How GDP new series changes the picture of Indian economy
Learnings from the National Statistics Committee report on GDP calculation and its impact on how we see the Indian economy
The Report of the Committee on Real Sector Statistics, containing the much-anticipated GDP back series data, is apparently just a draft report, according to the ministry of statistics and programme implementation. But while the numbers are yet to get the official imprimatur, the report does have a number of interesting things to say about the Indian economy. That’s apart from the puerile squabbling about growth rates under different governments.
How does the picture of the Indian economy change when we adopt the latest shiny new measuring rod—the GDP series with 2011-12 as base year?
Let’s take the fiscal year 2011-12 for illustrating the changes. First, the economy gets smaller. Under the new series, gross value added (GVA) at basic prices, (at current prices) was 3.4% lower than under the old series.
But it’s the composition of the economy that’s interesting. The share of industry in the economy has gone up when we use the new measuring rod and the share of services has shrunk. Recall the complaints about India’s unnatural dependence on services—well, the new series paints a slightly better picture, but not enough to dispel the gloom.
In 2011-12, using the new series, the share of industry in the Indian economy goes up to 23% from 19% under the old dispensation.
Manufacturing goes up from a measly 14.7% of total GVA to a bit more respectable 17.4%.
Correspondingly, the share of services shrinks when we use the new series. In 2011-12, services accounted for 63% of total GVA under the old measurements and 58% under the new one. The share of trade, hotels, transport and communication shrinks from 24.7% under the old series to 17.4% under the new one. That’s a substantial reduction, the reasons for which need to be explained. But the share of “financing, insurance, real estate and business services” goes up, as does the share of construction.
Agriculture was 17.8% of the economic pie in the old series and that increases to 18.5%. The details can be seen in the chart above. Note that the data pertains to a single year—2011-12—and the changes are due entirely to our using a new measuring rod to compute GDP and GVA.
How did the share of manufacturing change between 1993-94 and 2011-12? Under the old series, it fell from 15.4% of GVA to 14.7%; under the new series, it went up from 16.8% to 17.4%. So perhaps it’s not entirely premature de-industrialization, but stagnant industrialization? Hard to tell—in 2017-18, the share of manufacturing had slipped back to 16.7%.
What changes did the new series have on the expenditure-side GDP figures? For 2011-12, the biggest change is in gross fixed capital formation—it went up from 31.8% of GDP under the old yardstick to 34.3% using the new one. The share of capital formation is higher under the new series. On the other hand, the consumption share is only slightly different. The details are given in Chart 2.
There’s another curious feature about the new GDP back series. From 1994-95 to 2002-03, growth rates under the old series at constant prices were higher every year than in the new series. But from 2003-04 onwards, growth rates have consistently been higher under the new series than under the old one. What changed in 2003-04 to alter the trend? It’s possible that some activities were measured from 2003-04 onwards but not earlier. The final data release should tell us the reasons.
And finally, it’s important to note that even the data under the new back series suffers from serious limitations. A note to one of the tables on GDP in the report says, ‘‘Discrepancies found to be volatile and constant price series could not be estimated.”
Another table shows that discrepancies under the GDP data in the new back series range all the way from a negative ₹1.93 trillion in 2007-08 to a positive ₹1.14 trillion in 2013-14.
The report says, “when we look at the growth rates, there are some differences, although not significant and this is largely due to the ‘discrepancy’ variable, which is found to be highly volatile”.