There are methodological errors in estimating annual and quarterly GDP data, especially when there is a shock to the economy, by using projections from the previous year, dividing the annual estimates into the four quarters and using production targets as if they have been achieved, explains DR.ARUN KUMAR.
A sharp rise in growth after a steep fall in the preceding year is not a new phenomenon for the economy. Prior to 1999, only annual, not quarterly, data was available. Official data shows that the economy has risen sharply several times since independence: 1953-54 (6.2%), 1958-59 (7.3%), 1967-68 (7.7%), 1975-76 (9.2%) 1980-81 (6.8%), 1988-89 (9.4%) and 2010-11 (9.8%). The data after 2011-12 base revision was controversial. For instance, the new series shows a high growth rate of 8.3% for 2016-17 though it is well known that demonetisation devastated the economy.
If the new series, using 2011-12 as the base year, shows a high growth rate for 2016-17, the methodology is not right. This has been extensively discussed since 2015, when the series was announced. A major change has been the use of the data provided by the union ministry of corporate affairs, called the MCA-21 database, since 2015. But it has been pointed out that many of the companies in this database are shell firms and the government shut down several of them in 2018. Further, many companies were found to be missing.
Another problem pointed out, starting the year of demonetisation, is that the measurement of the contribution of the unorganised sector—which constitutes 45% of the GDP—is not based on independent data.
The data for the non-agriculture sector is collected during surveys every five years. In between these years, the organised sector is largely used as a proxy and projections are made from the past. Both these features of estimation pose a problem when there is a shock to the economy.
The demonetisation shock impacted the unorganised sector far more adversely than it did the organised sector. Hence, after demonetisation, the organised sector data should not have been used as a proxy to measure the contribution of the unorganised sector. Further, due to the shock, projections from the past will not be a valid procedure. This problem was accentuated by the implementation of the Goods and Services Tax (GST), which again impacted the unorganised sector more adversely.
Demand started to shift from the unorganised sector to the organized, making the situation even more adverse. For instance, e-commerce has severely impacted the neighbourhood stores and taxi aggregators have displaced the local taxi stands.
Due to the shocks, the earlier procedure of calculating GDP becomes invalid and should have been changed. Since this has not been done, in effect, the GDP data is measuring the organised sector and agriculture.
Thus, 31% of the economy is not being measured, and by all accounts, this part is declining, not growing. Therefore, GDP growth is far lower than what has been officially projected since 2016-17.
The pandemic and the lockdown have administered the biggest shock to the economy. But the organised sector was hit far less than the unorganised sector. The split between the two sectors has been far greater than due to demonetisation or GST. Therefore, there is an urgent need to revise the method of calculating GDP—also, projections from the past do not make sense.
The problem is even greater when projecting quarterly GDP growth. The data used is sketchier than the annual data. Not only most of the data for the unorganised sector is unavailable (except for agriculture), even the organised sector data is partial. For instance, the data for businesses is based on companies that declare their results in that quarter. Only a few hundred companies out of the thousands might be declaring such data.
Worse, the estimation is based on a) projections for the same quarter in the preceding year same quarter, b) in many cases, the projection is not just for the quarter but for the year as a whole and then it is divided into four to get the data for one quarter and c) cases where targets, not actual production data. are used to estimate the contribution to GDP.
Fishing and aquaculture, mining and quarrying, and quasi-corporate and the unorganised sector are a few sectors which belong to the first group. Some sectors belonging to the second category are other crops, major livestock products, other livestock products and forestry and logging. Livestock belongs to the third category, where annual targets/projections are used.
This procedure is clearly inadequate but maybe acceptable in a normal year. But when there is a shock to the economy, does it make sense? If there is a projection from the previous year, it is likely to give an upward bias since the economy was performing better in the preceding year. Further, projections have to be based on some indicators and the data on these indicators were only partially available due to the lockdown.
Finally, how can the annual projection be made and then divided into four to obtain the quarterly estimate when the economy is highly variable from quarter to quarter. In 2020, each quarter was very different from the previous one.
Next, if the data for 2020-21 is erroneous, when there was a massive slump in the economy, the shock continues into 2021-22. How can projections be made from the 2020-21 to 2021-22? Thus, there would be large errors in the quarterly data for the current year. This will then be fed into the data for 2022-23. Therefore, the shock to the economy will play itself out for several years.
Impact on other macro variables
Quarterly data are also published for other macro variables like consumption, and investment by public and private sectors. The government-related data is available in the Budget documents, but the private sector data poses a huge challenge. These estimates are, again, based on projections from the previous year, and in some cases, annual estimates are divided between quarters. Production data is also used to project consumption and investment by the private sector. So, if the former is incorrect, as pointed out above, then the estimates for the latter will also be erroneous.
The RBI’s survey of the organised sector showed that capacity utilisation was down to 63% in January 2021, but the official quarterly data was showing a growth of 1.3% rather than a decline of 10%. Thus, the quarterly data was not representative of even the organised sector.
Similarly, consumer sentiment was down to 55.5 compared to 105 a year back, implying that even the organised sector consumption had not recovered to the pre-pandemic levels. Both these variables were further dented in the second wave of COVID-19 in Q1 of 2021-22. The implication is that the data on these variables is also not reliable.
If the production data is an overestimate due to the use of projections from the last year, the consumption and investment data would also be over projections. The further implication is that if the data for 2020-21 is not right, the quarterly data for 2021-22, projected from the previous year, will also be erroneous and overestimate.
Analysis of macro variables for Q1 of 2021-22
For the moment, let us analyse the Q1 data leaving aside the errors pointed out above. When the economy was in decline in the preceding year, comparing rates of growth makes less sense than comparing the level of GDP.
On a low base of 2020-21 (-24.4%), the rate of growth for 2021-22 looks impressive (+20.1%). But it is 9.2% less than the pre-pandemic Q1 of 2019-20—i.e., the economy has not recovered to the pre-pandemic level.
Further, if the economy was growing at the pre-pandemic rate, the economy would have expanded another 7.5% in two years. Thus, compared to the possible level of GDP in 2021-22, it is down by about 16%.
Except for agriculture and the utilities sectors, data shows that none of the other sectors have recovered to the levels in 2019-20. Private final consumption expenditure is down by 11.9% and gross fixed capital formation by 17.1%. Government consumption expenditure and exports have increased compared to their levels in 2019-20. The former does give a boost to the economy by increasing demand but the latter does not since imports remain much higher than exports.
Therefore, out of the four sources of demand, only government expenditure has increased—but this is not enough to compensate for the decline in the other three and that is why the economy is still down compared to 2019-20.
It may be argued that over time, data undergoes revision as more data becomes available. But the situation now is unusual due to the pandemic. This necessitated a major revision in the methodology itself due to lack of data and consequent non-comparability across quarters and years.
(Dr. Arun Kumar is Malcolm Adiseshiah Chair Professor, Institute of Social Sciences. The views expressed are personal.)