GDP conundrum: What makes the changes in the new series so radical?

Dennis Rajakumar and S.L. Shetty of the EPW Research Foundation, present a detailed explanation of the wide-ranging changes in the new 2011-12 National Accounts Statistics series.

This is the first of a four-part series.

Rebasing economic statistics at appropriate intervals to capture structural changes and to incorporate newer sources of data is a noteworthy and healthy feature of any country’s statistical system. In line with the recommendations of the country’s National Statistical Commission, the Central Statistics Office (CSO) has been making attempts to rebase economic statistics once every five years. Although the CSO should have rebased National Accounts Statistics (NAS) from 2004-05 to 2009-10, the NAS has been rebased to 2011-12 for want of quality information with regard to employment. Besides, 2009 was not a normal year due to the global financial crisis, and drought.

Whenever Indian NAS gets rebased, some changes between the old and new series are expected because of taking recourse to a better method of estimation, adopting newer database, use of more recent price indices, and so on. The changes introduced in the earlier NAS revisions were more of a peripheral nature, but those that occurred in the latest 2011-12 series are quite distinctly large due to CSO’s adherence to a wide range of recommendations of the United Nations System of National Accounts, 2008 (UN SNA, 2008) and also because of the availability of a wider dataset in this age of information technology. As a result, the 2011-12 series has attracted rather unparalleled attention of both media and scholars.

Introduction of wide-ranging changes in NAS 2011-12

Based on the dissemination of official information by the CSO (2015 and 2015b), we have attempted to regroup the changes into: conceptual changes, improvement in the coverage of sectors, methodological changes, and presentation of NAS aggregates for the first time at the level of institutional sectors (Rajakumar and Shetty 2017, forthcoming).

Conceptual changes

Two major conceptual changes relate to:

the use of Gross Domestic Product (GDP) at market prices as a preferred income concept for the national economy as a whole instead of GDP at factor cost; and
use of basic prices concept for measuring gross value added (GVA)1 at aggregate as well as sectoral levels.

Thus, the GVA (formerly GDP) at factor cost has ceased to be a relevant indicator of growth; instead GDP at market prices is now used to measure growth of the economy both in real and nominal terms. Based on the UN SNA, the CSO has also introduced the new concept of GVA at basic prices, which falls between GVA at factor cost and GDP at market prices. Splitting indirect taxes and subsidies into production taxes and subsidies and product taxes and subsidies, CSO provides the following sequence at the aggregate and sectoral and institutional levels:

Also, departing from viewing GDP as a measure of welfare too, CSO now argues that “GDP as a statistical indicator has a limited purpose, which is to describe and quantify the process of value addition in the economy” (CSO, 2015a). The UN SNA (2008) earlier had argued that GDP at factor cost suffered from the serious limitation of not having outputs with corresponding observable vector of prices.

The GVA at basic prices are available by economic activities, that is, sectoral levels and, like in the past, GDP at market prices are available by its expenditure components.

Changes in the coverage of sectors:

The foremost change in this respect in the new series has been the use of data for a large number of companies, captured through e-filing portal of Ministry of Corporate Affairs (MCA), popularly known as MCA21, for building up population estimates for the private corporate sector. This has replaced the erstwhile use of the study of company finances by the Reserve Bank of India (RBI) based on a thin sample of around 4,500 companies. With a dynamic system of filing e-returns, the final count reaches 525,000 companies with about 85% coverage in terms of paid-up capital (PUC) for the first two years - 2011-12 and 2012-13 (CSO 2015a). The increase in coverage thus takes place in stages; the above final count is preceded by covering about 300,000 common companies for some years and basing it on advance filing of corporate accounts with stock exchanges for the advance and provisional results.

Coverage of financial corporations has been expanded to include mutual funds, stock brokers, stock exchanges, pension funds, and regulatory authorities. As for coverage of public sector, the CSO now has moved away from sample basis to complete enumeration of local bodies (both rural and urban) and autonomous institutions.

