Volunteer grocery basket index
A consumer group publishes a monthly 'basic food basket' price. The basket fixes 30 items and quantities chosen in 2022. Each month, volunteers price the basket at 12 stores across three cities. City A has only budget-format supermarkets; City B has a mix; City C has one budget and three full-service stores. If an item is unavailable, volunteers substitute a similar product. Potatoes and kumara are priced only in season; the group carries the last observed price forward otherwise. The group reports an 11.2% increase from June 2023 to June 2024, and says 'grocery prices are up 11.2% for everyone'. Internal notes show the June 2023 basket had a special on chicken, and the June 2024 basket included a larger pack size for rice. The group wants to know whether the 11.2% is a valid measure of inflation and how to improve it.
1. Critique the construction of the 11.2% figure. Explain how fixed quantities, substitution, carry-forward, store mix, and specials affect it.
Read the worked answer
The 11.2% is a fixed-basket price change, not a measure of inflation for everyone. Fixed 2022 quantities ignore substitution: if chicken is on special in June 2023, the base is artificially low and the later increase is overstated. Substituting similar products when items are unavailable can change quality and package size; the larger rice pack in 2024 may lower unit price or change comparability. Carrying the last price forward for seasonal potatoes and kumara assumes no change, which may understate or overstate. Store mix differs by city: City A has only budget stores, City C has mostly full-service, so a national average is sensitive to which stores are sampled. Specials are temporary and should not set a base without adjustment. The claim 'for everyone' ignores household baskets and store choice.
2. Propose an improved index and a method for handling missing seasonal prices and package-size changes.
Read the worked answer
Use a chained or annually reweighted index, with weights from household expenditure and separate strata by city and store type. Collect actual transaction prices, including unit prices, and record package size. For seasonal missing items, impute within item and store stratum using neighbouring months rather than carrying forward indefinitely; publish the imputation rate. For package-size changes, compare unit prices or use a hedonic adjustment. Treat temporary specials explicitly: use regular prices or a moving average, and report both. Publish a sensitivity range using different baskets and store weights. A superlative index (Fisher or Törnqvist) reduces substitution bias but needs more data.
3. How would you communicate the uncertainty and scope of the index to the public without overclaiming?
Read the worked answer
State the basket, the stores, the cities, the dates, and the weighting. Present the estimate with a range or confidence interval from alternative methods, and distinguish 'prices for this basket at these stores' from 'the cost of living for all households'. Report the share of missing/imputed prices and the effect of specials. Avoid 'for everyone'; instead give examples of how different household baskets or store choices could differ. Include a methods note and invite comparison with official statistics, explaining conceptual differences.