RBI faults in using y-o-y data for predicting the course of inflation
In its August 4 policy statement (hereafter, PS), RBI rigidly kept to its no-rate-cut stance and justified it on the grounds that y-o-y inflation in June had perked up to 5.4 % from 5.0 % the previous month. This is what the PS said: “Headline consumer price index (CPI) inflation rose for the second successive month in June 2015 to a nine-month high on the back of a broad based increase in upside pressures, belying consensus expectations. The sharp month-on-month increase in food and non-food items overwhelmed the sizable ‘base effect’ in that month.” (emphasis added).
To really warn the public about widespread inflation, the PS added: “Furthermore, excluding food and fuel, inflation rose in respect of all sub-groups other than housing….Inflation in CPI excluding food, fuel, petrol and diesel has been rising steadily since April and exceeded headline inflation through Q1. Near-term inflation expectations of households returned to double digits after two quarters…Rural wage growth was moderate but there are indications of incipient pressures from corporate staff costs”.
Readers of this column know that for more than a decade, I have warned against the dangers of using y-o-y data for making policy. Let me illustrate for the umpteenth time. Assume the CPI index is 100 in January 2014 and there is zero month-on-month inflation till June 2014. In July 2014, there is 10% inflation, and thereafter, again, there is 0% m-o-m inflation till October 2015. The RBI y-o-y inflation measure will show 10 % inflation till June 2015, even though for the last 12 months there has been zero inflation. Only after a year of wrong data and inference would the y-o-y data get back on track for showing the “correct” inflation.
Policy makers get around this problem by looking at seasonally adjusted data. RBI doesn’t officially use seasonally adjusted data—perhaps, with outside experts’ influence in the Monetary Policy Committee, it will begin to do so. The accompanying table shows inflation data for a wide variety of CPI indices, for the January-June 2015 period. The June RBI PS (when repo rates were cut by 25 basis points) was based on CPI data till April; the August statement had data available till June—hence, the grouping of data for January-April; May June data are reported separately in order to constructively assess RBI’s inference of an across the board acceleration of inflation.
The table shows annualised rates for both unadjusted and seasonally adjusted data. Look at CPI excluding food, fuel and light (FL), and petrol (or core CPI inflation). (FL includes electricity, kerosene, charcoal, etc, but excludes petroleum which has approximately a 2.4% weight in total consumption). Core CPI inflation shows a decline in the unadjusted data in June—only a 5.2 % rate compared to 5.9 % in May and an average of 5.6 % January through March. This certainly does not fit with the RBI perspective of across-the-board inflation increase. Why the different result? Because RBI persists with y-o-y data.
Further justification is provided by noting the marginal decline in the miscellaneous category of expenditures. This is part of core CPI and accounts for more than 28% of the total CPI. It shows only a marginal decline in June, to 9.3%, which by all accounts is very high inflation—up from an average of 3.4% between January-April. However, at least 40 % of this category (via NCAER input-output tables) is services. Service taxes went up in June 2015 from 12.36% to 14%. If prices remained constant, this would mean an increase of 1.4% in the price of a particular service, or an inflation rate of around 17% (since the tax increase happened in just one month). Adjusted for service tax, miscellaneous goods and services registered a maximum price increase of only 5.6% in June, a level almost half of the 10.4 % observed in May 2015.
Seasonally-adjusted data confirm the real story that inflation is falling, and falling across the non-food board. These data suggest that the outlier in the January-June inflation pack is May, and there appears to be a steep reversion to the mean in June. One reason for the high May inflation was that there was a generalised panic about El-Nino, and a possible drought for the second year running.
It is likely that the caution that Rajan and RBI exercised at their August meeting had a lot to do with the May inflation data rather than the latest June data. It is also likely that this is what Governor Rajan was hinting at when he said that one should not rule out an inter-meeting rate-cut. If July inflation data are as benign as June, (and we expect it to be so) then May is the exception in the last seven months. As such, like the rest of the world, India is also moving towards lower inflation, and lower repo rates.
What is consistent with this conclusion is the doveish anomaly in the otherwise static hawkish RBI policy statement. Because of softer crude prices and a near normal monsoon so far, RBI projects January-March 2016 inflation to be lower by 0.2%, i.e., at 5.8% rather than 6%. As several commentators have pointed out (including RBI), there is a very loose connection (if at all) between food production decline and food inflation. In its April and June policy statements, RBI correctly advised the government that it was watching “food management” as an instrument of inflation control. FAO food prices, as of June 2015, are down 21% y-o-y. This, along with the excess stocks of wheat and rice, means that the government can comfortably ride out any weather-induced price storm—if there is one. So, if RBI is lowering its year-end estimate, soft June data may be responsible!
One final observation about RBI and inflation forecasts. The RBI estimate for y-o-y inflation in January 2015, made in August last year, was 8 %. It turned out to be a grand miss with actual y-o-y inflation in January 2015 about 300 basis points lower, at 5.2 %. What happened? A favourite explanation of many commentators, inflation experts, and Congress party officials was that this decline had a lot to do with the decline in crude oil price, from $105/barrel (June 2014) to $47 a barrel (January 2015). That is a (log) decline of 80%. Petrol consumption is 2.4% of the total expenditure; so, this would have brought inflation down by 192 basis points. So, maybe RBI wasn’t that far-off from its estimate—they erred by only a 100 basis points. In the larger scheme of “forecasting”, this should not be so bad.
But it is. CPI reflects the cost of petrol you and I pay. As the international price of oil declined, the government (Centre and states) mopped up a large part in taxes! Domestic price of petrol fell by only (log) 15%; so, the approximate effect of the crude price fall on the CPI was only 36 basis points, i.e., adjusted for oil, inflation in January 2014 should have been 7.6%! RBI, even with the oil price decline, was off by 240 bps, in a forecast of y-o-y inflation only five to six months forward.
Will the RBI forecast for January-March inflation be off or will RBI reverse to the mean and be right? Time will tell, but all things considered, the odds are greater that January-March inflation will be closer to 4 % than RBI’s 5.8 %.
The author is chairman, Oxus Investments, and senior India analyst, Observatory Group, a New York-based macroeconomic policy advisory firm