OPINION | A new GDP series comes in with important changes to enhance accuracy

The expenditure side estimation of GDP

For a large and complex economy such as ours, getting economic data right is always a daunting task, especially when beset with unique issues- a large informal economy, lack of data, to name but a few.

Discrepancies in private consumption estimates encapsulated the challenge

For instance, for estimating the largest component of GDP viz. Personal Final Consumption Expenditure (PFCE) it had to use a combination of a bottom-up method (consumption surveys) and a top-down method (commodity flow). It often led to a significant divergence between the HCES (survey-based) data and the PFCE (National Accounts) estimates, going up to as much as two-thirds.

Likewise, estimating the contribution of the unorganised sector (put at 45% of GDP) has been done in a rather crude fashion by extrapolating organised sector data. These are some of the issues which the new GDP series, to be launched on February 27th, seeks to iron out.

The critical change in the 2015 revision

When the Central Statistical Office (CSO) last revised methodology in 2015, it brought in some significant changes- notably, GVA methodology shifted data collection from establishments (or factories) to enterprises (or firms), changed the way value added was calculated (from volume to value basis). These had wide-ranging effects in manufacturing.

In some sectors, value added grew faster than output, as data source changed from IIP/ASI to corporate financial data. In trade and services too, growth went up with the use of indirect taxes to measure changes in value added.

A more important change was in the database for GVA, shifting from the RBI's fixed sample of large private companies to the MCA21 database, while also using IIP extrapolations. The Index of industrial production (IIP) a crucial component used extensively for GVA estimation, also saw changes (base year, extension of coverage etc) but its principal shortcomings (non-representation of unregistered sector, majorly output based with only about 25% of items value based) remained.

2015 changes led to an upward revision of GDP growth

Overall, the changes led to significantly higher GDP growth numbers (e.g., FY14 growth was revised from 4.7% to 6.9%), sparking debate over possible over estimation by relying on corporate data as other indicators seemed to be out of sync.

The IMF’s recent certification of quality of India’s national accounts at “C.” (indicating shortcomings that somewhat hamper surveillance) is just a tad above being considered unusable for any serious work and an unflattering assessment of the world’s fourth largest economy. Which is perhaps why the current set of changes proposed (which also include the IIP and CPI indices) takes a more wide-ranging approach.

The two big changes expected in the 2026 revision

It may be in order to say that the two major reforms targeted are the elimination of “discrepancies” used to balance the expenditure side with the output side and a better estimation of the unorganised sector’s contribution to GDP. With these in view, it has proposed many changes both to database and methodology.

For output, given the infirmities of the present MCA21 base, the source data base is being expanded to new data sets such as the frame of active companies, Management & Administration related data (MGT-7/7A), frame of active Limited Liability Partnerships, GST, PFMS data, and e-Vahan, rather than relying solely on MCA-21. Many of the pitfalls in MCA 21 data (ghost companies, inactive firms etc) could go away.

But the biggest change could be in the estimation of the unorganised sector. In manufacturing and services, presently, such estimates are compiled indirectly through Effective Labour Input Method (ELI Method) but now with availability of Annual Survey on Unregistered Sector Enterprises (ASUSE) data, they will be generated annually as opposed to the indicator-based extrapolation approach. The erstwhile static NSS enterprise surveys will be replaced by dynamic ASUSE formula for unincorporated enterprises. Reportedly, these changes could result in the shrinking of the size of the informal economy from about 45% presently to about 40%.

Double deflator is not on the cards

For those who hope that the other major bugbear of GDP statistics viz. real numbers derived using a single deflator will be addressed there could be disappointment in store, as the NSO proposes to continue with single extrapolation/volume extrapolation method as an “acceptable second-best method” on the grounds that double deflation is data intensive. In manufacturing it will continue to use double deflation but for the other sectors, where single deflation was being used, use of volume extrapolation will be explored instead of double deflation.

Major benefit expected

The other major gain could be the elimination of the “discrepancy” between the production and expenditure approaches in final GDP estimates. Traditionally the expenditure side estimation of GDP was considered less reliable due to lack of timely data and GVA estimates are arrived from the production side, with the differences being reconciled using discrepancies as a balancing figure.

Interpreting quarterly estimates became problematic as discrepancies ran high. This problem is now sought to be done away with by directly integrating Supply and Use Tables (SUTs) into the annual compilation, such that production, intermediate use, and final expenditure (including exports) are fully reconciled. This compares with the earlier method of PFCE being extrapolated using the commodity flow approach.  While this may not completely eliminate discrepancies, they could be minimal, reducing to near zero in final estimates.

Changes proposed to IIP are important

There are also major changes proposed in the CPI index and the IIP, some of which could be game changers. For IIP, traditionally, compilations used a fixed-base framework in which sectoral and industry weights remain unchanged until a base-year revision. The introduction of chain linking is being considered which offers better capture of any recent production structure changes by allowing the increase and decrease in weights annually.

While the overall impact of the new series on top line growth, when they are released by end February 2026, may take a while to play out, if nothing else, the new numbers could enhance the credibility of national income statistics, as both data and methodology improve

(SA Raghu is a columnist who writes on economics, banking and finance.)

Views are personal and do not represent the stand of this organisation.