India's statistical system: institutions, reforms and credibility

Detail

1. Constitutional basis: who may collect statistics

• Union List Entry 94: Parliament can make laws on inquiries, surveys and statistics for Union subjects, such as defence, railways and foreign trade.
• Concurrent List Entry 45: both the Centre and the states can collect statistics for Concurrent subjects, such as education, labour and forests.
• Union List Entry 69: the census is only a Union subject. States cannot run their own census.
• What "decentralised system" means here:
• No single office makes all of India's numbers.
• Each ministry and each state produces statistics for its own work.

• MoSPI coordinates them and sets the standards.

• Why this matters: in a decentralised system, the same thing can be counted in different ways. Common definitions and one central coordinator are therefore very important.

2. The Centre: MoSPI and NSO

• MoSPI (Ministry of Statistics and Programme Implementation) was set up in 1999. It is the nodal ministry (the main ministry in charge) for developing India's statistical system [5].
• NSO (National Statistics Office) was formed in May 2019 by merging two older bodies:
• CSO (Central Statistical Office), set up in 1951. It produced the national accounts (GDP), the IIP (Index of Industrial Production, a monthly index of factory, mining and electricity output) and the CPI (Consumer Price Index, which measures the change in the prices that households pay).

• NSSO (National Sample Survey Office), which ran the NSS (National Sample Survey). The NSS began in 1950 on P.C. Mahalanobis's initiative.

• Sample survey vs census:

• A census counts every unit, for example every person in India.
• A sample survey studies a small, carefully chosen part and uses it to estimate the whole.

• Example: to estimate average household spending, the NSS visits about 1 lakh households, not all 30 crore. It then "blows up" (multiplies up) the sample results to cover all India.

• (NCERT outdated: the Class 11 chapter Collection of Data still lists the CSO and the NSS separately. Both have been part of the NSO since 2019.)

• What the NSO produces:
• GDP.
• IIP.
• CPI (Rural, Urban and Combined).
• ASI (Annual Survey of Industries, a yearly survey of registered factories).
• The NSS surveys, including the PLFS and the household consumption survey.
• The Economic Census, a count of all business establishments.

3. Oversight and law

• NSC (National Statistical Commission):
• Set up in 2005-06 on the advice of the Rangarajan Commission (2001).
• Its job is to act as the watchdog of core statistics.

• It is non-statutory: it was created by a government order, not by an Act of Parliament. So its independence is not protected by law.

• Collection of Statistics Act 2008, the legal backbone:

• It is Act No. 7 of 2009, in force from 7 January 2009 [6].
• It lets central, state and local governments collect statistics on social, economic, demographic, scientific and environmental matters [6].
• It allows statistics officers to be appointed to collect information [6].
• It has rules to keep the information collected secure and confidential [6].

• A 2017 amendment was brought to extend its reach [7].

• ISI (Indian Statistical Institute), Kolkata:

• Founded in 1931 by Mahalanobis.

• Became an Institute of National Importance under the ISI Act 1959.

• Statistics Day is 29 June, Mahalanobis's birthday.

4. Line agencies: who else makes India's numbers

Agency	Parent	Output
RGI (Registrar General of India)	MHA	Census; SRS (Sample Registration System, which gives birth and death rates); CRS (Civil Registration System); NPR (National Population Register)
DGCI&S (Kolkata, 1862)	Commerce	Foreign trade data (the source of NCERT's trade table)
Labour Bureau (1946, Shimla/Chandigarh)	Labour & Employment	CPI-IW (for industrial workers), CPI-AL (agricultural labourers), CPI-RL (rural labourers); establishment surveys
DES (Directorate of Economics & Statistics)	MoA&FW	Crop estimates; Agricultural Statistics at a Glance
Office of the Economic Adviser	DPIIT	WPI (Wholesale Price Index)
RBI	—	Handbook of Statistics; DBIE (its online database); forward-looking surveys, such as consumer confidence and inflation expectations
State DESs	States	State income (GSDP) and district data

• Trap: WPI is made by DPIIT, not the NSO. CPI-IW is made by the Labour Bureau, not the NSO.

5. Secondary-data publications

• Primary vs secondary data:
• Primary data are collected first-hand by the investigator.

• Secondary data are taken from someone else's published source.

• Sarvekshana: the NSS journal. NCERT calls it quarterly.

