Sharp GDP data revision triggers row

44 Views
May 31, 2020

Pronab Sen (PIB photo)

NEW DELHI: Sharp revisions in the GDP data for past quarters have once again raised doubts about the data quality with some economists demanding that the National Statistical Office (NSO) explain the methodology. But officials at the statistics ministry rejected the criticism saying they are in line with established practice.
The NSO on Friday had released the data for January-March quarter of 2019-20 and the full year which showed fourth quarter growth slowing to a 40-quarter low of 3.1% and 2019-20 GDP growth slowing to a 11-year low of 4.2%. The data also showed volatile revision to the previous quarters data, the second such revision after February. The NSO had also said that in view of the Covid-19 pandemic and the nationwide lockdown measures implemented since March, data flow from the economic entities were impacted. According to NSO some of these units are yet to resume operations and due to the extension of statutory time-lines for filing financial returns, these estimates are based on the available data and are likely to undergo revision.
“NSO has significantly revised the previous quarters’ growth rates (compared to Q3 release) which is quite puzzling and raises questions about data quality and remarkable volatility in the new series and we believe that a methodological note from NSO explaining the frequent revisions will be very useful,” Soumya Kanti Ghosh, group chief economic adviser at State Bank of India said India said in a note.
“In fact, in February, the quarterly numbers underwent significant upward revisions and such numbers have now been steeply revised downwards by an almost equal amount, within a span of 3 months. While it is customary to change the quarterly numbers in May when the 3rd estimate of FY20 is released, the extent of such revision reveals possibly the loss in Q4 because of lockdown may have been evenly distributed across quarters/Rs 1.18 lakh crore loss estimated and distributed across quarters in FY20 (Q4 accounted for only 50% of such), said Ghosh.
He said that on an unchanged Q1, Q2 and Q3 numbers, the Q4 ( Jan-March quarter) GDP growth comes at 1.2%. “As per our calculation only 18% of GVA (gross value added ) is exempted from the lockdown and NSO may release data for that segment only for a large part of Q1FY21, and hence we cannot rule out data issues even for Q1,” said Ghosh.
But officials at the statistics office dismissed the criticism saying a “mountain out of a molehill”, was sought to be created. “ Revisions take place based on the availability of data and revisions by the source agencies,” a senior MOSPI official said when asked about the sharp revisions.
In a report Madhavi Arora, economist at Edelweiss also said that had the last three quarters been unchanged in absolute levels, the fourth quarter FY20 growth would have been 1.5% or lower.
Pronab Sen, former chief statistician of India, said the standard operating procedure for quarterly data is that quarterly data for the first and third quarters are not changed until the fourth quarter data, and the provisional estimate for the full year is released. He said that at that point estimates for the first and third quarter are revised to maintain consistency with the annual data. “ What is not correct is the revision in February. The MOSPI should strictly adhere to the data revision calendar to maintain data credibility,” said Sen.
The data revision guidelines on the MOSPI website released on November 16, 2016 clearly states that the Q3 estimates of current year would be revised at the end of a financial year at the time of release of provisional estimates of current year on 31st May and also along with subsequent revisions in the annual estimates.
Another former chief statistician of India, who did not wish to be quoted said, large data revisions are in line with what is followed globally and it is an accepted practice with quarterly GDP data and any criticism is a “meaningless statement.” Asked whether deviation from the revision calendar is accepted, he said he did not have the full information to comment on the issue.


Source link

Leave a Reply

Your email address will not be published. Required fields are marked *