This is the first article in ‘India Black Boxed’. Read the series introduction here.
Controversy refuses to die down about the size of India’s GDP and its growth rate. It all started when the new GDP series with base 2011-12 was released in 2015. Not only did analysts point to problems, the government itself was unhappy that it showed a higher growth during the UPA’s ten years compared to the post-2014 NDA period.
The pandemic in 2020 severely dented the economy and the economy witnessed its steepest decline since Independence. The recovery from this low base was also steep. This has led to the official claim that India has done well in spite of the pandemic and the war in Ukraine to become the fastest growing major economy in the world. Is this the correct picture of the economy? That depends on the accuracy of the numbers and the policies formulated on that basis.
Pre-pandemic controversies
Doubts about the accuracy of data in the new series from 2011-12 have risen on several counts. To begin with, when the new series was announced in 2015, there was no back series to compare it with. It was said that the new series was based on the MCA21 data base of the industrial sector, which was more complete than what was used till then, the IIP data. It was stated that the back series could not be generated both because the MCA21 data base had not stabilised earlier and the relevant data on employment became available from 2011-12.
The next controversy was the government’s claim that the Indian economy grew at an average of about 7% during 2015-2020, which made it the fastest growing large economy in the world. This was undermined by A. Subramanian (2019). He showed that the growth rate was over estimated by up to 2.5% after 2014.
The next blow came when NSSO reported in 2019 that out of a sample of 35,456 companies taken from MCA21 data base, 38.7% were ‘out of survey’ units. These units are either not traceable or misclassified. So, data is either missing or mis-specified. Thus, the use of MCA21 for GDP calculation could be leading to errors in estimation.
The government argued that the inclusion of the ‘out of survey’ companies brings the output closer to the true production and there is no over-estimation of GDP.
A committee was set up to work out the missing back series. Its report showed that the rate of growth was higher during the UPA period compared to the NDA years. The government rejected it and in an unprecedented move, asked the NITI Ayog to rework the series. The NITI Ayog obliged and presented a back series showing that the rate of growth was higher during the NDA period compared to the UPA years.
Upward bias in GDP
The problem with the GDP data becomes clear when the official data shows that the highest rate of growth during the decade of the 2010-20 was in the year of demonetisation, 2016-17. From all accounts, starting November 2016, output was severely impacted in that year. Even if it is assumed that the output was growing up to October 2016, and declined after that, the average GDP growth became negative. This points to the flawed methodology used to measure GDP which gave an 8% upward bias to GDP in 2016-17. Even this flawed methodology showed the official growth rate declining from 8% in Q4 of 2017-18 to 3.1% in Q4 of 2019-20. So, the real actual rate of growth would have become negative even before the pandemic
Pandemic and the lockdown severely impacted the economy in 2020 and more particularly the unorganised sector. Subsequent recovery has been K-shaped – namely, some sectors growing while others (unorganised sector) declined. This decline has not been captured in data leading to over-estimation of the GDP. This becomes clear when one looks at the method of estimation of GDP, especially the quarterly GDP, which is what is usually discussed in public discourse.
Official methodology
I have previously analysed the official document which presents the `Methodology of Compiling Quarterly GDP Estimates’. It mentions three factors that need to be noted regarding the calculation of GDP from the supposedly more accurate production side:
- “The production approach used for compiling the QGVA estimates is broadly based on the benchmark-indicator method.”
- “In this method, for each of the industry-groups, estimates of GVA are compiled…”
- “In general terms, quarterly estimates of Gross Value Added (GVA) are extrapolations of annual series of GVA.”
These three points clarify that for the quarterly estimates of GDP based on the production approach, most current data are not available so, benchmark indicators from an earlier reference year have to be used. The last survey of unincorporated enterprises was carried out in 2015-16 so that the reference year is now dated and does not capture the current reality.
Further, the methodology states that current figures are obtained by extrapolations of the annual series of GVA of previous years. But if the previous year figures are incorrect, how can their extrapolation be correct? This has been the case post the demonetisation, introduction of the Goods and Services Tax and the lockdown. Each of these three occurrences administered a shock to the economy and caused disruption.
Finally, in some cases, the procedure adopted is to make annual projections and then to divide them by four to give the quarterly figures. Two problems arise. First, there are varying levels of activity in the different quarters. For instance, there is heightened activity during the festive season, while it is low at the start of the financial year. So, division by four cannot be correct. Second, errors in the figures of the previous year get projected to the next year.
