The study titled, 'Examining Gross Domestic Product Data Revisions in India', cited examples. For example, it said, GDP grew 7.6 per cent in 2015-16 according to CSO's first advance estimate. It was changed to 7.9 per cent in the first revised estimate and to 8.2 per cent in the second revised estimate.
“In view of multiple rounds of data revision, it may be confusing for data users to decide on the true state of the economy, and the real strength of growth momentum,” the study noted.
The dilemma regarding reliability is usually the greatest around the release of advance estimates, generally understood to be tentative and liable to change with the arrival of subsequent datasets, it said.
Using data from 2003-04, the study found CSO had revised real GVA growth estimates relative to advance estimates, upward in 10 years and downward in four. The average upward revision was around 70 basis points (bps); in the case of downward revision it was about 27 bps. In the case of real GDP, the CSO revised advance estimates upward in 12 years (average of 81 bps) and downward in two years (average of 204 bps).
The sharpest revisions were observed during growth cycle turns. For instance, during the upswing of 2005-06, the advance estimates initially pegged real GVA growth at 8.1 per cent. However, it was revised to 9.5 per cent, a difference of 140 basis points.
During the downswing of 2008-09, real GDP growth was revised downward from 7.1 per cent in advance estimates to 3.9 per cent in final estimates, an overestimation of around 320 basis points.
On the production side of GDP data, mining, quarrying and manufacturing showed significant revisions. On the expenditure side, private final consumption expenditure showed revision.
The explanation for these deviations stem from subsequent revisions of GDP data that are able to draw on more comprehensive and reliable data sources.
For example, in the case of the manufacturing, the major reason for the revisions is due to incorporation of data from the Annual survey of industries which comes with a lag of two years.
“It may be advisable for data users to read GDP growth numbers carefully, along with other high frequency indicators of the real economy,” it added. In the services sector, some of the high frequency indicators are commercial vehicle sales, domestic and international air passenger traffic, and foreign arrivals.