistics outside the Western world. Mahalonobis, with his famed closeness to Nehru, was also one of the members of independent India’s first Planning Commission and the statistical adviser to the first Cabinet, which set up the Central Statistical Organization (CSO) in 1950. In the early 1950s, Mahalonobis was already a globally-renowned figure on statistical sample surveys and under his influence, India was the first nation outside Europe and America to introduce a regular system of national sample surveys, under the fledgling National Sample Survey Organisation (NSSO).

‘Commanding heights’ of large-scale sample surveys and massive census operations

In the last seven decades, NSSO has conducted almost 80 national, large-scale surveys on issues such as employment, consumption, health, education, migration, land-holding, agriculture, civic amenities etc. — these sample surveys has been the mainstay for estimating national and state-level figures for poverty and employment/unemployment levels across the country for over 50 years. The decennial national census is also another remarkable, gargantuan month-long exercise where almost an entire army of schoolteachers, health workers, local government officials visit more than 200 million households and collect an enormous amount of information. In recent years, several other excellent surveys on health and education — in some cases (e.g. the National Family Health Surveys (NFHS), the Annual Status of Education Reports (ASER)) with technical and minor financial support from international agencies — have significantly improved availability of information on several key social indicators, some of which are essential for monitoring of national performance on global development goals, e.g. the Sustainable Development Goals of the UN.

Of blind spots and avoidable controversies

However, these surveys have been mostly discrete, independent exercises with little coordination between each other as well as with other similar surveys conducted by other agencies, at times even by the same department of the government! This causes costly overlaps, inefficient use of resources and naturally, common blind spots — subjects and quantities for which little information is available. Most of such inefficiencies and weaknesses are due to lack of a coherent, integrated statistical system across the country that should link different key sources of information (e.g. surveys, routine administrative information) as well as the different agencies (statistical cells of different state and central government departments; independent academic or non-governmental organizations) responsible for collecting these.

In recent years, there has been several controversies on availability and transparency of essential information on key aspects of the national economy in India, the most remarkable being the much-delayed release and then hurried withdrawal of data on consumption expenditure from 2017–18 round of the survey by the government, citing incomprehensible reasons. Similarly, the NSSO employment surveys has been discontinued since 2011–12 and replaced by Periodic Labour Force Survey, which has been criticised of limited coverage of the unorganised sector which accounts for the bulk of India’s workforce.

What lacunae in the statistical system did COVID-19 expose?

What the pandemic has brought to the fore is the glaring gaps and inconsistencies between different agencies in maintaining and reporting public health statistics. Most of this is due to absence of any standardized system of collecting, compiling and sharing information from patients across hospitals or in the community. This, in turn, follows from long negligence of the need for having appropriate public health surveillance systems that can act as early warning systems for disease outbreaks which is not entirely a rarity in the country.

Similarly, reflecting the general lack of organisation between public and private health care services in the country and no regulatory systems controlling patient flows into different health facilities, we are perhaps the only country of such a geographic and demographic spread that does not have anything close to patient health records, beyond archaic paper-based case-records which are not even standardized within a state, let alone nationally. In simpler terms, this means in times of need such as the COVID-19 pandemic, the government can hardly start with anything more than groping in the dark if it wants to have a reliable estimate of tests, treatment, cure and death statistics of infected patients across the country.

Most high-income countries in Europe, and even middle-income countries such as Brazil have excellent public health information systems that are integrated between local health authorities such as municipalities and city corporations, hospitals or physician practices and with a national health authority.

The curious case of the ‘missing’ dead

A glaring shortcoming of health information in India is the weak system of collecting and reporting mortality and causes of death data. In recent years, there has been significant improvements in registration of deaths, with most cremation or burials requiring death certificates. But the information recorded is not compiled and maintained appropriately mostly due to lack of technical capacity at local levels — an ignominy for a country that prides itself as an information superpower. Even states such as Kerala or Delhi with near-universal registration don’t have the information in standardized databases that one can access, although most registrations use a common proforma provided by the Registrar General of India.

Another weakness is extremely poor quality of cause of death statistics, which follows a globally standardized International Classification of Diseases (ICD-10). With only about one in every five deaths in India being medically-certified, this remains a major deficiency of mortality data. Often, even for deaths in hospitals, medical certification of death is not properly coded, or in ineligible handwriting which means most deaths are recorded as either too generic or unspecified causes. In fact, there have been controversies on several deaths during the current pandemic, with claims and counterclaims between direct and indirect or secondary cause of deaths, or co-morbidities. With robust, reliable mortality statistics considered globally as benchmark indicators of prevalent health conditions in any country, neglecting its development only bodes ill for the entire health information system in the country.

