First of all, very few states participate in collection of index of industrial product (IIP) data. As such they mainly rely on Central IIP, Sen says.
Secondly, although states do participate in the National Sample Survey (NSS) on unincorporated enterprises, they don’t process the data, says Sen, who is now programme director for the India programme of the International Growth Centre.
"There's a huge difference between the Central and state samples. Clearly the field investigators from states are not doing their job properly. The net result is that states have very foggy idea about what is happening within their boundaries, and even lesser within districts," he says.
So, when states compute GDP, they use Central data and extrapolate it to their individual proportions. States proportions are arrived at by using central data on IIP, and unincorporated enterprises state-wise.
"If you depend on Central level data, its samples are not large enough to give you estimates at the district level. The Central sample is small. It is representative at the national level and reasonably representative at the state level, but not at sub-state level. It is completely unreliable at the district level," says Sen.
That is why states are supposed to do matching samples. If they do those properly, states will have enough data to estimate district level GDP. Without that, it can’t be done, he says.
A former official at the ministry of statistics and programme implementation (MoSPI) says states have to rely on central data on corporate earnings but these are available at the national level as the corporate is an all-India entity.
The National Statistics Office (NSO) allocates these earnings to states on the basis of employment data of the organized sector. It is also possible for them to get data on manufacturing state-wise through annual survey of industries (ASI) data.
He said computing district level GDP is more complicated because you have to do further level of allocation. Many data is not available district-wise. For instance, IIP data is not available district wise, as also employment data.
Central samples are not very robust to give you reliable data at the disaggregated level, they can give you data but these are very coarse estimates, he said.
Indicus Analytics used to come out with district level GDP and did this work for ten years, but it stopped doing so in 2016.
Laveesh Bhandari of Indicus Analytics said," "(It is) impossible to do now properly as many data sources such as NSS employment and consumption surveys are not released."
The ministry of statistics and programme implementation (MOSPI) had in November last year decided “not to release the Consumer Expenditure Survey (CES) results of 2017-2018” in view of “data
The survey showed that inflation-adjusted average monthly per capita expenditure (MPCE) declined by 3.7 per cent in the country between 2011-12 and 2017-18, a first in four decades. The decline came because of an 8.8 per cent decline in rural MPCE even as urban MPCE increased by 2 per cent.
Earlier, the ministry had also withheld the NSSO (now NSS) report on unemployment. The report showed that the unemployment rate stood at 45-year high of 6.1 per cent during 2017-18. The report was later released by the ministry.
It is not that none of the states come out with district level GDP data. Many states such as Maharashtra, Karnataka, Kerala, Uttar Pradesh, Odisha, Rajasthan do.
For instance Mumbai, which is a red zone in Maharashtra, had gross value added at current prices to the extent of Rs 4.59 trillion in 2018-19. This represented 20 per cent of Maharashtra's GVA at Rs 22.95 trillion and 2.7 per cent of national GVA at Rs 171.4 trillion.