• The Statistical Year Book and Women and Men in India (gender-wise data) are both MoSPI publications.
• Economic Survey:
• Prepared by the DEA (Department of Economic Affairs), Ministry of Finance, under the Chief Economic Adviser.
• Tabled in Parliament just before the Union Budget.
• The first one was for 1950-51. It was delinked from the Budget in 1964, so it became a separate document.
• Its statistical appendix is a standard source of secondary data.
• Its policy role is covered in → economic-problem-systems.

6. Credibility flashpoints

(a) GDP series controversy

• Base year is the reference year against which quantities and prices are compared.

• Updating it brings in the new products, new industries and new price patterns of the economy.

• The 2015 series:

• Base year 2011-12.
• It used MCA-21 data, the Ministry of Corporate Affairs' online filings of company accounts, to measure the corporate sector.

• Critics said the sample of companies and the way the results were blown up overstated growth.

• 2018 back-series row:

• Two different back-series (old years recalculated on the new base) gave very different growth for the UPA years.
• One was prepared by an NSC committee. The other was released officially by NITI Aayog and CSO.

• The dispute raised questions about political influence.

• Overestimation claim: Arvind Subramanian (2019), a former CEA, argued that GDP growth for 2011-12 to 2016-17 was overstated.

(b) PLFS leak and NSC resignations

• PLFS (Periodic Labour Force Survey) is the NSO's main survey of jobs and unemployment. It started in 2017-18.
• The 2017-18 report was held back, and then it leaked.
• Two NSC members resigned in January 2019. They said the government had kept the Commission out of the decision.

(c) Junked CES 2017-18

• CES (Consumption Expenditure Survey) measures what households spend. Poverty lines and the CPI weights both depend on it.
• The government scrapped the 2017-18 results, citing "data quality" issues.
• Effect: India had no official poverty estimate for about a decade, from 2011-12 until the HCES 2022-23.

(d) Census delay

• The 2021 Census was postponed, first because of COVID and then again.
• What goes wrong:
• Frames go stale. A sampling frame is the list from which a survey sample is drawn. Surveys like the NSS use Census 2011 lists that are now out of date.
• Welfare coverage is outdated. For example, NFSA ration coverage is still worked out from 2011 population figures.
• Delimitation is stalled. This is the redrawing of constituency boundaries, which is tied to the first census after 2026.

(e) IMF grade

• The IMF Executive Board concluded India's 2025 Article IV consultation (the IMF's yearly health check of a member economy) in November 2025 [8].
• In that review, the IMF graded India's national accounts "C" (scaffold figure; verify current).
• A "C" grade means the data have gaps that somewhat hamper the IMF's monitoring. The weak points cited include the old base year and weak deflators (price indices used to turn nominal values into real values).
• The IMF gave India an overall data grade of "B".

7. Reforms and responses

(a) Institutional demands

• Give the NSC statutory backing through a law, so its independence cannot be taken away easily.
• Publish advance release calendars, so the dates for every data release are announced beforehand. This stops anyone from timing releases for political reasons. MoSPI now publishes one [9].
• Standing Committee on Statistics (2023):
• Chaired by Pronab Sen, to review survey design.
• Later restructured (verify current).

(b) eSankhyiki portal (2024)

• Launched on Statistics Day, 29 June 2024, by Arvind Panagariya, Chairman of the 16th Finance Commission [5].
• Its aim is a single data-management and sharing system, so that official statistics are easy to find and download [5].
• It has since been upgraded with AI features and an AI chatbot to help users find data [10].

(c) New base years (2026 rollout)

• Release schedule [2]:
• CPI (base 2024): 12 February 2026.
• GDP (base 2022-23): 27 February 2026.

• IIP (base 2022-23): May 2026.

• GDP new series [3]:

• Nominal GDP in the base year 2022-23 is ₹261.18 lakh crore.
• Real GDP was ₹280.01 lakh crore (2023-24) and ₹299.89 lakh crore (2024-25).

• This means growth of 7.2% (2023-24) and 7.1% (2024-25).

• Worked example: growth rate from real GDP

• Formula: Growth rate = (GDPₜ − GDPₜ₋₁) ÷ GDPₜ₋₁ × 100
• = (299.89 − 280.01) ÷ 280.01 × 100

• = 19.88 ÷ 280.01 × 100 ≈ 7.1%

• Method changes in the new GDP series [3][4]:

• Double deflation is now used for manufacturing. Output and inputs are deflated separately, each with its own price index, which gives a truer measure of real growth. The old series used single deflation, where only output was deflated.
• Deflators (CPI, WPI, unit value index) are now applied at the item-group level, not at the aggregate level.