Methodological changes:

Third, significant methodological changes have been introduced in the new series in four vital areas:

Enterprise approach

The approach to defining an institutional unit of observation (or analysis) has been changed to the ‘enterprise’ level. Earlier, CSO had relied on the ‘establishment’ approach whereby an institutional unit was seen confined to a geographical boundary (or physical structure or factory) in which that unit functioned. The Annual Survey of Industries (ASI), which was the principal source for estimating manufacturing GVA then, followed this ‘establishment’ approach, whereas now under the ‘enterprise’ approach, consolidated accounts of the parent organisations are taken into consideration. Use of ‘enterprise’ approach takes into account the services rendered by company headquarters in the form of marketing, accounting and other ancillary services, and hence the size of manufacturing GVA will be higher under this method. Adoption of the enterprise approach is seen in perfect alignment with the use of MCA21 database.

Effective labour input method

Significant changes have been introduced in the methodology of capturing income of the informal sector by moving away from the conventional labour input (LI) method to an ‘effective labour input’ (ELI) method. Under the LI method, various categories of labour such as owners, hired workers, and helpers were not differentiated by their marginal productivities. In sharp contrast, the ELI method accounts for differential productivities of the workforce; based on weights derived using a nested Cobb-Douglas production function².

Fixed capital formation

As for the estimation of gross fixed capital formation (GFCF), while CSO has retained two major nomenclatures, namely, construction and machinery and equipment, there have been some reallocations by creating two additional items, namely, ‘cultivated biological resources’ and ‘intellectual property products’. The capital expenditure on plantations earlier included in ‘construction’ and investment in livestock earlier covered under ‘plant & machinery’ have now been shifted to ‘cultivated biological resources’, and investment in software and Research and Development (R&D) expenditures have been placed under ‘intellectual property products’.

Valuables

In the treatment of valuables, some methodological changes have been made. In the 2004-05 series, the net acquisition of valuables by the economy as a whole, in the form of precious items like gold, gems, ornaments and precious stones, was included as part of gross capital formation (GCF), but that acquired by households was impliedly treated as part of their consumption expenditure. Now, while there is no change in the treatment of non-household net acquisition of valuables, the part acquired by households is treated as their physical savings, which is treated as part of GFCF.

NAS aggregates presented for institutional sectors for the first time

In the earlier series, the whole economy used to be divided by three institutional categories such as public sector, private corporate sector and household sector, for which savings and capital formation used to be provided - but no estimates for their value added or GVA. In the categorisation of sectors now, while the above broad categories are retained, there have been some changes within them. In the past, public sector consisted of departmental and non-departmental enterprises, but now they have been classified into public financial corporations, public non-financial corporations, and general government. Private corporate sector has been split into private financial corporations and private non-financial corporations. As for the household sector, the firms registered under Limited Liability Partnership (LLP) Act and quasi-corporations (QCs, firms with published accounts), which were hitherto part of households, are now included under private corporate sector.

Some traditional approaches remain

While the above narrative bring out that the changes introduced in the new NAS series that constitute major departures, there are still some traditional approaches being followed. First, the commodity flow approach is continued for estimating economy-wide capital formation. Second, subtracting from this total, the capital formation of public sector and private corporate sector, estimated using the budget documents and financial statements, the capital formation of household sector is arrived at. That is, the household sector has been continued to be treated as a residual sector. And, thirdly, the use of ‘blow-up’ method is continued for accounting for uncovered companies in the case of private sector companies. The CSO builds up population estimates for the corporate sector using companies’ data available as on 1 December from MCA21 data. It is very likely that several companies would not have completed online filing through MCA21 portal by then. To account for such companies, the estimates based on data extracted from MCA21 as on the cut-off date are scaled up using the ratio of population PUC to the PUC of covered companies. In the past, the RBI also relied on the same blow-up method for estimating corporate savings and capital formation. The difference being that, the CSO considers population PUC only for those companies which have filed their returns at least once during the preceding three years, whereas RBI depended on PUC numbers as supplied by the MCA. Thus, the chances of blow-up factor being higher were inherent in the case of RBI as the population PUC may include a company not in existence but not struck off by the Registrar of Companies.

Changes in the new series have been truly radical: How and why?

A pertinent question now is: Have these changes been radical? Our answer is ‘Yes’. Each time the base year gets shifted, not only do the levels of GDP and other macro aggregates change, but also their respective growth rates vary due to changes in scope, use of recent prices, and change of methods. But magnitudes of such changes have been minimal in the previous revisions. What is more, we have observed that several results such as percentage changes in the level of NAS aggregates, their respective growth rates, sectoral shares in GVA, savings and investment by institutional sectors, and savings and investment rates portray distinctly different pictures under the 2011-12 series compared to the 2004-05 series for common years, namely, 2011-12, 2012-13 and 2013-14 (Rajakumar and Shetty, 2017 forthcoming).

