This appeared in Financial Express today.
In India, many people look at year-on-year changes to track the state of the economy. This indicator has important weaknesses. It is the moving average of the change seen in the latest twelve months and is hence a sluggish indicator of the changes in the economy. In order to monitor current developments in the economy, it is preferable to look at month-on-month changes.
However, month on month changes are distorted by seasonal fluctuation. The solution lies in seasonal adjustment. The seasonally adjusted month on month changes provide more timely information about the state of the economy. Internationally, the standard procedure for examining and monitoring economic series uses seasonal adjustment.
The figure shows the familiar time-series of year-on-year growth of IIP. This shows that output in October 2009 was 10.3% bigger than the level of October 2008. This seems reassuring.
Far more informative is the time-series of month-on-month changes. Each of these values is the annualised month-on-month change in the seasonally adjusted IIP. The term applied is `SAAR change' which stands for the Seasonally Adjusted Annualised Rate of change. This shows an unhappy value of -5.12% for October 2009 (when compared with September 2009). The key strength of this approach is that we are discussing the change from September 2009 to October 2009, instead of the 12 changes from October 2008 till October 2009.
Is the picture so dismal? To answer this, we need to look into the non-economic factors which might influence this number. October 2009 was a month of festivities with fewer working day but enhanced purchases prior to Diwali. At the same time, Diwali does not occur in a fixed month every year. Hence, the simplest seasonal adjustment procedures will not remove these effects.
In the jargon of seasonal adjustment, Diwali is a `moving holiday'. It requires special care in seasonal adjustment. Hence, in our work on seasonal adjustment, we test for the impact of moving holidays such as Diwali and Id, and when these effects are statistically significant, we adjust for them. Through this procedure, we find that SAAR for IIP for October 2009 works out to +0.12%. In other words, correcting for Diwali yields a change from an estimate of -5.12% SAAR for October 2009 to an estimate of +0.12% SAAR.
The figure superposes the two time-series of SAAR IIP (without adjusting for Diwali) and SAAR IIP (with adjustment for Diwali). In most months, the two series are obviously identical. But in some months, the interpretation of the IIP data strongly requires care in treatment of Diwali as a moving holiday.
Many analysts warned about reading too much into the weak October 2009 IIP performance, on the grounds of a Diwali effect. We go from this broad but unspecific caution to a precise estimate of what happened to overall IIP and IIP-consumer goods in October 2009 when compared with September 2009, after adjusting for seasonality and Diwali. The result is a gloomy value of 0.12% SAAR for IIP in October 2009.
The biggest impact is visible on the IIP-consumer goods (figure above) which shows a pleasant value of 7.01% SAAR after the Diwali adjustment, while without this adjustment, there is a worrisome SAAR value of -25.07%.
The calculations reported here are updated every Monday at http://www.mayin.org/cycle.in on the world wide web. For all series, Diwali effect testing is done, and wherever the impact is statistically significant, it is adjusted for. It proves to be significant for IIP, IIP (Manufacturing) and IIP (Consumer goods).
takethetimes@gmail.com
ReplyDelete