Overestimation of inflation in Jan, Feb? RBI cites NSO methodology

Various analysts had earlier doubted the accuracy of retail price inflation because of “overestimation” of the rate of price rise in cereals in January and February this year. The Reserve Bank of India (RBI) in its recent paper attempted to provide an answer to this anomaly.

There were differences between the figures arrived at by the National Statistical Office (NSO) and analysts due to the way the NSO aggregates data from the item level upwards to the numbers at the national level.

The paper, put out in the latest RBI Bulletin, called for a change in the aggregation method to make CPI data more useful for monetary and other policies. It also urged for an expeditious revision of the base year of the consumer price index (CPI) from 2012, to guide the monetary policy correctly.

The overall CPI inflation rose to 6.52 per cent in January this year from 5.72 per cent in December last year, and then came down slightly to 6.44 per cent in February, 2023. Within that, cereal inflation rose to 16.2 per cent from 13.79 per cent and again went up further to 16.73 per cent in these months.

Aggregation of price indices in the CPI to arrive at the headline all-India index published by the NSO is done in two stages, the RBI paper pointed out.

In the first stage, price indices are calculated at the item level. In the second stage, these elementary indices are aggregated to obtain higher-level indices using consumption expenditure as weights.

By using this vertical aggregation method, various indices for each state and Union Territory are compiled by the NSO, separately for rural and urban sectors.

These indices are then aggregated as a weighted average, using the share of respective states (and UTs) in total expenditure of that item to compile all-India indices, respectively. This is called horizontal aggregation, explained the paper.

There is an alternative method, as well, which is called vertical aggregation. In it, the sub-group/group/overall index is derived by taking the weighted average of their constituent item indices.

Ideally, these two methods should give the same numbers. But, the problem arises when there are seasonal items, such as fruits, or when there are zero-price items, such as PDS cereals, part of which became free for 800 million people from January this year.

The substantial divergence observed in January and February between the published CPI-based headline inflation and that derived by vertical aggregation of item-level CPI data was due to a sharp divergence in inflation in the cereals sub-group.

Following the extant methodology, when PDS wheat and rice were made available for free, the weights of such zero-price items were apparently redistributed across the closest category in the CPI in states, implying higher weighting to non-PDS or open market prices.

In a scenario of a sharp increase in open market wheat and rice prices, cereal inflation may have been further accentuated in states, the article said.

Now, item-level indices were compiled by horizontal aggregation across states/UTs with weights for missing PDS prices in one state redistributed to PDS items of other states, and not between the weights of PDS and non-PDS items. As a result, when the state-level overall CPI datasets were aggregated to arrive at the all-India headline overall index and inflation, it turned out to be substantially higher than what was obtained from a vertical aggregation of the CPI item-level data, the RBI article said.

The paper called upon the Technical Advisory Committee on Statistics of Prices and Cost of Living (TAC on SPCL) that guides the NSO on the preparation of the CPI to consider a change in the aggregation method to make the index more usable for policy analysis.

Several analysts had earlier doubted the accuracy of cereal inflation and, hence, overall numbers. For instance, a note by Kotak Institutional Equities stated: “There appears to be some data anomaly in the cereals index for January.”

If this data print were to undergo revisions, the cereal index’s sequential momentum would be around 0.8 per cent, leading to food and beverage inflation at 5.75 per cent, compared to 6.19 per cent according to the official release, the note said.

Consequently, the headline CPI print would stand corrected at 6.29 per cent — 23 bps lower than the official release, it had opined. Many other analysts, too, doubted the data on similar lines.

However, the Ministry of Statistics and Programme Implementation (MoSPI) maintained that there was no anomaly in the data. Also, the final January headline inflation stood the same as the provisional one.