The gross domestic product (GDP) of a country reflects the size of its economy. It is its most important economic statistic, acting as a universal denominator for macro-economic ratios.[1] In this piece, we explain how Indian statisticians calculate the country's GDP, the methodological debates around them, and the revisions made in 2026.
The two sides of the macroeconomic ledger
There are two main ways to arrive at a country's GDP figures. One is to add up all that is produced across farms and factories. This is known as the production estimate. To eliminate double counting, statisticians record only the value added at each stage of production. So, in estimating the produce of a sugar mill, the raw material (sugarcanes) it purchases will be excluded from the calculations. Only the value of sugar produced, or the 'value added' by the mill, will be considered.
The total value added across the economy is known as the country's GVA, or gross value added, and is expressed in 'basic prices'.[2] This figure reflects the contribution of labour and capital in the production process, and excludes net taxes on products[3] appropriated by the government. Once net taxes on products are added to the GVA figure, we obtain the country's GDP 'at market prices'.[4]
The other way to arrive at the GDP estimate is to add up all 'final' spending of individuals, firms and the government in the country. This is known as the expenditure estimate. As in the case of the production approach, intermediate consumption (such as purchases of raw materials by firms) is excluded from the calculations to prevent double counting.
Just as a company's debit account should tally with its credit account, a country's production account should tally with its expenditure account.[5] In other words, the GDP estimate from the production side should either match or be close to the GDP estimate from the expenditure side.[6] The two sets of estimates (and their components) constitute the country's national accounts statistics (NAS).[7] These statistics are compiled in accordance with the United Nations System of National Accounts (SNA) manuals.[8]
Base-year and base-revision
To assess macroeconomic changes across time, it is essential to use the same measuring rod. So, the measurement methods and data sources used in computing national accounts statistics are fixed for a few years. The reference year for the start of a new national accounts series is called the base year.
All growth metrics are computed with reference to the base year. This allows for comparison in inflation-adjusted terms, with price changes since the base year stripped out of the current price estimates to arrive at the constant price estimates.The constant price estimates depict what the GDP figures would have been in a world of zero inflation.[9]
The base year needs to be periodically revised to ensure that changes in the economic structure of a country are reflected in the measurement framework.[10] During the base year revision exercise, statisticians incorporate new data sources,[11] update measurement methods, and construct a back series that allows for comparison with past national accounts data pegged to older base years. They also try to ensure that the methods align with the latest SNA guidelines.
Globally, base-year revisions tend to increase the size of a country's GDP,[12] as new economic activities get incorporated in the national accounts. This was true of India earlier. But in the last two revisions, the size of the Indian economy shrank when the GDP estimates were recomputed using new sources and methods.
After the base revision exercise conducted in 2026, the GDP estimates for the new base year, 2022-23, fell by Rs 7.7 trillion (as compared to the earlier estimates for the same year) to Rs 261 trillion. This translates to a decline of about 3%.
The big swings
The non-financial private corporate sector[13] accounted for most of the decline in the new GDP estimates. This sector shrank by nearly Rs 6.7 trillion in the new series to Rs 80 trillion, with its share in the GVA declining 1.5 percentage points to 33.9%.[14] Improvements in the source data for estimating corporate GVA and methodological refinements (see methods section below) led to this decline.
The estimated GVA of public corporations[15] has gone up marginally in the new series. The estimated GVA of the unorganised sector[16] and that of the government sector[17] have declined marginally. But their share in overall GVA has gone up since the denominator (overall GVA) is smaller than before.
The national accounts also provide the industrial split of GVA. This shows that the GVA share of the primary sector has gone up significantly, driven by agriculture and allied activities. The share of the secondary sector has increased marginally in the new series, driven by a small increase in the manufacturing sector's size. The share of the tertiary sector has fallen sharply, owing to a large contraction in the size of the trade, communication, hotels and restaurants sector.
Expenditure side estimates show that consumption spending has seen a sharp downward revision in the new series. Private final consumption expenditure estimates for 2022-23 declined 4.3% to Rs 149 trillion. Estimates of gross fixed capital formation, or investments, rose 1% to Rs 84.5 trillion. Other expenditure heads did not see major changes.
