Reading consumers’ minds: an analysis of inflation expectations

ABSTRACT

The heterogeneity in households’ consumption baskets is often deemed responsible for deviation of their inflation expectations from headline inflation. The paper verifies this by simulating population baskets and estimating mean inflation by sampling baskets. The estimated mean inflation fails to display closeness with survey numbers. Therefore, the paper proposes alternative logical methods for designing basket compositions and identifies the most suited method using which the estimated expectations are found to be close to and well-correlated with survey numbers. The findings suggest that a sudden rise in inflation of regular-use items can explain the deviation in households’ expectations from official inflation.

KEYWORDS:

• Consumer price index
• consumption basket
• inflation
• inflation expectations
• sampling

JEL CLASSIFICATION:

• D84
• E20
• E21
• E30
• E31

Acknowledgments

The author is thankful to the reviewers and the journal’s associate editor for providing valuable comments which helped in improving the manuscript.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1. The survey reports households’ annual expenditure on item groups in INR. From these figures, the shares of per capita expenditure across the groups are derived.

2. State-wise analysis is not performed as CPI-inflation figures for items are unavailable for the states.

3. Here, the numbers of baskets to be simulated in the population and proportions of items to be drawn from each of the consumption subgroups of items at each step are indicative. Other values may be considered.

4. Item-wise CPI indices are unavailable for the months March to May 2020 due to the COVID-19 pandemic and the resulting restrictions.

5. The bimonthly IESH conducted by the RBI forms one of the important inputs to the bimonthly Monetary Policy. Given the limited time frame, it covers about 6,000 households (survey centre Jammu, which was added recently, is not considered for the analysis) out of the total 1,77,28,937 households (as per Census 2011) in the 18 survey centres considered for the survey. Keeping this in consideration, a small sampling fraction is considered here for the numerical illustration.

6. For the samples drawn from the population baskets simulated using Method 3 (to be described later in the paper), as all figures are above $16$ per cent, these are not transformed into ranges.

7. Low fractions are considered here with the presumption that households can remember a limited number of items while making expectations.

8. Small percentages are considered here with a presumption that households may remember only a few items in mind while making their inflationary sentiments.

9. The tables on performances of Methods 2 to 4 and of eleven combinations of $\mathit{CPIUI}$ and $\mathit{CW}$ for Method 5 are refrained from reporting here to conserve space, but available on request.

Additional information

Notes on contributors

Purnima Shaw

Purnima Shaw is working as an Assistant Adviser in the Reserve Bank of India. She received her Ph.D. from West Bengal State University, India in 2021. Her research interests include survey sampling, inflation expectations, fan charts and official statistics. She has 21 publications in journals and bulletins including Journal of the American Statistical Association, Communications in Statistics – Theory and Methods, Handbook of Statistics, Irving Fisher Committee Bulletin and Reserve Bank of India Bulletin.