Bridging the gap between GDP estimates and the real GDP
The only consolation the CSO has is that the research reveals that the GDP data is revised upwards or downwards from initial numbers in many countries, so India is not alone in this respect. But having said that, it also makes it difficult to analyse GDP trends until the final revision is over.
In an earlier column, I had pointed out that the advance estimates of the Gross Domestic Product (GDP) released by the Central Statistical Office (CSO) are just what their name says they are - Estimates. They are often revised upwards or downwards after a period of time. And typically, the real GDP will be available only much later.
A research conducted on GDP revisions over the years by four researchers - Anupam Prakash, Avdesh Kumar Shukla, Anand Prakash Ekka and Kunal Priyadarshi - of the National Accounts Analysis Division of Department of Economic and Policy Research (DEPR) in Reserve Bank of India and published as a Mint Street Memo shows how much the variations have been over the years.
The researchers point out that an analysis of the revisions since 2003-04 for annual data and from 2002-03 for quarterly data shows that the CSO had to revise the real GVA upwards in 10 years, and downwards in the remaining four, in the subsequent revised estimates. More interestingly, the GVA estimates needed to be revised upwards in 12 of those 14 years, and revised downwards in two years.
As I had pointed out in the earlier column, it happens because the first advance estimates are typically based on extrapolations of the relatively small sample for which data is available at that time. As more data comes in, the GVA and GDP needs to be revised. The CSO releases the data as First Advance Estimates, Second Advance Estimates, Provisional Estimates, First Revised Estimates, Second Revised Estimates, and finally, the Third Revised Estimates at different points of time. In some years therefore, the GVA moves up or down four times or more after the First Advance Estimates.
The only consolation the CSO has is that the research reveals that the GDP data is revised upwards or downwards from initial numbers in many countries, so India is not alone in this respect. But having said that, it also makes it difficult to analyse GDP trends until the final revision is over. Because it might give a different picture.
The bigger problem though as the research makes clear is that the advanced estimates - especially the First Advanced Estimates - often lag behind in catching a reversal in trend. For example, in 2008-09, at the time of the global financial meltdown following the Lehman Brothers collapse, the first advance estimates completely failed to capture the extent of the problem to the Indian economy. The final revisions were huge.
Equally, the first advance estimates have also failed to capture the extent of a recovery, and the GVA needs to be revised upwards a year down the line very significantly.
The researchers point out that the only way out therefore is for analysts to use other high frequency data along with Advance Estimates to figure out whether the economy is improving or doing badly. Both the GDP and GVA first advance estimates are therefore fairly unreliable in spotting both economic improvement and economic distress.
Even otherwise, the GDP is of limited use when trying to see the well-being of the citizens of the country. The GDP does not capture a range of other things - from economic inequality to pollution to happiness. Which is why the government should be careful when it pins its achievements only on the basis of rate at which the GDP is growing.