Shocks undermine the method
The methodology outlined above relies on a smoothly functioning economy. But it will not apply when there are big unexpected changes, called a shock, like due to demonetisation or the sudden lockdown. The shocks impact the basic parameters of the economy. Like the ratio of the unorganised to the organised sector or the real output in the agriculture sector. So, with a shock, neither the ‘benchmark-indicators’ will be valid nor will it be correct to extrapolate from a normal year to the next one that has experienced a shock.
The Indian economy has suffered several shocks since 2016. Demonetisation in 2016 followed by the introduction of the structurally faulty GST in 2017, the NBFC (non-bank financial company) crisis in 2018 and finally the sudden lockdown in 2020. Each of them impacted the unorganised and the organised sectors differentially, thereby changing the ratio between the two and invalidating the old benchmark indicators.
Further issues with quarterly data
The problems related to methodological issues were compounded by the data deficiencies. Even for the organised sector, only limited data is available. For instance, the corporate sector data representing industry is available only for a few hundred firms. In the case of agriculture, it is assumed that targets set by the ministry are achieved. But that has not been the case in the last few years due to heat or late rains or inability of perishable crops to come to the market during the lockdown and demonetisation, so that it rotted in the fields and agricultural output declined while it was taken to have increased. The method for estimating the unorganised sector in the GDP needed to be modified, but this has not been done.
In brief, there are two inter-related problems with the GDP data. The infirmity in the data and the invalidity of the method to calculate the GDP.
The problem was further compounded by the government’s lack of faith in its own employment data which it rejected in 2019 because it showed that unemployment had reached a high of 45 years. Since employment data is used in the calculation of the GDP, if it is rejected, the GDP calculation also becomes unreliable.
To persist with the methodology in the 2017 official document, new indicators are required based on fresh surveys. But no new survey of the unorganised sector has been conducted since 2015. Even the Census has not been conducted in 2021 and that compounds the problem.
Further, each of the shocks listed above impacted the economy differently. So, without a change in the method and resolving the data issues, errors get compounded and reliable GDP numbers cannot be generated.
Stance of international agencies
The government claims that international agencies, like the IMF and the UN, have supported its claims on GDP. Their figures for GDP growth differ from the official figures by a small percent. But that is not surprising since these agencies are not data collecting agencies and use the official data. Even the RBI uses the official data on a host of macro variables.
Effectively, all of them reproduce the errors in the official data and none of them have more accurate data. The surprise is that all these agencies ignore the data-related issues when the errors are glaring. Worse, if Indian data has such huge errors, other developing countries are likely to have similar or even greater errors, making international comparisons meaningless.
Impact on other macro aggregates
GDP data is the base used to estimate other macro aggregates, like consumption and savings. These affect the measurement of poverty and growing inequality. If growth is strong then it would imply strong growth in employment. But this link is broken since growth is in the organised sector while the unorganised sector is declining. The former hardly creates employment while the latter which provides a bulk of the employment is losing employment. So, this lopsided growth has broken the link between growth and employment.
Further, if the unorganised sector declines then the overall demand becomes short, leading to low capacity utilisation and decline in the investment rate and even the organised sector rate of growth will fall. This was visible in the period 2017-18 and 2019-20 (before the pandemic).
The incorrect GDP numbers should impact the fiscal situation. This is reflected in the revenue and expenditures often missing the targets set in the budget. The final figures differ considerably from the budget and revised estimates. But these revisions are not as stark as the errors in the GDP data should lead to.
The reason for this smaller error is that the budget is largely for the organised sectors and of the organised sector. The revenue collection is largely from the organised sector. Most expenditures are also for the organised sector. Where the expenditures pertain to the unorganised sectors like on food, rural development, education and health, revisions are made when the deficit in the budget increases. Thus, the budgetary calculus is not as seriously impacted as the large errors in GDP data ought to lead to.
Conclusion
To conclude, India’s GDP numbers are vitiated due to methodological and data-related deficiencies. This suits the ruling party’s political narrative of a well-functioning economy. By continuing to harp on these incorrect numbers and hiding the true facts, it adds to the non-transparency in the government’s functioning.
Arun Kumar is the author of Understanding Black Economy and Black Money in India.