Why factory output figures are suspect

In early 2015, the Central Statistical Office (CSO) introduced a new series of National Accounts Statistics (NAS) with 2011-12 as the base year, replacing the earlier series with the base year 2004-05. It is the CSO's routine job to make such revisions, roughly once in a decade, to account for the changing economic structure, and relative prices.

Moreover, the revision allows for updating the statistical methods used, to keep up with the guidelines of UN System of National Accounts (SNA) and to introduce newer data series to improve the estimates. Usually, with base year revision the absolute GDP size gets slightly enlarged. But rarely, if ever, does the growth rate of GDP (or of its sectors) differ markedly between the new and the old series.

But this time it is different. For the base year 2011-12, the absolute size of India’s domestic output at current prices (at ₹82,06,398 crore) is smaller than in the previous series (₹83,91,691 crore) by 2.2 per cent; and, its growth rate in 2013-14 in real terms was higher at 6.6 per cent, compared to 4.7 per cent in the older series. (Note: GDP at factor cost in the older series is almost identical to gross value added or GVA, at basic prices; referred as GDP for simplicity).

For some sectors, even the direction of change has got altered after the revision. For instance, for 2013-14, for the manufacturing sector, real growth rate swung from (-) 0.7 per cent in the old series to (+) 5.3 per cent in the new series. Such drastic revision of industrial growth rates was met with considerable scepticism, as the revised (higher) estimates were quite at variance with other macroeconomic correlates, such as bank credit growth, or industrial capacity utilisation, or new investment projects launched.

Old and new methods

The CSO has, however, defended the new NAS by claiming that the new GDP estimates are an improvement over the earlier series as the revision has followed the global best practices of the UN System of National Accounts 2008 (SNA 2008), replacing the earlier template of SNA 1994. It has also argued that newer methodologies and much larger databases have been used for estimations.

The debate has remained unresolved. To understand why manufacturing sector GDP estimates appear impaired in the new series, one has to examine the assumptions that lie behind the methodological changes, and quality of the newer databases used in the revised series.

The older methodology: In the older series, the manufacturing sector consisted of two parts: registered (or organised) sector accounting for about two-thirds of manufacturing output, estimated using Annual Survey of Industries (ASI) – a survey of factories (or establishments) registered under the Factories Act. The output of unregistered (or unorganised) manufacturing was estimated using various NSS sample surveys.

The new methodology: The new NAS has practically abandoned ASI, replacing it with the Ministry of Corporate Affairs’ (MCA) company database obtained from the mandatory filing of financial returns by registered companies. It is a much larger and up-to-date database, available every quarter (though unaudited).

MCA versus ASI

Critics, however, have shown that the MCA database had many shortcomings of incompleteness and inconsistency, thus questioning the value of the corporate sector estimates in the new series (of which manufacturing accounted for about 50 per cent). Further, the CSO has used multipliers (or “blow up” factors) to arrive at the population estimates from the “sample” estimates of the MCA database. As the details of its blowing up methodology are not made public, serious apprehensions persist about the veracity and consistency of the procedures followed, and hence the question mark over the reliability of corporate sector estimates.

The crux of the problem is that different approaches are followed for data collection in ASI and company finances. MCA data follows an enterprise approach , which takes into account all activities undertaken by an enterprise. In contrast, a factory (or an establishment) is the primary unit of data collection in ASI. The CSO has contended that the ASI’s establishment approach , by definition, leaves out value addition taking place outside the factory premises in locations such as company head office, R&D centres, sales and service facilities. According to the CSO, this is the reason for under-estimation of manufacturing value added in the earlier series. The CSO has contended that the problem is now corrected by using the MCA database — hence the relative size and growth rate of manufacturing is higher in the new series.

False contention

Did the ASI really leave out non-factory value addition, as the CSO has claimed? We sought to verify the truth behind CSO’s contention — as reported in a paper published in Economic and Political Weekly , September 1, 2018 (co-authored with Ravindra Dholakia and Manish Pandya).

A close reading of the ASI’s field investigators’ manual — prepared by the CSO for the field investigators — showed that this is not true. Further, a perusal of a sample of ASI filled-in schedules confirmed that value addition taking place outside the factory premises are in fact recorded — thus, contradicting the CSO’s claims. The list of companies examined includes some of the leading firms from Gujarat and elsewhere. Therefore, the assumption under which the ASI was replaced with the MCA does not seem to hold true. Hence the critics’ contention that the changeover to the MCA database could be responsible for obtaining a faster growth rate of GDP in manufacturing seems to be valid.

To sum up, the new GDP estimates show a higher share and faster growth rate of manufacturing sector compared to the older series. The CSO has justified the shift — against public criticism — arguing that the newer methodology and database have overcome shortcomings of the older series. The new NAS has replaced ASI with the MCA’s company financial data on the assumption that the ASI was leaving out value addition taking place outside of factory premises, leading to an underestimation of manufacturing output. Is this claim true?

No, it is not, as evident from ASI’s field investigators’ manual, and from an examination of filled-in ASI schedules. Hence, the assumption underlying the methodological changes made in the new NAS seem questionable. The estimates of manufacturing sector output in new GDP series are suspect — supporting the widespread criticism.

The writer is Professor, Indira Gandhi Institute of Development Research, Mumbai

COMMENT NOW