The impact of seasonal changes
Economic statisticians are constantly concerned with deseasonalising effects. This is relatively easy when the effect is predictable. For example, there will be more demand for labour during a harvest season and there will be more consumption during Christmas. But, it is also often required to cope with 'one-off' events, which could range from a disaster like a cyclone or earthquake, to something like an Olympics.
A third type of deseasonalisation is required for festivals, which are not aligned to the solar calendar. The Chinese New Year festivities last for anything up to a fortnight and there are official holidays for several days. Being a lunar festival, this often falls across two solar months. The impact on economic data is tough to handle. The Chinese solved it pragmatically by always clubbing January and February data together. That way, regardless of the actual dates, the festival is always taken into account.
Indian statisticians should consider doing something similar with Durga Puja-Kali Puja or DussehraDiwali, as those festivals are also known. Sometimes both festivals fall in October; sometimes one festival falls in October and the other in November. This period is associated with holidays, which means less industrial production. It is also associated with bonuses and bargain offers, which means higher consumption.
When October is a double-festival month and it's compared to a single festival October in the previous year, the Index of Industrial Production (IIP) tends to dip due to the higher base. However, November in those years tends to show a sharp jump since the base is lower.
In 2015, Durga Puja was in October and Kali Puja in November. In 2016, both festivals were in October. Under normal circumstances, we would expect the base effect to cause a dip in the IIP of October 2016 and a rise in the IIP of November 2016. Circumstances were not normal due to the demonetisation of November 8.
A year-on-year comparison showed October 2016 was down by 1.8 per cent and November 2016 was up by 5.7 per cent. Now, November 2016 had 25 working days compared to 22 in November 2015. That in itself, would explain a difference of 14 per cent. Sugar production had a large contribution to growth as well, since the sugar harvest was up 21 per cent in 2016. Also, insulated cables production rose 185 per cent year-on-year(y-o-y), due to a low-base effect. That was another major contributor to 2016 growth. If the two months are clubbed together Chinese-style, the y-o-y change is 1.7 per cent. Also, the November 2016 index (175.8) is lower than the October 2016 index (178.1), remember October 2016 had both festivals, and hence, was actually depressed.
The cumulative growth for April-November 2016, using an averaged method, compared to AprilNovember 2015, is a negligible 0.37 per cent. Incidentally, there was a sharp pickup in December 2015 when the index hit 184. It is thus, extremely likely that December 2016 will see a sharp IIP decline.
The volatility of the IIP leaves us with few firm conclusions. But, the first market response — a relief rally on apparently strong growth in November — was due to a misreading of the y-o-y change. There has been a negative impact from demonetisation and that effect will probably look stronger when December comes around, because the base month of December 2015 was higher.
The other data coming out of the NovemberDecember period, such as the Purchase Managers' Index (PMI) and high-speed data such as automobile sales, FMCG (fast-moving consumer good) offtake, cement despatches all indicate the impact was fairly severe. The earnings and anecdotal reports from various sectors are more likely to confirm the negative viewpoint.