We have pointed out, here and here, that the standard narrative about 2016-17 Indian GDP, that its growth was unchanged in Q3 but fell in Q4, is based on a misreading of the data. An application of a standard seasonal adjustment method to the raw figures shows that GDP growth actually declined in Q3 and recovered weakly in Q4.
One could ask the question, does this even matter? The bottom line is to observe that GDP growth in 2016-17 was about 7%, the trend rate for the last decade. Or, you could observe that there was a perceptible slowing in the second half of 2016-17. Why split hairs about differences between Q3 and Q4 and apparently esoteric statistical procedures when the conclusion is entirely clear?
There is a reason why getting this right is important. The likely effect of the note ban on the economy has been a highly contested issue. Although most economists expected a sizable decline in GDP growth as a result of the note ban, a few economists, and some government officials, expected that the note ban would have barely any negative impact. Some even foreccast that it would have a paradoxical positive impact. The misreading of the GDP data that resulted in the idea that growth did not change in Q3 and fell in Q4 has fueled the belief that the no-negative-impact forecast has turned out right. It has also allowed government representatives to make spurious claims that it was not the note ban, but global economic weakness that caused the GDP slowdown. The correct reading of the data shows that the mainstream forecast of a decline in growth in Q3 and a weak rebound in Q4 is exactly what happened.
There is another reason for trying to correct this misreading. The Indian media and the public seem ready to swallow the rather absurd story that GDP growth did not decline in Q3, when the note ban threw the economy into considerable turmoil, but then GDP growth fell in Q4 when the situation went back more or less to normal. This unlikely account has led to much puzzlement and even allegations of statistical malfeasance, all of which can be cleared up through a simple and well known statistical procedure.