National income is notoriously hard to estimate, particularly so for economies like India where the informal sector accounts for much activity and employment, making it horrendously difficult to capture in official data. The assumptions and leaps of faith involved in producing the GDP numbers are numerous and enormous. All sorts of estimates must be made of the scale of various activities in agriculture, small-scale industry, and services – often based on employment and wage data that is itself limited, infrequent, and sometimes simply missing. Many different statistical adjustments have to be made to take into account gaps and discrepancies in the available data-sets.

This is even more the case for quarterly data – not only in the infamous “advance estimates” now regularly seized upon for political use, but also in the subsequent revisions and corrections that eventually lead up to the final estimates. Such quick data is even more difficult to generate without taking judgement calls on which estimate to use in what manner and what assumptions about relationships are to be made (say in terms of the relative shares of formal and informal activity).

All this means that producing these numbers is as much an art as a science. Indeed, it can never be a purely technocratic exercise as even slightly different assumptions can result in markedly different numbers.

This being the case, it is downright amazing that all of us remain so spellbound by the periodic release of quarterly data on national income. We celebrate when we become “the fastest-growing economy in the world” by achieving a growth rate half-a-percentage point higher than China’s on the basis of these rapid and often dubious estimates. We similarly get dejected when we are dethroned from that position by another set of data, even though the basic underlying economic processes have not changed at all.

Hold your horses

Obviously, all this reflects the fact that, however flawed, inadequate, and approximate such data might be, it is all we have in terms of hard numbers to interpret the economic reality around us, especially since employment data is simply unavailable.

The data just released by the Central Statistical Office – advance estimates of GDP in the fourth quarter of 2016-’17 and for the entire year – is, therefore, being seized upon to confirm or deny economic trends that appear obvious to some. The context is that of the drastic demonetisation of November, which surprisingly had no (or hardly any) impact on GDP in the third quarter of the year when it occurred. The estimates for the final quarter, however, indicate a deceleration, interpreted by many as delayed effect of the demonetisation.

The government, expectedly, is downplaying this assessment. As it happens, Finance Minister Arun Jaitley is absolutely right when he says that the slowdown in growth had begun well before the demonetisation debacle. From the peak of 9.21% growth (year on year) achieved in the last quarter of 2015-’16, GDP has declined continuously in every quarter since.

Furthermore, even the growth of 6.12% in the last quarter has been shored up by the performance of agriculture (mostly a gift of the rain gods) and by huge increases in public administration and defence, which contributed 36% to growth in that quarter. In terms of gross value added, rather than GDP (which excludes taxes net of subsidies), the slowdown has been even more marked: from 8.7% in the fourth quarter of 2015-’16 to only 5.6% in the fourth quarter of 2016-’17.

The folly of underestimation

But it could be argued that even these numbers are actually overstating the extent of real economic growth when all parts of the economy are included. Certainly, these figures are simply not in accordance with so much other economic information. Industrial production, even on the basis of the new index released a short while ago, has been mostly flat.

Much has been made of the increase in the index of industrial production in March, but even that made it only 2.3% higher than the level of the same month of the previous year. Credit disbursements are at an all-time low, reflecting not only the disruptions to banks caused by demonetisation, but also the reticence among prime borrowers to take on more credit and the unwillingness of banks to lend to those who remain rationed out of the system. Investment rates fell below 30% of GDP in the the second quarter of 2016-’17 and have remained low since then. So the bigger question could be, how come there is any growth at all?

The latest data is unlikely to be able to capture what exactly is going on in the Indian economy, precisely because we do not have any way of gauging what exactly has happened to informal activity. It is not just that the economy was dealt this huge blow (demonetisation), which seems to have accomplished little beyond the political aggrandisement of the leader. It is also that our statistical system is not equipped to capture the full effect of this on peoples’ incomes and living standards, precisely because of the size of the informal sector.

This brings up the point made at the beginning, that much of GDP calculation is not purely “technocratic” but relies on judgements and assumptions.

One critical assumption is of a stable value added per worker and a stable relationship between formal and informal activity. This is necessary because typically with respect to informal activity, there is often no way at all of getting the full picture. So, especially with quick estimates, the only way to get some idea of the aggregate is to assume that informal activity bears some relation to formal activity. Since the methodology used by the Central Statistics Office for this is opaque, we do not know how exactly they decide on this, but it is safe to assume that it would be a reasonably consistent and stable relationship that is assumed.

But what if something happens to radically disrupt such a relationship? For instance, the demonetisation of 86% of the value of the currency in circulation followed by replenishment so slow that that even six months later only 80% of that value has been replaced. What if its effects disproportionately fall on the informal economy, which is hugely if not completely dependent on cash? And what if that in turn affects production, supply chains, employment, and demand? Clearly, the standard assumptions would not at all be justified in calculating aggregate income.

So, quite possibly, the real impact on the economy and the actual degree of deceleration of economic activity are much greater. Unfortunately, as long as our system of national accounting does not clarify these important points, we will remain in the position of only knowing what we do not know. There are others, of course, who do not even know what they don’t know – and sadly, many of them are the ones deciding policy in India today.

This article first appeared on Quartz.