New data, new methods
Historically, the GDP estimates for the unorganised sector have been the weakest component of the national accounts series.[18] Since the mid-90s, unorganised sector estimates for the base year were based on workforce estimates from household surveys and productivity (GVA/worker) estimates from enterprise surveys conducted by the National Statistics Office (NSO). These surveys were conducted once in five years, typically close to the base year. Estimates for subsequent years were projected using past trends or proxy indicators.[19] In the new series, unorganised sector estimates are based on current data obtained from annual surveys.[20]
The second big change in the new series relates to the private corporate sector. Estimates for this sector have been a subject of controversy since the 90s, given that the estimates for the sector were based on a small sample of companies.[21] In the last base-revision exercise, conducted in 2015, a new corporate database, the MCA-21, was incorporated in the national accounts.[22] The manner in which the new database was plugged into the national accounts faced intense scrutiny after the base revision exercise.[23] Partly in response to those criticisms, the methodology has undergone significant changes in 2026,[24] leading to a correction in the corporate sector estimates.
The third big change in the new series has to do with the use of supply-use tables to reconcile the production side estimates against the expenditure side estimates, in line with SNA guidelines.[25] Supply-use tables show how much of each product is used in different industries as intermediate inputs, and how much is consumed by the end consumer. They can be used to eliminate discrepancies between the two sides of the macroeconomic ledger.[26] In India, the reconciliation framework has been applied only for the current price estimates so far.[27]
The fourth major change relates to the constant price estimates. To strip out the impact of inflation, the earlier series largely relied on a 'single-deflation' strategy in which the GVA of each sub-sector was deflated using a relevant output price index. The implicit assumption here was that input and output prices move in tandem.
In reality, trends in input prices often diverge from output prices. Hence, the preferred approach in the new series is to deflate gross value of output (GVO) using output price data, and use deflated GVO growth rates to project GVA growth. This 'volume extrapolation' strategy implicitly assumes that the GVO-GVA ratio remains unchanged across time.[28]
For several manufacturing sub-sectors in the new series, the 'double-deflation' strategy has been used, in which input costs and output prices are deflated separately (using respective price indices) to arrive at constant price GVA estimates.[29] The double-deflation strategy is the most preferred approach globally, with most advanced economies such as the USA and UK using this approach to estimate constant price estimates. Historically, the double-deflation approach was used only for agricultural GDP estimates in India, because of the lack of reliable input price indices for other sectors.[30] In the new series, the double-deflation approach has been extended to large parts of the manufacturing sector. It is likely to be extended to other sectors of the economy once the experimental input producer price index (PPI) series is deemed reliable.[31]
Quarterly estimates
In almost all countries, detailed data on production (such as company balance sheets) are available only on an annual basis, and often with a lag. So national accountants tend to rely on quick surveys and rough proxies to generate quarterly national accounts estimates. Quarter-wise estimates for the year-ago period are first generated, and these are moved forward to the current year using growth rates observed in high-frequency indicators such as tax, vehicle registration, and air traffic data.
The accuracy of quarterly estimates tends to be lower than those of annual estimates since these estimates are generated from a limited set of data. Quarterly estimates are aligned with the annual estimates through a process of benchmarking. In the new series, the process of benchmarking has been revised to bring them in line with international guidelines published by the International Monetary Fund (IMF).[32]
The new series also incorporates new data sources to improve the accuracy of the quarterly estimates. For most service industries, Goods and Services Tax (GST) data has been used to estimate growth figures. GST data has also been used to project growth for the unorganised manufacturing sector, replacing the index of industrial production (IIP) being used earlier. For the organised manufacturing sector, quarterly estimates continue to rely on earnings of listed firms, as in the old series. Earnings growth of listed firms are also being used for the IT and telecom sectors.[33]
The road ahead
While the National Statistics Office (NSO) has published discussion reports on the new methodology, and the reports of the sub-committees constituted by the Advisory Committee on National Accounts Statistics (ACNAS)[34] to guide the preparation of the new series, the detailed methodology document has not been released as of July 2026. It is scheduled to be released in August 2026, and the back series data is expected to be released later in 2026.
Given the increasing use of administrative databases in the national accounts series, there have been growing calls for opening up these datasets. Releasing the unit level records of these databases would bolster the credibility of the GDP estimates,[35] and help data users understand economic changes better.
[1] The investment rate, the fiscal deficit (difference between government spending and revenues), and the current account deficit (difference between import and export bills) are all expressed as a percentage of GDP. The reliability of these metrics depends on the accuracy of the GDP figures.
