Column The numbers game

Written by Mahesh Vyas | Updated: Sep 14 2008, 05:28am hrs
If we believe the official statistics of the recent months we should worry about the economy running into stagflation. Inflation in the WPI has been rising rapidly and growth in the IIP has been falling. A rise in inflation should indicate rising demand and during such times, production should be rising rapidly and companies should be investing aggressively to expand capacities. But, if production growth is declining when inflation is accelerating then, these are signs of an economy heading towards stagflationthe combination of high inflation and low production.

But, this is not true! The economy is working quite well and taking good care to control inflation in consumer prices. It is the data that is bad. Bad data is leading to bad diagnosis and further to the formulation of bad policy prescriptions.

The most commonly used measure of production is the Index of Industrial Production (IIP). The Index has a base of 1993-94. The set of products whose output is used to construct the Index every month in 2008 reflects the structure of Indias industrial economy before the boom of the mid-1990s, the slowdown that followed in the late-1990s and the revival since 2004. Production of new products such as mobile handsets are not included and many old ones like typewriters and sewing machines are included in the computation of todays IIP.

The production of typewriters and sewing machines has been falling rapidly as laptops and readymade garments replace them. The IIP fails to recognise these and other similar structural transformations of the industrial sector. It systematically assigns greater weight to products that have faded out and lower weight to those that have been rising rapidly. It therefore ends up systematically under-estimating growth.

The phenomenon of archaic composition and weights in an index is a common problem. The solution is to change the composition and weights as frequently as possible. Liberalisation has accelerated the pace of change in India since the early 1990s. Therefore, the need for change has been greater. The inability of the official statistical machinery to effect these changes in its computations is a disservice to the nation.

The Central Statistical Organisation has been trying for the past several years, to shift the base year. It tried to shift it to 1999-00. But, this task itself took so many years that in the meantime, the economy shifted gears and started to accelerate. Now, the CSO has given up its earlier efforts and is likely to shift the base to 2004-05.

Besides the base-year problems of outdated composition and weights, the IIP suffers seriously from poor response. Industrial units are often not cooperative in providing information on production although they are required, by law, to do so. But, this culprit is smaller than the inefficiency of the government machinery that collects data.

The result of the poor construction and implementation of the IIP leads to measurements that are unrealistic. For example, the IIP indicates that the manufacturing sector grew by 5.6% in the quarter ended June 2008. However, an alternate way of measuring the same outcome is the inflation-adjusted growth in sales of companies. This growth (based on public information of over 2,100 companies) was a robust 22.6%. The IIP thus under-estimated growth in the manufacturing sector in the June 2008 quarter by a whopping 17 percentage points.

The problem with the IIP is not limited to the damage it causes to perceptions of where the Indian economy is headed, every month. It also feeds into the computations of the quarterly GDP growth statistics released by the Central Statistical Organisation. The industrial sector has a share of over 20% in the GDP and thus the underestimation of the IIP has a corresponding impact upon the GDP computations. It is unfortunate that the government machinery does not use any independent mechanism to cross-check what is an obviously faulty measure. Steps need to be taken in two directions to correct the situation. First, methodological changes are needed. This requires discarding the Laspeyres fixed base index and the fixed composition of the basket of products or responding units. Currently, the list of products is fixed and so is the list of responding entities. The government needs to move to the Chain-linked index computation methodology and move to a dynamic set of products and dynamic set of responding units in the computation of the IIP.

Secondly, the government needs to move away from the statistical machinery relying heavily on the administrative machinery. That structure was a useful way to organise the statistical system when the government was deeply enmeshed in the economy. Liberalisation has taken away a lot of the power that the government had to demand for data.

Now, there is a need for professionalisation of the statistical machinery, an overhauling for the collection of data in the new dispensation. Until these changes to the statistical system take place, it is advisable to ignore the IIP as it is released in its current form.

The author heads Centre for Monitoring Indian Economy, Mumbai