The current series of the Consumer Price Index (CPI) has an old base of 2012=100. As CPI is one of the most extensively used indicator for public policy, to get a correct measurement of inflation (that is, year-on-year changes in prices) the country is awaiting the revised series with a new base. The National Statistics Office (NSO) sets up an expert group (EG) for CPI revision, which deliberates on switching the base to a recent period with the availability of results from the national household consumption expenditure surveys for 2022-23 and 2023-24. The price statistics division at NSO, under the guidance of the EG, is working towards presenting a revised series.
CPI base revision is an extensive exercise that takes into consideration the fixing of norms for selection of items, based on the results of the consumption expenditure survey; conduct of market survey for identification of popular markets; selection of shops for different commodities in the selected markets; determination of specifications of the commodities to be priced; and, finally, price collection for the base year from the markets in rural and urban areas across the country.
There is one other aspect of CPI compilation that needs attention — namely the mode of aggregation. The present CPI has 299 individual items (for example, rice, potato, house rent, electricity, medicine, tuition fee, and gold, among others) clubbed into 20 sub-groups (such as cereals, vegetables, health, education, and so on) and six groups (food, pan-tobacco, clothing-footwear, housing, fuel, and miscellaneous). The NSO provides the weights of these items based on the household consumption expenditure survey. The weights of items add up to their sub-groups, sub-groups to groups, and groups to headline at the all-India level. The NSO compiles the indices in two stages — first it arrives at the item/ sub-group/ group/ headline indices for each State/Union territory (UT), separately for rural and urban sectors; second, it horizontally aggregates the item/ sub-group/ group/ headline indices across sectors and State/UTs using item expenditure as weight to arrive at the corresponding all-India indices.
Horizontal aggregation often creates challenges in using the index for economic and monetary policy analysis. Inflation analysis employs vertical aggregation of item indices to sub-groups/ groups and headline indices to deep-dive into inflation drivers. Often, published inflation (arrived at by horizontal aggregation) and user-derived inflation (by vertical aggregation) do not tally, owing to aberrations from the underlying methodology. This divergence is prominent when prices of some products are not available from certain markets.
In India, for the bi-monthly policy decisions by the Monetary Policy Committee, under the flexible inflation targeting (FIT) framework, a proper assessment of sources of inflation and outlook is critical. The aggregation issues discussed above influence the following aspects of monetary policy analysis:
• Inflation assessment: To understand inflation sources, headline and groups/ sub-groups inflation contributions are dissected in various categories based on item-level data. For example, CPI is split into goods and services components, various core and non-core portions, food and non-food sections, and so on, by aggregating corresponding items. These categories usually do not add up with the published overall CPI. The residual is often substantial, impeding policy analysis.
• Inflation outlook: Forecasts prepared by most institutional analysts, including those in the Reserve Bank, have three broad components — nowcasting using high-frequency indicators on the price momentum of a large set of CPI items that become available ahead of the official data release; short-term projections from a bottom-up approach employing various time-series analysis techniques; and medium-term projections using nuanced econometric approaches. Importantly, the nowcasts feed into short-term projections, and further into the medium-term forecasts. The inflation nowcasts ahead of each official print by RBI’s forecasters use bottom-up vertical aggregation. Thus, aggregation becomes an integral part of the forecasting framework for policy inputs.
Under the horizontal aggregation followed in the present CPI series, often, even when the nowcasts tally with the corresponding published item indices, they do not add up to the corresponding groups, sub-groups and headline CPI. The gap between the NSO’s published and user-derived inflation has been material on several occasions. Such differences are found to be statistically significant in studies conducted earlier.
It will be valuable if NSO’s expert group deliberates on the aggregation method while changing the base year. In order to address the aggregation anomaly, the NSO may continue to arrive at individual item indices based on horizontal aggregation across sectors and States/UTs, as at present. Thereafter, the aggregation may be done vertically, at all-India level, so that weighted item indices aggregate to sub-groups, then groups, and headline. This will help maintain consistency in inflation measurement and forecasting, particularly for monetary policymaking under FIT.
(The writer is former head of the Monetary Policy Department, RBI, and Secretary to the MPC)