How on earth did India come up with these GDP numbers?

So maybe demonetisation never really happened. Maybe it was all a bad dream: the late evening announcement, the subsequent cash crunch, the regulatory chaos, the deaths because people could not get medical treatment with old notes. Maybe the reporters who described all the job losses and migrant workers forced to go back home and farmers unable to get their sowing done in time and so on were all affected by a mirage.

Maybe those who conducted surveys and found massive drops in sales, in consumer spending, in livelihoods of informal workers and self-employed people were similarly deluded. And of course, all the economists who predicted significant declines in economic activity as a result of this drastic measure were clearly caught in this mayajaal as well.

This is certainly what the Central Statistics Office’s latest GDP estimates for the October-December quarter of 2016 would suggest. The stately progress of the Indian economy appears to have been completely unaffected by demonetisation, if this data is to be believed. In the third quarter of the year, GDP is estimated to have grown at 7%, bringing the growth estimate for the full financial year to 7.1% – exactly what was predicted in the CSO’s advance estimate, which explicitly did not factor in demonetisation.

Effectively, what the CSO’s statisticians are telling us is that demonetisation had no impact on the economy, and the trajectory of economic activity in the quarter in which it occurred continued exactly the same as it would have done anyway. The CSO’s estimates throw up some real surprises.

How could this happen?

It was widely expected that agriculture would perform better than before, largely because of an excellent monsoon following upon a succession of near-drought years, but in fact the improvement, at least in terms of agricultural GDP, has been somewhat less than expected, at only 4.4% (compared to the same quarter in the previous year).

However, manufacturing, contrary to all expectations, grew 7.7% in that quarter in terms of constant prices, compared to the same quarter of the previous year, and by 10.3% in current price terms. This is definitely unexpected, and runs counter to all the evidence of depressed demand, small factories closing, workers losing not just daily wages but even their jobs altogether, and a significant hit especially on small-scale and informal manufactured goods producers.

What’s fuelling it?

This conundrum could well be solved by looking more closely into what is driving this estimate.

Ever since the CSO moved to its new method of calculating manufacturing GDP, there have been concerns that this could inflate manufacturing output. This new method is based on gross value-added data from company reports produced by the ministry of company affairs (plus taxes, minus subsidies) rather than on the index of industrial production, as it was previously.

Matters are even worse for these quick estimates because they include data only for listed companies. This obviously excludes all unregistered companies and informal manufacturing producers, but it also excludes a substantial number of registered firms that are not listed on the stock market. So all we can say is that the group of companies considered for providing these estimates may not be representative of the entire manufacturing sector.

Further, the differences between these national income estimates and the estimates of industrial production are stark. The IIP data show that manufacturing output in the October to December 2016 period increased by only 0.2% over the same period in the previous year – in other words, it was essentially flat. How this could translate into more than a 10% increase in nominal output is indeed a mystery.

Obviously, we ordinary mortals are not privy to data used by the CSO, or even know which manufacturing companies it has used to generate these higher output figures. But the other surprising thing is that even company results published for the third quarter do not appear to support these buoyant conclusions.

According to the Dalal Street Investment Journal, year-on-year sales of 4,220 companies in the third quarter of this year increased by only 5.74% – significantly lower than the 10.3% increase in nominal production declared by the CSO. And this group of companies includes oil and infrastructure companies that were presumably unaffected by the note ban and instead benefited from the rising energy prices and other changes in the economic environment. So they would push up the overall sales figures.

‘Channel stuffing’

One possibility that has been noted by Pronab Sen, the government’s former chief statistician, is that “channel stuffing” made the difference. In other words, in this quarter, due to demonetisation, manufacturers sold their supplies to trade channels like wholesalers and retailers, and were willing to take demonetised notes as payment, which they then used to pay off the excise duties and other taxes.

This is an ingenious take and could well be part of the explanation. But it is hard to believe that the distribution channels would be willing to keep purchasing this output and keep building up their stocks in a period when sales were clearly falling because of the note ban, and there was no immediate reason to expect a quick recovery of sales. In any case, the extent to which this actually operated will not be known for a while – indeed, given the opacity of the current process of working out these estimates, it may never be known.

Another two very surprising results from the CSO quick estimates relate to investment and private consumption. Since production estimates themselves are open to question, this makes them even less reliable.

Gross fixed capital formation, which had been declining since the middle of last year, has apparently revived and suddenly increased, despite the note ban! The data suggest that investment increased by 3.5% over the same quarter of the previous year. This is certainly unexpected, and also sits uneasily with IIP data on production of capital goods, which actually declined by 7.8% over the same period.

Similarly, private consumption expenditure supposedly increased by more than 10% in constant price terms over this period, once again hard to square with so much other information about declining demand and declining sales of both durable and non-durable consumer goods.

It is true that the expenditure side of the advance estimates should never be taken too seriously, because they are not derived independently but are computed from production estimates. And since the production estimates themselves are clearly open to question, this makes them even less reliable. But there is no doubt that these GDP estimates are politically a relief for the ruling government, and also come at a convenient time, allowing them to claim that the demonetisation, if not a success, at least did not cost the economy any lost income.

Many observers have noted that we should wait for the revised estimates to get a real sense of what actually happened to economic activity in this period. But that very possibility raises a disturbing thought. The revised estimates typically come out later, and attract much less attention than advance estimates that determine expectations about the economy.

It is possible to think of a scenario whereby the revised estimates get pushed down substantially without creating too much of a hue and cry, because that very downward revision in turn creates a lower base that enable a higher projection of growth for the next round of advance estimates! This would be post-truth taken to a whole new level.

This article first appeared on Quartz.