Passing the tax smell test
Subramanian’s growth claims would imply a strange boost in tax to GDP ratio.
Former Chief Economic Advisor Arvind Subramanian has released a working paper wherein he claims that the actual growth rate of the Indian economy has been overestimated by around 2.5 percentage points from 2011-12 to 2016-17: He has summarised his argument in these pages (‘India’s GDP growth: New evidence for fresh beginnings’, IE, June 11). Instead of the official growth rate of around 7 per cent, actual growth was between 3.5 per cent and 5.5 per cent with a high statistical “confidence interval”.
To arrive at these numbers, he looks at the correlation of growth with some “real indicators” such as two-wheeler and commercial vehicle sales, electricity, inflation-adjusted credit off-take, industrial indices and so on across different periods. He concludes that the economy in FY17 was “overstated by about 9-21 per cent” because of slower growth.
Similar real indicators have been used on the Chinese economy. Therefore, such an analysis on India, albeit a preliminary one, should be taken seriously and must be used as a springboard for further discussion. Moreover, the years under consideration cover both governments almost equally so this should not be taken as a partisan critique.
However, the critique is not without its flaws. Surprisingly, it does “not use tax indicators because of the major changes in direct and indirect taxes in the post-2011 period which render the tax-to-GDP relationship different and unstable, and hence make the indicators unreliable proxies for GDP growth.” However, GST (indirect tax reform) came into effect on July 1, 2017 (or FY18 while our analysis stops at FY17) and no radical direct tax overhaul has been implemented since 2011.
Why could this omission be important? Because, the total tax to GDP ratio (“ratio” hereafter) would be unusually higher if the GDP since FY12 was to be reduced by 2.5 per cent every year (cumulatively). If we focus only on the central kitty to begin with, the ratio would be more than 13 per cent in FY17 based on the Harvard paper but based on official estimates it was a bit more than 11 per cent.
The official number is closer to the extrapolated trend from FY02 to FY11: This is important as over time one does expect a secular though gradual increase in the ratio as an emerging economy grows and formalises. Yet, if the economy grew at half the rate of earlier years — one should ceteris paribus — expect a significant fall in the ratio as tax buoyancy is sensitive to growth. But based on the Subramanian GDP estimates, our tax to GDP ratio has even crossed the records of the FY08 boom. Remember, taxes are real cash and no one I know is saying even tax figures have been fudged for around a decade.
As central tax numbers above contain a portion that is devolved to the states and since devolution changes due to finance commissions, etc, I conducted a similar exercise with total central and state taxes just to be sure. Again, the same results. The official ratio for FY12 to FY17 is in line with historical estimates, and is in fact a bit below the extrapolated trend based on FY02 to FY11. Subramanian’s lower estimates for GDP would mean a 20 per cent ratio whereas the official is at 17 per cent, the trend suggests 18 per cent, and even the FY08 peak had not crossed 18 per cent. If growth really was so slow for around the first half of this decade, the ratio should be much lower.
None of this is, of course, the final word on the GDP debate. I am just raising some pertinent doubts about other valid doubts — even the Subramanian paper acknowledges the need to double check its thesis by cross verifying production and consumption numbers. However, the big question remains: How does one square the circle of the official numbers suggesting stable and high growth for almost two decades, with the second half, roughly speaking, seeing a slowdown in many real non-government indicators? One hypothesis is an increase in efficiency or total factor productivity (TFP) over and above hard inputs such as capital. This could be due to technology and/or a more literate workforce as well as other factors.
We are selling fewer cars because thanks to Ola and Uber the same number of cars are being used more aggressively instead of lying idle. Blue collar workers have opportunities to work as “delivery boys” almost 24×7 as opposed to jobs where there was often large, though intermittent downtime. The spread of cheap smartphones and the internet is forcing middlemen everywhere to “shape up or ship out”. Blue-collar real wages are up and often higher than entry-level white collar post-tax salaries. Farm mechanisation is proceeding despite agrarian distress. Some of it is caught in the numbers — not all of it, yet. And even if growth was overestimated, the extent suggested (150-350 basis points) does seem very high.
TFP is almost always measured as a “residual” — so I cannot further quantify my hypothesis here, but the tax to GDP “smell test” is very real and needs to be answered if the revisionist GDP estimates are to be considered genuinely robust.