After corrections, CPI inflation in India in FY16 was 3.8%, a full 110 bps below the official estimate of 4.9%
Breaking news: The GDP debate has now shifted to whether the GDP deflator is accurate or not. For more than a year now, there has been deep questioning about the reliability of the Indian GDP data after the “new” estimates were presented in January 2015. Of late, perhaps after noting that this was a losing cause (also see my article ‘GDP Debate: RIP’, The Financial Express, April 23), the GDP sceptics have turned their critique towards the possibility that the GDP price index (also called the deflator) is understating inflation. Since real GDP growth is the difference between nominal growth and deflator inflation, the intended outcome of the investigation is the same i.e., GDP growth in India is overstated—back to square one.
The main contention is that the GDP deflator is flawed because of its heavy reliance on the Wholesale Price Index (WPI). This questioning of the GDP deflator was started by the RBI in their September 2015 Monetary Policy Report (page 14): “This (WPI) tends to overstate the extent of price decline in the GDP deflator when WPI is in deflation.”
In its very recent report on India, the IMF regurgitates the RBI conclusion: “Constant price estimates of GDP deviate from the conceptual requirements of the national accounts, in part because the Wholesale Price Index (WPI) is used to derive volume estimates for many economic activities.” (IMF Country Report for India, March 2016)
The following two quotes from prominent Western experts (and check out the titles!) confirms the authoritarian Western seal that Indian GDP growth is overstated and maybe by as much as 250 basis points in FY16 i.e., rather than GDP growth being 7.6%, it “likely” was close to 5%. In other words, no improvement in the economy since Mr. Modi, and the BJP, won the election in May 2014.
The Economist (‘India’s GDP Data: The Elephant in the Stats’ , April 9) authoritatively and comprehensively states: “The blended inflation figure (WPI) used to deflate the nominal data may therefore be too low, making real GDP growth come out too high. Investors, at any rate, roundly disbelieve India’s growth figures.”
And on May 1, The Wall Street Journal delivered its verdict (‘On Close Inspection, India’s Sharp Growth Picture Gets Fuzzy’): “To strip price changes out of a wide swath of GDP, for instance, India uses its wholesale price index—which, thanks to lower oil prices, has been decreasing for 17 straight months.”
So you get the picture: Indian GDP is low, because deflator is underestimated, while the real indicator of inflation in the economy, the CPI, is rising at around 5%. To quote The Economist, “Despite this, deflation remains a distant dream for shoppers: the price of consumer staples is still rising by over 5% a year.”
What if, and I know it is heretic to say so, the authorities (and the journalist experts) have got the CPI inflation estimate wrong? There is considerable evidence that this is the case—and not by a small amount either, but by about 110 basis points in FY16. In other words, CPI inflation in India, correctly measured, is running at a 3.8% rate rather than the official 4.9% rate. How is this possible—pinch me, can this be right?
Yes, it is right. Before presenting the results, a little bit of dry history as to how the CPI is calculated. The CPI is based on price surveys conducted every month, and, such surveys, while subject to errors, are the best we can do. And therefore, we can say that there is no problem with the CPI data on prices. So how is the CPI mis-measuring inflation? Because it has got the expenditure weights wrong. These weights are obtained from the National Sample Survey of Consumer Expenditures, 2011-12.
Unfortunately, the NSS surveys have strangely and mysteriously been under-estimating average per capita consumption for three decades now. In 2011-12, the NSS was able to capture only 50% of personal expenditures, NSS estimate R1,627 per capita per month; National Accounts (NA) estimate for the same year: Rs 3,255. Although back in the 1960s, the NSS estimates of individual consumption were the gold standard, to the extent that the NA authorities used to obtain estimates of consumption from the NSS. (Side note—before one engages in another debate as to whether the NSS or NA is correct, the reader is referred to my book which discusses this subject in gory detail Imagine There’s No Country, 2002).
To make the point of flawed weights clearer, according to the NSS, a typical Indian consumer spends 46% of her monthly expenditure on food and beverages; 3.7% on household goods and services, 8.6% on transport and communication and zero percent on financial services (insurance, banking, other business services). The comparative shares in the NA data (i.e. private final consumption expenditure) are 31.4, 6.3, 16.8 and 7.2%—clearly, there is a large difference in the shares. More importantly, the CPI data also completely misses out on an increasingly significant part of each consumer’s life—financial and insurance services (zero in CPI and 7.2% in NA).
In other words, there is no question that the NSS 2011-12 based consumption weights are flawed. Our purpose here is to assess whether the CPI is measuring actual inflation correctly—but this solves only one half of the GDP deflator puzzle. GDP in India is (approximately) composed of 65% consumption and 35% investment. If one gets appropriate deflators for the two components, one can obtain an independent assessment of what deflator inflation looks like in India, and an estimate that does not erroneously use WPI prices for consumption expenditures (as pointed out by the RBI). In a subsequent piece, to answer the question as to whether the GDP deflator in India is measuring inflation correctly will be addressed.
Therefore, in this article, we restrict ourselves to correctly assessing the consumption component of the GDP deflator. By definition, one should use the consumption weights as in the national accounts, which is what we do. (Note that the US also uses the personal consumption deflator as its preferred estimate of consumer inflation). We use CPI prices for 25 broad items of expenditure (14 food and beverages and 11 non-food) and equivalent consumption shares as per private final consumption expenditure in the national accounts for 2011-12.
Inflation data are reported for three different consumer price series—official CPI as calculated by CSO, the first CPI adjusted measure or CPI1 (NA weights and CPI prices and zero weight for financial services), and the second CPI adjusted measure CPIf (CPI1 along with 7.2% weight for financial services). Price data for financial services are proxied via the banking service index, available on the Office of Economic Advisor website. Comparing with the first adjusted measure CPI1, CPI overstates inflation by a large 0.9 percentage points in FY16. In three of the last four years, and both in high and low inflation times, CPI has exceeded inflation (CPI1) to the order of 0.8 percentage points.
If consumer purchases are assumed to include financial services (which they are supposed to), then NA CPI was even lower by 20 basis points i.e. correctly adjusted, inflation (CPIf) in India in FY16 was a full 110 basis points below the official estimate, and an average of 130 basis points below the official estimate over the last four years.
The difference in inflation estimates are not small. If the adjusted CPI estimates do reflect the underlying reality of consumer inflation, then the results have strong implications for RBI policy (are Indian repo rates too high?) and GDP growth (are we underestimating GDP deflator inflation?). For these, and other conclusions, tune in again, next week…
The author is contributing editor, The Financial Express, and senior India analyst, The Observatory Group, a
New York based macro policy advisory group. Twitter: @surjitbhalla