In times of pandemics such as the present Covid-19, strong mortality statistics are not only essential for having a correct assessment of the casualties, but also to provide several other useful, policy-relevant information e.g. whether deaths are concentrated among people from a particular socioeconomic background. Additionally, a reliable system of mortality data are of direct relevance to assess if pandemics such as Covid-19 are responsible for ‘excess deaths’ — additional deaths compared to usual mortality patterns of earlier years — as have been observe din several major cities and countries worldwide. In UK, availability of mortality statistics according to geographic locations and in some cases ethnicity of the deceased have revealed a disproportionate mortality burden among blacks, Asians and other ethnic minorities, and in regions where these minority ethnic groups are predominant. Availability of such granulated statistics are essential inputs to allow public discourses on the whys and hows of such inequalities. In its absence, reliance on government-fed statistics limit the potential for debates, and scope for democratic public action.

Efficient monitoring public health indicators are not limited to indicators related to health alone. Of the closest relevance is the ability to track movements of people between provinces and cities, and as they change schools, employers or grow older. This remains a murky issue in India with major debates around human rights and data protection guarantees, particularly around the use of the biometric identification system of AADHAR and its seeding with documentation relevant to personal identity. Some of these concerns are legitimate and related to questions of citizen trust and credibility of governments, but there are established

Key imperatives for a modern, responsive statistical system …. If lessons from COVID matters

As several studies and observers have noted, a good statistical system is the one which allows robust, reliable information that are readily relevant for policy decision-making. In addition, it should allow researchers and other technical specialists to dig deeper into the data to highlight other dimensions beyond the obvious such as the socioeconomic disparities in disease and death. In democratic societies, legitimacy of governments and the State at large, rests on transparency and accountability — credible statistics on key questions of the day only cements it.

Better financing of the statistical system

While, the immediate public health implications of the COVID-19 pandemic might have been relatively modest in India, frequent reports of confusion, debates and inconsistencies on key statistics have revealed gaping holes and crippling limitations in the statistical system. A major reason has been chronic underfunding of the statistical system, both under the central and state governments. While the government statistical architecture has come a long way from the colonial times, notably with the creation of a separate Ministry of Statistics in 1999 and setting up separate directorates of statistics in almost all the states and even district statistical offices, this have been largely devoid of adequate resources. Naturally, technical skills, capacities and applications of and in these offices has been below par.

India boasts of national statistical civil service cadre — the Indian Statistical Service — one of the few cadres that requires university degrees in specific subjects. However, such talented mass of technical expertise is largely underutilized, with limited mandate for bringing in innovative, modern reforms, largely due to lack of political interest in this cadre. Statistics make noise when its absence is explicit, in normal times poor quality of statistics is good enough to tick the required boxes. The National Statistical Commission established in 2005 to provide the required oversight to reforming the statistical system has remained little more than a formality. Some efforts towards initiating the much-needed reform, and more importantly, a convergence across the myriad statistical streams in the country, has been initiated in recent years, but has largely been inadequate so far.

Hopes from a redesigned National Statistical Commission?

The proposed NSC Bill 2019 talks about a few organisational changes but skirts the core issue of inadequate funding and technical limitations plaguing the system. Also, it proposes little to breathe in much-needed autonomous, independent spirit in the nation’s official statistics which has taken a costly dent in its reputation in recent years. An important first step would be to develop the blueprint for integrating the routine statistical system — surveys like the NFHS and NSS are costly and have limitations — across the key departments such as health, education, employment, access to public utilities. A key to this is effective decentralisation of the statistical system but with a common format of collecting information. Where such common format already exists such as for death registrations, the quality of reported data needs to be significantly improved with training the ground-level officials who are at the first-line of collecting and entering the information.

Statistics has been always and will continue to be political, as it is not the numbers themselves, but the use they are put into that matters. While debating and discussing numbers and their movements over time in the times of the pandemic, it is also the time to push for public awareness and demand on ensuring its credibility and appropriate use keeping up to its potential.

[i] The records included a series on ‘Statement Exhibiting the Moral and Material Progress and Condition of India’ as the official report of the colonial administration to the British Parliament, and had none other than the budding economist of the day, John Maynard Keynes as its editor in 1906–07 in one of his first post-doctoral assignment. He was, however, critical of the quality of the data being reported, including on mortality, and argued for the appointment of a trained statistician to oversee and harmonise the efforts. He was reportedly asked to revise his report! (Ravi Kanbur’s presentation at the IMF Statistical Forum)

[ii] A Department of Commerce and Industry created in 1905 was the first move towards having some identity of the collection and compilation of statistics across different other departments, except Home department which retained the responsibility of conducting the census exercise, and under a Director of Commercial Intelligence with a handful of staff. But appointment of professional statisticians was turned down a number of times until 1914, when Mr Findlay Shirras was appointed as Director of Statistics and had about 100 staff working in 10 sections (See Subramanian)

[iii] Subramanian, S (1960). A Brief History of the Organisation of Official Statistics in India during the British Period, Sankhyā: The Indian Journal of Statistics (1933–1960), Vol. 22, №1/2 (Jan, 1960), pp. 85–118

Views are personal.



Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store
Sumit Mazumdar

Sumit Mazumdar


Health Economics, Centre for Health Economics, University of York Health system financing, Political economy of health reforms Follow: @sumitmazumdar