NSC junked idea to calibrate NSSO's labour survey with population numbers

The National Statistical Commission (NSC) did not favour the idea of calibrating the estimates of the National Sample Survey Office’s (NSSO’s) first periodic labour force survey (PLFS) with the official population projection of the country.

The issue of linking the population estimates with the size of the workforce, the unemployed, among others, in the NSSO's survey report of 2017-18 was discussed in the NSC's last meeting on December 5.

The NSC had approved the NSSO’s report, which showed unemployment rate at a 45-year high of 6.1 per cent, while rejecting the idea of calibrating the estimates in the report with population projections based on Census figures. The NSC felt that in a way this exercise would have required a back-series on the NSSO’s jobs data so that the figures are comparable with the past estimates produced by the NSSO in its quinquennial reports.

“There were two issues: One, the data would have become incomparable with the past estimates of the NSSO on employment and unemployment. Second, a similar exercise on the past reports of the NSSO since 1972-73 could have been done but that would have delayed the process,” former NSC chairman P C Mohanan, who had chaired the December 5 meeting, said. Chief statistician Pravin Srivastava was also a part of this meeting, along with NSSO officials.

Mohanan, along with a member of the NSC J V Meenakshi, had resigned in January, expressing their disappointment over a delay in releasing the NSSO’s PLFS report and the functioning of the commission.

The Ministry of Statistics and Programme Implementation (MoSPI) had given a project, known as Structured Preserving Estimation Procedures (or SPREE), to Kolkata-based Indian Statistical Institute (ISI) for coming up with a methodology to calibrate the NSSO estimates with the population projections based on the Census numbers, last released for the year 2011.

The ISI had given a presentation to the NSC in its December meeting and had said the changes in the final estimate after the calibration "would be cosmetic", sources said. The idea presented was to use the NSSO rates for a sample of households and adjust the blowing up according to the NSSO design.

Mohanan confirmed that a decision was taken to not attempt calibration in the NSSO’s PLFS report for 2017-18 as “it will create confusion among researchers while using the unit-level data and the figures will not be comparable with the past rounds.” The NSC, however, suggested that a separate report could be brought out with calibrated figures.

The issue here is that the NSSO reports over the years have been underestimating the country’s population when compared to the population projections made through the Census data. The difference essentially comes because the NSSO estimates the population based on its sample size of 100,000-odd households. “This is true for a long time when it comes to the NSSO surveys. But the surveys are not designed to measure population as it derives the ratios. The estimates are based on a sample and when you calibrate them with the actual or projected population numbers, you are fiddling with the unit-level data. That is unacceptable,” former chief statistician Pronab Sen said.

Sources said according to the Census data, the projected population for 2017 stood at 1.3 billion. The NSSO’s estimate of population stood at 1.07 billion — around 20 per cent lower than the Census’s projected population, according to the source.

The NSSO’s 2011-12 report had underestimated the rural population by 5 per cent and the urban population by 18 per cent.

The NSSO report has been withheld for release by the government, despite the approval of the NSC and a standing committee in the first week of December. The government has termed it a “draft” report. A copy of the NSSO report reviewed by Business Standard showed unemployment rate at a 45-year high of 6.1 per cent.

Data deviation

• NSC discussed calibrating estimates of the NSSO’s periodic labour force survey 2017-18 with projected population of the Census
• It rejected the idea in its December meet as it felt the move may fiddle with the unit level data and create confusion among researchers
• The commission felt a separate report may be released with calibrated figures
• NSSO’s estimates of the population are always lower than the Census figures as the former's calculation is based on a sample