[2] The UN SNA (System of National Accounts) 2008 manual defines basic price as the amount a producer receives for selling a unit of good or service, excluding any tax payable and including any government subsidy linked to its production or sale.
[3] Net product taxes refer to taxes producers pass on to the government minus the subsidies they receive. These taxes (such as GST) are levied on the value or volume of produce being sold, and are excluded while calculating GVA. Some minor taxes such as stamp duties and professional taxes are included in GVA. These taxes are called 'production taxes' and are levied on all eligible producers, regardless of how much output they produce.
[4] The SNA defines market prices as 'amounts of money that willing buyers pay to acquire something from willing sellers'. It includes net taxes paid on the product.
[5] The two estimates should ideally match the income side estimates, which involves adding up the total incomes of all the people in the country from all possible sources (rent, wages, interest and profits). Developed economies such as the UK and the US use their detailed tax records to compute income side GDP estimates. In developing economies such as India, tax records are available for only a small minority of the population. Given the absence of credible income data for the rest of the population, income side estimates are not prepared in India.
[6] Differences between the production and expenditure side estimates are denoted as statistical discrepancies in the national accounts tables published by the National Statistics Office (NSO).
[7] These include tables on major expenditure heads (private expenditure, government expenditure, investments by public corporations, investments by private companies, etc.), GVA contribution of different industrial sectors (agriculture, mining, manufacturing, etc.), GVA contribution of different institutional sectors (private corporate sector, unorganised sector, government sector, etc.), composition of private consumption expenditure (showing breakup across individual items of consumption), and composition of capital formation (by industrial use).
[8] The UN SNA (System of National Accounts) manual was first prepared in 1953. It was revised in 1968, 1993, 2008, and 2025. The 2025 version is yet to be implemented, and most countries (including India) follow the 2008 SNA guidelines.
[9] Changes in current price estimates partly reflect changes in volume of output and partly reflect change in prices. Year-on-year percentage changes in current price estimates are referred to as the nominal growth rates. Changes in constant price estimates reflect just the changes in volume of (value-added in) output, and are used to derive the 'real' growth rates.
[10] "One consequence of preparing national accounts on a continuing basis over a number of years is that data sources change and improve," the 2008 SNA said. "Intermittent sources, such as a survey held only every five years, may become available and indicate that the earlier assumptions based on projecting the previous survey were flawed. In such a case it is not sufficient to simply replace the data for the most recent period (or even from the date of the new survey forward) but to ensure that the whole time series is suitably adjusted in order to portray the best possible evolution of the series in question over as long a period as possible. Failure to do so results in inappropriate discontinuities in the series that can be seriously misleading to analysts unaware that the source of the underlying data has changed." While the 2008 SNA does not prescribe any fixed time frame for base year revisions, the earlier manual (SNA 1993) recommended that constant price estimates should be rebased every five years.
[11] In most countries, statisticians tend to rely on rates and ratios to compute GVA estimates for several sub-sectors of the economy. These ratios are derived from benchmark surveys, and these surveys typically take place close to the base-year. For instance, a 2019-20 study on estimates of feed and fodder quantity for different categories of livestock was used to update the animal feed consumption rates for the livestock sector in the 2022-23 series. The 2011-12 series had used the results of a study conducted in 2010. To estimate the unorganised construction sector GVA, ratios (on use of inputs; and output-GVA ratio) from a 2025 NSS pilot survey on the construction sector have been deployed in the new series.
[12] Among 78 economies that saw revisions in recent years, the median economy expanded 3.5% after the base revision exercise, a 2022 IMF working paper showed.
[13] The non-financial private corporate sector includes all privately-owned companies excluding those belonging to the financial sector (banks, insurance firms etc.).
[14] In contrast, in the earlier base-revision exercise, the size of the private (non-financial) corporate sector had expanded by nearly 10% while the GVA of the unorganised (household) sector fell by 22% in the base year 2011-12 (as compared to the earlier estimates for the same year). In the latest base-revision, the size of the unorganised sector has fallen marginally in 2022-23 (as compared to the earlier estimates for the same year). Within the unorganised sector, the agricultural sector has expanded while the size of the non-agricultural unorganised sector has shrunk, leaving the overall numbers for the unorganised sector broadly unchanged.
[15] Public corporations include state-owned companies (public sector enterprises run by the Union or state governments) and departmental enterprises such as the railways.