• New data sources and the latest survey data have been added, and weights have been updated.

• Why the base year matters:

• Take a price index built on an old base year, such as the Laspeyres index. This index uses base-year quantities as its weights: Index = Σp₁q₀ ÷ Σp₀q₀ × 100.
• Old weights keep counting goods people no longer buy, such as landline phones. They leave out new goods such as mobile data.
• So the index can misread both inflation and growth.

(d) Monthly PLFS

• The sampling methodology of PLFS was revamped from January 2025 [11].
• It now gives monthly estimates of the LFPR, WPR and UR for rural and urban areas, in CWS, at the all-India level [11].
• LFPR (Labour Force Participation Rate) is the share of people who are working or looking for work.
• WPR (Worker Population Ratio) is the share of people who are working.
• UR (Unemployment Rate) is the share of the labour force without work.

• CWS (Current Weekly Status) measures a person's activity during the last 7 days.

• The first monthly bulletin (for April 2025) was released in May 2025 [11].

• UR (CWS, age 15+) [12][13]:
• 5.1% (April 2025)
• 5.6% (May 2025)
• 5.6% (June 2025)

• 5.2% (July 2025)

• Formula: UR = Unemployed ÷ Labour force × 100.

• Example: 52 unemployed out of a labour force of 1,000 gives UR = 5.2%.

(e) Administrative and big data

• Administrative data are records that governments create while running schemes and collecting taxes, not through surveys. Examples:
• GSTN (GST returns), which tracks business turnover.
• EPFO (payroll additions), which tracks formal jobs.
• UPI (digital payments), which tracks spending.

• Satellite imagery, used for crop area and night-lights.

• Benefits: the data come quickly, cover large numbers, and cost little to collect.

• Risks:
• They cover only the formal economy, not the informal sector.

• They can reveal private details about people.

• Privacy is governed by the DPDP Act 2023 (Digital Personal Data Protection Act).

8. The thread of this note

• Every number carries its design choices:
• Frame: which list the sample was drawn from.
• Sample: how many units, and how they were chosen.
• Recall period: how far back respondents are asked to remember, for example 7 days or 30 days.

• Response: who refused to answer, and how the missing answers were handled.

• Credibility needs these choices to be transparent and the institutions to be independent.

• The Economic Survey 2018-19 described this as treating "data as a public good", meaning data made by the state should be open for everyone's benefit.

Prelims Hooks

• Census is in Union List Entry 69. Statistics are in Union List Entry 94 and Concurrent List Entry 45.
• NSO was formed in May 2019 by merging the CSO (1951) and the NSSO. The NSS began in 1950 on Mahalanobis's initiative.
• The NSC (2005-06, Rangarajan Commission 2001) is non-statutory. Trap: it was not created by the Collection of Statistics Act 2008.
• The Collection of Statistics Act 2008 is Act No. 7 of 2009 [6].
• WPI comes from the Office of the Economic Adviser, DPIIT. CPI-IW comes from the Labour Bureau. Foreign trade data come from DGCI&S, Kolkata (1862).
• Statistics Day is 29 June. The eSankhyiki portal was launched on Statistics Day 2024 [5].
• New base years: GDP and IIP are 2022-23. CPI is 2024 [2].
• The new GDP series uses double deflation for manufacturing [3].
• The PLFS has given monthly CWS estimates since the January 2025 revamp [11].
• The Economic Survey is prepared by the DEA under the CEA. The first was for 1950-51, and it was delinked from the Budget in 1964.

Mains Points

• Independence vs control: the PLFS leak, the junked CES and the NSC resignations all show that a non-statutory NSC cannot protect data from political pressure.

• Reform: a law-backed NSC, a fixed advance release calendar [9], and publishing all survey results, including unwelcome ones.

• Stale frames weaken policy:

• With no census since 2011, sample frames, NFSA coverage and delimitation all rest on old numbers.

• The 2026 base-year revision [2][3] fixes the weights, but not the population frame.

• Administrative and big data are a trade-off:

• GSTN, EPFO and UPI data are fast and cheap, but they miss the informal sector and raise privacy issues under the DPDP Act 2023.

• Best use: to complement surveys, not replace them.

• Credibility is economic capital:

• Doubtful GDP numbers (the 2019 overestimation debate, the IMF's "C" grade [8]) raise the risk premium that investors demand.
• They also weaken policymaking. The RBI, the Finance Commission and welfare targeting all depend on trusted data, which is the idea of "data as a public good".