Recommended Sponsor Painted-Moon.com - Buy Original Artwork Directly from the Artist

Source: MIL-OSI Submissions

Impacts of the COVID-19 lockdown on methodology for food price index April 2020  Media release

13 May 2020

This page summarises the impacts of the COVID-19 lockdown on methodology used for the April 2020 food price index (FPI).

Lockdown conditions under alert level 4 in April have made in-store price collection impossible due to widespread business closures and the emphasis on physical distancing to limit the spread of the virus.

Stats NZ has adopted different approaches to methodology for the April 2020 FPI to ensure the index continues to reflect the prices consumers pay for food.

Food price index: April 2020 will be published on 20 May 2020.

Data collection

The usual sample for the FPI consists of about 10,000 prices (2,500 collected weekly) for fresh fruit and vegetables, and about 9,000 prices collected monthly.

We could not collect about 2,000 of the monthly prices because businesses were closed, or prices were not available on their websites.

For the same reasons about 3,500 of the weekly prices were not collected.

For the April 2020 FPI we collected food prices from online supermarket sites, by making phone calls to businesses, and by using administrative data, from some supermarkets.

Data imputation

We identified two different scenarios for imputing unavailable prices in the April 2020 FPI, that is, substituting the missing prices with other values.

The first scenario is for items that were available, which are goods and services that could  be purchased by consumers. Second is for unavailable items, which are goods and services that could no longer be purchased because the industry had shut down (for example, restaurants and take-away stores).

For available items, low stock or lack of access to price data led to smaller sample sizes. For items that we were unable to collect prices for, we imputed data based on the prices of other similar items in the same item group. All samples sizes for the items priced were deemed large enough for confidence in the imputation method.

For unavailable items, we imputed the price movement based on the monthly movement from the overall FPI for available items. This approach is preferred as it means the monthly calculation of the all-items index is unaffected by unavailable items, so the final FPI will only reflect price movements of the goods that consumers can purchase. This approach is consistent with guidelines from international statistical organisations and other statistical agencies.

When New Zealand moved to alert level 3 on 28 April, we decided not to include prices from takeaway stores due to the small number of days they were open (3 out of 30 days, from 28–30 April) and the difficulty of collecting prices over the phone from extremely busy stores. We suppressed the series for indexes where the sample size is less than 50 percent. This will affect all the indexes in the restaurant and take-aways subgroup.

For supermarkets in the South Island that did not have online prices, we imputed prices from another branch of the supermarket chain as a proxy.

Alternative data source

This month we incorporated price data supplied directly from some supermarkets as an alternative data source. This data replaces all the items that would have previously been manually priced in-store or collected online, and will be a method we will continue to use.

MIL OSI