CPI is a vital statistic to measure pulse of the economy. It helps RBI decide on interest rates and assists GoI in planning its budget. To make sure this data reflects today's consumption, GoI is updating the index's base from 2012 to 2024 through a reformative and consultative process.

Within this update, the methodological approach to measuring housing services is of strategic importance. It carries a significant weight of 10.07% in all-India CPI basket, and a commanding 21.67% in urban CPI. Recognising its prominence, MoSPI released a discussion paper in October 2025 to propose a significant overhaul of the housing index compilation methodology. Proposals include:

Expanding housing index coverage to include rural areas will fill in a significant data gap.

Rents of employer provided dwellings were being measured by house rent allowance surrendered, plus licence fee paid. As these didn't represent market rental values and caused distortion in housing index, excluding employer-provided dwellings from the sample is apposite.

Using census 2011 housing frame will not only provide a detailed frame of dwellings, but also ensure inclusion of weight of employer-provided accommodation in housing index (part of rented dwellings), and implicitly provide rental equivalence for employer-provided dwellings.

The current methodology has resulted in a 'sudden upward shift' and 'introduced an unexplainable movement' in the index. These have propelled a drastic methodological change in the compilation of housing index. But the existing methodology is scientifically robust and its analysis shows:

Sharp rise in index observed in June 2013 was not a methodological flaw. Rather, it was an 'implementation choice' to hold the index fixed at 100 for the first 5 mths of the series despite incremental availability of rental data.

Current chain-based method is mathematically equivalent to the fixed-base method - it's transitive. This establishes that the chain-based approach could not have introduced the rhythmic, 6-mthly dips in the index. The dips likely stemmed from an initial calculation error unrelated to the core methodology.

Existing 6-mth moving panel survey, used for collecting house rents, is a robust and standard international practice for measuring rent changes, while minimising costs and respondent burden.

A major change suggested is to survey over 24,000 dwellings every month, instead of the current approach that requires surveying about 4,000 a month by visiting a dwelling every 6 mths. This would lead to a sharp upsurge in data collection costs, and increase in respondent burden with minimal informational gain, given that rent prices don't change frequently. Visiting a sample dwelling every month for rent collection is neither supported by sampling theory nor global practice. If resources are available, it's better to increase coverage of dwellings, instead.

If dwelling availability is a concern, a shift to a 3-mth moving panel would be more efficient than a complete overhaul. The proposal is to use geometric mean for the first level of aggregation and arithmetic mean thereafter. But to arrive at the elementary index, geometric mean is the recommended metric for aggregating rent-relatives (current month rent to rent 6 mths ago) across different dwelling categories and ownership categories.

Geometric mean accurately captures the true overall rent-relative, and is also robust against outliers. This significant benefit is lost with an arithmetic mean, which can be heavily skewed by isolated, large rent-relatives in certain categories of dwellings and, thus, distort the index for the average consumer.

The existing panel-based methodology for housing index compilation is theoretically sound. The proposed overhaul appears predicated on a misdiagnosis of historical issues. Given the significance of housing component in CPI, and importance of correct measurement of inflation, the proposed changes need to be methodologically robust. This will preserve the accuracy of India's inflation measurement, which informs monetary policy.