Did India’s manufacturing sector grow by an abysmal 0.1% in 2016-’17 over the previous year or by a decent 4.9%? This question, with obvious political undertones, has been occupying the minds of economists ever since the government revised the index of industrial production in May 2017.

The index, which had previously 2004-’05 as the base year, was recalibrated to use 2011-’12 at the base and its constituents overhauled .

In a study released last week, three economists examined whether the healthy growth of the manufacturing sector in 2016-’17 indicated by the new index actually reflected conditions on the ground. The conclusion reached by Radhika Pandey and Pramod Sinha from the National Institute of Public Finance along with Amey Sapre of Indian Institute of Technology: the new index, showing industries in India growing at a steady click in 2016-’17, may be badly designed to reflect reality in times when prices are too high or low in the market. The index shows a brighter picture of the economy than is accurate when prices in general are low – such as now.

How does the index of industrial production work?

The index of industrial production is an aggregate measure that reflects the situation of factories. In India, more than three lakh firms are registered under the Factories Act, 1948. The index of industrial production reflects the manufacturing activity from these units each month.

The index is compiled every month by the Central Statistics Office, but is realised after a two-month lag. About one-fourth of the registered firms are sent questionnaires about their production numbers at the end of each month. These numbers are then sorted into categories and subcategories and assigned weights.

The index is compiled in two steps from here. A commodity group’s total production is multiplied by its respective weight. The sum total of all calculations across commodity groups is the resulting index of industrial production number.

With the previous index, manufacturing in 2004-’05 was given a value of 100. Changes in the index value in relation to this was measured for each year. Hypothetically, if the score in 2006-’07 was 110 and the next year 115, it could be concluded that manufacturing had grown over the year by 4.55%.

Economists agree that it is a good practice to revise this base year every decade or so. That was done in May, along with many other changes to the composition of the index. However, Pandey, Sinha and Sapre have found that the new index is prone to depicting a more rosy picture of manufacturing than the actual output from factories, when the inflation rate is low.

The new index of industrial production series is in line with the Gross Domestic Product (which is the total value of goods produced and services provided in a country during a year) and the inflation numbers. This is important since both GDP and index of industrial production include big manufacturing components, so both should generally point in the same direction.

Using the new index, the Indian economy seems to have performed much better in the last financial year than it would come across if data provided by the old series was taken into account.

The revised index of industrial production also presented a more rosy picture of 2013-’14 than previously painted. The new series revised the annual growth figures from 0.8% to 3.6% for 2013-’14.

What changed after the revision?

While the change of base year from 2004 to 2011 has made the index of industrial production series more comparable to the GDP and other indices in the country, the real change in the way numbers are produced has come from the composition of the index. Pandey said that the composition of the items used to calculate the index of industrial production has undergone a substantial change.

On an aggregate level, 149 items were added in the new index series while 124 items were deleted. Among the items included were refined palm oil and surgical clinkers. The deletions included toothbrushes, fans, and calculators.

At the same time, the new index captures only the formal economy while the earlier index captured unorganised enterprises too through the Unincorporated Enterprises Survey of National Sample Survey Office.

The index of industrial production is divided into three major sectors: manufacturing, electricity and mining. In the new series, the weight of these sectors was changed, making it diverge from the old index, Sinha said.

“The weights have been slightly revised from the older series to the new series,” Sinha said. “Electricity was earlier 10%, now it is 8%. Mining was 14.3%, now it is 14.1%. Manufacturing was 75.5% and now it is 76.6%.”

Why does inflation impact the index of industrial production numbers?

Since the index of industrial production has traditionally been compiled on the basis of total volumes produced by factories rather than the market price of these goods, it used to be less prone to price changes in the economy. However, with the revision of the index, it has become a predominantly value-based measure, said Pandey. Close to 22% of commodities counted in the index of industrial production basket are reported in terms of value terms – their production units multiplied by current prices.

“Earlier, it was 54 value-based items and now there are 109 value-based items,” Pandey said. “In terms of the weight, these value-based items constitute 19.22% and in terms of number, they are close to 25%. Now, one-fourth of index of industrial production is value based. It’s a significant number to contribute to an index. Now you need to deflate it to arrive at volume because value is simply volume multiplied by price. So you use the wholesale price index to divide the total value to reach the volume index.”

The purchase value of money can change with time. A fall in the purchase value is referred to as inflation and a rise as deflation. To compare the actual value of a good over different periods of time, it is necessary to smother out the effect of deflation or inflation on money value. This is done by using an index of how value of money have changed over time. In this case the index of wholesale prices is used to do so.

So when the government’s statistical organisation began collecting more industrial data from factories in terms of the value of goods sold (and not the volume), the only way to compare this data fairly across time periods, the researchers assessed, was to factor out from the value the levels of inflation. This is where the researchers claim that the consequences of a more value-based index become questionable.

In periods of low inflation, they argue, the index of industrial production numbers could look much better than the ground reality because the deflator is smaller. While numerator of the calculation may not change, the denominator becomes smaller, artificially magnifying the output numbers.

“When the stake is as high as 19% and you observe so much volatility in Wholesale Price Index, then it is on the account of your methodology you are seeing growth in index of industrial production rather than a real increase in activity in production,” Sinha said. “It gives you a misleading growth rate which is not on account of production but on account of the methodology that we have chosen to arrive at it.”

A comparison of IIP and WPI shows how falling WPI results in a sharply rising IIP. Source: Ideas for India

This could mean that the index of industrial production numbers alone cannot be relied upon to provide an accurate picture of the manufacturing economy, especially in the periods of very low or high inflation as seen in the chart above. The chart shows how the index of industrial production rose to great heights when wholesale inflation numbers were low and vice versa.

How can the problem be fixed?

Pandey and Sinha feel that there should be a wider debate on whether using the wholesale price index to deflate indices like the gross domestic product or the index of industrial production is the right way to go. They said that many advanced economies use a hybrid indicator that is derived from both consumer and wholesale prices to give a better picture of the economy.

“Now the debate is about what measure to use to deflate these numbers – whether to use Wholesale Price Index or Consumer Price Index or a hybrid of the two indicators, as other countries do,” Sinha said.

Pandey added that India is not alone in facing this problem of statistics. Most emerging economies grapple with these issues and suggested that a better way to go about it is to perhaps look at other indicators like car sales and production data from a sector specific standpoint in addition to the index of industrial production numbers.

“Turkey also faced similar issues because they also rebased their series last year and it didn’t match with the high-frequency indicators [data released regularly],” she said. “They are carrying on with it but they are looking at other indicators too such as firm level indicators. They are looking at companies’ data such as sales and output of companies. This problem is not typical to India, it is faced by a number of emerging economies. One needs to look at how deflation is addressed in other countries. Otherwise, we have moved to a value based measure which is an internationally comparable measure but the methodology needs to be nuanced.”