How the GDP back series may have been calculated – and what’s wrong with it

The creation of a “back series” whenever gross domestic product (the value of goods and services produced by a country in a year) computations are rebased is a normal, if complex process. If the methodology of calculation is changed, the process of creating a back series is more complicated. But it is also necessary. We cannot make sense of the new methodology unless we compare it with the prior methodology over a long period.

Now suppose, purely as a theoretical exercise, you wish to show that GDP growth in a given period was lower than growth in another given period. It does not matter what the actual GDP was in either period.

Let’s say “Period A” should have lower GDP growth rates than “Period B”. How do you manipulate the data to get the results you want? You need to find “fuzzy” data you can fiddle with, or “rebalance”, if we are going to be polite about this exercise.

Finding fuzzy data

An economy is divided into three sectors – the Primary Sector is agriculture and mining, the Secondary Sector is manufacturing and industrial activity, and the Tertiary Sector is services.

The Primary Sector offers some scope for fiddling but not too much. Agricultural output is known and so are agro prices. But mining data can, perhaps, be rebalanced.

It is extremely difficult to change Secondary Sector data retrospectively. There are many (a non-exhaustive list follows) data series in the Secondary Sector that are correlated to GDP and released regularly. These are updated at monthly or fortnightly intervals and, hence, called “high-speed” data.

High-speed data consists, for example, of car, two-wheeler, tractor and truck sales. These numbers are released every month and, therefore, known for both Period A and Period B and cannot be changed. Bank credit (both agricultural and non-food) is known every fortnight and cannot be changed. Tax collections are known. Exports are known. Imports are known. Railway freight numbers are known. Power consumption is known. Air cargo and cargo passing through ports are all known. So, it is difficult to fiddle with Secondary Sector data.

The Tertiary Sector, services, is the easiest area to fiddle with. India, in particular, has a large unorganised services sector. So we can use various methodological changes to shrink or expand the services sector in size, as required.

What the diligent student will do to complete this exercise is reduce the services component for Period A and perhaps increase it for Period B. The student might also reduce the mining sector output for Period A.

This will automatically mean that the contribution of the Secondary Sector to total GDP rises and so does the contribution of the Primary Sector. The growth rate of the Tertiary Sector would be “massaged down” for Period A, and this would, of course, result in lower overall GDP growth.

The calculation

Coming from the theoretical to the political, this seems to be pretty much what has been done with the latest “back series” calculation of GDP. According to a new back series released by the Central government’s policy think tank Niti Aayog on November 28, Period A (2004-2005 to 2013-2014) had lower GDP growth than Period B (2014-2015 to 2017-2018). It may not be entirely coincidental that the Congress-led United Progressive Alliance was in power during Period A, and the current regime took over in 2014.

Quoting from the Ministry of Statistics and Programme Implementation’s note, in the new back series, we find “the share of Primary Sector in total GVA [gross value added, which is GDP minus net tax collections] is higher than that in the earlier 2004-2005 series primarily due to changes in the data sources. In the mining and quarrying sector, regular annual returns of public sector have been used instead of Indian Bureau of Mines data in the 2004-2005 base”. The Primary Sector contribution in the new series is consistently 1% higher, as share of gross value added in the back series.

Again quoting the note, “The share of Secondary Sector in total GVA has increased in the back series compared to the 2004-2005 series. The increase is largely due to use of MCA [Ministry of Corporate Affairs] data and public sector data in organised electricity and manufacturing sectors which was earlier sourced from annual reports of private electricity companies registered with the Central Electricity Authority and Annual Survey of Industries respectively.” The contribution of the Secondary Sector is consistently 4%-5% higher as share of gross value added.

The note goes on to say, “The share of Tertiary Sector in overall GVA has reduced in the back series compared to the 2004-2005 series. This decrease is largely on account of the use of revised methodology and latest survey data sources of unorganised sector in the new base. In the 2004-2005 base, the main data sources for unorganised sector were the NSS [National Sample Survey] informal sector survey of 1999-2000 for the trade sector, unorganised enterprise survey results of NSS 63rd round (2006-2007) for remaining non-financial service sectors and the Employment and Unemployment Survey (EUS) of NSS 61st round (2004-2005). In the 2011-2012 base, the main data source for unorganised non-financial service sector has been the results of unorganised enterprise survey of NSS 67th round (2010-2011) and the EUS of NSS 68th round (2011-2012).”

The contribution of the Tertiary Sector is consistently about 6% lower as share of gross value added.

Many inconsistencies

There are multiple other inconsistencies that make the back series hard to believe. For example, the government had repeatedly said there was a problem with creating back series because the Ministry of Corporate Affairs database was “unstable” before 2011-2012. How then did it suddenly become usable for this exercise? The Committee on Real Sector Statistics chaired by economist Sudipto Mundle had previously attempted to calculate a back series for the National Statistical Commission, and it came up with very different results, which were far more consistent with the earlier calculations.

Most damning, every high-speed data series mentioned above shows way higher growth for Period A than for Period B. The discrepancies are so large that it makes the new back series calculations embarrassing. One indicator being inconsistent with the overall trend for a couple of years would be explicable. But every measurable indicator showing that growth across the entire decade of Period A was way, way higher than in Period B is plain absurd.

Here is a comparison of growth rates on various indicators by the economist Vivek Kaul:

Assuming one is to take the new back series seriously, the calculations could be extended further backwards to see what other discrepancies arise in earlier periods. A “front series” could also be created, using the old methodology to see if growth did miraculously accelerate during Period B, even though people bought fewer vehicles, paid less tax, despatched less cargo, exported less, invested less, and so on.

The production of reliable statistics and government information is one of the key signs of a free society. Until last week, India’s Central Statistics Office was considered an independent, professional body that did the best job it could, given an economy with large gaps in data coverage. That is no longer the case.

There are studies suggesting that authoritarian regimes cook economic data far more consistently than democratic nations.

We have seen obfuscations and delays in releasing data before – for example, with the demonetised cash returned to the Reserve Bank of India and with the creation of the back series itself, and with Census caste data. This back series just adds to the impression that this government often finds the truth inconvenient.

Maybe it is time to consider where exactly this country is going.