Thread by @pramit_b

7 years after the new #GDP series was unveiled, it continues to baffle analysts, investors, and citizens of the country.
Rahul's #thread below is right on some aspects of the GDP calculations and not on others.
Hence this thread of mine! (1/n)

https://twitter.com/RahulBajoria_/status/1566352857175302146

First, I must point out that the very fact that we are still trying to figure out what the GDP data actually conveys so many years after the revision was made tells us something.
Had @GoIStats' transparency and documentation levels been higher, this wouldn't have happened. (2/n)

Second, Rahul is right about deflators being a key problem in the GDP calculations. However, the impression being conveyed that earlier it was all volume-based is not true. Even earlier, it was a mix of volume and value-based calculations. (3/n)

The mix varies depending on what indicator is available for which sector. This is true even of IIP by the way; not all sub-indices are volume indices. Some are calculated from output values using sectoral deflators. (4/n)

But it is indeed true that the reliance on volume based indicators came down, and that on value based indicators went up, increasing the importance of deflators in the new series. Had this been accompanied by new prices series data, things would have been fine. (5/n)

But that didn't happen. So we don't have sector-wise input and output price data - which is what you need to really deflate nominal input and output values before you get 'real value added'. Typically, you get these price series from the PPI... (6/n)

The PPI or the producer price index is what most large (G-20) economies rely on to produce their national accounts data. Not India. We have been talking about one since the past two decades. But we continue to use WPI, which is a very very flawed proxy. (7/n)

That is why the deflators are so erratic and fail to do what they are supposed to do: remove the price volatility that occurs in commodity/business cycles to tell us what's happening to 'real' levels of output/value-added. (8/n)

If you think that okay we should be careful about interpreting the official GDP numbers when commodity prices move up or down sharply and that's all we need to worry about, you will be wrong again. Because deflators are the second-biggest problem today. (9/n)

The biggest problem in the GDP calculations lies in how we estimate informal sector growth. For agriculture, we have actual data (even if not fully adequate). For the rest, it is growth by assumption! (10/n)

When a base-year change occurs, there is usually a benchmark survey that tells statisticians how much value-add is happening in the informal sector. The growth rate observed between that and the previous survey was earlier used to estimate future growth. (11/n)

In the 2011-12 base change, this procedure was changed. Statisticians shifted to formal sector proxies to estimate informal sector growth, leading to significant chances of over-estimation. At least one ex-CSO person (G C Manna) wrote a paper saying this wasn't kosher. (12/n)

Perhaps the most under-appreciated aspect of the new series was what it meant for the distribution of output/growth. The use of the MCA-21 database meant that it was difficult to figure out which sectors and which states the firms belonged to. (13/n)

So they had to use ASI data to apportion the value-added estimated using MCA-21 to figure out the sectoral shares of each industry (at least till 2018, as per records).
This is great juggad but the NSC committee on real sector statistics was not very happy about it... (14/n)

... and wanted MCA to share more details with the statistics ministry. I have not been able to establish if this has been done or not so far. But usually such changes only happen during the next base revision (which is still far away). (15/n)

For states too, ASI is being used to apportion national value-added across states, leading to huge swings in state GDP figures after the new series was implemented. (16/n)
livemint.com/politics/polic…

The Dholakia committee report on sub-national accounting submitted in 2020 asked Mospi to end this 'allocation' business and estimate each state's share directly but this too will likely have to wait till the next revision cycle. (17/n)

So the idea that the MCA-21 database has allowed our statisticians to segregate output more finely is completely wrong. If anything, it has made that task more difficult. (18/n)

There are other issues as well but since I have tested your patience enough, I will sign off with one piece from a few years ago which captures some of the key issues (including some discussed above). (19/n)
livemint.com/news/india/the…

If you still have patience and want to know why our statistical system is failing to generate the data we need, here is a thread from 2019 with links that may be of interest to you:

https://twitter.com/pramit_b/status/1197797718682292226?s=20&t=pJ7B1ujxCETxsEZl50HcDA

(n/n)