What does the NSS data imply for GDP estimates?

Just as Judas betrayed Jesus, it may well be the case that the new National Sample Survey (NSS) consumption data betrays the measurement problems in the National Income Accounts (NIA). Consider how.

Recently, there has been considerable discussion of the wide gap in the NIA between private consumption and gross domestic product (GDP) growth over the past few quarters. According to the NIA, the former is weak even as GDP growth is strong. Some people have seen this as a goo­d sign, indicating that India’s growth is at long last being fuelled by investment, rather than consumption. Other people, such as oursel­ves, have taken a different view: That the gap is a sign of serious measure­me­nt problems.

Why do we think that there are measurement problems? To answer this question, consider not just the last few quarters but the broad trends of the past two decades. Recall that India’s NIA data after 2011 were subject to a new methodology. Recall, too, that this methodology was used to re-estimate (back-cast) NIA estimates for the period 2004-05 to 2011-12. These new figures were very different from the old ones — and were very difficult to understand.

One of the biggest puzzles related to consumption. The old numbers had shown that consumption had exploded during the boom of 2004-11 and decelerated thereafter. But the new numbers showed the opposite, namely that consumption accelerated after the boom years ended, growing strongly all the way through 2022. This seemed odd. Apparently, the successive shocks of demonetization, GST and Covid took no toll on last decade’s consumption, even as they damaged the GDP.

Of course, it is possible for households to step up their consumption, even as their incomes are slowing. But this would be very strange behaviour, at odds with the way people normally operate. It would also be inconsistent with other available macro indicators, such as exports, imports,investment, and credit, all of which slowed sharply in the post-2011 period. Even the available proxy for consumption, the Index of Industrial Production for consumer goods, decelerated after 2011.

Rather than strange household behaviour, the more likely explanation is that the Central Statistical Office (CSO) has encountered problems measuring consumption under the new methodology. If so, it would hardly be surprising. After all, it is inherently very difficult to monitor the spending habits of 1½ billion people.

Of course, the CSO knows that private consumption is the weak link in NIA measurement. That is why it measures GDP from the production side rather than the demand side. And that is why it regularly shows large “discrepancies” in the NIA — an acknowledgment that it cannot reconcile the difference between its production and demand-side estimates. For all these reasons, the periodic household surveys by the NSS are truly useful, as they can help the CSO get a better handle on consumption trends.

What can we learn from the latest survey? Before we answer this question, we need to recognise the limitations of comparing NSS figures with the ones from the NIA. There are and always will be level differences between the two, notably because the NSS survey seriously ignores or under-counts consumption of the rich. For the same reason, even growth rates are typically not the same.

That said, it is reasonable to compare directional trends, on the assumption that the distortions caused by undercounting don’t really change from survey to survey. Figure 1 shows that indeed the NSS and NIA consumption growth rates move almost in parallel from 1983 to 2011, under the old NIA methodology. Figure 1 adds the private consumption figures estimated by the Sudipto Mundle commission, which suggested that the old methodology had under-estimated consumption and GDP growth during 2004-11. It tracks the NSS even better than the old NIA estimates.

How does the new methodology compare? The answer is given by Figure 2, which plots the NSS and NIA consumption figures for the period 2004-2022. The difference with Figure 1 is striking:

The NSS growth figures no longer track the counterpart NIA figures. For the period 2004-2011, the NSS shows an acceleration in consumption growth (from -0.4 per cent to 4.6 per cent), while the NIA numbers show a much slower growth acceleration (from 2.9 per cent to 4.5 per cent). Even more dramatically, post-2011 the two growth numbers don’t even move in the same direction. NSS consumption growth decelerates sharply (from 4.6 per cent to 2.8 per cent) while NIA consumption growth keeps accelerating (from 4.5 per cent to 4.8 per cent). Put simply, the NSS data contradicts the NIA data. That means there are two possibilities: Either the NSS estimates are better than the NIA estimates; or the opposite. Surely, both sets of estimates have their flaws, which means that different analysts will have different opinions. We favor the former view. That is, we believe that private consumption did indeed slow down after 2011. We believe this because it is the only view consistent with logic, with the slowdown in the other major indicators — and now with the NSS data.

In some interpretations of the Christian Bible, Judas has tragic-heroic status because he took on the burden of betraying Jesus, which then led to the crucifixion and the miracle of the resurrection.

Perhaps we should be similarly grateful for the NSS numbers — not just because they shed light on poverty developments, but also because they allow us to see more clearly the NIA numbers for what they are. And perhaps, if we see that clearly, we can mobilise efforts to improve them.

Josh Felman is Principal, JH Consulting; Arvind Subramanian is Senior Fellow, Peterson Institute for International Economics