Devil in the detail: Choice of base year in new GDP series and revisions need clarity

While revisions in GDP estimates are common globally, in India’s case, the magnitude of recent GDP revisions is somewhat higher than has been the case in the past.

Consider this: At the launch, new series (base year 2011-12) estimated the 2013-14 GDP growth at 6.9% compared with 4.7% as per the old series. Further, the advanced estimates under the new series suggested a 7.5% growth in the third quarter of 2014-15. Subsequently, it was revised down substantially to 6.6%.

However, it may not be appropriate to put the entire blame on specific methodology adopted or on inclusion of new data sources. Various reasons have been attributed to the divergence in the GDP growth as perceived by major constituents of the economy vis-à-vis the growth as estimated in the new series (base Year 2011-12) by CSO.

These reasons include quality of information in new data sources such as the ministry of corporate affairs’ MCA21 database (read here) to efficacy of statistical techniques used to determine suitability of new data sources as well as their weightage in the calculation of overall GDP. However, debates on suitability of statistical techniques are often inconclusive.

One tentative explanation for these sharp revisions in the new series may be the choice of the base-year – the timing of the launch of the new series as well as limitation in existing data infrastructure. Let us explore these aspects.

The base year: The GDP series with base year 2004-05 was launched in 2010, while the series with base year 1999-2000 was launched in 2006. Going by the past, the base year for the new series may have been 2009-10 instead of 2011-12. However, 2009-10 was an atypical and may not be an ideal choice for the ‘base year’. The most obvious reason being, that the Indian GDP slipped in 2008-09 after four years of 8% plus growth. The growth revived to 8% plus levels in 2009-10 and 2010-11.

As continuing with the old series (base year 2004-05) was untenable and the new series had to be launched, there were hardly too many options of base year to choose from. With the benefit of hind sight it appears that 2011-12 was an inflection point when economic activity started slipping. The choice of 2011-12 as base year, when a lot of macro trends started exhibiting trend reversal, may present some analytical challenges.

Now that 2011-12 is chosen as the base year, it may be worthwhile to reflect how this choice may potentially impact GDP estimates. The year 2011-12 was when the GDP deflator – an indicator of system inflation used to convert nominal GDP to real GDP – as the highest in the last 10 years. The inflation has cooled off thanks to a combination of global commodity price correction and moderation in industrial demand (WPI).

If inflation is benchmarked to 2011-12, it may numerically look somewhat lower than otherwise, to that extent, numerically , the real GDP may look higher. However, the more important factor that may drive GDP numbers as well as any potential future GDP revisions is the availability of certain critical information used in GDP calculations in the new Series.

Normal data limitations: The limitations of data availability are chronic in most nations including India. To address this well-acknowledged problem, the GDP for a period is first published as a provisional estimate. Then as more data from economy wide surveys as well as corporate balance sheet come in, the provisional estimates are revised approximately one year later and there may be a second revision after another two years. For instance, the 2014-15 GDP published in May 2015 may be revised in 2016. It is only in 2017 one may finally have a reasonably good idea of the 2014-15 GDP. Of course, this is a global practice and not a revelation.

Not surprisingly, these data issues also persist in the new series that was launched in 2015. At the time of the launch, the Annual Survey of Industries (ASI) as of 2011-12 was available. ASI is available with a two-year lag. Employment and Unemployment Surveys (EUS) of the National Sample Survey Organisation (NSSO) conducted once in five years was available for 2011-12. MCA21 data base used for full year estimates is typically available with an 8-9 month lag. For example, for 2012-13, the data available was of November-December 2013.

In other words, the GDP calculation for 2014-15 would have used inputs from ASI 2011-12, EUS 2011-12 and MCA21 data for the year 2013-14. These data inputs are suitably extrapolated for the period 2014-15. However, for extrapolation, high frequency data such as IIP, credit growth, vehicle sales for the 2014-15 are also used to capture the ‘recency’ impact. In addition, the quarterly GDP estimates also benefit from the RBI Study on Company Finances, which considered financial results of around 2,500 companies. However, at least a quarterly lag remains in usage of corporate data as well.

How Data Limitations Impact GDP calculation: Gross value added (GVA), which is a critical component of the GDP as per the new methodology, can be directly estimated for large listed entities. As such GVA is given as the sum ofa) compensation of employees; b) consumption of fixed capital which is often captured from ‘depreciation’ in corporate profit and loss (P&L) account; c) operating surplus which may be approximated by profit after tax (PAT) adjusted for property income and current transfers to trade channels and d) Production taxes less subsidies.

For small economic entities such as unregistered businesses, the GVA is calculated using statistical techniques, the input to which comes from surveys such as ASI and EUS. ASI and household sector account for roughly one-fourth of the critical manufacturing sector GVA, as per CSO.

Given the lag with respect to data availability, the GDP estimates for the most recent period (advanced and provisional estimate) have to depend on extrapolations based on corporate performance and also past trends reflected by the surveys.

For instance, the Q4 2014-15 GDP may not have incorporated the worrisome corporate performance of that quarter. Similarly, some of the constituents of GDP may have benefited from improving trends exhibited until ASI 2011-12.

Guessing the direction of future revisions: As per ASI, annual growth in employee compensation ranged between 15% and 24% for the period from 2005-06 to 2011-12. Comparable growth trends were also observed for depreciation and fixed capital growth, suggesting good economic activity during 2005-06 to 2011-12.

To the extent the GVA calculation of smaller entities is based on these numbers they may show an upward bias. However, as of March 2015 the Volume I of ASI 2012-13 is made available. ASI 2012-13 suggests that not only aggregate worker’s compensation growth has fallen to 11% (comparable growth rate last seen in 2004-05) but total number of workers has also de-grown. Likewise aggregate growth rate of fixed capital and depreciation have also shown muted growth rate as per ASI 2012-13.

The latest ASI will be utilised for revised GDP estimates for all periods from 2012-13 till date and is expected to be published in 2016. It is possible that these revised growth rates for the period 2013-14 and 2014-15 may be more closely aligned to the popular perception of the economic activity.

Removing the urgency from policy actions: GDP estimates are essential to provide decision makers with a reasonable, if not precise, idea of the current level of economic activity and how that level compares with the activity in previous time periods. Any confusion created in this regard may potentially lead to ineffective and even counter-productive policy decisions.

There is a broad consensus that the worst in terms of economic activity is clearly behind. Sporadic signs of recovery such as higher commercial vehicle sales, improved air passengers or higher electricity production are undeniable.

However, doubt remains in certain quarters about the pace of recovery. Those doubts are likely to be resolved when the revised estimates for the 2014-15 GDP are published in 2016 and then further revised in 2017.

If the revised GDP estimates are broadly in line with the provisional estimates then it will be a relief. On the contrary, if the magnitude of the revision is comparable with the revisions of Q3 2014-15, it is worrisome because then it may imply that India is losing out on the urgent concerted action required both by the government as well as the RBI. A 7.4% can only provide a false sense of complacency.

The author is senior director – corporate director, India Ratings. Views are personal. This is the second in a two-part part series on new GDP numbers. Read the first part here.

[“source – firstpost.com”]