A comparison of the magnitude of such changes when the previous revision was effected from 1999-2000 to 2004-05 shows that the new NAS series have witnessed radical changes. The GDP at market prices for the year 2011-12, for instance, has been lowered by (-)3.0% under the 2011-12 series compared to its corresponding estimates under 2004-05 series and by (-)1.6% for 2012-13 and by 0.7% for 2013-14. In contrast, in the previous revision, GDP at market prices was uniformly higher ranging from 2.9% in 2004-05 to 5.6% in 2007-08 than its corresponding estimates under the 1999-2000 series. The point differences in growth rates of 2012-13, the first year under the new series, seems of a higher order over the 2004-05 series — nearly 1.7 percentage points higher across all income aggregates. In sharp contrast, the percentage point differences in growth rates in the first year under 2004-05 series were very meagre — nil or higher by just 0.3 percentage point over the 1999-2000 series in most of income aggregates. At the sectoral level, the revision had impacted not only sectoral shares, but also their respective growth rates. The GVA of manufacturing sector has been revised upward and so also is the sector’s growth rate in the latest revised series. Savings and capital formation of the private corporate sector have also been revised upward rather sizeably.

Some brief scope for misgivings

Undoubtedly, the 2011-12 NAS revision exercise has been commendable with significant adoption of the conceptual contours of UN SNA 2008 and simultaneous use of newer sources of data, methodological changes and refinements in the dissemination of NAS aggregates and sectoral and institutional details. However, on a closer examination of the end results, we have discerned some scope for misgivings. We pose them here only in the spirit of making constructive suggestions so that the contribution of NAS estimates in providing insights into the working of the national economy is further strengthened.

First, while the adoption of the enterprise approach with the use of MCA21 corporate sector data has been very novel, it has apparently given rise to not only inflated estimates of manufacturing GVA, but also some exaggerated annual growth rates therein shown persistently. Moreover, a sharp divergence is noticed in the growth rate of manufacturing as per the Index of Industrial Production (IIP) and as per the GVA at basic prices under the new NAS series. This divergence is explained by the difference between the measure of value added which is the basis of the new series and that of physical output.

Second, the new data sources used at the stages of Advance Estimates (AEs) and Provisional Estimates (PEs)3 appear to be far removed from and inconsistent with the results of final revisions and hence the intervening results, particularly relating to the publicised growth rates. This seems to raise serious questions of credibility.

Third, the phenomenon of discrepancies (that is, the difference between supply and demand sides), which was large at constant prices but negligible at current prices, inevitably suggests the issues related to deflators. The GDP deflator is a conglomerate of Wholesale Price Index (WPI), Consumer Price Index (CPI), and Volume Index, but WPI is mostly used. The IIP and WPI are still based for 2004-05 and so it is possible that infirmities inherent in these indices would have caused such deviation and magnitude of discrepancies. Once they are rebased to 2011-12, some of these misgivings may disappear.

Finally, the National Accounts literature has recognised that GDP has been, “one of the great inventions of the 20th century” and all great minds who have laboured on it have concluded that GDP measures total incomes earned by households by summing wages and salaries, rents, profits and interest, that is, factor incomes. It is an economic concept, a concept of economic welfare - not just an accountant’s bookkeeping. In this light, the neglect on the part of the CSO to disseminate regular time series on GDP (or GVA) at factor cost is indefensible.

Notes:

Gross value added (GVA) is broadly defined as the value of output less the value of intermediate consumption. It measures the contribution to an economy of an individual producer, industry, sector or region.
In economics, the Cobb–Douglas production function is a particular functional form of the production function, widely used to represent the technological relationship between the amounts of two or more inputs, particularly physical capital and labour, and the amount of output that can be produced by those inputs.
The AE is done for a financial year before the closure of that year using a number of lead indicators, whereas PE of a financial year is done subsequently within two months of the completion of that year using further improved lead indicators for the full year. For instance, AE for 2015-16 was released on 7 February 2016 and PE for 2015-16 was released on 29 May 2016. The release of AE for the current year, however, has been advanced by a month in view of the early budget presentation.