NSSO report: GDP data was bound to be questioned due to methodological issues; it's time CSO redefines sample

GDP data in India has always had problems of computation due to the predominance of the unorganised sector for which data is not readily available. This has meant that there are imputations made regularly for these segments which can never be accurate. With 40-50 percent of the GDP in these pockets, it is always a challenge to find the right proxies. While all analysts have been aware of these problems, they were accepted without much debate as the system was consistent.Things have changed post demonetisation and GST where the perception is that there is a conscious effort to show the economy in a better light. This has been done by having several top economists also showing how the economy has become really robust post these two reforms and that the economy is galloping. This has led to other data sets being brought in like employment, direct tax collections (number of assesses), GST collections etc. As inconsistencies came in like ground level reality on jobs or currency coming back into the system, the narrative became more aggressive than defensive. This made it convenient to reduce it to a political tool on both sides.The recent controversy on NSSO finding 36 percent of companies used in calculating GDP to be non-existent has quite needlessly stoked umbrage. What exactly is the issue here? When GDP is calculated we look at Gross value added (GVA) in every sector. GVA simply put is the return on land, labour, capital and enterprise as per the text book. In terms of corporate lexicon, this would be the sum of gross profits and salaries and wages. Hence GVA is calculated for some sector based on this number. The sectors covered which use GVA as the indicator are manufacturing (organised which is 75 percent of total), hotels and restaurants, real estate, mining, etc. These are proxies used for calculating GVA for these sectors, and this is probably the right way to go about it.[caption id="attachment_4288537" align="alignleft" width="380"]

Representational image. PTI[/caption]Now, where is the problem? To get hold of data is an extremely daunting task as such information does not flow easily. If one looks at all the commercial well-established databases on corporate results, the coverage is normally not more than 50,000 companies. Further, if one wants to analyse data for, say, 4 quarters, which is required for calculating GVA growth, the sample becomes much smaller and would get limited to 5000. But even though this number is small it is a good indicator for the overall sector as all the leading companies are included. In fact, the 80-20 rule holds here where 80 percent of GVA comes from 20 percent of the top companies in this set of 5000.In the revised methodology, the CSO chose the database with MCA which has around 5 lakhs companies. In the earlier scenario, the CSO relied on data from RBI which was used in its studies where the sample never exceeded 2,500. This jump from 2,500 to 5 lakhs was significant and fraught with doubt. The question always was that which were these companies whose results never appeared in the normal course to the public? But as mentioned earlier, these companies do not matter as once we move from 5,000 to 5 lakhs, the size of the company would be too small to influence the overall GVA. In fact, it was more likely that they would be making losses and the GVA would be negative.The NSSO has now pointed out that over a third of these companies are missing from the calculation now. This has raised controversy once again on the veracity of data. Again it should be reiterated that this was bound to happen and would not have actually created this kind of stir but for the political overtones. The CSO has an agreed and admissible approach to tackle this issue which is to use proxies of paid up capital to derive the output. This is now problematic. If we assume that every company with a paid up capital of Rs 1 core had a turnover of say Rs 5 crore, we could be pushing up the GDP growth substantially – which is a legitimate argument. This appears sharp because these smaller companies would tend to be in the SME segment which was affected quite perniciously by demonetisation and GST. The employment numbers here have decreased (an impressionistic statement as there is no reliable data on this variable). Yet the GVA has increased as per the CSO data.There is definitely a need to redefine proxies used in calculating GVA as the present system while being transparently defined may not be appropriate. As 5 lakhs companies are considered to be organised this number does matter. It must be remembered that for the unorganised segment of this sector, the IIP growth rate is used as a proxy. In fact, if the use of proxies based on paid-up capital had an upper bias then it explains well the divergence which is often seen in the growth in IIP and GVA in manufacturing. GVA growth this year is to be 8.1 percent and it is very unlikely that IIP growth will even reach 5 percent.Therefore, maybe when the series is being revised with a new base year, more representative proxies must be used so as to ensure consistency in data. Also as all other related variables like employment, tax collections etc. are linked finally with GDP numbers, the data set will always be examined more than closely for faults. The goal should be to reduce them.(The writer is chief economist, CARE Ratings)