By Abhijit Banerjee & Esther Duflo
It was the best of times, it was the worst of times, it was the age of breakneck growth, it was the age of inevitable slowdowns, the economy had finally been unshackled, the economy had nothing to lose except its chains —in short, a time of great confusion.
Were the past five years a good time for the Indian economy? Or a bad time? Consider just the financial year 2013-14. At the time it was considered the nadir of the UPA raj. Growth was 4.7 per cent, courtesy the indecisive policies of poor Manmohan Singh. Voters came to believe that India needed a reset and Narendra Modi stepped up to deliver it.
In early 2015, something strange happened. The statistics ministry put out revised GDP numbers, and lo and behold, 2014, far from being in the doldrums, had a growth rate of 6.9 per cent, one of the highest in the world for that year. Singh’s legacy was rescued, perhaps a year too late, courtesy the NDA. There were, inevitably, many sceptics who pointed to the low capacity utilisation in manufacturing, slowing orders from the bigger firms, drooping corporate profits and falling investment.
The debate got even more rancorous in 2016, when the 10-plus per cent claimed growth in manufacturing output seemed to fly in the face of every other piece of evidence, prompting claims that the GDP data were made up or a ‘hoax’. We think that it is, therefore, worth emphasising that a lot of this disagreement is rooted in the nature of GDP measurement.
The first problem — called, somewhat unhelpfully, the problem of index numbers — affects all countries and is simple. In 1900, many people owned horses but no one had a cellphone. The reverse is true today. Measured by the number of cellphones the average person can buy, the Indian population is almost unfathomably richer today. But who these days can afford to keep a horse? So, are they actually richer? Or poorer?
Or to take a less extreme example. Between 2002 and 2012 the prices of grains in India grew twice as fast as that of ‘other foods’ (milk, eggs, meat, biscuits, etc.). Our nation’s purchasing power may well have gone up in terms of ‘other foods’, but down in terms of grains.
Our estimate of growth depends on the choice of the benchmark commodities. The standard practice is to use a bundle of what is commonly consumed, which inevitably changes over time, requiring updates every few years. Whenever this adjustment is carried out (every five years or so), the current GDP tends to be revised upwards, because it is valued more in terms of commodities that are currently popular, precisely because they are cheap.
A part of what happened in 2013-14 was this. In 2015 and 2016, there was another problem. The data reported to the statistics ministry is in terms of value of sales and the associated costs, both of which are subject to inflation. Inflation creates an illusory sense of progress, which needs to be purged before we can know the growth rate.
The trouble is that the Indian statistical system, unlike their counterparts in the developed world, only collects output prices and not input prices, and uses output price inflation to purge inflation from both sales, where it is appropriate, and costs, where it is not. When, as in 2015 and 2016, input prices grew much more slowly than output prices, mainly because of the global oil price collapse, growth was exaggerated, especially in manufacturing.
Essentially because the assumed input prices were too high, the firms looked like they were using less inputs to produce the same output than they were actually, making them look more productive. Pranjul Bhandari at HSBC computed that this might have added 4.5 per cent to the manufacturing growth rate in 2016. By the same logic, when the input price lag is reversed, as seems to be the case this year, growth may be underestimated.
This is a fixable problem — it just needs some more effort (and money) for collecting input prices. But there is another problem, more specific to developing countries like India that will be with us for some time. About 30 per cent of our GDP is produced in the informal industrial sector.
Which means that they are in every nook and cranny, often trying to fly under the radar. How is the statistics ministry supposed to figure out how much they produced last quarter?
National Sample Survey
The main source of information about the scope of these firms is the National Sample Survey (NSS), which, every few years, asks households how many hours they spent working in informal businesses. This is combined with the average hourly productivity in various informal trades from the Annual Survey of Industries, which captures a sample of these firms, to get GDP of the informal sector.
Unfortunately, this NSS data is collected every five years. For other years, the GDP of the informal sector is mostly made up by assuming that it grows apace with the rest of the economy.
One implication of this way of computing GDP is that when there is a shock primarily to the informal sector, like demonetisation, the GDP numbers will not pick it up immediately. When Modi triumphantly announced that demonetisation had no negative impact on growth, he may not have realised that this was a pure artifact of the way we compute GDP.
Although the statistics will eventually detect any ripple effect on the formal sector (indeed, this seems to be happening now), the true impact on the informal sector will probably never show up in GDP data. By the time the numbers are updated the unfortunate episode would have hopefully passed.
A second implication is that at a time like now, when the formal sector is expanding at the expense of the informal sector — tax rolls and participation in provident fund are growing fast — the GDP estimates will tend to jump around quite a bit. When more firms join the formal sector, GDP will tend to jump up because formal sector output goes up, but the informal sector output is not properly adjusted for the loss of these firms.
Of course, when the new informal sector numbers finally get collected, the loss of these firms will have an impact, and we will see a downward revision in the GDP numbers. Perhaps, this is what happened when the 2013-14 numbers eventually got downgraded again, just a little bit, to 6.5 per cent.
The final, and most important point, is that it makes no sense to get too emotionally involved with individual GDP numbers. Did India really grow faster than China for a couple of quarters? Or was that just a statistical blip? This is a question we will never satisfactorily answer.
What is measured as GDP is, at best, the correct application of the currently accepted methodology to the available data — a summary of what is known now, not some absolute truth. At the same time, it is worth being patient with the poor folks in the statistics ministry who have to wrestle with all these imperfections and come up with something — at least until the country is willing to spend the money to upgrade its statistical apparatus to match its aspirations.
(The writers are Professors of Economics, Massachusetts Institute of Technology, US)
DISCLAIMER : Views expressed above are the author’s own.
Recommended By Colombia