NSS at 75: Strengthening statistical systems for Viksit Bharat by 2047

The strong connection with academia that existed in the first 20 years needs to be revitalised to support effective policymaking in India's journey towards a Viksit Bharat

This month marks the beginning of the 75th year of the National Sample Survey (NSS). It is an opportune moment to not only acknowledge those who contributed to development of survey design and sampling methods but also reflect on the way forward.

The first survey conducted by the NSS 74 years ago, from October 1950 to March 1951, was a multi-topic survey covering only rural India. It was only in the third round, from August to November 1951, that the survey was extended to cover urban India.  The NSS General Report No 1 mentions that the need to conduct surveys was recognised by both the Standing Committee of Departmental Statistics and the National Income Committee, constituted in 1948 and 1949, respectively. The immediate objective was to fill data gaps for estimating national income. As many as 10 rounds of surveys were conducted between October 1950 and May 1956.

Since the NSS operated from the Indian Statistical Institute, Calcutta, its link with academia was strong. Later, in March 1970, National Sample Survey Organisation (NSSO) came into existence as a unit in the Department of Statistics.

In the initial years, there were many protagonists. P C Mahalanobis needs no introduction. V K R V Rao, who worked extensively on national income accounts, was also the founder of the Delhi School of Economics. D R Gadgil was the founder-director of the Gokhale Institute of Politics and Economics, Pune, an institute that had a long tradition of conducting surveys even before Independence. Anyone who has taken a course in statistics will have learned about C R Rao through the Rao-Blackwell theorem or the Cramer-Rao bound. But who was Debabrata Lahiri? He was in charge of the sampling design of NSS and his contribution is acknowledged by Mahalanobis in the introduction to the 5th NSS report. Lahiri is credited with working out how to select villages from where the households will be surveyed. He suggested a method for systematically selecting villages, whereby those with larger populations had a higher likelihood of being included in the sample.

Administratively, important changes came to effect in March 1970 when NSSO came into existence as a unit in the Department of Statistics. A governing council of the NSSO was formed, chaired by  V M Dandekar. It recommended a 10-year survey cycle, within which major surveys would be conducted. For example, surveys of employment-unemployment and consumer expenditure, and self-employment in the non-agricultural sector would be conducted every five years, while others, such as debt and investment surveys, would be conducted every 10 years. The core surveys would take seven years and the remaining three years were kept free to accommodate requests from the central and state governments.

The periodicity has changed in recent times. Since 2017-18, the NSSO has been conducting the Periodic Labour Force Survey annually, and from 2021-22, the Survey of Unincorporated Sector Enterprises is also being conducted annually. In its 75th year, for the first time, a forward-looking survey on private corporate sector capital expenditure (Capex) investment intentions is being launched, along with a pilot study of an annual survey of service sector enterprises. All surveys are conducted under the supervision of the National Statistical Commission. The methods and procedures followed in surveys comply with the guidelines and best practices suggested in the manuals of the United Nations Statistics Division.     

While the NSSO conducts the all-India survey, any state government can also undertake the same survey. The former is known as the central sample and the latter the state sample. State governments participated in a survey for the first time in 1954, specifically in the 8th round, which included a survey on land holdings as part of the World Agricultural Census. The participation of states in data collection ensured an adequate sample size for generating state-level estimates.

Today, there is a growing recognition of the need for district-level estimates. In August 2024, at the 28th Conference of Central and State Statistical Organisations, state governments expressed interest to participate in surveys of NSSO.

However, in order to generate district-level estimates, the number of households surveyed in each district needs to be higher than what it is currently. Only a reasonably large sample will ensure that the estimates of the district-level indicators of interest are measured with reasonable precision. There are two options: Either the central sample must be increased, or the data from the central and state samples must be combined to generate district-level estimates.

Ideally, the NSSO needs an updated master list of geographical units and population of each unit. Till the next census is conducted, the source for this will continue to be the Census of India 2011. In the absence of up-to-date information on district or state-level populations, states will need to rely on the 2011 figures to derive estimated numbers of workers and the unemployed, by multiplying the population figures by the estimates of labour force participation rate and unemployment rate, respectively.

However, effective communication of survey results in a non-technical manner to a wide range of audiences also requires careful consideration. A beginning has been made with the conduct of data user conferences following the release of survey results. Recent criticisms of the NSS can be turned into actionable points by having a constructive conversation. The strong connection with academia that existed in the first 20 years needs to be revitalised. It is in everyone’s interest to invest in and strengthen a statistical system that will provide evidence to support effective policymaking in India’s journey towards Viksit Bharat by 2047.

The writers are, respectively, professor, Indira Gandhi Institute of Development Research, Mumbai, and deputy director general, survey design & research division, National Sample Survey Office, Kolkata. The views are personal