[16] All enterprises that do not furnish annual or quarterly returns to regulatory authorities are classified as unorganised enterprises by national accountants in India. The unorganised sector, also referred to as the 'household sector' in national accounts tables, includes unincorporated enterprises and most of the agricultural sector. Although there is an overlap between the unorganised sector and the 'informal sector', they are not identical concepts. Unincorporated enterprises refer to enterprises not registered with the Companies Act, 1956 or Companies Act, 2013 or Limited Liability Partnership (LLP) Act, 2008.
[17] The general government sector includes Union, state, local governments, and autonomous bodies. It excludes state-owned firms.
[18] Among unorganised enterprises, annual estimates have been available only for the agricultural sector (major crops) in the past. Since the launch of the Annual Survey of Unincorporated Sector Enterprises (ASUSE) in 2021-22, output estimates are now available for the rest of the unorganised sector.
[19] The exact method used to estimate GVA differed across sub-sectors. For some sub-sectors, inter-survey growth rates were used to project benchmark output estimates. In others, organised sector proxies were used. In many sectors, population projections were used to project workforce estimates.
[20] The Annual Survey of Unincorporated Sector Enterprises (ASUSE) and the Periodic Labour Force Survey (PLFS) are being used to estimate unorganised sector GVA in the 2022-23 series. While ASUSE data is being used to estimate worker productivity (GVA per worker), PLFS data is being used to estimate worker counts (labour input).
In the 2011-12 series, worker productivity was calculated separately for different categories of workers (owners, hired workers, unpaid helpers) in the base year. Only larger enterprises (employing six or more workers) were considered while preparing the benchmark (base year) estimates for the urban sector. For subsequent years (post 2011-12), proxy indicators were used to project growth.
In the 2022-23 series, worker productivity is calculated on an aggregate basis for all workers. All enterprises with hired workers are now included in the calculations for both the rural and urban sectors. Own account enterprises (that do not employ any hired worker) are excluded from the calculations. Enterprises with negative GVA are also excluded. Instead of proxy indicators, current data from surveys are used to compute estimates for each year in the new series. For sectors in which survey estimates are volatile, a three-year moving average of productivity figures may be used to compute GVA, said the sub-committee on methodological improvements constituted by the Advisory Committee on National Accounts Statistics (ACNAS), Ministry of Statistics and Programme Implementation (MoSPI).
[21] The 2001 report of the National Statistical Commission (headed by C. Rangarajan) pointed out that the corporate sector estimates were based on a small and unrepresentative sample of companies included in the RBI studies on company finances. "In the absence of a reliable population frame, the RBI is not in a position to apply suitable sampling techniques," the report said. "Further, the RBI is also constrained by the poor response from companies and non-receipt of annual reports directly from the ROCs (Registrar of Companies) . . . The reliability of the estimates of gross savings and investment in the private corporate sector arrived at by blowing up the sample results available from the RBI's studies in proportion to the coverage of the paid-up capital (PUC) of the sample companies to the PUC of all companies has been questioned time and again." Paid-up capital refers to the amount of money raised by a company from its shareholders in exchange for equity shares.
[22] MCA-21 is a digitised database of regulatory filings maintained by the Ministry of Corporate Affairs (MCA). Every registered company is expected to file financial returns in a format prescribed by the ministry.
[23] For methodological criticisms about the corporate sector estimates, see Seeds of Doubts on New GDP Numbers (2015), R Nagaraj, Economic and Political Weekly; An Investigation into Some Contentious Issues of GDP Estimation (2017), G.C. Manna, Journal of Indian School of Political Economy; Some areas of concern about Indian manufacturing sector GDP estimation (2018), Amey Sapre and Pramod Sinha NIPFP Policy Brief No. 36; Report of the Committee on Real Sector Statistics (2018), National Statistical Commission.
[24] The manner in which corporate data is adjusted to account for non-reporting firms has changed significantly in the new series. In the earlier series, an economy-wide Paid-Up Capital (PUC) based multiplier (or blow-up factor) was used to scale up the estimates of reporting companies, to account for the contribution of 'active companies' that had not filed their returns (but for which PUC data was available). In the new series, disaggregated PUC-based multipliers have been applied separately to different size classes of firms across each industry group, based on the recommendations of the sub-committee on methodological improvements constituted by the Advisory Committee on National Accounts Statistics (ACNAS), Ministry of Statistics and Programme Implementation (MoSPI).
