As prices normalise, GDP overestimation will narrow
It’s GDP season again. India’s “new” GDP series has kept us on our toes, springing a new surprise with every release. Over the last several quarters, the statistics agency, the ministry of statistics and programme implementation, overestimated economic growth; however, as prices normalise over the next few quarters, the overestimation will likely narrow. Users of the data should not confuse it with renewed weakness on the ground.
For most sectors, the statistics agency starts off by calculating “gross value added” (GVA), a measure of economic growth, in nominal terms. Put simply: GVA=economic output-input.
Subsequently it deflates nominal GVA to get an estimate of real growth. And this is where the “trouble” starts. Instead of double deflation, it resorts to single deflation. Instead of deflating output and input with their respective prices, the statistics agency deflates both with a common output deflator.
“Single” instead of “double” deflation was overstating GDP prints
This practice did not cause much problem when input and output prices were moving in tandem. However, over the last few quarters, as global commodity prices plummeted, input prices in India fell rapidly, much more than the output prices. The rising difference between CPI (a proxy for output prices) and WPI (a proxy for input prices) inflation is a reflection of this phenomenon, as is the difference between our estimates of WPI output and input prices. The practice of single deflation at a time when output inflation was trending well above input inflation is likely to have exaggerated real GVA growth.
It is tempting to suspect that much of this exaggeration came from the manufacturing sector. After all, advance estimates peg manufacturing GVA growth at a significant 9.5% in FY16, (much higher than the 5.5% growth of FY15), while other volume measures of manufacturing output stagnated or even deteriorated over the same period. To get more conviction on our suspicion, a quantification of the likely overestimation in manufacturing growth has been attempted.
The exercise begins with an attempt to get a handle on input prices. This data is not readily available and needs to be calculated,by breaking down WPI inflation into its input constituents.
Next, manufacturing growth estimates are worked out under scenarios of single and double deflation, respectively. The exercise shows that for FY15, manufacturing growth may have been exaggerated by 120bp. So, instead of the 5.5% y-o-y reported growth in manufacturing, actual growth on the ground is likely to have been closer to 4.3% y-o-y.
Unfortunately, the breakdown of GVA into output and input is not yet available for FY16, making the calculation of the likely overestimation difficult. Thankfully, past data can be used to calculate the sensitivity between relative prices and growth overestimation. Applying the derived sensitivity to current prices suggests that manufacturing growth was exaggerated by 450bp in FY16. This means that instead of the 9.5% y-o-y reported growth in manufacturing, actual growth on the ground is likely to have been closer to 5% y-o-y. Translating this to overall GDP growth suggests an overestimation of 80bp; not surprising, given the divergence between output and input prices reached a historic high that year, amplifying the error of using single instead of double deflation
This could now reverse (hello from the other side…)
However, all of this could well be a story of the past. Going ahead, thanks largely to a low base, input price inflation could climb, even if commodity prices do not rise dramatically. Already, the gap between CPI and WPI inflation is closing, as is the gap between WPI output and input prices.
Forecasts suggest that the gap in relative prices is likely to go back to normal (long-term average) by mid-2017. This implies that the 450bp overestimation in manufacturing growth should gradually narrow over the next six quarters.
This is evident from past experience. After the 2008 global financial crisis (GFC), global commodity prices plummeted for a few quarters, giving rise to a growing divergence between output and input prices. The practice of single deflation threw up an exaggerated manufacturing growth of 11.3% y-o-y in FY10, only to “crash” to -3.1% y-o-y two years down the line.
This time around, as manufacturing GDP begins to normalise and track the action on the ground more accurately, the moderation in growth prints should not be interpreted as renewed weakness.
What should we believe, where should we look?
So, as India’s new GDP data normalises from the large global commodities supply shock, what should one believe to be the actual growth on the ground? Here, users will have to resort to volume indicators. Thankfully, there are a ton of them, ranging from industrial production index, PMI manufacturing, car sales and transport services.
The statistics agency could also make life easier by releasing a volume growth index every quarter. It already uses various volume indicators to estimate quarterly GVA, so releasing it as a consolidated index in a timely fashion may not be too much extra effort.
Future of India’s growth
As the overestimation narrows, will there be staggering fall in overall GDP prints? Not necessarily. While there is a likelihood to see manufacturing growth soften over the next six quarters, a similar quantum of rise in agricultural growth on the back of normal rains could offset the fall. After all, the share of manufacturing and agriculture in India’s GDP is not too different (17% and 15%, respectively).
All said, the headline GDP growth print will be flat at 7.4% y-o-y in FY17. While normalising manufacturing growth and banking sector stress are likely to be a drag, a bounce-back in agriculture alongside the ongoing consumption revival could just about offset the slowdown.
Consensus estimates that are pegging India’s GDP growth at much higher levels over the next year could be in for some disappointment.
The author is chief India economist, HSBC. Column is excerpted from HSBC note: India’s GDP to auto-correct