Differing accounting practices employed by the govt and FCI, at a time when past years’ subsidy arrears are being paid, could be distorting GDP estimates
Furthermore, a negative growth clip in Q4 does not sit well with the high frequency indicators, which continue to tick higher.
By Pranjul Bhandari, Aayushi Chaudhary & Priya Mehrishi
A puzzle of sorts has arisen (again) in India’s GDP estimates. The statistics office is estimating that after expanding 0.4% yoy in Q3, growth will go back into contraction mode in Q4, contracting 1.1% yoy.
This should have implications for investors, who are broadly expecting that India is out of the woods and will grow positively over the next few quarters. Furthermore, a negative growth clip in Q4 does not sit well with the high frequency indicators, which continue to tick higher.
Further investigation reveals that there are some methodological issues in the calculation of growth, which are perhaps not just distorting current growth estimates, but could even leave an imprint on growth data for the next few years. It could even have distorted growth estimates in the past few years.
1. Which data points alerted us to the problem? Q3FY21 GDP came in weaker than expected, but Q3 GVA was not as weak. There was a surprisingly sharp difference between the two measures of growth (0.4% GDP growth versus 1% y-o-y GVA growth). Note that the difference between GDP and GVA growth is generally lower, at 0.2ppt (10-year average).
This sharp difference also extended to the FY21 full-year advance estimate of the statistics office (-8% GDP versus -6.5% GVA growth). Compared to the previous advance estimate which was released in January, GVA was revised up (to -6.5% versus -7% previously), but GDP was revised down (to -8% versus -7.8% previously).
Working out the Q4FY21 estimate from the full-year one suggests that the statistics office expects GDP growth to contract 1.1%, but GVA to expand 2.5%. This got us confused. Is growth ticking up, or sliding down?
2. What explains the divergence between GDP and GVA? We know that GDP = GVA + indirect taxes – subsidies. We also know that indirect taxes grew sharply in Q3 (GST grew +8% y-o-y; central government indirect taxes grew 33% y-o-y). So, for GDP to grow at a much slower pace than GVA, subsidies would have had to grow rather strongly. But why would that be? Because the budget on February 1, made it all too clear that over two years, the government intends to pay off past unpaid dues to FCI, the intermediary for food subsidies. To be precise, the plan is to pay 0.9% of GDP in FY21 and 0.3% of GDP in FY22. Repayment of some of these resulted in bloated subsidy growth, thereby depressing Q3, Q4 and FY21 growth estimates.
3. But shouldn’t the GDP methodology have a way to take care of this? Yes, if every economic agent used the same accounting methodology. Let us explain. For simplicity sake, let us work with two economic agents—the FCI and the central government. If both did “accrual accounting”, we would not have a problem. The FCI would account for subsidies in the year they accrued, and the government would account for them in the same year too. In this situation, GDP would be a better indicator of underlying growth, rather than GVA (because it strips off the subsidy payments, which tend to inflate GVA).
The problem arises when the FCI and the government follow different accounting practices—for instance, if the FCI does “accrual accounting” while the government does “cash accounting”. In this case, there could be a problem, for instance, if the subsidies in FCI’s books accrued (say) last year, but the government only paid up in the current year. To arrive at the GDP number, the statistics office would end up subtracting from current GVA more subsidies than what accrues in the current year. This could lead to an underestimation.
And, indeed, this is what we think is going on. The FCI has clarified in its accounts manual that “all expenses are accounted for on an accrual basis.” It has gone on to explain that “income is recognized when there is reasonable assurance for its realisation and is earned (usually when goods are transferred or services rendered), no matter when cash is received.” On the other hand, we know that the Indian government records expenditures on a cash basis.
4. Were previous-year growth numbers impacted too? Yes, likely. If we are right with our analysis so far, there is a chance that past numbers have been impacted too. The government has been owing money to the FCI over the last several years, and the amount picked up rapidly from FY18.
Over those years, the FCI would have accounted for the subsidies in its books and this would show up in GVA. However, the government did not pay up on time. As such, a smaller cash subsidy amount was deducted from GVA to arrive at GDP, thereby potentially overstating growth.
Indeed, we find that over FY18 and FY19, GDP growth was 0.6ppt higher than GVA growth (versus the 0.2ppt average in the previous five years). True, some of it could be because of other factors like a fall in fuel subsidies or a rise in indirect taxes; but, regardless, we think this deserves investigation.
5. By how much could future growth numbers be impacted? The government plans to repay the dues for subsidies over the next two years, FY21 and FY22. Its FY21 advance estimate for GDP growth is -8%, lower than the GVA advance estimate of -6.5%.
About half of the 1.5ppt wedge, by our calculation, is because of the distortions created by the payment of past subsidy dues, and does not reflect actual weakness in economic activity.
Moreover, it ends up inflating the GDP growth number in FY22, because of a low statistical base. However, some of the base effects could be offset by the balance payment of past subsidy dues (budgeted at 0.3% of GDP in FY22), which could end up depressing GDP for the reasons explained.
We work all this out carefully and find that, on net, the positive base effect impact overshadows the negative payment of subsidy dues impact, leading to GDP growth being overstated by 1ppt in FY22. Finally, the repayment of balance dues in FY22 could impact FY23, again due to low base effects. We calculate that GDP growth may be overestimated by 0.5ppt in FY23.
6. What, then, is the ‘true’ economic growth in India? In normal times, GDP is a more wholesome indicator of economic growth than GVA, because it includes the government as well. But with GDP impacted by payment of previous-year subsidy dues, we think GVA will better reflect economic growth, not just in FY21 but also over FY22 and FY23.
7. What happens to HSBC growth estimates? As explained above, the difference in accounting procedures by the FCI and the government may overstate GDP growth in FY22 and FY23. But it is also possible that the statistics Office corrects this problem over time. In the past, the statistics office has tried to improve its methodology, even if it means making sharp changes to GDP history. We will therefore wait and watch, at least until the next GDP release scheduled for May, when many revisions tend to happen.
For now, we will stick to our previously published forecasts of -6.3% y-o-y GDP growth in FY21, 11.2% in FY22, and 5.8% in FY23, which we think reflect the ‘true’ growth on the ground.
Excerpted from HSBC Global Research, Economics India report (March 4)
Bhandari is chief economist, Chaudhary is economist & Mehrishi an associate, HSBC Securities and Capital Markets (India)