[25] "Only supply and use tables provide a sufficiently rigorous framework to eliminate discrepancies . . . to ensure the alternative measures of GDP converge to the same value," the 2008 SNA manual said.
[26] The SNA manual lays out two possible approaches to deal with discrepancies. One is to openly acknowledge the gap between the production and expenditure estimates of GDP, and ascribe the discrepancy to the less reliable of the two estimates. This was the approach followed in India prior to the launch of the 2022-23 series, when discrepancies were included as a component of the expenditure side estimates. The production side estimates were deemed to be the more reliable of the two estimates. The second approach is to "remove the discrepancy by examining the data in the light of the many accounting constraints in the SNA, making the best judgement possible about where the errors are likely to have arisen and modifying the data accordingly". The SNA recommends the use of the supply and use tables to achieve this end. This latter approach has been adopted in the 2022-23 GDP series.
[27] The SNA recommends that supply and use tables should be constructed at both current and constant prices (or volumes) and balanced (or reconciled) simultaneously. However, this requires detailed item-wise input and output price data which are still unavailable for many sectors in India. Hence, the constant price GDP estimates on the production side do not match the expenditure side estimates, with discrepancies still showing up as a part of the expenditure components.
[28] "Volume extrapolation implicitly assumes that the input-output ratio (in volume terms) is constant in the short term, which is often a more reasonable assumption than single deflation's implicit assumption that the price of inputs moves in sync with the price of output," according to the NSO discussion paper on compilation of aggregates of national accounts (production side estimates).
[29] In manufacturing sub-sectors without adequate price indices, or substantial imported inputs, the volume extrapolation strategy has been used. Industries where this approach has been applied include manufacture of electronics, computer equipment, communications equipment, chemicals and pharmaceuticals.
[30] "While the double deflation method is theoretically sound, the resulting estimates are subject to the errors of measurement in the volume estimates of both output and intermediate consumption," the 2008 SNA manual said. "…Because value added is the relatively small difference between two much larger figures, it is extremely sensitive to error." When reliable price or volume metrics are unavailable, the SNA recommends extrapolation of value added in proportion to the volume changes in the corresponding levels of output (volume extrapolation approach).
[31] The Office of Economic Advisor, Department for Promotion of Industry and Internal Trade (DPITT) released output producer price indices and trial input producer price indices in June 2026. While the output PPI indices are likely to replace the wholesale price index (WPI) in the GDP calculations soon, it may be a while before the input PPIs are used in these calculations.
[32] In the earlier (2011-12) series, benchmark quarterly estimates were prepared using a simple pro-rata distribution of annual estimates across four quarters. This method introduces artificial discontinuities in the quarterly series across years (called the step problem). To avoid the step problem, IMF's 2017 manual on quarterly estimates recommends alternate methods such as the proportional Denton method or the Cholette-Dagum method, which avoid discontinuities across successive years. In the 2022-23 series, the proportional Denton method has been used for benchmark estimates, based on the recommendations of the sub-committee on methodological improvements constituted by the Advisory Committee on National Accounts Statistics (ACNAS), ministry of statistics and programme implementation (MoSPI).
[33] The GVA share of listed firms in the manufacturing, IT and telecom sectors was deemed adequate by the sub-committee on methodological improvements constituted by the Advisory Committee on National Accounts Statistics (ACNAS), ministry of statistics and programme implementation (MoSPI). In other sectors, listed firms account for a small part of the GVA.
[34] The statistics ministry set up five sub-committees under ACNAS in 2024 to advise the ministry on five specific issues pertaining to the base-revision exercise: i) the use of new databases, rates, and ratios; ii) methodological improvements; iii) estimation of constant price estimates; iv) regional accounts; and v) SNA 2025 update. While the reports of the first three sub-committees have been published, the ones on regional accounts and SNA 2025 have not been released yet.
[35] Two former MoSPI officials have suggested that detailed data on national accounts should be made available on MoSPI's microdata site. - Reliability of Our Statistical System (2024), Kumar and Sharma, Economic and Political Weekly. Two other statisticians have asked for more detailed disclosures on administrative datasets to bolster the credibility of the national accounts estimates. - New Data, New Estimates (2026), Singh and Kulashreshtha, Economic and Political Weekly.
