Why effect of demonetisation on GDP did not show up now
Economists have been scratching their heads to explain quarterly GDP (qGDP) growth figures for Q3 FY17 released by the government that are much higher than expected in the wake of demonetisation. Most experts had expected a growth rate of 6.5 per cent, and have been surprised by a figure of 7 per cent.
Does this mean that demonetisation has had little or no impact on the economy? Not quite.
While a lot of economists have simply thrown their hands up and said the data is simply unreliable and untrustworthy, I don’t believe questioning the government and insinuating some sort of China-esque black-box style politically motivated cooking up of numbers is the correct way to go. Instead, the key lies in understanding how qGDP growth rates are constructed and what they actually mean.
The first thing to understand is that they are calculated quarter on quarter. A figure of 7 per cent indicates that GDP was higher by 7 per cent in Q3 FY17 when compared with Q3 FY16; it does not serve as a comparison between this quarter and the last. Incidentally, the government also revised downward the qGDP growth estimates for Q3 FY16 by 0.3 percentage points. Most estimates took into account the earlier higher Q3 FY16 figure. When compared to the earlier Q3 FY16 figure, the present growth would read 6.7 per cent, a lot closer to what experts had expected.
The second thing to understand is agricultural production operates with a 3-6 month time lag. If you sow something in October-November, you harvest it in March-April (rabi crop), and if you sow something in June-July, you harvest it in September-October (kharif crop). The crop harvested during Q3 FY17 was the kharif crop planted in June-July 2016, which turned out to be a bumper crop because of the first good monsoon in three years. This caused the jump in agricultural output. The true effects of demonetisation on agriculture can only be gauged when the rabi crop, that was being planted when demonetisation was announced, is harvested in Q4 FY17 and Q1 FY18.
The third thing, and this is by far the most important, is that very little actual data is available on production as recent as Q3 FY17. These figures are only estimates calculated mostly on trends and assumptions.
If the RBI still doesn’t know precisely how many new notes it itself has printed, then how can you expect the government to accurately know the number and value of nails produced all over India by a vast network of small enterprises, many operating in far-flung rural areas?
A quick reading of the Central Statistical Office’s methodology document on estimating quarterly GDP shows that most sector GDP estimates are based either on past trends, or on other government estimates, which themselves are based on past trends. The little actual data that is available relates mostly to that part of the organised sector that is fully digitised, which, naturally, would have felt the least impact of demonetisation. For instance:
1) In agriculture, ministry of agriculture estimates are used.
2) In manufacturing, the figure is arrived at by inflating last year’s figure using the Index of Industrial Production (IIP), which measures how much production has increased or decreased year on year by looking at production in a sample of factories in each industry that account for a large share of total production. In other words, the IIP looks only at large formal factories and ignores the informal sector completely, where the impact of demonetisation would have been felt the most.
3) In real estate, ownership of dwellings and business services, the informal sector part is based only on trends in past data.
4) In other services, the private sector part is purely based on past trends.
5) In mining, railways, banking, electricity and some other sectors, actual data is used as it is readily available.
In short, data on the parts of the economy that are most susceptible to a move like demonetisation is not yet available. These overlap mostly with the informal sector, where data availability is limited and true production is measurable only after a monumental effort from enumerators, which leads to a large time lag.
Estimates are made based mainly on past trends. These estimates are routinely revised even two years after they are first made as more data is collected and becomes available.
This methodology is the best we can do given data availability, and is accurate to capture gradual, incremental changes. It is very poor, however, at capturing the results of a one-off shock, resulting in an understatement either way.
If you’re really interested in the impact of demonetisation, revisit this figure after the revisions over the next two years, once data on the informal sector is available. The answer right now is – it’s too early to tell.